Research ArticleOptimization of LED Lighting for Clinical Settings

Snjezana Soltic 1 and Andrew Chalmers 2

1Professional Engineering Manukau Institute of Technology Manukau 2241 New Zealand2Institute of Biomedical Technologies Auckland University of Technology Auckland 1142 New Zealand

Correspondence should be addressed to Snjezana Soltic ssolticmanukauacnz

Received 20 December 2018 Revised 4 July 2019 Accepted 1 August 2019 Published 27 August 2019

Academic Editor Hasan Ayaz

Copyright copy 2019 Snjezana Soltic and Andrew Chalmers +is is an open access article distributed under the Creative CommonsAttribution License which permits unrestricted use distribution and reproduction in anymedium provided the original work isproperly cited

+e advent of the LED light source has promoted the concept of human-centric lighting (HCL)+e LED has also been responsiblefor increases in the electrical efficiency of lighting systems coupled with recent improvements in their colour properties We havefound that it is also possible to create a lit environment with enhanced clinical attributes by providing a source spectrum thatmeets the requirements of the Cyanosis Observation Index (COI) +is paper describes the use of a differential evolution (DE)algorithm for the spectral design of a mixed LED light source capable of meeting COI recommendations as well as HCLperformance criteria

1 Introduction

Researchers worldwide have for several years been attemptingto quantify the benefits of human-centric lighting (HCL) Areport for the European Commission has identified a range ofpotential human-health issues arising from LED lighting [1](and onemay add itsmisapplication) HCL attempts to avoidthe pitfalls by the creation of a lit environment that emulatesnatural lighting and promotes the human circadian response+e relationship between light and the levels of wellbeing ofthe users was expanded when a third photoreceptor in ad-dition to rods and cones was identified in the human eyeearly in the new century +ese additional light-sensitivephotoreceptors in the retinal ganglion cell layer (pRGCs) aredirectly sensitive to light and primarily responsible for me-diating these responses [2ndash4]+ese cells are most sensitive toshort-wavelength light with a peak sensitivity in the visibleblue light range (446ndash477 nm) [5]

With that discovery the effects on circadian rhythms (iemental physical and biological changes) could be correlatedto specific lighting conditionsmdashin particular the properties ofthe light spectrum as well as the absolute level of the lightingA number of distinct purposes for HCL are now emerging ofwhich three are readily discerned biologically effective

lighting to improve cognitive performance emotionally ef-fective lighting to create stimulating environments andclinically effective lighting to improve health care andtreatment facilities in hospitals clinics and retirement homesLED lighting appears to be the driving force for new researchinto HCL and its applications in a wide range of appropriatesettings LEDs enable control of the lighting with the prospectof independently varying intensity spectrum and colour

As stated in [6] ldquowhat we are really defining here is theability to influence human experience physiology andorbehaviour through the delivery of carefully chosen lightqualities with controlrdquo In our approach HCL harnesses theoptimization of the light-source spectrum to achieve max-imum comfort and efficiency for the occupants In theclinical context this requires good colour rendition ingeneral as well as the facility for effective observation ofpatients by clinic staff

Cyanosis (or perhaps more correctly its absence) isregarded by clinicians as an important indicator of a pa-tientrsquos health It refers to a blue coloration of skin and lipsand results from poor oxygenation of the blood which can bean indicator for a range of conditions [7ndash9] +e COI hasbeen developed as a lighting design objective that enhancesthe clinicianrsquos ability to discern differences between blue and

HindawiJournal of Healthcare EngineeringVolume 2019 Article ID 5016013 8 pageshttpsdoiorg10115520195016013

red colour casts of the skin+e cyanosis observation criteriaset by the ANZS standard for hospital lighting [10] are3300leCCTle 5300K plus the COIlt 33

+e following lighting design objectives flow from theseconcepts (i) variable CCT (correlated colour temperature)to stimulate the circadian response (ii) the best achievableCOI and colour rendition properties at each CCT (iii) thehighest achievable luminous efficacy (for best system effi-ciency) consistent with objectives (i) and (ii)

+e purpose of this project is to derive the SPDs (spectralpower distributions) of LED mixtures that are capable ofproviding optimum cyanosis observation along with ex-cellent colour rendition properties as well as the highestachievable luminous efficacy of radiation Since these threeobjectives impose their own unique demands on the dis-tribution of spectral power this requires a balancing (andoptimization) of their contravariant effects

2 Lighting Background

21 Correlated Colour Temperature Correlated colourtemperature (CCT) refers to the colour appearance of thesource viewed directly It is expressed on a numerical scalethat represents the temperature of a Planckian radiatorhaving the colour that matches (as nearly as possible) thecolour of the test source [11] In contradistinction to theCCT value the ambience created by different sourcecolours is described as ldquowarmrdquo for lower CCTs and ldquocoolrdquofor the higher CCTs Source CCTs between 3300 K and5300 K have been found to be best for COI purposes [7] Ingeneral lighting practice it is usual to specify CCTs be-tween 2700 K (approximately the colour of a tungstenfilament lamp) and 6500 K (matching noon daylight on aclear summer day) We also allow for the possibility ofincreasing the CCT to 7500 K to emulate sky light duringthe daytime hours

22 Luminous Efficacy Luminous efficacy (LE) refers to thecapacity of the source to produce visible light output effi-ciently and is measured in lumens per watt +e choice towork with LEDs was influenced by their generally highperformance in this context In the simulations reportedhere we quote the LER (luminous efficacy of the radiation)of each LEDmixture which needs to be as high as possible toprovide a modern energy-efficient lighting design

+e LER is defined in equation (1) and the overall lu-minous efficacy (LE) in equation (2)

LER Km1113938λV(λ)S(λ)dλ

1113938 S(λ)dλ (1)

LE Km1113938λV(λ)S(λ)dλ

PE

(2)

Km is the maximum luminous efficacy of radiation (asymp683lumen per watt) S(λ) is the spectral distribution of the lightsource and V(λ) is the CIE spectral sensitivity function forhuman photopic vision Hence the LER assesses the

ldquolighting contentrdquo of the spectrum by comparing the visiblelight output (in lumens) against the total radiant output (inwatts) LE on the contrary is based on the electrical powerconsumption PE which needs to supply the conversion lossesin the light source as well as the total radiant output asshown in the following equation

PE CL + 1113946 S(λ)dλ (3)

where CL represents the conversion losses which depend onthe physical processes in the lamp

It is clear that the denominator in equation (2) is alwaysgreater than that in equation (1) and so LE is always lessthan LER However the LER is simpler to predict since itdepends solely on the spectrum of the source

23 Colour Rendition +is refers to the ability of the sourceto illuminate surfaces such that their colours appear asnatural as possible [12] +e current internationally agreedmethod for the classification of the colour rendition prop-erties of a source is the CIE colour rendering index (CRI)also known as Ra [13] For exacting applications such as thisstudy we have supplemented this with the IESNArsquos TM-30-15 method for the specification of colour fidelity Rf [14] andwe quote both Ra and Rf (and several subsidiary indices) inour simulations Both systems employ sets of test colourswhich are illuminated in turn by test and reference sourcespectra (+is is a conceptual comparison which is carriedout using numerical calculations) +e reference spectrum isalways selected to have the same CCT as the test spectrum+e colour difference for each sample as viewed under thetwo sources is calculated numerically and the results aremanipulated and averaged to provide an index on a scalehaving a maximum of 100 (which would mean perfectcolour conformance between the test and reference sources)In practice the aim is to achieve the highest possible Ra andRf at a given CCT and we do not recommend spectraldesigns having Ra or Rf values below 83

24 Clinical Lighting Practice +e lighting of hospitals andother healthcare facilities is guided by the standards or codesof practice adopted for use in different countries +ese areselected as the main focus for this discussion since theyrepresent the distilled wisdom of groups of expert practi-tioners in the countries concerned

+e major concerns of such documents are the levels oflighting (eg lux levels or illuminances) as well as glarecontrol of the light sources While these properties arenaturally of great importance they are not matters ofconcern in the present paper since our focus is on the colourand colour-rendition properties of the light sources

In this context themajority of hospital and clinical lightinghas up to now been based on the use of tubular fluorescent(and in some cases compact fluorescent) lamps LED lightinghas only recently come into consideration and the concept ofusing LEDs to achieve HCL is a very recent development (andwill take time to become reflected in codes and standards)

2 Journal of Healthcare Engineering

Colour and colour rendition are important properties ofany light source and recommendations on these are nor-mally provided in codes and standards as summarized inTable 1

It is noteworthy that the American recommendations arelargely descriptive and are less prescriptive than the others inTable 1 However all the listed recommendations are seen tobe in general accord with one another and differ only inparticular details

Various authors have commented on the desirable colourproperties for hospital lighting Alzubaidi and Soori [18]describe the need for a quality-lighting environment callingfor the CRI (Ra) to be ge80 in all areas of hospitals Lecceseet al writing in Italy [19] also quote this figure as well asRage 90 for examination and treatment preop and recoveryrooms operating theatres and colour inspection areas+ornLighting [20] also refers to these values for Ra and adds thatthe most common CCT used in UK healthcare spaces is4000K while 2700K may be used where it is desired toprovide a ldquomore homely feelrdquo Mehrotra et al [21] describinghospital lighting in India also call for a minimum Ra of 80 (ingeneral) or 90 (for examination and treatment etc)

Similar to the ASNZS requirement for COI lighting butfollowing different paths Bartczak et al [22] have devised anLED-based tunable illuminator that provides improvedcontrast between skin and veins as compared to daylightwhile Litorja and Ecker [23] have designed a 3-wavelengthLED lamp that can enhance visual contrast when viewingveins on the back of human hands

We note that our proposals will be taking hospitallighting design into new directions Our work aims to meetas far as possible the intentions expressed in Table 1 butwith the additional target of meeting the requirements forCOI wherever possible and also to provide the capability forcircadian-based HCL

We have aimed to meet or exceed the following nu-merical indicators

Rage 80 (CIE 133)Rfge 80 Rfskinge 90 (IES TM-30-15)COIle 33 with 3300KleCCTle 5300K (ASNZS168025 2018)2700KleCCTle 7500K (range for HCL)

+e intention is to satisfy the Ra Rf and Rfskin re-quirements at all times and to comply with COI conditions atall times except when extreme CCTvalues are called for by theneeds of HCLWe argue that a requirement for Rage 90 (as forexamination and treatment areas) can be satisfied by acombination of Rage 80 Rfge 80 and Rfskinge 90 since it is thecolour of human skin that is most critical in such situations Avalue over 90 for the skin index when coupled with COIcompliance in the mid-CCT range will ensure very satis-factory skin rendition and patient condition monitoring

25 Mixed LED Lighting We have investigated white-lightmixtures using a range of seven coloured-LED sources [24]+e mixtures will use four LEDs at a time Fewer than 4 lead

to inferior colour properties while more than 4 are assessedas too cumbersome for effective control If a suitable LEDcombination can be established then it becomes possible toachieve variations in the CCT of the mixture simply byvarying the proportions of each constituent in the mixture+e benefit of such an approach is the simplification of thedesign of HCL for clinical settings

We are not alone in making proposals for the imple-mentation of colour-mixed LED lighting Chiang and Chien[25] for example have devised a ldquopentachromatic RGBACW platform suitable for clinic userdquo +ey propose an al-gorithm that permits their multispectral LED cluster toproduce CCTs in the range 2800K to 8000K together with ahigh Colour Quality Scale (CQSgt 85 points)mdashsee [26]

One of the concerns with the use of LED mixtures iswhether the mixture is capable of maintaining its targetsettings when subjected to the rigours of field use Llenas andCarreras [27] have demonstrated that the problems thatresult from LED aging and temperature variations can beeffectively eliminated in practice by the use of suitablefeedback systems

26 Optimization Tool We have designed an optimizationtool to find spectra that best meet the performance re-quirements for COI colour rendition and LER It is basedon a Matlabreg program that derives the best 4-band mixturesof LEDs from a total set of seven available narrow-band LEDspectra Optimizations are performed by use of a DE (dif-ferential evolution) type of algorithm [28 29] which hasbeen demonstrated to be effective in several other spectraldesign settings [30 31]

+e search for an optimal SPD starts with a population ofP randomly created candidate-solution vectors each of whichrepresents a possible combination of LED spectra +e valueof P is kept constant during the optimization process +ecandidate vectors undergo mutation crossover evaluationand selection over a number of generations G Both thepopulation sizeP and the number of generationsG depend onthe problem to be optimised In practice it has been foundthat P must be gt4 to provide sufficient mutually differentsolution vectors for the algorithm to function properly Notethat G is a user-selectable quantity and larger values for G

inevitably lead to longer execution times for the programOur algorithm also includes selections for two further

DE settings F is a mutation weight (between 0 and 2) whichinfluences the magnitudes of mutations in the evolutionprocess and CR (from 0 to 1) is a crossover constant en-suring that offspring solution vectors always differ from theirparents A sorting process ensures that only the fitter off-spring vectors are moved to the succeeding generation

We experimented with different settings for the DEparameters P F and CR and found that the values of theseparameters had only a minor influence on the optimizationresults Our experience was that the process of choosing thecontrol variables to achieve good optimization wasstraightforward After some experimentation we set P 10and F and CR to the values suggested in [29] ie F 05 andCR 01

Journal of Healthcare Engineering 3

Every DE process has to be controlled by a fitnessfunction that determines how closely the members of a givengeneration have approached the required objective(s) In thiscase the fitness function was set equal to the COI (equation(4)) and the set objective was that ffit be minimised +esolution vector (in this case SPD) having the lowest ffit aswell as CCT within range (ie 3300leCCTle 5300K) at theend of the final generation is regarded as the optimum

ffit COI (4)

For the purposes of this project we performed over 140independent optimization runs We kept P 10 throughoutand G of different durations (10 20 50 100 and 200) Eachvalue of G was applied in at least five separate optimizationruns Of the total number of runs G of 10 was applied 50times and G of 200 was used 65 times

27 Selection of LED Spectra We had access to the SPDs ofseven coloured LEDs in the Lumileds LUXEON range [24]For the first 75 optimization runs the program was set tomake a random selection of any 4 LEDs (from the 7available) to form each of the members of the initialpopulation for that run In reviewing all 75 runs it becameevident that there were 4 specific LED spectra that con-sistently gave a range of ldquogoodrdquo results as judged on thebasis of colour rendition and LER performance as well asCOI and CCT +ese four LED spectra then became thefocus of the next 65 optimization runs from which a finaltotal of 33 SPDs were selected as having potentially usefulproperties

Since we wish to make a practical proposal for actualclinical lighting we have subsequently made a selection of 8

mixtures (from the set of 33) to serve as basis for ourrecommendations

3 Results and Discussion

Our selection of the four final LED colours (with peakwavelengths in brackets) was royal blue (450 nm) green(525 nm) amber (590 nm) and red (640 nm) and the resultsshown in Table 2 were obtained using these colours foroptimum mixtures at specific CCTs Note that the blueconstituent of our mixture has its peak at 450 nm that iswithin the range for the stimulation of the alerting responseof the circadian system [5]

31Optimizing forCOI +eCCTs were selected on the basisof (a) the acceptable range of CCT to provide good COI and(b) the ability to provide a CCT range that will support thecircadian response +e selection of mixtures for CCTsoutside of the preferred range for COI is covered in a latersubsection

+e 33 optimised mixtures that conform to COI re-quirements are summarized in Table 2 which shows thevarious colorimetric and operational parameters for eachmixture sorted in order of increasing CCT Note that severaladditional parameters have been listed in the table to providemore detailed information on the colour properties of eachSPD

By definition the CRI value Ra [13] is based on the averagecolour difference for eight moderate-chroma colours evenlyspread around the hue circle +e CIE has also defined asupplementary set of six colours to provide more detail wherenecessary Among these are the saturated red sample (index

Table 1 Recommended source colour properties in codes and standards

Country or region Australia and New Zealand[10]

United Kingdom [15] andEuropean Union [16] United States [17]

Colour appearance (CCT) 3300KleCCTle 5300K (allareas) No general recommendation 2700K night lighting No other

general recommendationColour rendering indexmdashgeneraland ward areas Rage 80 Rage 80 No general recommendation

Colour renderingindexmdashexamination andtreatment areas

Rage 85 (eg dermatology) Rage 90 (all examination areas) Rage 80 for exam areasRagt 90 for surgery

Special colour requirements Cyanosis observationCOIle 33

Examination rooms (generallighting)

