IEEE SMART CITIES INITIATIVE - TRENTO WHITE PAPER 1 Smart

IEEE SMART CITIES INITIATIVE - TRENTO WHITE PAPER 1

Smart Health from Big Data to System Medicine:the Network Physiology Paradigm

Martina Valente, IEEE EMBS Student Member, and Giandomenico Nollo, IEEE EMBS Member

Abstract—The technological development of the last yearsand recent advances in scientific research have favoured theemergence of new models of healthcare in smart cities. Fosteredby the decentralization of clinical practice outside hospitals andthe massive amount of physiological data that are recorded daily,new approaches that aim to go beyond traditional medicine areemerging. At the forefront of numerate disciplines, biologicalresearch and clinics, Network Physiology is a new field in sciencethat proposes to take the distance from traditional segregationand look at human physiology from a holistic perspective.According to this approach, systems composing the humanbody are studied in their interactions, through newly developedtheoretical frameworks and cutting edge data analysis tools. Theaim is to introduce new biomarkers of physiological functioningthat will improve our knowledge about the human body, helpclinicians in diagnoses and enhance prevention of critical events.This white paper elaborates on the impact of Network Physiologyin future smart cities and presents how diversified fields of scienceare implementing this paradigm to help scientific and clinicalprogress.

I. INTRODUCTION

The application of the Smart City concept have becomea trend in today’s urban environments, favored by recent ad-vances in information and communication technologies (ICTs)and fostered by the increasing number of people living in cities[1]. In this context, the smartness challenge implies that thelatest technological improvements are used at the service ofthe citizens, with the purpose of improving their daily livingand their quality of life.

Of broad relevance, the rapid technological developmentof the last years have seen the use of wearable sensors tobecome widespread in our daily activities, coming either inthe form of self-contained devices or as an integrating partof mobile phones [2]. Wearables are usually small and lighthigh-tech devices, implementing a wide range of functions andoften representing pieces of design, such as bracelets, glassesor necklaces. Mostly, they are used for fitness purpose, soto monitor personal physical performance and tune exerciseto reach personalized goals. However, their potentialities aremuch wider and new configurations of wearable devices arestarting to enter very diversified applications [3]. Let us men-tion for instance movement data capturing devices that allowrehabilitation to be performed at home [4], elderly peoplemonitors that favor remote assistance [5], heart rate sensorsthat allow immediate emergency alerting for people sufferingfrom cardiac disease [6] and insulin pumps that deliver insulin

M. Valente is with the Dept. of Information Engineering and ComputerScience, University of Trento, Italy.

G. Nollo is with the IRCS-FBK and BIOtech, Dept. of Industrial Engineer-ing, University of Trento, Italy.

by continuous infusion mimicking physiological demands [7].The massive use of wearable monitors results in the dailycollection of a huge amount of data, so that it is possibleto talk about a new type of big data: physiological big data.

At the same time, a trend is developing within the clinicalsetting, that aims to shift from a traditional reductionistapproach to a modern system approach, known as systemmedicine [8]. The idea behind system medicine is to considerthe human organism as part of an integrated whole, whereinteractions among different constituents are studied in theattempt to provide a wider perspective on the holistic natureof physiological conditions of health or disease. Contrarilyto modern medicine, that focuses on single organ systems asseparated entities, system medicine aims to consider dynamicinteractions between different elements composing the humanbody in the attempt to understand how they affect the systemas a whole. Indeed, according to this perspective, existingnetworks manifest emergent properties that define the wholethat are not simply the sum of features of the component parts.

Physiological big data constituted by recordings from di-versified physiological districts would represent an invaluablesource of information for system medicine to improve ourknowledge about the functioning of the human body, however,there is the need to develop advanced tools that are able todeal with a wide variety of data types, happening at differenttime scales, showing non linear interactions and representingvery different mechanisms. This is the main goal of NetworkPhysiology, a new interdisciplinary field at the forefrontbetween scientific research and clinical practice, whose aim isto develop a theoretical framework and practical analytic toolsfor the treatment of big data acquired from the human body,shifting the focus from single organ systems to networks oforgan interactions, in the attempt to better understand humanphysiology in an integrated way and predict clinical eventsand diseases [9].

II. NETWORK PHYSIOLOGY: ORIGIN AND FOCUS

Network Physiology was introduced as a new field inscience by Professor Plamen Ivanov and its research group atBoston University as a result of an innovative study regardingthe communication between organ systems across sleep stages[10]. They analyzed full-night sleep recordings from 36 sub-jects consisting of the measurements of the electrical activityof the heart and the brain, the ventilation of the lungs and themovements of eyes, legs and chin (Figure 1). They looked atcorrelations between the activity coming from different bodyparts, and what they found was quite revolutionary. Whileduring deep sleep the physiological systems appeared to be

2 IEEE SMART CITIES INITIATIVE - TRENTO WHITE PAPER

Fig. 1. Schematic representation of the acquisition process in a NetworkPhysiology perspective, involving simultaneous recording of biomedical sig-nals from different body parts (here: brain, heart, vasculature, lungs, muscles,eye) and follwing extraction of time series representing in a compact way theactivity of the monitored physiological districts over time.

somehow disconnected, stronger coupling appeared as soonas the subjects transitioned to REM sleep, and even strongergoing to light sleep and finally to awake, in contrast to thetraditional indices computed on the dynamics of individualsystems, that show high similarity between wake and REMsleep, compared to non-REM sleep. Even more surprisingly,the network reorganized very fast (within seconds) duringthe transitions from one sleep stage to another, showingan amazing flexibility of the mechanism involved in theoptimization and coordination of organs behavior. Whereasit was known that sleep regulation influenced the controlmechanisms of individual physiological systems, this studyfurther revealed that sleep stages are able to modify topologyand coupling strength of the physiological network composedby all the observed systems, in a way that cannot be directlyinferred by individual dynamics, thus setting the ground forthe development of new and revolutionary approaches to lookat the human body from an integrated perspective.

Network Physiology is presented as an interdisciplinaryfield, that combines efforts of scientists in statistical physics,applied mathematics, computer and data science, in the attemptto unravel the complex mechanisms that regulate the com-munications between organ systems. In other works, NetworkPhysiology aims to study the horizontal integration that existsbetween systems that are commonly treated vertically (Figure2). Whereas it is common practice to study organs from themolecular composition all the way up to the anatomical struc-ture (vertical perspective), the relationships that exists betweendifferent organs (horizontal perspective) are almost never takeninto account by modern medicine: the cardiologist studies theheart, the pneumologist studies the lungs, and so on. However,it is argued that a lot of information about the functioningof our body may be lost if interactions between organs arenot considered. Paradigmatically, one may think at multipleorgans failure, a clinical condition in which critically ill peopledie, without any organ being directly responsible. The lackof a single point of failure suggests that the cause of thesystem breakdown may be a dysfunction of the mechanismsthat organs use to synchronize themselves. According to the

Fig. 2. Graphics illustrating the concepts of vertical and horizontal integrationin the study of the systems composing the human body. Vertical integrationinvolves the study of an organ system from its subcellular componentsup to the overall functions. Horizontal integration takes into account therelationships that exist between different organ systems.

Network Physiology perspective, if the focus is shifted to thestudy of the interactions among organs, it might be possiblein the future to predict critical events such as multiple organsfailure, and hopefully prevent them from happening.

III. FUTURE PERSPECTIVES

Drawing inspiration by existing atlases of human anatomy,mapping bones, muscles, nerves and vasculature, the finalgoal of Network Physiology would be to map the informationbeing circulated in the human body and build an atlas of thedynamical interactions between organ systems under healthyconditions, pathological and physiological states, differentiatedby age or population group. This atlas will be constituted bya catalogue of reference maps of dynamical organs interac-tions extracted from continuous parallel recordings of organsystems. The availability of such an atlas would dramaticallychange the way in which physiological regulatory mechanismsare understood and revolutionize and innovate many aspectsof health provision. First, new biomarkers related to organnetwork interactions could be defined in order to provideclinicians with a wider picture of the physiological statusof patients. Then, the atlas may be a helpful resource fornew types of physicians, having expertise in the field ofsystem medicine. Moreover, a comprehensive assessment ofthe effects of drugs of pharmaceuticals on the whole humanorganism might be also favored by knowing how organsreciprocally affect each other. Finally, algorithms extractingmaps of physiological interactions or indicators of systemfunctioning may be implemented in next generation ICUs and,in a later future, in personalized health monitors, allowing acomprehensive evaluation of the patient status based on severalinnovative physiological indices.