4000KleCCTle 5000KColour inspection (labs and

pharmacies)6000KleCCTle 6500K

Consider the appropriate specialcolour rendering index Ri foraccurate rendition of coloured

objects or human skin

Surgery and examination rooms4000KleCCTle 5000K

Human-centric lighting(reinforcing circadianentrainment)

Not includedLighting may have a significantinfluence on human circadian

rhythms

Lighting practitioners to be awareof principles of the circadian cycle

Note +is table represents a distillation by the authors of detailed and sometimes complex requirements or recommendations Please refer to the originaldocuments for full details and explanations

4 Journal of Healthcare Engineering

R9) and Caucasian skin colour analogue (index R13) and wehave included these two indices along with Ra

In a similar way the colour fidelity Rf is defined as anaverage colour difference for 99 different colours Indices forspecific colours can be extracted from the data available forthe calculation of Rf Here we have included Rfskin which theIESNA [14] defines as the average of two indices (for samplenumbers 15 and 18) representing light and dark skin tonesIn addition we have extracted the index (Rfmin) and itssample number (imin) for the sample having the lowest indexvalue for a particular SPD since this indicates the worst-rendered colour under the specific source

+e eight SPDs in our final recommendation were se-lected for their superior properties in terms of colour per-formance and LER To assist in the selection we formulateda simple Overall Index (OI) as a general figure of merit asdefined in equation (5) Each of the selections was made onthe basis of the highest OI value when compared with othersfully meeting the COI criteria in the immediate CCTvicinity

OI Ra + R9 + R13 + Rf + Rfskin + Rfmin +LER5

1113874 1113875 (5)

Focussing now on the recommended eight SPDs Fig-ure 1 shows their SPDs on a single set of axes All eight havebeen normalized to the red (640 nm) peak which all share astheir common maximum

In practice suitable controls will need to be imple-mented to adjust the relative contributions to the mixture byeach of the four LED colours +e controls should also bedesigned to enable dimming so as to permit the setting ofdifferent light levels at any selected CCT thereby enhancingthe HCL effects at appropriate times of the day

It is evident from Table 2 that the LER and all significantcolour parameters are dependent on the CCT Figure 2indicates the variations in a selection of these parametersfor the recommended range of CCT settings +e curvesshown are best-fit polynomials (order 2)

+e COI (for which lower values are preferable) showedan upward (ie deteriorating) trend with CCT and amaximum COI result of 30 for a CCT just below 5300K Allthe other depicted parameters also showed a deteriorating(now downward) trend +e LER was 354 lmW at 3300Kand dropped about 17 to the region of 300 lmW above4500K +e colour fidelity metrics Ra and Rf both dropped

Table 2 33 optimised mixtures of the four selected LED colours

SPD no CCT (K) COICIE CRI results IES TM-30-15 results

Surface colour giving Rfmin OIRa R9 R13 LER (lmw) Rf Rg Rfskin Rfmin imin

1 3304 096 91 49 95 354 90 99 91 75 40 Dark greenish-grey (F) 5622 3311 149 90 40 92 362 89 95 89 74 42 Olive-green (F) 5463 3365 122 93 61 98 339 89 103 91 67 20 Orange-brown (F) 5674 3381 088 92 59 97 344 90 102 92 73 20 Dark orange (F) 5725 3401 22 94 88 99 339 91 103 96 72 42 Olive-green (F) 6086 3444 096 93 68 98 344 90 102 93 74 42 Olive-green (F) 5857 3462 085 93 69 99 336 90 104 93 68 20 Orange-brown (F) 5798 3488 129 94 81 97 326 89 106 93 62 20 Orange-brown (F) 5819 3526 117 94 91 97 331 90 105 96 71 42 Olive-green (F) 60510 3607 149 92 66 97 182 90 102 93 74 42 Olive-green (F) 54811 3608 127 94 85 96 324 89 107 94 62 20 Orange-brown (F) 58512 3716 086 92 98 94 317 89 108 96 63 20 Dark orange (F) 59513 3744 071 92 96 95 322 89 107 97 67 42 Olive-green (F) 60014 3744 311 93 81 95 317 88 108 92 55 20 Dark orange (F) 56715 3753 117 91 84 90 310 88 110 97 62 20 Orange-brown (F) 57416 3880 124 93 91 96 323 89 107 96 65 20 Orange-brown (F) 59517 3895 141 93 94 95 319 89 108 96 63 20 Orange-brown (F) 59418 4060 217 93 80 99 329 89 105 95 68 20 Dark orange (F) 59019 4199 249 80 24 79 286 83 115 92 55 42 Olive-green (F) 47020 4222 149 89 79 93 316 88 108 98 64 42 Olive-green (F) 57421 4277 092 90 79 91 309 87 109 98 62 42 Olive-green (F) 56922 4308 333 92 97 93 193 88 108 96 59 20 Orange-brown (F) 56423 4431 083 88 69 90 306 87 110 97 61 42 Olive-green (F) 55324 4669 15 82 32 81 286 83 114 93 54 42 Olive-green (F) 48225 4694 105 87 61 90 303 86 110 96 59 42 Olive-green (F) 54026 4944 133 83 33 82 287 83 114 93 54 42 Olive-green (F) 48527 5111 082 83 37 84 291 84 113 94 54 42 Olive-green (F) 49428 5129 14 83 38 87 295 84 112 94 56 42 Olive-green (F) 50129 5148 244 83 44 83 287 83 113 94 53 42 Olive-green (F) 49730 5196 144 77 4 76 277 81 116 90 49 42 Olive-green (F) 43231 5211 077 82 31 84 290 83 114 93 54 42 Olive-green (F) 48532 5243 3 88 75 91 301 86 110 97 58 42 Olive-green (F) 55533 5331 337 88 77 91 301 86 110 97 57 42 Olive-green (F) 556Final recommendation SPD nos 1 9 13 17 20 25 28 and 32 Note that SPDs 22 and 33 are slightly outside the COI recommendations Colour descriptionsare our own interpretations of TM-30-15 test colours type F are print colours

Journal of Healthcare Engineering 5

from the low-90s region to approximately the mid-80s +eskin fidelity metric R13 fell from around 95 to about 89 andRfmin (an indicator of the worst-case fidelity) fell from 75 toroughly 57 Note that Rfskin was ge91 throughout and

together with the high R13 values indicates that skin colourrendition will be good at all CCT settings

Taken together these results indicate that operation at3300K gives the best economy as well as the best COI and

00102030405060708091

380 400 420 440 460 480 500 520 540 560 580 600 620 640 660 680 700 720 740 760 780

3304K3526K3744K

3895K4222K4694K

5129K5243K

Figure 1 SPDs of the eight recommended LED mixtures

0051

152

253

35

3000 3500 4000 4500 5000 5500

(a)

290300310320330340350360

3000 3500 4000 4500 5000 5500

(b)

82

84

86

88

90

92

94

96

3000 3500 4000 4500 5000 5500

(c)

838485868788899091

3000 3500 4000 4500 5000 5500

(d)

86

88

90

92

94

96

98

3000 3500 4000 4500 5000 5500

(e)

50

55

60

65

70

75

80

3000 3500 4000 4500 5000 5500

(f )

Figure 2 Relationship of source parameters to correlated colour temperature (CCT) Dependence of (a) COI (b) LER (c) Ra (d) Rf (e) R13and (f) Rfmin

6 Journal of Healthcare Engineering

best general colour fidelity properties It should be notedhowever that the above properties are in fact acceptable forthe full range of CCTs depicted and there are likely to besituations when higher-CCT operation may be desired egto satisfy patient preferences or for HCL purposes which callfor lower CCT at night and higher CCT in the morning

32 Optimizing for HCL HCL (human-centric lighting)objectives are not restricted to the CCT range desired forCOI and we note that it is possible in principle to set theCCT of the mixture to values that fall outside the recom-mended range for optimum cyanosis observation For ex-ample it may be desirable in a hospital ward to set both CCTand lumen output to low values to promote a restful eveningenvironment or to high values for a more stimulatingmorning and daytime environment

A further optimization experiment was therefore con-ducted to find acceptable mixtures to achieve such variationswithout changing the constituent wavelengths in the mix-ture +e method was an adaptation of our earlier work [32]using differential evolution (DE) with a fitness functiondesigned for optimization in the CIE 133 domain

Table 3 gives a selection of the results that can beachieved and illustrates the feasibility of the approach Itshows that although outside the acceptable range of CCTforCOI purposes very good results are obtainable for Ra and Rfas well as for the skin colour indices R13 and Rfskin +e chiefdrawback of the high-range CCTs is their rather poor Rfminperformance

Note that 7500K is higher than normally recommendedfor interior lighting as it represents the stronger blue lightcomponent of the sky during daylight but it may be usedwhen appropriate for an enhancedHCL stimulation response

4 Conclusions

We have explained the importance of lighting from theperspective of human health and have shown how thehuman circadian system can best be stimulated for goodhealth effects We have also explained the value of cyanosisas a general indicator of health in the clinical setting andhave shown how good spectral design of the lighting is bestable to enhance cyanosis observation

We have developed an optimization tool in Matlabregwhich invokes the Cyanosis Observation Index (COI) toprovide the facility to optimize the lighting and so enhance

the caregiverrsquos ability to observe cyanosis+e recommendedrange of CCT for effective cyanosis observation is 3300ndash5300K and we have made recommendations for eightdifferent mixtures of four LED spectra that give good valuesfor the COI (ie lt33) over this range

We have utilized a previously developed CRI optimi-zation tool which has enabled us also to recommend fivefurther LED mixtures (using the same 4 LEDs as before) forCCTs which are below 3300K or above 5300K +ese willallow the clinician to provide a wider range of different CCTs(when cyanosis observation is not continuously required) tohelp harness the benefits of human-centric lighting (HCL)

It is noteworthy that all the recommended mixturesprovide excellent LER (285le LERle 354 lmW) togetherwith good colour rendition in terms of the CRI (Ra and R13)of the CIE as well as the colour fidelity index (Rf ) and colourskin index (Rfskin) of the IESNA

We note that the blue LED in our recommended mix-tures has a peak wavelength of 450 nm and a bandwidth(FWHM) of 18 nm (441ndash459 nm) It is therefore capable ofproviding significant energy for effective HCL activity withinthe recommended (446ndash477 nm) bandmdashparticularly whenoperating in the higher-CCTmixtures Conversely for low-CCTmixtures (in which the blue content is significantly cut)there will be low activation as required for the creation of arestful night-time ambience

+ere are numerous other ways in which lighting is ableto enhance the patient experience and health outcomes [33]and we have not attempted to include them all in this studyWe have focussed here on the more general areas of patientlighting as they may apply in both day-stay and overnighthospital wards or clinics

Data Availability

+e data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

+e authors declare that there are no conflicts of interestregarding the publication of this paper

Acknowledgments

+e second author Andrew Chalmers thanks ProfessorAhmed Al-Jumaily Director of IBTec at AUT for his

Table 3 Optimised mixtures for low and high CCTs

CCT (K)CIE CRI results

LER (lmw)IES TM-30-15 results

Surface colour giving Rfmin OIRa R9 R13 Rf Rg Rfskin Rfmin imin

Low2759 95 87 98 332 89 100 94 71 25 Yellowish-brown (A) 6003051 94 74 100 340 90 101 94 75 42 Olive-green (F) 595

High5734 90 95 98 309 86 107 97 57 75 Dark blue (F) 5856563 89 95 100 305 86 106 97 53 75 Dark blue (F) 5817539 86 71 90 285 83 110 96 50 75 Dark blue (F) 533

Colour descriptions are our own interpretations of the TM-30-15 test colours type A are from nature type F are print colours

Journal of Healthcare Engineering 7

support and the provision of facilities +is work wassupported by the School of Professional Engineering and theManukau Institute of Technology Research Fund

References

[1] SCHEER (Scientific Committee on Health Environmentaland Emerging Risks) Opinion on Potential Risks to HumanHealth of Light Emitting Diodes (LEDs) European Com-mission Luxembourg 2018 httpseceuropaeuhealthsiteshealthfilesscientific_committeesscheerdocsscheer_o_011pdf

[2] C Genova ldquoNormal responses to non-visual effects of lightretained by blind humans lacking rods and conesrdquo MedicalNews Today (Online) December 2007 httpswwwmedicalnewstodaycomreleases91836php

[3] W van Bommel and G van den Beld ldquoLighting for work areview of visual and biological effectsrdquo Lighting Research ampTechnology vol 36 no 4 pp 255ndash266 2004

[4] C S Pechacek M Andersen and S W Lockley ldquoPreliminarymethod for prospective analysis of the circadian efficacy of(day) light with applications to healthcare architecturerdquoLEUKOS vol 5 no 1 pp 1ndash26 2008

[5] M Andersen J Mardaljevic and S W Lockley ldquoA frame-work for predicting the non-visual effects of daylight Part 1photobiology-based modelrdquo Lighting Research amp Technologyvol 44 no 4 pp 37ndash53 2012

[6] C Meadows ldquoSetting expectations on human-centric light-ingrdquo LEDs Magazine (Online) December 2017 httpswwwledsmagazinecomarticles201712setting-expectations-on-human-centric-lightinghtml

[7] N A Midolo and L Sergeyeva ldquoLighting for clinical ob-servation of cyanosisrdquo e Australian Hospital Engineervol 30 no 2 pp 38ndash43 2007

[8] M Changizi and K Rio ldquoHarnessing color vision for visualoximetry in central cyanosisrdquo Medical Hypotheses vol 74no 1 pp 87ndash91 2010

[9] H L Snider ldquoCyanosisrdquo in Clinical Methods e HistoryPhysical and Laboratory Examinations H K WalkerW D Hall and J W Hurst Eds Butterworths Boston MAUSA 3rd edition 1990 httpwwwthornlightingcomdownloadHandbook5-6pdf

[10] ASNZS 1680252018 AustralianNew Zealand StandardInterior and Workplace Lighting Part 25 Hospital andMedical Tasks Section 73 Cyanosis Observation Lightingand Appendix G Method for Determination of CyanosisObservation Index of a Light Source

[11] Commission Internationale de lrsquoEclairage Colorimetry CIEPublication 153 Vienna Austria 3rd edition 2003

[12] Commission Internationale de lrsquoEclairage InternationalLighting Vocabulary CIE publication DIS 017E2016 ViennaAustria 2017

[13] Commission Internationale de lrsquoEclairage Method of Mea-suring and Specifying Colour Rendering Properties of LightSources CIE Publication 133 Vienna Austria 1995

[14] Illuminating Engineering Society of North America IESMethod for Evaluating Light Source Color Rendition Illumi-nating Engineering Society New York NY USA 2015

[15] +e Society of Light and Lighting Lighting Guide 2 Hospitalsand Health Care Buildings CIBSE London UK 2008

[16] BSI Standards Publication BS EN 12464-12011 Light andLightingmdashLighting of Work Places Part 1 IndoorWork PlacesBSI London UK 2011

[17] ANSIIES RP-29-16 Lighting for Hospitals and HealthcareFacilities Illuminating Engineering Society of North AmericaNew York NY USA 2016

[18] S Alzubaidi and P K Soori ldquoEnergy efficient lighting systemdesign for hospitals diagnostic and treatment roommdasha casestudyrdquo Journal of Light amp Visual Environment vol 36 no 1pp 23ndash31 2012

[19] F Leccese C Montagnani S Iaia M Rocca and G SalvadorildquoQuality of lighting in hospital environments a wide surveythrough in situ measurementsrdquo Journal of Light amp VisualEnvironment vol 40 pp 52ndash65 2016

[20] +orn Lighting Ltd orn Applications and TechniquesHealthcare +orn Lighting Spennymoor UK 2019

[21] S Mehrotra S Basukala and S Devarakonda ldquoEffectivelighting design standards impacting patient care a systemsapproachrdquo Journal of Biosciences and Medicines vol 3 no 11pp 54ndash61 2015

[22] P Bartczak A Gebejes P Falt and M Hauta-Kasari ldquoAnLED-based tunable illumination for diverse medical appli-cationsrdquo in Proceedings of the IEEE Symposium on Computer-Based Medical Systems pp 292-293 IEEE Dublin IrelandAugust 2016

[23] M Litorja and B Ecker ldquoUse of a spectrally-tunable source toexplore improvement in chromatic contrast for illuminationof tissuesrdquo in Proceedings of the SPIEmdashe InternationalSociety for Optical Engineering vol 7596 Bellingham WAUSA December 2010