IV. CURRENT DEVELOPMENTS IN RESEARCH ANDCLINICAL PRACTICE

If the future perspectives that Network Physiology offers arevery appealing, it is also true that at the present time efforts


Fig. 3. Paradigmatic example of the spectrum of the heart rate variabilitysignal recorded from a subject at rest. Two peaks are clearly identifiable inthe spectrum: the low frequency peak (LF) can be related to the autonomicdrive of the heart rate (that has an effect on slow timescales), whereas thehigh frequency peak (HF) reflects respiratory influences (the frequency of arespiratory cycle is usually in the HF band).

in various fields are being made in order to take conceptsthat belong to fields such as complex systems theory, non-linear systems and complex network analysis to mention some,and apply them to the study of the behaviors that emergefrom networks of physiological systems. Prominent examplesrepresenting current developments in the field of NetworkPhysiology will be presented in the following.

A. Cardiorespiratory interactions

Following a research line that started in the 80s [11], [12],the study of the dynamical interactions between heart rate andrespiration is a prominent example of how the behavior ofa systems may be strongly influenced by the dynamics ofanother. This type of coupling can be probed by everyone ofus, simply measuring the pulse during one whole respirationcycle (inspiration and expiration). This simple experiment willmake it possible to verify that during inspiration the heart rateaccelerates a bit, while during expiration it slows down again.This phenomenon is known as respiratory sinus arrhythmia(RSA), and explains the presence of oscillatory componentsat the respiratory frequencies that are found in the spectrumof the heart rate variability signal (Figure 3).

However, the physiological mechanisms that drive RSA arevery complex and not fully understood yet. They comprisethe action of the autonomic nervous system as a mediator,possible mechanical effects of the rib cage and may also beaffected the activity of higher brain centers. Most importantly,the interaction between respiration and heart rate varies acrossphysiological states and may be impaired as a consequence ofpathological conditions [13]. A typical example is the study ofthe response to an orthostatic stress, that results in a strongreduction of RSA with a concomitant simplification of theheart rate dynamics as a result of autonomic effects. This

Fig. 4. Example of respiratory traces acquired from a healthy subject (toppanel) and a sleep apnoea patient (bottom panel). In the bottom panel, it ispossible to observe how the irregularities in the nasal (NAF), thoracic (Thorax)and abdominal (Abd) air flow that characterize sleep apnoea affect the oxigensaturation, that oscillates between 80 and 100 percent during apnoea episodes,while it is stable in healthy conditions.

healthy behavior is not observed in patients suffering fromorthostatic syncope [14].

Going beyond the interaction between heart and lungs,recent approaches make use of extended set of signals recordedfrom different districts (thorax, fingers, limbs, head) in orderto shed light on the neuroautonomic regulation of cardiovas-cular and cardiorespiratory activity. Different body parts aremodelled as interdependent on one another as nodes of anunderlying network and analyzed in an integrated fashion [15].

B. Sleep regulation

Besides being elucidatory on the role of sleep stages inchanging the topology of the physiological network, the studyof physiological systems’ interactions during sleep also provedto be useful for the evaluation of the consequences of sleepdisorders. As an example, we report the results of a studyon sleep apnoea, a sleep disorder characterized by repetitivecessations of respiratory flow for at least 10 seconds [16](Figure 4). The study shows that the cardiorespiratory couplingthat is commonly observed in deep sleep is lost with apnoea.This is particularly critical because a function of deep sleepis physical recreation in a low-energy consumption setting,that is favored by a high degree of synchronization betweenrespiration and heart beat. If this synchronization is lost withapnoeas, it is possible that physical recreation is impaired, thusincreasing the possibility to develop very serious consequencesin the mid-long term [17].

C. Brain connectivity

Founding concepts that set the basis of Network Physiologysee their applicability for the study of the most complexorgan of our body, that is the human brain. Understanding the


Fig. 5. Examples of signals recorded from the brain. In the top panel, anEEG trace in shown, representing the electrical activity of the brain recorded atdifferent locations on the scalp. In the bottom panel, activation maps extractedfrom fMRI recordings are shown, where colors are indicative of the level ofpositive or negative activation/correlation of the underlying brain area, whenthe subject is undergoing a language task or in resting state [18].

mechanisms that govern information processing in the brainis one of the biggest challenges of our century, favored bythe development and improvement of investigation methodsthat allow to study this organ with an increasing level ofdetail. Among the most commonly used techniques it is worthmentioning EEG, fMRI, MEG, NIRS, fNIRS, PET (Figure5). Signals recorded using one or a combination of thesetechniques allow to build what is known as a functionalconnectivity map, that is a special type of graph, where thenodes represent brain areas, while the links represent thedynamical interactions between them [19].

Although functional networks partly reflect the properties ofthe underlying structural network that composes the brain, it isimportant to point out that nodes in a functional network canbe linked even though there is no physical structure connectingthem, thus reflecting a common functionality. Unlike structuralconnections, functional links are highly dynamics and subjectto changes that depend on the level of attention or arousal, onthe emotional state, on the task that is performed and also onpathological conditions.

The study of the functional connectivity maps allows toshed light on neurodegenerative diseases: it was discoveredthat regions of high connectivity in the human brain, calledhubs, are more likely to be affected by Alzheimer’s disease,probably because of their continuous high baseline activity[20]; moreover, recent evidence regarding the study of epilepsysuggests the existence of an epileptic network in contrast to thetraditional epileptic focus hypothesis, arguing that seizures arenot relegated to a circumscribed area of the brain, but rather

Fig. 6. Thermographic image showing how the heat is distributed in thesubject’s hands.

large functionally and anatomically connected brain structuresmay be involved [21].

D. Cardiovascular networksCardiovascular network health is of paramount importance

for the life and function of all systems composing the humanbody. The functioning of the of cardiovascular network can beprobed by measuring blood perfusion, that is also an indicatorof hormonal, circulatory and tissue health in humans. How-ever, providing a reliable measurement of this quantity canbe challenging. Different non-invasive monitoring systems areavailable, relying on mechanical (plethysmography), optical(photoplethysmography, laser Doppler flowmetry), acoustic(ultrasound) or thermal (infrared thermography) physicalprinciples. Whereas optical and acoustic techniques are par-ticularly effective in terms of spatio-temporal mapping, theyshow an extreme sensitivity to motion, requiring immobiliza-tion of the subjects during the measurements. On the otherhand, thermal techniques, that measure blood flow from thethermal response of the skin, are less sensitive to motion, thusproviding a valid alternative in scenarios that require move-ment. A further advantage is that these techniques allow tomap the functional architecture of cutaneous and subcutaneousvessels simultaneously (Figure 6).

Thermography has been applied to a range of differentstudies as a novel means to provide additional information withrespect to more common practice indicators. In [22], thermog-raphy has been successfully used during brain surgery on theexposed cerebral cortex, in order to map the distribution of cor-tical activation during language and motor tasks, going aboveand beyond the information provided by standard functionalimaging techniques. In [23], infrared imaging is employedduring renal transplant in order to assess noninvasively kidneyreperfusion, as a novel objective means to evaluate the effectsof ischemia in real time. This is a particularly critical indicatorbecause ischemia results in an increased risk of rejection aftertransplant as a consequence of ischemic damage. In [24],thermography is used to evaluate vascular dysfunction as anindicator of sickle cell disease.


Fig. 7. Examples of amorphous (a) and star-like (b) network topologies.

In addition, thermography also provides blood perfusionindicators related to a variety of mechanisms happening atdifferent frequencies, including heart activity, respiration andsympathetic activity.

E. Genomics, metabolomics and transcriptomics

While structural networks characterize physical connectionsbetween units (vascular tree, brain connectome, airway tree)and functional networks reveal the temporal correlations be-tween the activity of the nodes under study (cardiovascularand cardiorespiratory interactions, brain function), there existclasses of biological data that lack both physical and timedependencies, for which it does not make sense to computestructural or functional connectivity. One may think for exam-ple at blood test results, genes or metabolites. For this type ofdata, adopting a network approach might be helpful in order tounveil the presence of reference relationships between nodes.Recently, a new type of network for the representation ofisolated and heterogeneous values was defined, called paren-clitic network, in which nodes represent features extractedfrom data, whereas links quantify the deviation (parenclisis)of the two connected features from a reference model [25]. Thetopology of the resulting network is then informative about thesystem under study.