[24] Lumileds Lighting ldquoLUXEONreg K2 emitterrdquo in TechnicalDatasheet DS51 Lumileds Lighting US LLC San Jose CAUSA 2006

[25] S B Chiang and K Chien ldquo+e optimization of color-mixedLED lightingrdquo in Proceedings of the 2014 IEEE IndustryApplication Society Annual Meeting Vancouver CanadaOctober 2014

[26] W Davis and Y Ohno ldquoColor quality scalerdquo Optical Engi-neering vol 49 no 3 article 033602 2010

[27] A Llenas and J Carreras ldquoArbitrary spectral matching usingmulti-LED lighting systemsrdquo Optical Engineering vol 58no 3 article 035105 2019

[28] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient adaptive scheme for global optimization over con-tinuous spacesrdquo Technical report TR-95-012 InternationalComputer Science Institute Berkley CA USA 1995

[29] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient heuristic for global optimization over continuousspacerdquo Journal of Global Optimization vol 11 no 4pp 341ndash359 1997

[30] S Soltic and A Chalmers ldquoDifferential evolution for theoptimisation of multi-band white LED light sourcesrdquo LightingResearch amp Technology vol 44 no 2 pp 224ndash237 2012

[31] A N Chalmers and S Soltic ldquoLight source optimizationspectral design and simulation of four-band white-lightsourcesrdquo Optical Engineering vol 51 no 4 article 0440032012

[32] S Soltic and A N Chalmers ldquoDifferential evolution and itsapplication to intelligent spectral designrdquo in Proceedings of the16th Electronics New Zealand Conference (ENZCon) Uni-versity of Otago Dunedin New Zealand November 2009

[33] L J M Schlangen e Role of Lighting in Promoting Well-Being and Recovery Within Healthcare Philips White Pa-per Koninklijke Philips Electronics NV AmsterdamNetherlands 2010 httpswwwitdoescoukwpontentuploads201604Role-of-lighting-published-by-philips-as-a-whitepaper-2010pdf

8 Journal of Healthcare Engineering

Colour and colour rendition are important properties ofany light source and recommendations on these are nor-mally provided in codes and standards as summarized inTable 1

It is noteworthy that the American recommendations arelargely descriptive and are less prescriptive than the others inTable 1 However all the listed recommendations are seen tobe in general accord with one another and differ only inparticular details

Various authors have commented on the desirable colourproperties for hospital lighting Alzubaidi and Soori [18]describe the need for a quality-lighting environment callingfor the CRI (Ra) to be ge80 in all areas of hospitals Lecceseet al writing in Italy [19] also quote this figure as well asRage 90 for examination and treatment preop and recoveryrooms operating theatres and colour inspection areas+ornLighting [20] also refers to these values for Ra and adds thatthe most common CCT used in UK healthcare spaces is4000K while 2700K may be used where it is desired toprovide a ldquomore homely feelrdquo Mehrotra et al [21] describinghospital lighting in India also call for a minimum Ra of 80 (ingeneral) or 90 (for examination and treatment etc)

Similar to the ASNZS requirement for COI lighting butfollowing different paths Bartczak et al [22] have devised anLED-based tunable illuminator that provides improvedcontrast between skin and veins as compared to daylightwhile Litorja and Ecker [23] have designed a 3-wavelengthLED lamp that can enhance visual contrast when viewingveins on the back of human hands

We note that our proposals will be taking hospitallighting design into new directions Our work aims to meetas far as possible the intentions expressed in Table 1 butwith the additional target of meeting the requirements forCOI wherever possible and also to provide the capability forcircadian-based HCL

We have aimed to meet or exceed the following nu-merical indicators

Rage 80 (CIE 133)Rfge 80 Rfskinge 90 (IES TM-30-15)COIle 33 with 3300KleCCTle 5300K (ASNZS168025 2018)2700KleCCTle 7500K (range for HCL)

+e intention is to satisfy the Ra Rf and Rfskin re-quirements at all times and to comply with COI conditions atall times except when extreme CCTvalues are called for by theneeds of HCLWe argue that a requirement for Rage 90 (as forexamination and treatment areas) can be satisfied by acombination of Rage 80 Rfge 80 and Rfskinge 90 since it is thecolour of human skin that is most critical in such situations Avalue over 90 for the skin index when coupled with COIcompliance in the mid-CCT range will ensure very satis-factory skin rendition and patient condition monitoring

25 Mixed LED Lighting We have investigated white-lightmixtures using a range of seven coloured-LED sources [24]+e mixtures will use four LEDs at a time Fewer than 4 lead

to inferior colour properties while more than 4 are assessedas too cumbersome for effective control If a suitable LEDcombination can be established then it becomes possible toachieve variations in the CCT of the mixture simply byvarying the proportions of each constituent in the mixture+e benefit of such an approach is the simplification of thedesign of HCL for clinical settings

We are not alone in making proposals for the imple-mentation of colour-mixed LED lighting Chiang and Chien[25] for example have devised a ldquopentachromatic RGBACW platform suitable for clinic userdquo +ey propose an al-gorithm that permits their multispectral LED cluster toproduce CCTs in the range 2800K to 8000K together with ahigh Colour Quality Scale (CQSgt 85 points)mdashsee [26]

One of the concerns with the use of LED mixtures iswhether the mixture is capable of maintaining its targetsettings when subjected to the rigours of field use Llenas andCarreras [27] have demonstrated that the problems thatresult from LED aging and temperature variations can beeffectively eliminated in practice by the use of suitablefeedback systems

26 Optimization Tool We have designed an optimizationtool to find spectra that best meet the performance re-quirements for COI colour rendition and LER It is basedon a Matlabreg program that derives the best 4-band mixturesof LEDs from a total set of seven available narrow-band LEDspectra Optimizations are performed by use of a DE (dif-ferential evolution) type of algorithm [28 29] which hasbeen demonstrated to be effective in several other spectraldesign settings [30 31]

+e search for an optimal SPD starts with a population ofP randomly created candidate-solution vectors each of whichrepresents a possible combination of LED spectra +e valueof P is kept constant during the optimization process +ecandidate vectors undergo mutation crossover evaluationand selection over a number of generations G Both thepopulation sizeP and the number of generationsG depend onthe problem to be optimised In practice it has been foundthat P must be gt4 to provide sufficient mutually differentsolution vectors for the algorithm to function properly Notethat G is a user-selectable quantity and larger values for G

inevitably lead to longer execution times for the programOur algorithm also includes selections for two further

DE settings F is a mutation weight (between 0 and 2) whichinfluences the magnitudes of mutations in the evolutionprocess and CR (from 0 to 1) is a crossover constant en-suring that offspring solution vectors always differ from theirparents A sorting process ensures that only the fitter off-spring vectors are moved to the succeeding generation

We experimented with different settings for the DEparameters P F and CR and found that the values of theseparameters had only a minor influence on the optimizationresults Our experience was that the process of choosing thecontrol variables to achieve good optimization wasstraightforward After some experimentation we set P 10and F and CR to the values suggested in [29] ie F 05 andCR 01

Journal of Healthcare Engineering 3

Every DE process has to be controlled by a fitnessfunction that determines how closely the members of a givengeneration have approached the required objective(s) In thiscase the fitness function was set equal to the COI (equation(4)) and the set objective was that ffit be minimised +esolution vector (in this case SPD) having the lowest ffit aswell as CCT within range (ie 3300leCCTle 5300K) at theend of the final generation is regarded as the optimum

ffit COI (4)

For the purposes of this project we performed over 140independent optimization runs We kept P 10 throughoutand G of different durations (10 20 50 100 and 200) Eachvalue of G was applied in at least five separate optimizationruns Of the total number of runs G of 10 was applied 50times and G of 200 was used 65 times

27 Selection of LED Spectra We had access to the SPDs ofseven coloured LEDs in the Lumileds LUXEON range [24]For the first 75 optimization runs the program was set tomake a random selection of any 4 LEDs (from the 7available) to form each of the members of the initialpopulation for that run In reviewing all 75 runs it becameevident that there were 4 specific LED spectra that con-sistently gave a range of ldquogoodrdquo results as judged on thebasis of colour rendition and LER performance as well asCOI and CCT +ese four LED spectra then became thefocus of the next 65 optimization runs from which a finaltotal of 33 SPDs were selected as having potentially usefulproperties

Since we wish to make a practical proposal for actualclinical lighting we have subsequently made a selection of 8

mixtures (from the set of 33) to serve as basis for ourrecommendations

3 Results and Discussion

Our selection of the four final LED colours (with peakwavelengths in brackets) was royal blue (450 nm) green(525 nm) amber (590 nm) and red (640 nm) and the resultsshown in Table 2 were obtained using these colours foroptimum mixtures at specific CCTs Note that the blueconstituent of our mixture has its peak at 450 nm that iswithin the range for the stimulation of the alerting responseof the circadian system [5]

31Optimizing forCOI +eCCTs were selected on the basisof (a) the acceptable range of CCT to provide good COI and(b) the ability to provide a CCT range that will support thecircadian response +e selection of mixtures for CCTsoutside of the preferred range for COI is covered in a latersubsection

+e 33 optimised mixtures that conform to COI re-quirements are summarized in Table 2 which shows thevarious colorimetric and operational parameters for eachmixture sorted in order of increasing CCT Note that severaladditional parameters have been listed in the table to providemore detailed information on the colour properties of eachSPD

By definition the CRI value Ra [13] is based on the averagecolour difference for eight moderate-chroma colours evenlyspread around the hue circle +e CIE has also defined asupplementary set of six colours to provide more detail wherenecessary Among these are the saturated red sample (index

Table 1 Recommended source colour properties in codes and standards

Country or region Australia and New Zealand[10]

United Kingdom [15] andEuropean Union [16] United States [17]

Colour appearance (CCT) 3300KleCCTle 5300K (allareas) No general recommendation 2700K night lighting No other

general recommendationColour rendering indexmdashgeneraland ward areas Rage 80 Rage 80 No general recommendation

Colour renderingindexmdashexamination andtreatment areas

Rage 85 (eg dermatology) Rage 90 (all examination areas) Rage 80 for exam areasRagt 90 for surgery

Special colour requirements Cyanosis observationCOIle 33

Examination rooms (generallighting)

4000KleCCTle 5000KColour inspection (labs and

pharmacies)6000KleCCTle 6500K

Consider the appropriate specialcolour rendering index Ri foraccurate rendition of coloured

objects or human skin

Surgery and examination rooms4000KleCCTle 5000K

Human-centric lighting(reinforcing circadianentrainment)

Not includedLighting may have a significantinfluence on human circadian

rhythms

Lighting practitioners to be awareof principles of the circadian cycle

Note +is table represents a distillation by the authors of detailed and sometimes complex requirements or recommendations Please refer to the originaldocuments for full details and explanations

4 Journal of Healthcare Engineering

R9) and Caucasian skin colour analogue (index R13) and wehave included these two indices along with Ra

In a similar way the colour fidelity Rf is defined as anaverage colour difference for 99 different colours Indices forspecific colours can be extracted from the data available forthe calculation of Rf Here we have included Rfskin which theIESNA [14] defines as the average of two indices (for samplenumbers 15 and 18) representing light and dark skin tonesIn addition we have extracted the index (Rfmin) and itssample number (imin) for the sample having the lowest indexvalue for a particular SPD since this indicates the worst-rendered colour under the specific source

+e eight SPDs in our final recommendation were se-lected for their superior properties in terms of colour per-formance and LER To assist in the selection we formulateda simple Overall Index (OI) as a general figure of merit asdefined in equation (5) Each of the selections was made onthe basis of the highest OI value when compared with othersfully meeting the COI criteria in the immediate CCTvicinity

OI Ra + R9 + R13 + Rf + Rfskin + Rfmin +LER5

1113874 1113875 (5)

Focussing now on the recommended eight SPDs Fig-ure 1 shows their SPDs on a single set of axes All eight havebeen normalized to the red (640 nm) peak which all share astheir common maximum

In practice suitable controls will need to be imple-mented to adjust the relative contributions to the mixture byeach of the four LED colours +e controls should also bedesigned to enable dimming so as to permit the setting ofdifferent light levels at any selected CCT thereby enhancingthe HCL effects at appropriate times of the day

It is evident from Table 2 that the LER and all significantcolour parameters are dependent on the CCT Figure 2indicates the variations in a selection of these parametersfor the recommended range of CCT settings +e curvesshown are best-fit polynomials (order 2)

+e COI (for which lower values are preferable) showedan upward (ie deteriorating) trend with CCT and amaximum COI result of 30 for a CCT just below 5300K Allthe other depicted parameters also showed a deteriorating(now downward) trend +e LER was 354 lmW at 3300Kand dropped about 17 to the region of 300 lmW above4500K +e colour fidelity metrics Ra and Rf both dropped

Table 2 33 optimised mixtures of the four selected LED colours

SPD no CCT (K) COICIE CRI results IES TM-30-15 results

Surface colour giving Rfmin OIRa R9 R13 LER (lmw) Rf Rg Rfskin Rfmin imin

1 3304 096 91 49 95 354 90 99 91 75 40 Dark greenish-grey (F) 5622 3311 149 90 40 92 362 89 95 89 74 42 Olive-green (F) 5463 3365 122 93 61 98 339 89 103 91 67 20 Orange-brown (F) 5674 3381 088 92 59 97 344 90 102 92 73 20 Dark orange (F) 5725 3401 22 94 88 99 339 91 103 96 72 42 Olive-green (F) 6086 3444 096 93 68 98 344 90 102 93 74 42 Olive-green (F) 5857 3462 085 93 69 99 336 90 104 93 68 20 Orange-brown (F) 5798 3488 129 94 81 97 326 89 106 93 62 20 Orange-brown (F) 5819 3526 117 94 91 97 331 90 105 96 71 42 Olive-green (F) 60510 3607 149 92 66 97 182 90 102 93 74 42 Olive-green (F) 54811 3608 127 94 85 96 324 89 107 94 62 20 Orange-brown (F) 58512 3716 086 92 98 94 317 89 108 96 63 20 Dark orange (F) 59513 3744 071 92 96 95 322 89 107 97 67 42 Olive-green (F) 60014 3744 311 93 81 95 317 88 108 92 55 20 Dark orange (F) 56715 3753 117 91 84 90 310 88 110 97 62 20 Orange-brown (F) 57416 3880 124 93 91 96 323 89 107 96 65 20 Orange-brown (F) 59517 3895 141 93 94 95 319 89 108 96 63 20 Orange-brown (F) 59418 4060 217 93 80 99 329 89 105 95 68 20 Dark orange (F) 59019 4199 249 80 24 79 286 83 115 92 55 42 Olive-green (F) 47020 4222 149 89 79 93 316 88 108 98 64 42 Olive-green (F) 57421 4277 092 90 79 91 309 87 109 98 62 42 Olive-green (F) 56922 4308 333 92 97 93 193 88 108 96 59 20 Orange-brown (F) 56423 4431 083 88 69 90 306 87 110 97 61 42 Olive-green (F) 55324 4669 15 82 32 81 286 83 114 93 54 42 Olive-green (F) 48225 4694 105 87 61 90 303 86 110 96 59 42 Olive-green (F) 54026 4944 133 83 33 82 287 83 114 93 54 42 Olive-green (F) 48527 5111 082 83 37 84 291 84 113 94 54 42 Olive-green (F) 49428 5129 14 83 38 87 295 84 112 94 56 42 Olive-green (F) 50129 5148 244 83 44 83 287 83 113 94 53 42 Olive-green (F) 49730 5196 144 77 4 76 277 81 116 90 49 42 Olive-green (F) 43231 5211 077 82 31 84 290 83 114 93 54 42 Olive-green (F) 48532 5243 3 88 75 91 301 86 110 97 58 42 Olive-green (F) 55533 5331 337 88 77 91 301 86 110 97 57 42 Olive-green (F) 556Final recommendation SPD nos 1 9 13 17 20 25 28 and 32 Note that SPDs 22 and 33 are slightly outside the COI recommendations Colour descriptionsare our own interpretations of TM-30-15 test colours type F are print colours

Journal of Healthcare Engineering 5

from the low-90s region to approximately the mid-80s +eskin fidelity metric R13 fell from around 95 to about 89 andRfmin (an indicator of the worst-case fidelity) fell from 75 toroughly 57 Note that Rfskin was ge91 throughout and

together with the high R13 values indicates that skin colourrendition will be good at all CCT settings