A prominent result was obtained for the early diagnosisof obstructive neuropathy from metabolites and regulators[26]. Obstructive neuropathy is a complex disease affectingchildren, that is characterized by a partial or total obstructionof urinary tract that may result in kidney damage. Despitethe gravity of this disease, the pathological mechanisms arenot fully understood yet. Parenclitic networks allow a fur-ther insight into the genetic and metabolic elements thatare responsible for the malfunction: whereas the topology ofthe parenclitic network of control subjects was amorphous,patients suffering from obstructive neuropathy showed star-like topologies centred on the most abnormal metabolite orregulator (Figure 7).

In another study, a different approach based on linearalgebra techniques for dimensionality reduction was appliedto transcriptomic data from adipocytes (fat cells) in orderto study the alterations in regulatory and metabolic activi-ties in obesity [27]. In order to provide a more robust andsignificant result, different transcriptomic datasets in literaturewere combined, so to remove the bias deriving from single ex-

periments. The dimensionality reduction technique applied ontranscriptomic big data allowed to find a signature of 38 genesthat was able to discriminate between obese and lean subjects,even when tested on independent databases. Interestingly, thegenes in the signature were involved in typical dysfunctionsthat are associated with obesity, such as those related to thecentral nervous system (anxiety, depression), digestive system(salivary secretion, digestive problems), diabetes or fertility. Inaddition, a similar transcriptomic signature was found whencomparing data from healthy breast tissue with data frombreast cancer tissue, validating the strong association betweenobesity and breast cancer.

V. CONCLUSION

This white paper illustrated the role of the emerging fieldof Network Physiology in fostering and shaping innovativemodels of healthcare in smart cities. Relying on the availabilityof big data coming either from hospital recordings or fromphysiological measurements performed by wearable devices,and following the lines of system medicine in treating thehuman organism as a connected whole, Network Physiologyis proving the effectiveness of its distinctive approach in avariety of different clinical settings, ranging from cardiovascu-lar impairments to neurological diseases, from sleep disordersto genetic diseases. Current efforts in developing theoreticalframeworks and practical analytical tools will give the possi-bility to revolutionize the way in which the functioning ofthe human body is understood, fostering a more informedtreatment of diseases, a more effective prediction of criticalepisodes and new ways of prevention.

GLOSSARY

Network Physiology A new interdisciplinary field of science that is emergingat the forefront of statistical physics, biomedical engineering,applied physiology and clinical medicine, that aims to developa theoretical framework and system-wide analysis tools to studyhow different organ systems, each characterized by its ownregulatory mechanism and physiological functioning, dynamicallyinteract with each other and change their interconnected behaviorsin conditions of health and disease.

hubs A hub is a concept belonging to network theory that indicates a nodecharacterized by high connectivity, i.e. with a number of linksthat is much larger than the average.

infrared thermography Infrared thermography is an imaging technique thatdetects the infrared radiation emitted from objects, converts it totemperature and outputs an image of the heat distribution.

metabolites Metabolites are the intermediate products of metabolic reactionsthat occur within cells. Primary metabolites are indispensablefor the cells growth and include amino acids, vitamins, alcohols,organic acids, nucleotides. Secondary metabolites are not requiredfor primary metabolic processes, but are involved in other func-tions such as protection, competition, species interaction.

orthostatic stress A type of physiological stress that is caused by assumingthe standing position. It is characterized by a lowering of thearterial blood pressure that in turn triggers the autonomic ner-vous system response. Orthostatic stress is induced in laboratoryexperiments via the head-up tilt (HUT) test.

parenclitic network A parenclitic network is a novel network representationof a set of items that unveils the presence of relationships betweenfeatures of these items. Nodes in parenclitic networks representfeatures, while links encode the deviation from a reference model.

physiological big data Big data consisting of recordings of human bodyfunction from a variety of districts, acquired either in hospitalsor through the use of modern portable systems that incorporate


sensors into wearable devices. Physiological big data is charac-terized by a high degree of heterogeneity in terms of physicalquantities, timescales and activity patterns.