Taken together these results indicate that operation at3300K gives the best economy as well as the best COI and

00102030405060708091

380 400 420 440 460 480 500 520 540 560 580 600 620 640 660 680 700 720 740 760 780

3304K3526K3744K

3895K4222K4694K

5129K5243K

Figure 1 SPDs of the eight recommended LED mixtures

0051

152

253

35

3000 3500 4000 4500 5000 5500

(a)

290300310320330340350360

3000 3500 4000 4500 5000 5500

(b)

82

84

86

88

90

92

94

96

3000 3500 4000 4500 5000 5500

(c)

838485868788899091

3000 3500 4000 4500 5000 5500

(d)

86

88

90

92

94

96

98

3000 3500 4000 4500 5000 5500

(e)

50

55

60

65

70

75

80

3000 3500 4000 4500 5000 5500

(f )

Figure 2 Relationship of source parameters to correlated colour temperature (CCT) Dependence of (a) COI (b) LER (c) Ra (d) Rf (e) R13and (f) Rfmin

6 Journal of Healthcare Engineering

best general colour fidelity properties It should be notedhowever that the above properties are in fact acceptable forthe full range of CCTs depicted and there are likely to besituations when higher-CCT operation may be desired egto satisfy patient preferences or for HCL purposes which callfor lower CCT at night and higher CCT in the morning

32 Optimizing for HCL HCL (human-centric lighting)objectives are not restricted to the CCT range desired forCOI and we note that it is possible in principle to set theCCT of the mixture to values that fall outside the recom-mended range for optimum cyanosis observation For ex-ample it may be desirable in a hospital ward to set both CCTand lumen output to low values to promote a restful eveningenvironment or to high values for a more stimulatingmorning and daytime environment

A further optimization experiment was therefore con-ducted to find acceptable mixtures to achieve such variationswithout changing the constituent wavelengths in the mix-ture +e method was an adaptation of our earlier work [32]using differential evolution (DE) with a fitness functiondesigned for optimization in the CIE 133 domain

Table 3 gives a selection of the results that can beachieved and illustrates the feasibility of the approach Itshows that although outside the acceptable range of CCTforCOI purposes very good results are obtainable for Ra and Rfas well as for the skin colour indices R13 and Rfskin +e chiefdrawback of the high-range CCTs is their rather poor Rfminperformance

Note that 7500K is higher than normally recommendedfor interior lighting as it represents the stronger blue lightcomponent of the sky during daylight but it may be usedwhen appropriate for an enhancedHCL stimulation response

4 Conclusions

We have explained the importance of lighting from theperspective of human health and have shown how thehuman circadian system can best be stimulated for goodhealth effects We have also explained the value of cyanosisas a general indicator of health in the clinical setting andhave shown how good spectral design of the lighting is bestable to enhance cyanosis observation

We have developed an optimization tool in Matlabregwhich invokes the Cyanosis Observation Index (COI) toprovide the facility to optimize the lighting and so enhance

the caregiverrsquos ability to observe cyanosis+e recommendedrange of CCT for effective cyanosis observation is 3300ndash5300K and we have made recommendations for eightdifferent mixtures of four LED spectra that give good valuesfor the COI (ie lt33) over this range

We have utilized a previously developed CRI optimi-zation tool which has enabled us also to recommend fivefurther LED mixtures (using the same 4 LEDs as before) forCCTs which are below 3300K or above 5300K +ese willallow the clinician to provide a wider range of different CCTs(when cyanosis observation is not continuously required) tohelp harness the benefits of human-centric lighting (HCL)

It is noteworthy that all the recommended mixturesprovide excellent LER (285le LERle 354 lmW) togetherwith good colour rendition in terms of the CRI (Ra and R13)of the CIE as well as the colour fidelity index (Rf ) and colourskin index (Rfskin) of the IESNA

We note that the blue LED in our recommended mix-tures has a peak wavelength of 450 nm and a bandwidth(FWHM) of 18 nm (441ndash459 nm) It is therefore capable ofproviding significant energy for effective HCL activity withinthe recommended (446ndash477 nm) bandmdashparticularly whenoperating in the higher-CCTmixtures Conversely for low-CCTmixtures (in which the blue content is significantly cut)there will be low activation as required for the creation of arestful night-time ambience

+ere are numerous other ways in which lighting is ableto enhance the patient experience and health outcomes [33]and we have not attempted to include them all in this studyWe have focussed here on the more general areas of patientlighting as they may apply in both day-stay and overnighthospital wards or clinics

Data Availability

+e data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

+e authors declare that there are no conflicts of interestregarding the publication of this paper

Acknowledgments

+e second author Andrew Chalmers thanks ProfessorAhmed Al-Jumaily Director of IBTec at AUT for his

Table 3 Optimised mixtures for low and high CCTs

CCT (K)CIE CRI results

LER (lmw)IES TM-30-15 results

Surface colour giving Rfmin OIRa R9 R13 Rf Rg Rfskin Rfmin imin

Low2759 95 87 98 332 89 100 94 71 25 Yellowish-brown (A) 6003051 94 74 100 340 90 101 94 75 42 Olive-green (F) 595

High5734 90 95 98 309 86 107 97 57 75 Dark blue (F) 5856563 89 95 100 305 86 106 97 53 75 Dark blue (F) 5817539 86 71 90 285 83 110 96 50 75 Dark blue (F) 533

Colour descriptions are our own interpretations of the TM-30-15 test colours type A are from nature type F are print colours

Journal of Healthcare Engineering 7

support and the provision of facilities +is work wassupported by the School of Professional Engineering and theManukau Institute of Technology Research Fund

References

[1] SCHEER (Scientific Committee on Health Environmentaland Emerging Risks) Opinion on Potential Risks to HumanHealth of Light Emitting Diodes (LEDs) European Com-mission Luxembourg 2018 httpseceuropaeuhealthsiteshealthfilesscientific_committeesscheerdocsscheer_o_011pdf

[2] C Genova ldquoNormal responses to non-visual effects of lightretained by blind humans lacking rods and conesrdquo MedicalNews Today (Online) December 2007 httpswwwmedicalnewstodaycomreleases91836php

[3] W van Bommel and G van den Beld ldquoLighting for work areview of visual and biological effectsrdquo Lighting Research ampTechnology vol 36 no 4 pp 255ndash266 2004

[4] C S Pechacek M Andersen and S W Lockley ldquoPreliminarymethod for prospective analysis of the circadian efficacy of(day) light with applications to healthcare architecturerdquoLEUKOS vol 5 no 1 pp 1ndash26 2008

[5] M Andersen J Mardaljevic and S W Lockley ldquoA frame-work for predicting the non-visual effects of daylight Part 1photobiology-based modelrdquo Lighting Research amp Technologyvol 44 no 4 pp 37ndash53 2012

[6] C Meadows ldquoSetting expectations on human-centric light-ingrdquo LEDs Magazine (Online) December 2017 httpswwwledsmagazinecomarticles201712setting-expectations-on-human-centric-lightinghtml

[7] N A Midolo and L Sergeyeva ldquoLighting for clinical ob-servation of cyanosisrdquo e Australian Hospital Engineervol 30 no 2 pp 38ndash43 2007

[8] M Changizi and K Rio ldquoHarnessing color vision for visualoximetry in central cyanosisrdquo Medical Hypotheses vol 74no 1 pp 87ndash91 2010

[9] H L Snider ldquoCyanosisrdquo in Clinical Methods e HistoryPhysical and Laboratory Examinations H K WalkerW D Hall and J W Hurst Eds Butterworths Boston MAUSA 3rd edition 1990 httpwwwthornlightingcomdownloadHandbook5-6pdf

[10] ASNZS 1680252018 AustralianNew Zealand StandardInterior and Workplace Lighting Part 25 Hospital andMedical Tasks Section 73 Cyanosis Observation Lightingand Appendix G Method for Determination of CyanosisObservation Index of a Light Source

[11] Commission Internationale de lrsquoEclairage Colorimetry CIEPublication 153 Vienna Austria 3rd edition 2003

[12] Commission Internationale de lrsquoEclairage InternationalLighting Vocabulary CIE publication DIS 017E2016 ViennaAustria 2017

[13] Commission Internationale de lrsquoEclairage Method of Mea-suring and Specifying Colour Rendering Properties of LightSources CIE Publication 133 Vienna Austria 1995

[14] Illuminating Engineering Society of North America IESMethod for Evaluating Light Source Color Rendition Illumi-nating Engineering Society New York NY USA 2015

[15] +e Society of Light and Lighting Lighting Guide 2 Hospitalsand Health Care Buildings CIBSE London UK 2008

[16] BSI Standards Publication BS EN 12464-12011 Light andLightingmdashLighting of Work Places Part 1 IndoorWork PlacesBSI London UK 2011

[17] ANSIIES RP-29-16 Lighting for Hospitals and HealthcareFacilities Illuminating Engineering Society of North AmericaNew York NY USA 2016

[18] S Alzubaidi and P K Soori ldquoEnergy efficient lighting systemdesign for hospitals diagnostic and treatment roommdasha casestudyrdquo Journal of Light amp Visual Environment vol 36 no 1pp 23ndash31 2012

[19] F Leccese C Montagnani S Iaia M Rocca and G SalvadorildquoQuality of lighting in hospital environments a wide surveythrough in situ measurementsrdquo Journal of Light amp VisualEnvironment vol 40 pp 52ndash65 2016

[20] +orn Lighting Ltd orn Applications and TechniquesHealthcare +orn Lighting Spennymoor UK 2019

[21] S Mehrotra S Basukala and S Devarakonda ldquoEffectivelighting design standards impacting patient care a systemsapproachrdquo Journal of Biosciences and Medicines vol 3 no 11pp 54ndash61 2015

[22] P Bartczak A Gebejes P Falt and M Hauta-Kasari ldquoAnLED-based tunable illumination for diverse medical appli-cationsrdquo in Proceedings of the IEEE Symposium on Computer-Based Medical Systems pp 292-293 IEEE Dublin IrelandAugust 2016

[23] M Litorja and B Ecker ldquoUse of a spectrally-tunable source toexplore improvement in chromatic contrast for illuminationof tissuesrdquo in Proceedings of the SPIEmdashe InternationalSociety for Optical Engineering vol 7596 Bellingham WAUSA December 2010

[24] Lumileds Lighting ldquoLUXEONreg K2 emitterrdquo in TechnicalDatasheet DS51 Lumileds Lighting US LLC San Jose CAUSA 2006

[25] S B Chiang and K Chien ldquo+e optimization of color-mixedLED lightingrdquo in Proceedings of the 2014 IEEE IndustryApplication Society Annual Meeting Vancouver CanadaOctober 2014

[26] W Davis and Y Ohno ldquoColor quality scalerdquo Optical Engi-neering vol 49 no 3 article 033602 2010

[27] A Llenas and J Carreras ldquoArbitrary spectral matching usingmulti-LED lighting systemsrdquo Optical Engineering vol 58no 3 article 035105 2019

[28] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient adaptive scheme for global optimization over con-tinuous spacesrdquo Technical report TR-95-012 InternationalComputer Science Institute Berkley CA USA 1995

[29] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient heuristic for global optimization over continuousspacerdquo Journal of Global Optimization vol 11 no 4pp 341ndash359 1997

[30] S Soltic and A Chalmers ldquoDifferential evolution for theoptimisation of multi-band white LED light sourcesrdquo LightingResearch amp Technology vol 44 no 2 pp 224ndash237 2012

[31] A N Chalmers and S Soltic ldquoLight source optimizationspectral design and simulation of four-band white-lightsourcesrdquo Optical Engineering vol 51 no 4 article 0440032012

[32] S Soltic and A N Chalmers ldquoDifferential evolution and itsapplication to intelligent spectral designrdquo in Proceedings of the16th Electronics New Zealand Conference (ENZCon) Uni-versity of Otago Dunedin New Zealand November 2009

[33] L J M Schlangen e Role of Lighting in Promoting Well-Being and Recovery Within Healthcare Philips White Pa-per Koninklijke Philips Electronics NV AmsterdamNetherlands 2010 httpswwwitdoescoukwpontentuploads201604Role-of-lighting-published-by-philips-as-a-whitepaper-2010pdf

8 Journal of Healthcare Engineering

Every DE process has to be controlled by a fitnessfunction that determines how closely the members of a givengeneration have approached the required objective(s) In thiscase the fitness function was set equal to the COI (equation(4)) and the set objective was that ffit be minimised +esolution vector (in this case SPD) having the lowest ffit aswell as CCT within range (ie 3300leCCTle 5300K) at theend of the final generation is regarded as the optimum

ffit COI (4)

For the purposes of this project we performed over 140independent optimization runs We kept P 10 throughoutand G of different durations (10 20 50 100 and 200) Eachvalue of G was applied in at least five separate optimizationruns Of the total number of runs G of 10 was applied 50times and G of 200 was used 65 times

27 Selection of LED Spectra We had access to the SPDs ofseven coloured LEDs in the Lumileds LUXEON range [24]For the first 75 optimization runs the program was set tomake a random selection of any 4 LEDs (from the 7available) to form each of the members of the initialpopulation for that run In reviewing all 75 runs it becameevident that there were 4 specific LED spectra that con-sistently gave a range of ldquogoodrdquo results as judged on thebasis of colour rendition and LER performance as well asCOI and CCT +ese four LED spectra then became thefocus of the next 65 optimization runs from which a finaltotal of 33 SPDs were selected as having potentially usefulproperties

Since we wish to make a practical proposal for actualclinical lighting we have subsequently made a selection of 8

mixtures (from the set of 33) to serve as basis for ourrecommendations

3 Results and Discussion

Our selection of the four final LED colours (with peakwavelengths in brackets) was royal blue (450 nm) green(525 nm) amber (590 nm) and red (640 nm) and the resultsshown in Table 2 were obtained using these colours foroptimum mixtures at specific CCTs Note that the blueconstituent of our mixture has its peak at 450 nm that iswithin the range for the stimulation of the alerting responseof the circadian system [5]

31Optimizing forCOI +eCCTs were selected on the basisof (a) the acceptable range of CCT to provide good COI and(b) the ability to provide a CCT range that will support thecircadian response +e selection of mixtures for CCTsoutside of the preferred range for COI is covered in a latersubsection

+e 33 optimised mixtures that conform to COI re-quirements are summarized in Table 2 which shows thevarious colorimetric and operational parameters for eachmixture sorted in order of increasing CCT Note that severaladditional parameters have been listed in the table to providemore detailed information on the colour properties of eachSPD

By definition the CRI value Ra [13] is based on the averagecolour difference for eight moderate-chroma colours evenlyspread around the hue circle +e CIE has also defined asupplementary set of six colours to provide more detail wherenecessary Among these are the saturated red sample (index

Table 1 Recommended source colour properties in codes and standards

Country or region Australia and New Zealand[10]

United Kingdom [15] andEuropean Union [16] United States [17]

Colour appearance (CCT) 3300KleCCTle 5300K (allareas) No general recommendation 2700K night lighting No other

general recommendationColour rendering indexmdashgeneraland ward areas Rage 80 Rage 80 No general recommendation

Colour renderingindexmdashexamination andtreatment areas

Rage 85 (eg dermatology) Rage 90 (all examination areas) Rage 80 for exam areasRagt 90 for surgery

Special colour requirements Cyanosis observationCOIle 33

Examination rooms (generallighting)

4000KleCCTle 5000KColour inspection (labs and

pharmacies)6000KleCCTle 6500K

Consider the appropriate specialcolour rendering index Ri foraccurate rendition of coloured

objects or human skin

Surgery and examination rooms4000KleCCTle 5000K

Human-centric lighting(reinforcing circadianentrainment)

Not includedLighting may have a significantinfluence on human circadian

rhythms

Lighting practitioners to be awareof principles of the circadian cycle

Note +is table represents a distillation by the authors of detailed and sometimes complex requirements or recommendations Please refer to the originaldocuments for full details and explanations

4 Journal of Healthcare Engineering

R9) and Caucasian skin colour analogue (index R13) and wehave included these two indices along with Ra

In a similar way the colour fidelity Rf is defined as anaverage colour difference for 99 different colours Indices forspecific colours can be extracted from the data available forthe calculation of Rf Here we have included Rfskin which theIESNA [14] defines as the average of two indices (for samplenumbers 15 and 18) representing light and dark skin tonesIn addition we have extracted the index (Rfmin) and itssample number (imin) for the sample having the lowest indexvalue for a particular SPD since this indicates the worst-rendered colour under the specific source