sleep stages The different stages of sleep that cyclically repeat during sleep.Sleep stages are identified by the corresponding predominantfrequency of brain wave activity that occurs during each stage.Five different sleep stages are commonly identified: light sleep 1and 2, deep sleep 1 and 2 and rapid eye movement (REM).

system medicine A newly-proposed holistic approach to medical educationand practice, that aims to take the distance from the reductionistframework of modern medicine, by studying interactions amongdifferent body parts to provide a wider perspective on physiolog-ical functioning in conditions of health or disease.

transcriptomic data The transcriptome is the set of all RNA molecules inone cell or in a population of cells, reflecting all the genes thatare being actively expressed at any given time.

ACKNOWLEDGMENT

This project is supported by the IEEE Smart Cities Trento,Student Grant Program (Second Call, May 2016).

REFERENCES

[1] United Nations, Department of Economic and Social Affairs, PopulationDivision, World Urbanization Prospects: The 2014 Revision, 2015.

[2] S. C. Mukhopadhyay, “Wearable sensors for human activity monitoring:A review,” IEEE sensors journal, vol. 15, no. 3, pp. 1321–1330, 2015.

[3] M. Valente and G. Nollo, “Wireless body area networks: a new paradigmof personal smart health,” IEEE Smart Cities Initiative - Trento whitepaper.

[4] M. Sung, C. Marci, and A. Pentland, “Wearable feedback systems forrehabilitation,” Journal of NeuroEngineering and Rehabilitation, vol. 2,no. 1, p. 17, 2005.

[5] T. Pisoni, N. Conci, F. De Natale, M. De Cecco, G. Nollo, A. Frattari,and G. Guandalini, “Ausilia: Assisted unit for simulating independentliving activities,” 2016.

[6] P. Rubel, J. Fayn, G. Nollo, D. Assanelli, B. Li, L. Restier, S. Adami,S. Arod, H. Atoui, M. Ohlsson, L. Simon-Chautemps, D. Tlisson,C. Malossi, G.-L. Ziliani, A. Galassi, L. Edenbrandt, and P. Chevalier,“Toward personal ehealth in cardiology. results from the epi-medicstelemedicine project.,” Journal of electrocardiology, vol. 38, pp. 100–106, Oct 2005.

[7] B. H. McAdams and A. A. Rizvi, “An overview of insulin pumps andglucose sensors for the generalist,” Journal of clinical medicine, vol. 5,no. 1, p. 5, 2016.

[8] H. J. Federoff and L. O. Gostin, “Evolving from reductionism to holism:is there a future for systems medicine?,” Jama, vol. 302, no. 9, pp. 994–996, 2009.

[9] P. C. Ivanov, K. K. Liu, A. Lin, and R. P. Bartsch, “Network physiology:From neural plasticity to organ network interactions,” in EmergentComplexity from Nonlinearity, in Physics, Engineering and the LifeSciences, pp. 145–165, Springer, 2017.

[10] A. Bashan, R. Bartsch, J. Kantelhardt, S. Havlin, and P. Ivanov,“Network physiology reveals relations between network topology andphysiological function,” Nature Communications, vol. 3, 2012.

[11] S. Akselrod, D. Gordon, F. A. Ubel, D. C. Shannon, A. C. Barger,and R. J. Cohen, “Power spectrum analysis of heart rate fluctuation:a quantitative probe of beat-to-beat cardiovascular control,” science,pp. 220–222, 1981.

[12] M. Pagani, F. Lombardi, S. Guzzetti, O. Rimoldi, R. Furlan, P. Pizzinelli,G. Sandrone, G. Malfatto, S. Dell’Orto, and E. Piccaluga, “Powerspectral analysis of heart rate and arterial pressure variabilities as amarker of sympatho-vagal interaction in man and conscious dog.,”Circulation research, vol. 59, no. 2, pp. 178–193, 1986.

[13] A. Porta, M. Di Rienzo, N. Wessel, and J. Kurths, “Addressing thecomplexity of cardiovascular regulation,” 2009.

[14] L. Faes, L. Widesott, M. Del Greco, R. Antolini, and G. Nollo, “Causalcross-spectral analysis of heart rate and blood pressure variability fordescribing the impairment of the cardiovascular control in neurally me-diated syncope,” IEEE transactions on biomedical engineering, vol. 53,no. 1, pp. 65–73, 2006.