+e eight SPDs in our final recommendation were se-lected for their superior properties in terms of colour per-formance and LER To assist in the selection we formulateda simple Overall Index (OI) as a general figure of merit asdefined in equation (5) Each of the selections was made onthe basis of the highest OI value when compared with othersfully meeting the COI criteria in the immediate CCTvicinity

OI Ra + R9 + R13 + Rf + Rfskin + Rfmin +LER5

1113874 1113875 (5)

Focussing now on the recommended eight SPDs Fig-ure 1 shows their SPDs on a single set of axes All eight havebeen normalized to the red (640 nm) peak which all share astheir common maximum

In practice suitable controls will need to be imple-mented to adjust the relative contributions to the mixture byeach of the four LED colours +e controls should also bedesigned to enable dimming so as to permit the setting ofdifferent light levels at any selected CCT thereby enhancingthe HCL effects at appropriate times of the day

It is evident from Table 2 that the LER and all significantcolour parameters are dependent on the CCT Figure 2indicates the variations in a selection of these parametersfor the recommended range of CCT settings +e curvesshown are best-fit polynomials (order 2)

+e COI (for which lower values are preferable) showedan upward (ie deteriorating) trend with CCT and amaximum COI result of 30 for a CCT just below 5300K Allthe other depicted parameters also showed a deteriorating(now downward) trend +e LER was 354 lmW at 3300Kand dropped about 17 to the region of 300 lmW above4500K +e colour fidelity metrics Ra and Rf both dropped

Table 2 33 optimised mixtures of the four selected LED colours

SPD no CCT (K) COICIE CRI results IES TM-30-15 results

Surface colour giving Rfmin OIRa R9 R13 LER (lmw) Rf Rg Rfskin Rfmin imin

1 3304 096 91 49 95 354 90 99 91 75 40 Dark greenish-grey (F) 5622 3311 149 90 40 92 362 89 95 89 74 42 Olive-green (F) 5463 3365 122 93 61 98 339 89 103 91 67 20 Orange-brown (F) 5674 3381 088 92 59 97 344 90 102 92 73 20 Dark orange (F) 5725 3401 22 94 88 99 339 91 103 96 72 42 Olive-green (F) 6086 3444 096 93 68 98 344 90 102 93 74 42 Olive-green (F) 5857 3462 085 93 69 99 336 90 104 93 68 20 Orange-brown (F) 5798 3488 129 94 81 97 326 89 106 93 62 20 Orange-brown (F) 5819 3526 117 94 91 97 331 90 105 96 71 42 Olive-green (F) 60510 3607 149 92 66 97 182 90 102 93 74 42 Olive-green (F) 54811 3608 127 94 85 96 324 89 107 94 62 20 Orange-brown (F) 58512 3716 086 92 98 94 317 89 108 96 63 20 Dark orange (F) 59513 3744 071 92 96 95 322 89 107 97 67 42 Olive-green (F) 60014 3744 311 93 81 95 317 88 108 92 55 20 Dark orange (F) 56715 3753 117 91 84 90 310 88 110 97 62 20 Orange-brown (F) 57416 3880 124 93 91 96 323 89 107 96 65 20 Orange-brown (F) 59517 3895 141 93 94 95 319 89 108 96 63 20 Orange-brown (F) 59418 4060 217 93 80 99 329 89 105 95 68 20 Dark orange (F) 59019 4199 249 80 24 79 286 83 115 92 55 42 Olive-green (F) 47020 4222 149 89 79 93 316 88 108 98 64 42 Olive-green (F) 57421 4277 092 90 79 91 309 87 109 98 62 42 Olive-green (F) 56922 4308 333 92 97 93 193 88 108 96 59 20 Orange-brown (F) 56423 4431 083 88 69 90 306 87 110 97 61 42 Olive-green (F) 55324 4669 15 82 32 81 286 83 114 93 54 42 Olive-green (F) 48225 4694 105 87 61 90 303 86 110 96 59 42 Olive-green (F) 54026 4944 133 83 33 82 287 83 114 93 54 42 Olive-green (F) 48527 5111 082 83 37 84 291 84 113 94 54 42 Olive-green (F) 49428 5129 14 83 38 87 295 84 112 94 56 42 Olive-green (F) 50129 5148 244 83 44 83 287 83 113 94 53 42 Olive-green (F) 49730 5196 144 77 4 76 277 81 116 90 49 42 Olive-green (F) 43231 5211 077 82 31 84 290 83 114 93 54 42 Olive-green (F) 48532 5243 3 88 75 91 301 86 110 97 58 42 Olive-green (F) 55533 5331 337 88 77 91 301 86 110 97 57 42 Olive-green (F) 556Final recommendation SPD nos 1 9 13 17 20 25 28 and 32 Note that SPDs 22 and 33 are slightly outside the COI recommendations Colour descriptionsare our own interpretations of TM-30-15 test colours type F are print colours

Journal of Healthcare Engineering 5

from the low-90s region to approximately the mid-80s +eskin fidelity metric R13 fell from around 95 to about 89 andRfmin (an indicator of the worst-case fidelity) fell from 75 toroughly 57 Note that Rfskin was ge91 throughout and

together with the high R13 values indicates that skin colourrendition will be good at all CCT settings

Taken together these results indicate that operation at3300K gives the best economy as well as the best COI and

00102030405060708091

380 400 420 440 460 480 500 520 540 560 580 600 620 640 660 680 700 720 740 760 780

3304K3526K3744K

3895K4222K4694K

5129K5243K

Figure 1 SPDs of the eight recommended LED mixtures

0051

152

253

35

3000 3500 4000 4500 5000 5500

(a)

290300310320330340350360

3000 3500 4000 4500 5000 5500

(b)

82

84

86

88

90

92

94

96

3000 3500 4000 4500 5000 5500

(c)

838485868788899091

3000 3500 4000 4500 5000 5500

(d)

86

88

90

92

94

96

98

3000 3500 4000 4500 5000 5500

(e)

50

55

60

65

70

75

80

3000 3500 4000 4500 5000 5500

(f )

Figure 2 Relationship of source parameters to correlated colour temperature (CCT) Dependence of (a) COI (b) LER (c) Ra (d) Rf (e) R13and (f) Rfmin

6 Journal of Healthcare Engineering

best general colour fidelity properties It should be notedhowever that the above properties are in fact acceptable forthe full range of CCTs depicted and there are likely to besituations when higher-CCT operation may be desired egto satisfy patient preferences or for HCL purposes which callfor lower CCT at night and higher CCT in the morning

32 Optimizing for HCL HCL (human-centric lighting)objectives are not restricted to the CCT range desired forCOI and we note that it is possible in principle to set theCCT of the mixture to values that fall outside the recom-mended range for optimum cyanosis observation For ex-ample it may be desirable in a hospital ward to set both CCTand lumen output to low values to promote a restful eveningenvironment or to high values for a more stimulatingmorning and daytime environment

A further optimization experiment was therefore con-ducted to find acceptable mixtures to achieve such variationswithout changing the constituent wavelengths in the mix-ture +e method was an adaptation of our earlier work [32]using differential evolution (DE) with a fitness functiondesigned for optimization in the CIE 133 domain

Table 3 gives a selection of the results that can beachieved and illustrates the feasibility of the approach Itshows that although outside the acceptable range of CCTforCOI purposes very good results are obtainable for Ra and Rfas well as for the skin colour indices R13 and Rfskin +e chiefdrawback of the high-range CCTs is their rather poor Rfminperformance

Note that 7500K is higher than normally recommendedfor interior lighting as it represents the stronger blue lightcomponent of the sky during daylight but it may be usedwhen appropriate for an enhancedHCL stimulation response

4 Conclusions

We have explained the importance of lighting from theperspective of human health and have shown how thehuman circadian system can best be stimulated for goodhealth effects We have also explained the value of cyanosisas a general indicator of health in the clinical setting andhave shown how good spectral design of the lighting is bestable to enhance cyanosis observation

We have developed an optimization tool in Matlabregwhich invokes the Cyanosis Observation Index (COI) toprovide the facility to optimize the lighting and so enhance

the caregiverrsquos ability to observe cyanosis+e recommendedrange of CCT for effective cyanosis observation is 3300ndash5300K and we have made recommendations for eightdifferent mixtures of four LED spectra that give good valuesfor the COI (ie lt33) over this range

We have utilized a previously developed CRI optimi-zation tool which has enabled us also to recommend fivefurther LED mixtures (using the same 4 LEDs as before) forCCTs which are below 3300K or above 5300K +ese willallow the clinician to provide a wider range of different CCTs(when cyanosis observation is not continuously required) tohelp harness the benefits of human-centric lighting (HCL)

It is noteworthy that all the recommended mixturesprovide excellent LER (285le LERle 354 lmW) togetherwith good colour rendition in terms of the CRI (Ra and R13)of the CIE as well as the colour fidelity index (Rf ) and colourskin index (Rfskin) of the IESNA

We note that the blue LED in our recommended mix-tures has a peak wavelength of 450 nm and a bandwidth(FWHM) of 18 nm (441ndash459 nm) It is therefore capable ofproviding significant energy for effective HCL activity withinthe recommended (446ndash477 nm) bandmdashparticularly whenoperating in the higher-CCTmixtures Conversely for low-CCTmixtures (in which the blue content is significantly cut)there will be low activation as required for the creation of arestful night-time ambience

+ere are numerous other ways in which lighting is ableto enhance the patient experience and health outcomes [33]and we have not attempted to include them all in this studyWe have focussed here on the more general areas of patientlighting as they may apply in both day-stay and overnighthospital wards or clinics

Data Availability

+e data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

+e authors declare that there are no conflicts of interestregarding the publication of this paper

Acknowledgments

+e second author Andrew Chalmers thanks ProfessorAhmed Al-Jumaily Director of IBTec at AUT for his

Table 3 Optimised mixtures for low and high CCTs

CCT (K)CIE CRI results

LER (lmw)IES TM-30-15 results

Surface colour giving Rfmin OIRa R9 R13 Rf Rg Rfskin Rfmin imin

Low2759 95 87 98 332 89 100 94 71 25 Yellowish-brown (A) 6003051 94 74 100 340 90 101 94 75 42 Olive-green (F) 595

High5734 90 95 98 309 86 107 97 57 75 Dark blue (F) 5856563 89 95 100 305 86 106 97 53 75 Dark blue (F) 5817539 86 71 90 285 83 110 96 50 75 Dark blue (F) 533

Colour descriptions are our own interpretations of the TM-30-15 test colours type A are from nature type F are print colours

Journal of Healthcare Engineering 7

support and the provision of facilities +is work wassupported by the School of Professional Engineering and theManukau Institute of Technology Research Fund

References

[1] SCHEER (Scientific Committee on Health Environmentaland Emerging Risks) Opinion on Potential Risks to HumanHealth of Light Emitting Diodes (LEDs) European Com-mission Luxembourg 2018 httpseceuropaeuhealthsiteshealthfilesscientific_committeesscheerdocsscheer_o_011pdf

[2] C Genova ldquoNormal responses to non-visual effects of lightretained by blind humans lacking rods and conesrdquo MedicalNews Today (Online) December 2007 httpswwwmedicalnewstodaycomreleases91836php

[3] W van Bommel and G van den Beld ldquoLighting for work areview of visual and biological effectsrdquo Lighting Research ampTechnology vol 36 no 4 pp 255ndash266 2004

[4] C S Pechacek M Andersen and S W Lockley ldquoPreliminarymethod for prospective analysis of the circadian efficacy of(day) light with applications to healthcare architecturerdquoLEUKOS vol 5 no 1 pp 1ndash26 2008

[5] M Andersen J Mardaljevic and S W Lockley ldquoA frame-work for predicting the non-visual effects of daylight Part 1photobiology-based modelrdquo Lighting Research amp Technologyvol 44 no 4 pp 37ndash53 2012

[6] C Meadows ldquoSetting expectations on human-centric light-ingrdquo LEDs Magazine (Online) December 2017 httpswwwledsmagazinecomarticles201712setting-expectations-on-human-centric-lightinghtml

[7] N A Midolo and L Sergeyeva ldquoLighting for clinical ob-servation of cyanosisrdquo e Australian Hospital Engineervol 30 no 2 pp 38ndash43 2007

[8] M Changizi and K Rio ldquoHarnessing color vision for visualoximetry in central cyanosisrdquo Medical Hypotheses vol 74no 1 pp 87ndash91 2010

[9] H L Snider ldquoCyanosisrdquo in Clinical Methods e HistoryPhysical and Laboratory Examinations H K WalkerW D Hall and J W Hurst Eds Butterworths Boston MAUSA 3rd edition 1990 httpwwwthornlightingcomdownloadHandbook5-6pdf

[10] ASNZS 1680252018 AustralianNew Zealand StandardInterior and Workplace Lighting Part 25 Hospital andMedical Tasks Section 73 Cyanosis Observation Lightingand Appendix G Method for Determination of CyanosisObservation Index of a Light Source

[11] Commission Internationale de lrsquoEclairage Colorimetry CIEPublication 153 Vienna Austria 3rd edition 2003

[12] Commission Internationale de lrsquoEclairage InternationalLighting Vocabulary CIE publication DIS 017E2016 ViennaAustria 2017

[13] Commission Internationale de lrsquoEclairage Method of Mea-suring and Specifying Colour Rendering Properties of LightSources CIE Publication 133 Vienna Austria 1995

[14] Illuminating Engineering Society of North America IESMethod for Evaluating Light Source Color Rendition Illumi-nating Engineering Society New York NY USA 2015

[15] +e Society of Light and Lighting Lighting Guide 2 Hospitalsand Health Care Buildings CIBSE London UK 2008

[16] BSI Standards Publication BS EN 12464-12011 Light andLightingmdashLighting of Work Places Part 1 IndoorWork PlacesBSI London UK 2011

[17] ANSIIES RP-29-16 Lighting for Hospitals and HealthcareFacilities Illuminating Engineering Society of North AmericaNew York NY USA 2016

[18] S Alzubaidi and P K Soori ldquoEnergy efficient lighting systemdesign for hospitals diagnostic and treatment roommdasha casestudyrdquo Journal of Light amp Visual Environment vol 36 no 1pp 23ndash31 2012

[19] F Leccese C Montagnani S Iaia M Rocca and G SalvadorildquoQuality of lighting in hospital environments a wide surveythrough in situ measurementsrdquo Journal of Light amp VisualEnvironment vol 40 pp 52ndash65 2016

[20] +orn Lighting Ltd orn Applications and TechniquesHealthcare +orn Lighting Spennymoor UK 2019

[21] S Mehrotra S Basukala and S Devarakonda ldquoEffectivelighting design standards impacting patient care a systemsapproachrdquo Journal of Biosciences and Medicines vol 3 no 11pp 54ndash61 2015

[22] P Bartczak A Gebejes P Falt and M Hauta-Kasari ldquoAnLED-based tunable illumination for diverse medical appli-cationsrdquo in Proceedings of the IEEE Symposium on Computer-Based Medical Systems pp 292-293 IEEE Dublin IrelandAugust 2016

[23] M Litorja and B Ecker ldquoUse of a spectrally-tunable source toexplore improvement in chromatic contrast for illuminationof tissuesrdquo in Proceedings of the SPIEmdashe InternationalSociety for Optical Engineering vol 7596 Bellingham WAUSA December 2010

[24] Lumileds Lighting ldquoLUXEONreg K2 emitterrdquo in TechnicalDatasheet DS51 Lumileds Lighting US LLC San Jose CAUSA 2006

[25] S B Chiang and K Chien ldquo+e optimization of color-mixedLED lightingrdquo in Proceedings of the 2014 IEEE IndustryApplication Society Annual Meeting Vancouver CanadaOctober 2014

[26] W Davis and Y Ohno ldquoColor quality scalerdquo Optical Engi-neering vol 49 no 3 article 033602 2010

[27] A Llenas and J Carreras ldquoArbitrary spectral matching usingmulti-LED lighting systemsrdquo Optical Engineering vol 58no 3 article 035105 2019

[28] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient adaptive scheme for global optimization over con-tinuous spacesrdquo Technical report TR-95-012 InternationalComputer Science Institute Berkley CA USA 1995

[29] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient heuristic for global optimization over continuousspacerdquo Journal of Global Optimization vol 11 no 4pp 341ndash359 1997

[30] S Soltic and A Chalmers ldquoDifferential evolution for theoptimisation of multi-band white LED light sourcesrdquo LightingResearch amp Technology vol 44 no 2 pp 224ndash237 2012

[31] A N Chalmers and S Soltic ldquoLight source optimizationspectral design and simulation of four-band white-lightsourcesrdquo Optical Engineering vol 51 no 4 article 0440032012

[32] S Soltic and A N Chalmers ldquoDifferential evolution and itsapplication to intelligent spectral designrdquo in Proceedings of the16th Electronics New Zealand Conference (ENZCon) Uni-versity of Otago Dunedin New Zealand November 2009

[33] L J M Schlangen e Role of Lighting in Promoting Well-Being and Recovery Within Healthcare Philips White Pa-per Koninklijke Philips Electronics NV AmsterdamNetherlands 2010 httpswwwitdoescoukwpontentuploads201604Role-of-lighting-published-by-philips-as-a-whitepaper-2010pdf

8 Journal of Healthcare Engineering

R9) and Caucasian skin colour analogue (index R13) and wehave included these two indices along with Ra

In a similar way the colour fidelity Rf is defined as anaverage colour difference for 99 different colours Indices forspecific colours can be extracted from the data available forthe calculation of Rf Here we have included Rfskin which theIESNA [14] defines as the average of two indices (for samplenumbers 15 and 18) representing light and dark skin tonesIn addition we have extracted the index (Rfmin) and itssample number (imin) for the sample having the lowest indexvalue for a particular SPD since this indicates the worst-rendered colour under the specific source

+e eight SPDs in our final recommendation were se-lected for their superior properties in terms of colour per-formance and LER To assist in the selection we formulateda simple Overall Index (OI) as a general figure of merit asdefined in equation (5) Each of the selections was made onthe basis of the highest OI value when compared with othersfully meeting the COI criteria in the immediate CCTvicinity

OI Ra + R9 + R13 + Rf + Rfskin + Rfmin +LER5

1113874 1113875 (5)

Focussing now on the recommended eight SPDs Fig-ure 1 shows their SPDs on a single set of axes All eight havebeen normalized to the red (640 nm) peak which all share astheir common maximum

In practice suitable controls will need to be imple-mented to adjust the relative contributions to the mixture byeach of the four LED colours +e controls should also bedesigned to enable dimming so as to permit the setting ofdifferent light levels at any selected CCT thereby enhancingthe HCL effects at appropriate times of the day

It is evident from Table 2 that the LER and all significantcolour parameters are dependent on the CCT Figure 2indicates the variations in a selection of these parametersfor the recommended range of CCT settings +e curvesshown are best-fit polynomials (order 2)

+e COI (for which lower values are preferable) showedan upward (ie deteriorating) trend with CCT and amaximum COI result of 30 for a CCT just below 5300K Allthe other depicted parameters also showed a deteriorating(now downward) trend +e LER was 354 lmW at 3300Kand dropped about 17 to the region of 300 lmW above4500K +e colour fidelity metrics Ra and Rf both dropped

Table 2 33 optimised mixtures of the four selected LED colours

SPD no CCT (K) COICIE CRI results IES TM-30-15 results

Surface colour giving Rfmin OIRa R9 R13 LER (lmw) Rf Rg Rfskin Rfmin imin

1 3304 096 91 49 95 354 90 99 91 75 40 Dark greenish-grey (F) 5622 3311 149 90 40 92 362 89 95 89 74 42 Olive-green (F) 5463 3365 122 93 61 98 339 89 103 91 67 20 Orange-brown (F) 5674 3381 088 92 59 97 344 90 102 92 73 20 Dark orange (F) 5725 3401 22 94 88 99 339 91 103 96 72 42 Olive-green (F) 6086 3444 096 93 68 98 344 90 102 93 74 42 Olive-green (F) 5857 3462 085 93 69 99 336 90 104 93 68 20 Orange-brown (F) 5798 3488 129 94 81 97 326 89 106 93 62 20 Orange-brown (F) 5819 3526 117 94 91 97 331 90 105 96 71 42 Olive-green (F) 60510 3607 149 92 66 97 182 90 102 93 74 42 Olive-green (F) 54811 3608 127 94 85 96 324 89 107 94 62 20 Orange-brown (F) 58512 3716 086 92 98 94 317 89 108 96 63 20 Dark orange (F) 59513 3744 071 92 96 95 322 89 107 97 67 42 Olive-green (F) 60014 3744 311 93 81 95 317 88 108 92 55 20 Dark orange (F) 56715 3753 117 91 84 90 310 88 110 97 62 20 Orange-brown (F) 57416 3880 124 93 91 96 323 89 107 96 65 20 Orange-brown (F) 59517 3895 141 93 94 95 319 89 108 96 63 20 Orange-brown (F) 59418 4060 217 93 80 99 329 89 105 95 68 20 Dark orange (F) 59019 4199 249 80 24 79 286 83 115 92 55 42 Olive-green (F) 47020 4222 149 89 79 93 316 88 108 98 64 42 Olive-green (F) 57421 4277 092 90 79 91 309 87 109 98 62 42 Olive-green (F) 56922 4308 333 92 97 93 193 88 108 96 59 20 Orange-brown (F) 56423 4431 083 88 69 90 306 87 110 97 61 42 Olive-green (F) 55324 4669 15 82 32 81 286 83 114 93 54 42 Olive-green (F) 48225 4694 105 87 61 90 303 86 110 96 59 42 Olive-green (F) 54026 4944 133 83 33 82 287 83 114 93 54 42 Olive-green (F) 48527 5111 082 83 37 84 291 84 113 94 54 42 Olive-green (F) 49428 5129 14 83 38 87 295 84 112 94 56 42 Olive-green (F) 50129 5148 244 83 44 83 287 83 113 94 53 42 Olive-green (F) 49730 5196 144 77 4 76 277 81 116 90 49 42 Olive-green (F) 43231 5211 077 82 31 84 290 83 114 93 54 42 Olive-green (F) 48532 5243 3 88 75 91 301 86 110 97 58 42 Olive-green (F) 55533 5331 337 88 77 91 301 86 110 97 57 42 Olive-green (F) 556Final recommendation SPD nos 1 9 13 17 20 25 28 and 32 Note that SPDs 22 and 33 are slightly outside the COI recommendations Colour descriptionsare our own interpretations of TM-30-15 test colours type F are print colours

Journal of Healthcare Engineering 5

from the low-90s region to approximately the mid-80s +eskin fidelity metric R13 fell from around 95 to about 89 andRfmin (an indicator of the worst-case fidelity) fell from 75 toroughly 57 Note that Rfskin was ge91 throughout and

together with the high R13 values indicates that skin colourrendition will be good at all CCT settings

Taken together these results indicate that operation at3300K gives the best economy as well as the best COI and

00102030405060708091

380 400 420 440 460 480 500 520 540 560 580 600 620 640 660 680 700 720 740 760 780

3304K3526K3744K

3895K4222K4694K

5129K5243K

Figure 1 SPDs of the eight recommended LED mixtures

0051

152

253

35

3000 3500 4000 4500 5000 5500

(a)

290300310320330340350360

3000 3500 4000 4500 5000 5500

(b)

82

84

86

88

90

92

94

96

3000 3500 4000 4500 5000 5500

(c)

838485868788899091

3000 3500 4000 4500 5000 5500

(d)

86

88

90

92

94

96

98

3000 3500 4000 4500 5000 5500

(e)

50

55

60

65

70

75

80

3000 3500 4000 4500 5000 5500

(f )

Figure 2 Relationship of source parameters to correlated colour temperature (CCT) Dependence of (a) COI (b) LER (c) Ra (d) Rf (e) R13and (f) Rfmin

6 Journal of Healthcare Engineering

best general colour fidelity properties It should be notedhowever that the above properties are in fact acceptable forthe full range of CCTs depicted and there are likely to besituations when higher-CCT operation may be desired egto satisfy patient preferences or for HCL purposes which callfor lower CCT at night and higher CCT in the morning

32 Optimizing for HCL HCL (human-centric lighting)objectives are not restricted to the CCT range desired forCOI and we note that it is possible in principle to set theCCT of the mixture to values that fall outside the recom-mended range for optimum cyanosis observation For ex-ample it may be desirable in a hospital ward to set both CCTand lumen output to low values to promote a restful eveningenvironment or to high values for a more stimulatingmorning and daytime environment

A further optimization experiment was therefore con-ducted to find acceptable mixtures to achieve such variationswithout changing the constituent wavelengths in the mix-ture +e method was an adaptation of our earlier work [32]using differential evolution (DE) with a fitness functiondesigned for optimization in the CIE 133 domain

Table 3 gives a selection of the results that can beachieved and illustrates the feasibility of the approach Itshows that although outside the acceptable range of CCTforCOI purposes very good results are obtainable for Ra and Rfas well as for the skin colour indices R13 and Rfskin +e chiefdrawback of the high-range CCTs is their rather poor Rfminperformance

Note that 7500K is higher than normally recommendedfor interior lighting as it represents the stronger blue lightcomponent of the sky during daylight but it may be usedwhen appropriate for an enhancedHCL stimulation response

4 Conclusions

We have explained the importance of lighting from theperspective of human health and have shown how thehuman circadian system can best be stimulated for goodhealth effects We have also explained the value of cyanosisas a general indicator of health in the clinical setting andhave shown how good spectral design of the lighting is bestable to enhance cyanosis observation

We have developed an optimization tool in Matlabregwhich invokes the Cyanosis Observation Index (COI) toprovide the facility to optimize the lighting and so enhance

the caregiverrsquos ability to observe cyanosis+e recommendedrange of CCT for effective cyanosis observation is 3300ndash5300K and we have made recommendations for eightdifferent mixtures of four LED spectra that give good valuesfor the COI (ie lt33) over this range

We have utilized a previously developed CRI optimi-zation tool which has enabled us also to recommend fivefurther LED mixtures (using the same 4 LEDs as before) forCCTs which are below 3300K or above 5300K +ese willallow the clinician to provide a wider range of different CCTs(when cyanosis observation is not continuously required) tohelp harness the benefits of human-centric lighting (HCL)

It is noteworthy that all the recommended mixturesprovide excellent LER (285le LERle 354 lmW) togetherwith good colour rendition in terms of the CRI (Ra and R13)of the CIE as well as the colour fidelity index (Rf ) and colourskin index (Rfskin) of the IESNA

We note that the blue LED in our recommended mix-tures has a peak wavelength of 450 nm and a bandwidth(FWHM) of 18 nm (441ndash459 nm) It is therefore capable ofproviding significant energy for effective HCL activity withinthe recommended (446ndash477 nm) bandmdashparticularly whenoperating in the higher-CCTmixtures Conversely for low-CCTmixtures (in which the blue content is significantly cut)there will be low activation as required for the creation of arestful night-time ambience

+ere are numerous other ways in which lighting is ableto enhance the patient experience and health outcomes [33]and we have not attempted to include them all in this studyWe have focussed here on the more general areas of patientlighting as they may apply in both day-stay and overnighthospital wards or clinics

Data Availability

+e data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

+e authors declare that there are no conflicts of interestregarding the publication of this paper

Acknowledgments

+e second author Andrew Chalmers thanks ProfessorAhmed Al-Jumaily Director of IBTec at AUT for his

Table 3 Optimised mixtures for low and high CCTs

CCT (K)CIE CRI results

LER (lmw)IES TM-30-15 results

Surface colour giving Rfmin OIRa R9 R13 Rf Rg Rfskin Rfmin imin

Low2759 95 87 98 332 89 100 94 71 25 Yellowish-brown (A) 6003051 94 74 100 340 90 101 94 75 42 Olive-green (F) 595

High5734 90 95 98 309 86 107 97 57 75 Dark blue (F) 5856563 89 95 100 305 86 106 97 53 75 Dark blue (F) 5817539 86 71 90 285 83 110 96 50 75 Dark blue (F) 533

Colour descriptions are our own interpretations of the TM-30-15 test colours type A are from nature type F are print colours

Journal of Healthcare Engineering 7

support and the provision of facilities +is work wassupported by the School of Professional Engineering and theManukau Institute of Technology Research Fund

References

[1] SCHEER (Scientific Committee on Health Environmentaland Emerging Risks) Opinion on Potential Risks to HumanHealth of Light Emitting Diodes (LEDs) European Com-mission Luxembourg 2018 httpseceuropaeuhealthsiteshealthfilesscientific_committeesscheerdocsscheer_o_011pdf

[2] C Genova ldquoNormal responses to non-visual effects of lightretained by blind humans lacking rods and conesrdquo MedicalNews Today (Online) December 2007 httpswwwmedicalnewstodaycomreleases91836php

[3] W van Bommel and G van den Beld ldquoLighting for work areview of visual and biological effectsrdquo Lighting Research ampTechnology vol 36 no 4 pp 255ndash266 2004

[4] C S Pechacek M Andersen and S W Lockley ldquoPreliminarymethod for prospective analysis of the circadian efficacy of(day) light with applications to healthcare architecturerdquoLEUKOS vol 5 no 1 pp 1ndash26 2008

[5] M Andersen J Mardaljevic and S W Lockley ldquoA frame-work for predicting the non-visual effects of daylight Part 1photobiology-based modelrdquo Lighting Research amp Technologyvol 44 no 4 pp 37ndash53 2012

[6] C Meadows ldquoSetting expectations on human-centric light-ingrdquo LEDs Magazine (Online) December 2017 httpswwwledsmagazinecomarticles201712setting-expectations-on-human-centric-lightinghtml

[7] N A Midolo and L Sergeyeva ldquoLighting for clinical ob-servation of cyanosisrdquo e Australian Hospital Engineervol 30 no 2 pp 38ndash43 2007

[8] M Changizi and K Rio ldquoHarnessing color vision for visualoximetry in central cyanosisrdquo Medical Hypotheses vol 74no 1 pp 87ndash91 2010

[9] H L Snider ldquoCyanosisrdquo in Clinical Methods e HistoryPhysical and Laboratory Examinations H K WalkerW D Hall and J W Hurst Eds Butterworths Boston MAUSA 3rd edition 1990 httpwwwthornlightingcomdownloadHandbook5-6pdf

[10] ASNZS 1680252018 AustralianNew Zealand StandardInterior and Workplace Lighting Part 25 Hospital andMedical Tasks Section 73 Cyanosis Observation Lightingand Appendix G Method for Determination of CyanosisObservation Index of a Light Source

[11] Commission Internationale de lrsquoEclairage Colorimetry CIEPublication 153 Vienna Austria 3rd edition 2003

[12] Commission Internationale de lrsquoEclairage InternationalLighting Vocabulary CIE publication DIS 017E2016 ViennaAustria 2017

[13] Commission Internationale de lrsquoEclairage Method of Mea-suring and Specifying Colour Rendering Properties of LightSources CIE Publication 133 Vienna Austria 1995

[14] Illuminating Engineering Society of North America IESMethod for Evaluating Light Source Color Rendition Illumi-nating Engineering Society New York NY USA 2015

[15] +e Society of Light and Lighting Lighting Guide 2 Hospitalsand Health Care Buildings CIBSE London UK 2008

[16] BSI Standards Publication BS EN 12464-12011 Light andLightingmdashLighting of Work Places Part 1 IndoorWork PlacesBSI London UK 2011

[17] ANSIIES RP-29-16 Lighting for Hospitals and HealthcareFacilities Illuminating Engineering Society of North AmericaNew York NY USA 2016

[18] S Alzubaidi and P K Soori ldquoEnergy efficient lighting systemdesign for hospitals diagnostic and treatment roommdasha casestudyrdquo Journal of Light amp Visual Environment vol 36 no 1pp 23ndash31 2012

[19] F Leccese C Montagnani S Iaia M Rocca and G SalvadorildquoQuality of lighting in hospital environments a wide surveythrough in situ measurementsrdquo Journal of Light amp VisualEnvironment vol 40 pp 52ndash65 2016

[20] +orn Lighting Ltd orn Applications and TechniquesHealthcare +orn Lighting Spennymoor UK 2019

[21] S Mehrotra S Basukala and S Devarakonda ldquoEffectivelighting design standards impacting patient care a systemsapproachrdquo Journal of Biosciences and Medicines vol 3 no 11pp 54ndash61 2015

[22] P Bartczak A Gebejes P Falt and M Hauta-Kasari ldquoAnLED-based tunable illumination for diverse medical appli-cationsrdquo in Proceedings of the IEEE Symposium on Computer-Based Medical Systems pp 292-293 IEEE Dublin IrelandAugust 2016