[15] L. Faes, A. Porta, G. Nollo, and M. Javorka, “Information decompositionin multivariate systems: definitions, implementation and application tocardiovascular networks,” Entropy, vol. 19, no. 1, p. 5, 2016.

[16] T. Penzel, N. Wessel, M. Riedl, J. W. Kantelhardt, S. Rostig, M. Glos,A. Suhrbier, H. Malberg, and I. Fietze, “Cardiovascular and respiratorydynamics during normal and pathological sleep,” Chaos: An Interdisci-plinary Journal of Nonlinear Science, vol. 17, no. 1, p. 015116, 2007.

[17] J. M. Marin, S. J. Carrizo, E. Vicente, and A. G. Agusti, “Long-term cardiovascular outcomes in men with obstructive sleep apnoea-hypopnoea with or without treatment with continuous positive airwaypressure: an observational study,” The Lancet, vol. 365, no. 9464,pp. 1046–1053, 2005.

[18] B. Goodyear, E. Liebenthal, and V. Mosher, “Active and passive fmri forpresurgical mapping of motor and language cortex,” in Advanced BrainNeuroimaging Topics in Health and Disease - Methods and Applications(T. D. Papageorgiou, G. I. Christopoulos, and S. M. Smirnakis, eds.),ch. 18, Rijeka: InTech, 2014.

[19] E. Bullmore and O. Sporns, “Complex brain networks: graph theoreticalanalysis of structural and functional systems,” Nature reviews. Neuro-science, vol. 10, no. 3, p. 186, 2009.

[20] R. L. Buckner, J. Sepulcre, T. Talukdar, F. M. Krienen, H. Liu,T. Hedden, J. R. Andrews-Hanna, R. A. Sperling, and K. A. Johnson,“Cortical hubs revealed by intrinsic functional connectivity: mapping,assessment of stability, and relation to alzheimer’s disease,” Journal ofNeuroscience, vol. 29, no. 6, pp. 1860–1873, 2009.

[21] K. Lehnertz, G. Ansmann, S. Bialonski, H. Dickten, C. Geier, andS. Porz, “Evolving networks in the human epileptic brain,” Physica D:Nonlinear Phenomena, vol. 267, pp. 7–15, 2014.

[22] A. M. Gorbach, J. Heiss, C. Kufta, S. Sato, P. Fedio, W. A. Kammerer,J. Solomon, and E. H. Oldfield, “Intraoperative infrared functionalimaging of human brain,” Annals of neurology, vol. 54, no. 3, pp. 297–309, 2003.

[23] A. Gorbach, D. Simonton, D. A. Hale, S. J. Swanson, and A. D. Kirk,“Objective, real-time, intraoperative assessment of renal perfusion usinginfrared imaging,” American Journal of Transplantation, vol. 3, no. 8,pp. 988–993, 2003.

[24] A. M. Gorbach, H. C. Ackerman, W.-M. Liu, J. M. Meyer, P. L. Littel,C. Seamon, E. Footman, A. Chi, S. Zorca, M. L. Krajewski, et al.,“Infrared imaging of nitric oxide-mediated blood flow in human sicklecell disease,” Microvascular research, vol. 84, no. 3, pp. 262–269, 2012.

[25] M. Zanin, J. M. Alcazar, J. V. Carbajosa, D. Papo, M. G. Paez,P. Sousa, E. Menasalvas, and S. Boccaletti, “Parenclitic networks: amultilayer description of heterogeneous and static data-sets,” arXivpreprint arXiv:1304.1896, 2013.

[26] M. Zanin and S. Boccaletti, “Complex networks analysis of obstructivenephropathy data,” Chaos: An Interdisciplinary Journal of NonlinearScience, vol. 21, no. 3, p. 033103, 2011.

[27] F. Font-Clos, S. Zapperi, and C. A. La Porta, “Integrative analysis ofpathway deregulation in obesity,” NPJ Systems Biology and Applica-tions, vol. 3, 2017.

Documents

IEEE SMART CITIES INITIATIVE - TRENTO WHITE PAPER 1 Smart