[23] M Litorja and B Ecker ldquoUse of a spectrally-tunable source toexplore improvement in chromatic contrast for illuminationof tissuesrdquo in Proceedings of the SPIEmdashe InternationalSociety for Optical Engineering vol 7596 Bellingham WAUSA December 2010

[24] Lumileds Lighting ldquoLUXEONreg K2 emitterrdquo in TechnicalDatasheet DS51 Lumileds Lighting US LLC San Jose CAUSA 2006

[25] S B Chiang and K Chien ldquo+e optimization of color-mixedLED lightingrdquo in Proceedings of the 2014 IEEE IndustryApplication Society Annual Meeting Vancouver CanadaOctober 2014

[26] W Davis and Y Ohno ldquoColor quality scalerdquo Optical Engi-neering vol 49 no 3 article 033602 2010

[27] A Llenas and J Carreras ldquoArbitrary spectral matching usingmulti-LED lighting systemsrdquo Optical Engineering vol 58no 3 article 035105 2019

[28] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient adaptive scheme for global optimization over con-tinuous spacesrdquo Technical report TR-95-012 InternationalComputer Science Institute Berkley CA USA 1995

[29] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient heuristic for global optimization over continuousspacerdquo Journal of Global Optimization vol 11 no 4pp 341ndash359 1997

[30] S Soltic and A Chalmers ldquoDifferential evolution for theoptimisation of multi-band white LED light sourcesrdquo LightingResearch amp Technology vol 44 no 2 pp 224ndash237 2012

[31] A N Chalmers and S Soltic ldquoLight source optimizationspectral design and simulation of four-band white-lightsourcesrdquo Optical Engineering vol 51 no 4 article 0440032012

[32] S Soltic and A N Chalmers ldquoDifferential evolution and itsapplication to intelligent spectral designrdquo in Proceedings of the16th Electronics New Zealand Conference (ENZCon) Uni-versity of Otago Dunedin New Zealand November 2009

[33] L J M Schlangen e Role of Lighting in Promoting Well-Being and Recovery Within Healthcare Philips White Pa-per Koninklijke Philips Electronics NV AmsterdamNetherlands 2010 httpswwwitdoescoukwpontentuploads201604Role-of-lighting-published-by-philips-as-a-whitepaper-2010pdf

8 Journal of Healthcare Engineering

from the low-90s region to approximately the mid-80s +eskin fidelity metric R13 fell from around 95 to about 89 andRfmin (an indicator of the worst-case fidelity) fell from 75 toroughly 57 Note that Rfskin was ge91 throughout and

together with the high R13 values indicates that skin colourrendition will be good at all CCT settings

Taken together these results indicate that operation at3300K gives the best economy as well as the best COI and

00102030405060708091

380 400 420 440 460 480 500 520 540 560 580 600 620 640 660 680 700 720 740 760 780

3304K3526K3744K

3895K4222K4694K

5129K5243K

Figure 1 SPDs of the eight recommended LED mixtures

0051

152

253

35

3000 3500 4000 4500 5000 5500

(a)

290300310320330340350360

3000 3500 4000 4500 5000 5500

(b)

82

84

86

88

90

92

94

96

3000 3500 4000 4500 5000 5500

(c)

838485868788899091

3000 3500 4000 4500 5000 5500

(d)

86

88

90

92

94

96

98

3000 3500 4000 4500 5000 5500

(e)

50

55

60

65

70

75

80

3000 3500 4000 4500 5000 5500

(f )

Figure 2 Relationship of source parameters to correlated colour temperature (CCT) Dependence of (a) COI (b) LER (c) Ra (d) Rf (e) R13and (f) Rfmin

6 Journal of Healthcare Engineering

best general colour fidelity properties It should be notedhowever that the above properties are in fact acceptable forthe full range of CCTs depicted and there are likely to besituations when higher-CCT operation may be desired egto satisfy patient preferences or for HCL purposes which callfor lower CCT at night and higher CCT in the morning

32 Optimizing for HCL HCL (human-centric lighting)objectives are not restricted to the CCT range desired forCOI and we note that it is possible in principle to set theCCT of the mixture to values that fall outside the recom-mended range for optimum cyanosis observation For ex-ample it may be desirable in a hospital ward to set both CCTand lumen output to low values to promote a restful eveningenvironment or to high values for a more stimulatingmorning and daytime environment

A further optimization experiment was therefore con-ducted to find acceptable mixtures to achieve such variationswithout changing the constituent wavelengths in the mix-ture +e method was an adaptation of our earlier work [32]using differential evolution (DE) with a fitness functiondesigned for optimization in the CIE 133 domain

Table 3 gives a selection of the results that can beachieved and illustrates the feasibility of the approach Itshows that although outside the acceptable range of CCTforCOI purposes very good results are obtainable for Ra and Rfas well as for the skin colour indices R13 and Rfskin +e chiefdrawback of the high-range CCTs is their rather poor Rfminperformance

Note that 7500K is higher than normally recommendedfor interior lighting as it represents the stronger blue lightcomponent of the sky during daylight but it may be usedwhen appropriate for an enhancedHCL stimulation response

4 Conclusions

We have explained the importance of lighting from theperspective of human health and have shown how thehuman circadian system can best be stimulated for goodhealth effects We have also explained the value of cyanosisas a general indicator of health in the clinical setting andhave shown how good spectral design of the lighting is bestable to enhance cyanosis observation

We have developed an optimization tool in Matlabregwhich invokes the Cyanosis Observation Index (COI) toprovide the facility to optimize the lighting and so enhance

the caregiverrsquos ability to observe cyanosis+e recommendedrange of CCT for effective cyanosis observation is 3300ndash5300K and we have made recommendations for eightdifferent mixtures of four LED spectra that give good valuesfor the COI (ie lt33) over this range

We have utilized a previously developed CRI optimi-zation tool which has enabled us also to recommend fivefurther LED mixtures (using the same 4 LEDs as before) forCCTs which are below 3300K or above 5300K +ese willallow the clinician to provide a wider range of different CCTs(when cyanosis observation is not continuously required) tohelp harness the benefits of human-centric lighting (HCL)

It is noteworthy that all the recommended mixturesprovide excellent LER (285le LERle 354 lmW) togetherwith good colour rendition in terms of the CRI (Ra and R13)of the CIE as well as the colour fidelity index (Rf ) and colourskin index (Rfskin) of the IESNA

We note that the blue LED in our recommended mix-tures has a peak wavelength of 450 nm and a bandwidth(FWHM) of 18 nm (441ndash459 nm) It is therefore capable ofproviding significant energy for effective HCL activity withinthe recommended (446ndash477 nm) bandmdashparticularly whenoperating in the higher-CCTmixtures Conversely for low-CCTmixtures (in which the blue content is significantly cut)there will be low activation as required for the creation of arestful night-time ambience

+ere are numerous other ways in which lighting is ableto enhance the patient experience and health outcomes [33]and we have not attempted to include them all in this studyWe have focussed here on the more general areas of patientlighting as they may apply in both day-stay and overnighthospital wards or clinics

Data Availability

+e data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

+e authors declare that there are no conflicts of interestregarding the publication of this paper

Acknowledgments

+e second author Andrew Chalmers thanks ProfessorAhmed Al-Jumaily Director of IBTec at AUT for his

Table 3 Optimised mixtures for low and high CCTs

CCT (K)CIE CRI results

LER (lmw)IES TM-30-15 results

Surface colour giving Rfmin OIRa R9 R13 Rf Rg Rfskin Rfmin imin

Low2759 95 87 98 332 89 100 94 71 25 Yellowish-brown (A) 6003051 94 74 100 340 90 101 94 75 42 Olive-green (F) 595

High5734 90 95 98 309 86 107 97 57 75 Dark blue (F) 5856563 89 95 100 305 86 106 97 53 75 Dark blue (F) 5817539 86 71 90 285 83 110 96 50 75 Dark blue (F) 533

Colour descriptions are our own interpretations of the TM-30-15 test colours type A are from nature type F are print colours

Journal of Healthcare Engineering 7

support and the provision of facilities +is work wassupported by the School of Professional Engineering and theManukau Institute of Technology Research Fund

References

[1] SCHEER (Scientific Committee on Health Environmentaland Emerging Risks) Opinion on Potential Risks to HumanHealth of Light Emitting Diodes (LEDs) European Com-mission Luxembourg 2018 httpseceuropaeuhealthsiteshealthfilesscientific_committeesscheerdocsscheer_o_011pdf

[2] C Genova ldquoNormal responses to non-visual effects of lightretained by blind humans lacking rods and conesrdquo MedicalNews Today (Online) December 2007 httpswwwmedicalnewstodaycomreleases91836php

[3] W van Bommel and G van den Beld ldquoLighting for work areview of visual and biological effectsrdquo Lighting Research ampTechnology vol 36 no 4 pp 255ndash266 2004

[4] C S Pechacek M Andersen and S W Lockley ldquoPreliminarymethod for prospective analysis of the circadian efficacy of(day) light with applications to healthcare architecturerdquoLEUKOS vol 5 no 1 pp 1ndash26 2008

[5] M Andersen J Mardaljevic and S W Lockley ldquoA frame-work for predicting the non-visual effects of daylight Part 1photobiology-based modelrdquo Lighting Research amp Technologyvol 44 no 4 pp 37ndash53 2012

[6] C Meadows ldquoSetting expectations on human-centric light-ingrdquo LEDs Magazine (Online) December 2017 httpswwwledsmagazinecomarticles201712setting-expectations-on-human-centric-lightinghtml

[7] N A Midolo and L Sergeyeva ldquoLighting for clinical ob-servation of cyanosisrdquo e Australian Hospital Engineervol 30 no 2 pp 38ndash43 2007

[8] M Changizi and K Rio ldquoHarnessing color vision for visualoximetry in central cyanosisrdquo Medical Hypotheses vol 74no 1 pp 87ndash91 2010

[9] H L Snider ldquoCyanosisrdquo in Clinical Methods e HistoryPhysical and Laboratory Examinations H K WalkerW D Hall and J W Hurst Eds Butterworths Boston MAUSA 3rd edition 1990 httpwwwthornlightingcomdownloadHandbook5-6pdf

[10] ASNZS 1680252018 AustralianNew Zealand StandardInterior and Workplace Lighting Part 25 Hospital andMedical Tasks Section 73 Cyanosis Observation Lightingand Appendix G Method for Determination of CyanosisObservation Index of a Light Source

[11] Commission Internationale de lrsquoEclairage Colorimetry CIEPublication 153 Vienna Austria 3rd edition 2003

[12] Commission Internationale de lrsquoEclairage InternationalLighting Vocabulary CIE publication DIS 017E2016 ViennaAustria 2017

[13] Commission Internationale de lrsquoEclairage Method of Mea-suring and Specifying Colour Rendering Properties of LightSources CIE Publication 133 Vienna Austria 1995

[14] Illuminating Engineering Society of North America IESMethod for Evaluating Light Source Color Rendition Illumi-nating Engineering Society New York NY USA 2015

[15] +e Society of Light and Lighting Lighting Guide 2 Hospitalsand Health Care Buildings CIBSE London UK 2008

[16] BSI Standards Publication BS EN 12464-12011 Light andLightingmdashLighting of Work Places Part 1 IndoorWork PlacesBSI London UK 2011

[17] ANSIIES RP-29-16 Lighting for Hospitals and HealthcareFacilities Illuminating Engineering Society of North AmericaNew York NY USA 2016

[18] S Alzubaidi and P K Soori ldquoEnergy efficient lighting systemdesign for hospitals diagnostic and treatment roommdasha casestudyrdquo Journal of Light amp Visual Environment vol 36 no 1pp 23ndash31 2012

[19] F Leccese C Montagnani S Iaia M Rocca and G SalvadorildquoQuality of lighting in hospital environments a wide surveythrough in situ measurementsrdquo Journal of Light amp VisualEnvironment vol 40 pp 52ndash65 2016

[20] +orn Lighting Ltd orn Applications and TechniquesHealthcare +orn Lighting Spennymoor UK 2019

[21] S Mehrotra S Basukala and S Devarakonda ldquoEffectivelighting design standards impacting patient care a systemsapproachrdquo Journal of Biosciences and Medicines vol 3 no 11pp 54ndash61 2015

[22] P Bartczak A Gebejes P Falt and M Hauta-Kasari ldquoAnLED-based tunable illumination for diverse medical appli-cationsrdquo in Proceedings of the IEEE Symposium on Computer-Based Medical Systems pp 292-293 IEEE Dublin IrelandAugust 2016

[23] M Litorja and B Ecker ldquoUse of a spectrally-tunable source toexplore improvement in chromatic contrast for illuminationof tissuesrdquo in Proceedings of the SPIEmdashe InternationalSociety for Optical Engineering vol 7596 Bellingham WAUSA December 2010

[24] Lumileds Lighting ldquoLUXEONreg K2 emitterrdquo in TechnicalDatasheet DS51 Lumileds Lighting US LLC San Jose CAUSA 2006

[25] S B Chiang and K Chien ldquo+e optimization of color-mixedLED lightingrdquo in Proceedings of the 2014 IEEE IndustryApplication Society Annual Meeting Vancouver CanadaOctober 2014

[26] W Davis and Y Ohno ldquoColor quality scalerdquo Optical Engi-neering vol 49 no 3 article 033602 2010

[27] A Llenas and J Carreras ldquoArbitrary spectral matching usingmulti-LED lighting systemsrdquo Optical Engineering vol 58no 3 article 035105 2019

[28] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient adaptive scheme for global optimization over con-tinuous spacesrdquo Technical report TR-95-012 InternationalComputer Science Institute Berkley CA USA 1995

[29] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient heuristic for global optimization over continuousspacerdquo Journal of Global Optimization vol 11 no 4pp 341ndash359 1997

[30] S Soltic and A Chalmers ldquoDifferential evolution for theoptimisation of multi-band white LED light sourcesrdquo LightingResearch amp Technology vol 44 no 2 pp 224ndash237 2012

[31] A N Chalmers and S Soltic ldquoLight source optimizationspectral design and simulation of four-band white-lightsourcesrdquo Optical Engineering vol 51 no 4 article 0440032012

[32] S Soltic and A N Chalmers ldquoDifferential evolution and itsapplication to intelligent spectral designrdquo in Proceedings of the16th Electronics New Zealand Conference (ENZCon) Uni-versity of Otago Dunedin New Zealand November 2009

[33] L J M Schlangen e Role of Lighting in Promoting Well-Being and Recovery Within Healthcare Philips White Pa-per Koninklijke Philips Electronics NV AmsterdamNetherlands 2010 httpswwwitdoescoukwpontentuploads201604Role-of-lighting-published-by-philips-as-a-whitepaper-2010pdf

8 Journal of Healthcare Engineering

best general colour fidelity properties It should be notedhowever that the above properties are in fact acceptable forthe full range of CCTs depicted and there are likely to besituations when higher-CCT operation may be desired egto satisfy patient preferences or for HCL purposes which callfor lower CCT at night and higher CCT in the morning

32 Optimizing for HCL HCL (human-centric lighting)objectives are not restricted to the CCT range desired forCOI and we note that it is possible in principle to set theCCT of the mixture to values that fall outside the recom-mended range for optimum cyanosis observation For ex-ample it may be desirable in a hospital ward to set both CCTand lumen output to low values to promote a restful eveningenvironment or to high values for a more stimulatingmorning and daytime environment

A further optimization experiment was therefore con-ducted to find acceptable mixtures to achieve such variationswithout changing the constituent wavelengths in the mix-ture +e method was an adaptation of our earlier work [32]using differential evolution (DE) with a fitness functiondesigned for optimization in the CIE 133 domain

Table 3 gives a selection of the results that can beachieved and illustrates the feasibility of the approach Itshows that although outside the acceptable range of CCTforCOI purposes very good results are obtainable for Ra and Rfas well as for the skin colour indices R13 and Rfskin +e chiefdrawback of the high-range CCTs is their rather poor Rfminperformance

Note that 7500K is higher than normally recommendedfor interior lighting as it represents the stronger blue lightcomponent of the sky during daylight but it may be usedwhen appropriate for an enhancedHCL stimulation response

4 Conclusions

We have explained the importance of lighting from theperspective of human health and have shown how thehuman circadian system can best be stimulated for goodhealth effects We have also explained the value of cyanosisas a general indicator of health in the clinical setting andhave shown how good spectral design of the lighting is bestable to enhance cyanosis observation

We have developed an optimization tool in Matlabregwhich invokes the Cyanosis Observation Index (COI) toprovide the facility to optimize the lighting and so enhance

the caregiverrsquos ability to observe cyanosis+e recommendedrange of CCT for effective cyanosis observation is 3300ndash5300K and we have made recommendations for eightdifferent mixtures of four LED spectra that give good valuesfor the COI (ie lt33) over this range

We have utilized a previously developed CRI optimi-zation tool which has enabled us also to recommend fivefurther LED mixtures (using the same 4 LEDs as before) forCCTs which are below 3300K or above 5300K +ese willallow the clinician to provide a wider range of different CCTs(when cyanosis observation is not continuously required) tohelp harness the benefits of human-centric lighting (HCL)

It is noteworthy that all the recommended mixturesprovide excellent LER (285le LERle 354 lmW) togetherwith good colour rendition in terms of the CRI (Ra and R13)of the CIE as well as the colour fidelity index (Rf ) and colourskin index (Rfskin) of the IESNA

We note that the blue LED in our recommended mix-tures has a peak wavelength of 450 nm and a bandwidth(FWHM) of 18 nm (441ndash459 nm) It is therefore capable ofproviding significant energy for effective HCL activity withinthe recommended (446ndash477 nm) bandmdashparticularly whenoperating in the higher-CCTmixtures Conversely for low-CCTmixtures (in which the blue content is significantly cut)there will be low activation as required for the creation of arestful night-time ambience

+ere are numerous other ways in which lighting is ableto enhance the patient experience and health outcomes [33]and we have not attempted to include them all in this studyWe have focussed here on the more general areas of patientlighting as they may apply in both day-stay and overnighthospital wards or clinics

Data Availability

+e data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

+e authors declare that there are no conflicts of interestregarding the publication of this paper

Acknowledgments

+e second author Andrew Chalmers thanks ProfessorAhmed Al-Jumaily Director of IBTec at AUT for his

Table 3 Optimised mixtures for low and high CCTs

CCT (K)CIE CRI results

LER (lmw)IES TM-30-15 results

Surface colour giving Rfmin OIRa R9 R13 Rf Rg Rfskin Rfmin imin

Low2759 95 87 98 332 89 100 94 71 25 Yellowish-brown (A) 6003051 94 74 100 340 90 101 94 75 42 Olive-green (F) 595

High5734 90 95 98 309 86 107 97 57 75 Dark blue (F) 5856563 89 95 100 305 86 106 97 53 75 Dark blue (F) 5817539 86 71 90 285 83 110 96 50 75 Dark blue (F) 533

Colour descriptions are our own interpretations of the TM-30-15 test colours type A are from nature type F are print colours

Journal of Healthcare Engineering 7

support and the provision of facilities +is work wassupported by the School of Professional Engineering and theManukau Institute of Technology Research Fund

References

[1] SCHEER (Scientific Committee on Health Environmentaland Emerging Risks) Opinion on Potential Risks to HumanHealth of Light Emitting Diodes (LEDs) European Com-mission Luxembourg 2018 httpseceuropaeuhealthsiteshealthfilesscientific_committeesscheerdocsscheer_o_011pdf

[2] C Genova ldquoNormal responses to non-visual effects of lightretained by blind humans lacking rods and conesrdquo MedicalNews Today (Online) December 2007 httpswwwmedicalnewstodaycomreleases91836php

[3] W van Bommel and G van den Beld ldquoLighting for work areview of visual and biological effectsrdquo Lighting Research ampTechnology vol 36 no 4 pp 255ndash266 2004

[4] C S Pechacek M Andersen and S W Lockley ldquoPreliminarymethod for prospective analysis of the circadian efficacy of(day) light with applications to healthcare architecturerdquoLEUKOS vol 5 no 1 pp 1ndash26 2008

[5] M Andersen J Mardaljevic and S W Lockley ldquoA frame-work for predicting the non-visual effects of daylight Part 1photobiology-based modelrdquo Lighting Research amp Technologyvol 44 no 4 pp 37ndash53 2012

[6] C Meadows ldquoSetting expectations on human-centric light-ingrdquo LEDs Magazine (Online) December 2017 httpswwwledsmagazinecomarticles201712setting-expectations-on-human-centric-lightinghtml

[7] N A Midolo and L Sergeyeva ldquoLighting for clinical ob-servation of cyanosisrdquo e Australian Hospital Engineervol 30 no 2 pp 38ndash43 2007

[8] M Changizi and K Rio ldquoHarnessing color vision for visualoximetry in central cyanosisrdquo Medical Hypotheses vol 74no 1 pp 87ndash91 2010

[9] H L Snider ldquoCyanosisrdquo in Clinical Methods e HistoryPhysical and Laboratory Examinations H K WalkerW D Hall and J W Hurst Eds Butterworths Boston MAUSA 3rd edition 1990 httpwwwthornlightingcomdownloadHandbook5-6pdf

[10] ASNZS 1680252018 AustralianNew Zealand StandardInterior and Workplace Lighting Part 25 Hospital andMedical Tasks Section 73 Cyanosis Observation Lightingand Appendix G Method for Determination of CyanosisObservation Index of a Light Source

[11] Commission Internationale de lrsquoEclairage Colorimetry CIEPublication 153 Vienna Austria 3rd edition 2003

[12] Commission Internationale de lrsquoEclairage InternationalLighting Vocabulary CIE publication DIS 017E2016 ViennaAustria 2017

[13] Commission Internationale de lrsquoEclairage Method of Mea-suring and Specifying Colour Rendering Properties of LightSources CIE Publication 133 Vienna Austria 1995

[14] Illuminating Engineering Society of North America IESMethod for Evaluating Light Source Color Rendition Illumi-nating Engineering Society New York NY USA 2015

[15] +e Society of Light and Lighting Lighting Guide 2 Hospitalsand Health Care Buildings CIBSE London UK 2008

[16] BSI Standards Publication BS EN 12464-12011 Light andLightingmdashLighting of Work Places Part 1 IndoorWork PlacesBSI London UK 2011

[17] ANSIIES RP-29-16 Lighting for Hospitals and HealthcareFacilities Illuminating Engineering Society of North AmericaNew York NY USA 2016

[18] S Alzubaidi and P K Soori ldquoEnergy efficient lighting systemdesign for hospitals diagnostic and treatment roommdasha casestudyrdquo Journal of Light amp Visual Environment vol 36 no 1pp 23ndash31 2012

[19] F Leccese C Montagnani S Iaia M Rocca and G SalvadorildquoQuality of lighting in hospital environments a wide surveythrough in situ measurementsrdquo Journal of Light amp VisualEnvironment vol 40 pp 52ndash65 2016

[20] +orn Lighting Ltd orn Applications and TechniquesHealthcare +orn Lighting Spennymoor UK 2019

[21] S Mehrotra S Basukala and S Devarakonda ldquoEffectivelighting design standards impacting patient care a systemsapproachrdquo Journal of Biosciences and Medicines vol 3 no 11pp 54ndash61 2015

[22] P Bartczak A Gebejes P Falt and M Hauta-Kasari ldquoAnLED-based tunable illumination for diverse medical appli-cationsrdquo in Proceedings of the IEEE Symposium on Computer-Based Medical Systems pp 292-293 IEEE Dublin IrelandAugust 2016

[23] M Litorja and B Ecker ldquoUse of a spectrally-tunable source toexplore improvement in chromatic contrast for illuminationof tissuesrdquo in Proceedings of the SPIEmdashe InternationalSociety for Optical Engineering vol 7596 Bellingham WAUSA December 2010

[24] Lumileds Lighting ldquoLUXEONreg K2 emitterrdquo in TechnicalDatasheet DS51 Lumileds Lighting US LLC San Jose CAUSA 2006

[25] S B Chiang and K Chien ldquo+e optimization of color-mixedLED lightingrdquo in Proceedings of the 2014 IEEE IndustryApplication Society Annual Meeting Vancouver CanadaOctober 2014

[26] W Davis and Y Ohno ldquoColor quality scalerdquo Optical Engi-neering vol 49 no 3 article 033602 2010

[27] A Llenas and J Carreras ldquoArbitrary spectral matching usingmulti-LED lighting systemsrdquo Optical Engineering vol 58no 3 article 035105 2019

[28] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient adaptive scheme for global optimization over con-tinuous spacesrdquo Technical report TR-95-012 InternationalComputer Science Institute Berkley CA USA 1995

[29] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient heuristic for global optimization over continuousspacerdquo Journal of Global Optimization vol 11 no 4pp 341ndash359 1997

[30] S Soltic and A Chalmers ldquoDifferential evolution for theoptimisation of multi-band white LED light sourcesrdquo LightingResearch amp Technology vol 44 no 2 pp 224ndash237 2012

[31] A N Chalmers and S Soltic ldquoLight source optimizationspectral design and simulation of four-band white-lightsourcesrdquo Optical Engineering vol 51 no 4 article 0440032012

[32] S Soltic and A N Chalmers ldquoDifferential evolution and itsapplication to intelligent spectral designrdquo in Proceedings of the16th Electronics New Zealand Conference (ENZCon) Uni-versity of Otago Dunedin New Zealand November 2009

[33] L J M Schlangen e Role of Lighting in Promoting Well-Being and Recovery Within Healthcare Philips White Pa-per Koninklijke Philips Electronics NV AmsterdamNetherlands 2010 httpswwwitdoescoukwpontentuploads201604Role-of-lighting-published-by-philips-as-a-whitepaper-2010pdf

8 Journal of Healthcare Engineering

support and the provision of facilities +is work wassupported by the School of Professional Engineering and theManukau Institute of Technology Research Fund

References

[1] SCHEER (Scientific Committee on Health Environmentaland Emerging Risks) Opinion on Potential Risks to HumanHealth of Light Emitting Diodes (LEDs) European Com-mission Luxembourg 2018 httpseceuropaeuhealthsiteshealthfilesscientific_committeesscheerdocsscheer_o_011pdf

[2] C Genova ldquoNormal responses to non-visual effects of lightretained by blind humans lacking rods and conesrdquo MedicalNews Today (Online) December 2007 httpswwwmedicalnewstodaycomreleases91836php

[3] W van Bommel and G van den Beld ldquoLighting for work areview of visual and biological effectsrdquo Lighting Research ampTechnology vol 36 no 4 pp 255ndash266 2004

[4] C S Pechacek M Andersen and S W Lockley ldquoPreliminarymethod for prospective analysis of the circadian efficacy of(day) light with applications to healthcare architecturerdquoLEUKOS vol 5 no 1 pp 1ndash26 2008

[5] M Andersen J Mardaljevic and S W Lockley ldquoA frame-work for predicting the non-visual effects of daylight Part 1photobiology-based modelrdquo Lighting Research amp Technologyvol 44 no 4 pp 37ndash53 2012

[6] C Meadows ldquoSetting expectations on human-centric light-ingrdquo LEDs Magazine (Online) December 2017 httpswwwledsmagazinecomarticles201712setting-expectations-on-human-centric-lightinghtml

[7] N A Midolo and L Sergeyeva ldquoLighting for clinical ob-servation of cyanosisrdquo e Australian Hospital Engineervol 30 no 2 pp 38ndash43 2007

[8] M Changizi and K Rio ldquoHarnessing color vision for visualoximetry in central cyanosisrdquo Medical Hypotheses vol 74no 1 pp 87ndash91 2010

[9] H L Snider ldquoCyanosisrdquo in Clinical Methods e HistoryPhysical and Laboratory Examinations H K WalkerW D Hall and J W Hurst Eds Butterworths Boston MAUSA 3rd edition 1990 httpwwwthornlightingcomdownloadHandbook5-6pdf

[10] ASNZS 1680252018 AustralianNew Zealand StandardInterior and Workplace Lighting Part 25 Hospital andMedical Tasks Section 73 Cyanosis Observation Lightingand Appendix G Method for Determination of CyanosisObservation Index of a Light Source

[11] Commission Internationale de lrsquoEclairage Colorimetry CIEPublication 153 Vienna Austria 3rd edition 2003

[12] Commission Internationale de lrsquoEclairage InternationalLighting Vocabulary CIE publication DIS 017E2016 ViennaAustria 2017

[13] Commission Internationale de lrsquoEclairage Method of Mea-suring and Specifying Colour Rendering Properties of LightSources CIE Publication 133 Vienna Austria 1995

[14] Illuminating Engineering Society of North America IESMethod for Evaluating Light Source Color Rendition Illumi-nating Engineering Society New York NY USA 2015

[15] +e Society of Light and Lighting Lighting Guide 2 Hospitalsand Health Care Buildings CIBSE London UK 2008

[16] BSI Standards Publication BS EN 12464-12011 Light andLightingmdashLighting of Work Places Part 1 IndoorWork PlacesBSI London UK 2011

[17] ANSIIES RP-29-16 Lighting for Hospitals and HealthcareFacilities Illuminating Engineering Society of North AmericaNew York NY USA 2016

[18] S Alzubaidi and P K Soori ldquoEnergy efficient lighting systemdesign for hospitals diagnostic and treatment roommdasha casestudyrdquo Journal of Light amp Visual Environment vol 36 no 1pp 23ndash31 2012

[19] F Leccese C Montagnani S Iaia M Rocca and G SalvadorildquoQuality of lighting in hospital environments a wide surveythrough in situ measurementsrdquo Journal of Light amp VisualEnvironment vol 40 pp 52ndash65 2016

[20] +orn Lighting Ltd orn Applications and TechniquesHealthcare +orn Lighting Spennymoor UK 2019

[21] S Mehrotra S Basukala and S Devarakonda ldquoEffectivelighting design standards impacting patient care a systemsapproachrdquo Journal of Biosciences and Medicines vol 3 no 11pp 54ndash61 2015

[22] P Bartczak A Gebejes P Falt and M Hauta-Kasari ldquoAnLED-based tunable illumination for diverse medical appli-cationsrdquo in Proceedings of the IEEE Symposium on Computer-Based Medical Systems pp 292-293 IEEE Dublin IrelandAugust 2016

[23] M Litorja and B Ecker ldquoUse of a spectrally-tunable source toexplore improvement in chromatic contrast for illuminationof tissuesrdquo in Proceedings of the SPIEmdashe InternationalSociety for Optical Engineering vol 7596 Bellingham WAUSA December 2010

[24] Lumileds Lighting ldquoLUXEONreg K2 emitterrdquo in TechnicalDatasheet DS51 Lumileds Lighting US LLC San Jose CAUSA 2006

[25] S B Chiang and K Chien ldquo+e optimization of color-mixedLED lightingrdquo in Proceedings of the 2014 IEEE IndustryApplication Society Annual Meeting Vancouver CanadaOctober 2014

[26] W Davis and Y Ohno ldquoColor quality scalerdquo Optical Engi-neering vol 49 no 3 article 033602 2010

[27] A Llenas and J Carreras ldquoArbitrary spectral matching usingmulti-LED lighting systemsrdquo Optical Engineering vol 58no 3 article 035105 2019

[28] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient adaptive scheme for global optimization over con-tinuous spacesrdquo Technical report TR-95-012 InternationalComputer Science Institute Berkley CA USA 1995

[29] R Storn and K Price ldquoDifferential evolutionmdasha simple andefficient heuristic for global optimization over continuousspacerdquo Journal of Global Optimization vol 11 no 4pp 341ndash359 1997

[30] S Soltic and A Chalmers ldquoDifferential evolution for theoptimisation of multi-band white LED light sourcesrdquo LightingResearch amp Technology vol 44 no 2 pp 224ndash237 2012

[31] A N Chalmers and S Soltic ldquoLight source optimizationspectral design and simulation of four-band white-lightsourcesrdquo Optical Engineering vol 51 no 4 article 0440032012

[32] S Soltic and A N Chalmers ldquoDifferential evolution and itsapplication to intelligent spectral designrdquo in Proceedings of the16th Electronics New Zealand Conference (ENZCon) Uni-versity of Otago Dunedin New Zealand November 2009

[33] L J M Schlangen e Role of Lighting in Promoting Well-Being and Recovery Within Healthcare Philips White Pa-per Koninklijke Philips Electronics NV AmsterdamNetherlands 2010 httpswwwitdoescoukwpontentuploads201604Role-of-lighting-published-by-philips-as-a-whitepaper-2010pdf

8 Journal of Healthcare Engineering

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom