Decreasing Sedentary Behavior in Overweight Youth:
A Real-Time Mobile Intervention
Donna Spruijt-Metz, Gillian O'Reilly, Shrikanth Narayanan, Murali Annavaram, Ming Li,
Sangwon Lee, Cheng Kun [email protected]
www.metzlab.net
BMI ≥ 85th Percentile (age 4-19)
Overweight has doubled in 30 years, and
Obesity has TRIPLED
Data from the National Longitudinal Survey of Youth 1986-1998, NHANES 1998-2010
Childhood Obesity: Psychosocial Consequences
• Proximal consequences: – Negative stereotyping– Teasing– Fewer friends– Poor body image
• Distal consequences:– Lower educational
attainment– Discrimination
(apartment rentals, college admissions)
– Higher poverty
Childhood Obesity: Metabolic Consequences
Visceral Fat
Insulin Sensitivity
Leptin Resistance
Inflammation
DiabetesCVD
Some Cancers
Physical activity
• Decreases adiposity (Lazaar, 2007; Baxter-Jones, 2008)
• Improves insulin sensitivity (Imperatore, 2006; Carrel, 2009)
• Protects against breast, colon and other cancers (Bernstein, 2004; Percik, 2009)
• Improves lipid and cholesterol profiles (Tolfrey, 2000; Pam, 2008)
• Alleviates stress and depression (Ortega, 2008; Dockray, 2009)
• May improve cognitive function (van Praag, 2009; Li, 2008) and academic performance (Trudeau, 2008; Datar, 2008)
Sedentary Behavior
• Increases risk for metabolic syndrome (Edwardson, Gorely, Davies, et al, 2012)
• Increases risk for cardiovascular disease (Saunders, 2011)
• Increases risk for obesity (Saunders, 2011)
• Increases risk for hyperglycemia & type II diabetes (Dunstan, Salmon, Owen, et al, 2007)
• Related to premature mortality (Healy, Matthews, Dunstan, Winkler, Owen, 2003)
• Increases likelihood of participating in risky behaviors in adolescents (Nelson, Gordon-Larsen, 2006)
Decline in Moderate to Vigorous Physical Activity (MVPA) by Age
(Males + Females, n = 4867)Troiano, Berrigan, et al. MSSE 20087
Min
utes
/day
30 minutes a day is only 50% of recommended MVPA
mHealth: Measure, Understand,
Intervene• Bypass bias – measure
ubiquitously and well• Real or near-time data• Understand behavior in
time & in context• Share with participants,
clinicians […]• Real-time, personalized,
tailored & adaptive interventions
• Communication exchangeAnnavaram, et al 2008, Thatte et al 2009, 2009b,Lee et al 2009, Emken et al 2012
KNOWME Networks
• A suite of mobile, Bluetooth-enabled, wireless, wearable sensors
• That interface with a mobile phone and secure server
• To process data in real time, • Designed specifically for use in
overweight minority youth
KNOWME System
Database Server
Web Server
SQL with Encrypted Data
End-to-end encryption of sensitive data
Check right to use systemFilter noisy updates
Web enable data access
3GGSMWi-Fi
Doctor
Data with Secure connection
Decryption KeysNokia N95
2 Alive Heart Rate Monitor/Accelerometers
(ECG/ACC)
KNOWME NETWORKS
In-lab Development: Behavioral Pattern
Recognition
In-Lab Physical Activity Detection
Developing and testing algorithms to accurately classify types of PA in 20 overweight Hispanic youth 10 F/10M; 14.6 ± 1.8 years old; BMI %tile 96 ± 4
Protocol: 9 activities, 7 minutes/activity
3 sessions develop algorithm/1 session test algorithm
Lie Down
Sit Still
Sit &
Fidget
Stand Still
Stand &
Fidget
Stand &
Play Wii
Slow Walk
Fast Walk
Run
FREE LIVING
Li et al 2010
Predicted by the Model (84-94% accuracy)
Actu
al
% Normalized
Across Each Row
Accuracy of State Detection
• Overall accuracy: – 84% with all 9 activities– 94% when collapsed to 7 activities
• IMPORTANT FACT: If you are going to ‘share’ data with your participants it needs to be ACCURATE (How accurate??? And what does this mean in the eyes of the participant?)
KNOWME: Great idea!
BUT WILL THEY WEAR IT????
ECG/ACC
ACC
Structure of Data Collecting Software
End-to-end Encryption of Sensitive Data
Device Manager
Local Storage[User Configuration]
[Analyzed Data][Raw Data]
Transmitter[Encrypt/Decrypt]
Analyzer[Plug-in
modules]
GPS ACC ECG
Data Collector
Service Manager
Client Application with GUI
Local Socket or IPC
KNOWME NETWORKS
Free-Living Out-of-Lab Feasibility Study
Will/Can youth use this system
in the real? Subjects: 12 overweight Hispanic youth
5F/7M; 14.8 ± 1.9 years old; BMI %tile 97 ± 3 Protocol:
Results/day Wore KNOWME for 11.4 ± 2.0 hours Phone battery life 9.2 ± 2.6 hours 8 SMS sent to us / 9 SMS received from us
In-home training
Wear KNOWME for 2 days
Remote monitoring
Trouble-shooting via text
Exit Interview
SunSat
SMS
KNOWME Knows you…
KNOWME NETWORKS
Real-Time Intervention to decrease Sedentary Time
Your Activity Meter
Sedentary = lying down, sitting, sitting & fidgeting, standing, standing & fidgeting Active = standing playing Wii, slow walking, brisk walking, running
Battery Indicator for Each Device
Sedentary Time (since the last reset)
Active Time in the Last 60 Minutes
Total Active Time
Each bar = 30 seconds20 bars = 10 minutes
Total Elapsed Time
The intervention: 1) When the gauge reaches 120 minutes of sedentary time, the phone will automatically begin delivering “Move!!” messages.
2) The sedentary gauge will automatically reset to 0 minutes:
1) Following 10 minutes of active time within a 60 minute period; or
2) One hour after 140 minutes of sedentary time is reached if participant doesn’t respond (time out)
3) Researchers are notified at each reset.
3) Personalized text messages can be sent from the website monitoring team
Sedentary Intervention
Design & Participant Demographics
Baseline ~3-day baseline Actigraph Accelerometer wear
Intervention ~3-day KNOWME+ Accelerometer wear
Sample Size 10
Mean Age 16.3 ±1.7 years
Mean BMI Percentile 97.2 ± 4.4
Sex 50% Female
Ethnicity 100% Hispanic
Baseline accelerometer versus KNOWME Wear Time
There was no significant difference in the duration of baseline accelerometer and KNOWME system wear
Mean Minutes (±SD)
Mean Hours (±SD)
Accelerometer 2035.5 (214.7) 33.9 (3.6)
KNOWME System 1417.6 (332.8) 23.6 (5.6)
Intervention Activity Measurement Accelerometer vs. KNOWME System
Mean Minutes by Actigraph*
(±SD)
Mean Minutes by KNOWME
(±SD)t-value p-value*
Sedentary 1594.7 (208.3) 987.2 (272.2) 9.27 <0.0002
Light 413.4 (163.4) 309.4 (102.2) 2.54 0.04
MVPA 1.8 (3.2) 129.6 (53.9) -7.72 <0.0002
*Evansen (J Sports Sci. 2008) cutpoints used to reduce accelerometry data
Baseline vs. Intervention Activity Levels Measured by Accelerometer
Mean Minutes Baseline (±SD)
Mean Minutes Intervention
(±SD)t-value p-value*
Sedentary 1765.5 (357.7) 1594.7 (208.3) 1.28 0.1Light 436.6 (222.3) 413.4 (163.4) 0.75 0.2MVPA 0.3 (0.6) 1.8 (3.2) -1.45 0.09
*1-tailed, significance level set at 0.1
SMS Messaging During Intervention
Mean # SMS Messages (±SD)
Sent by Participants 33 (15)
Sent by Research Staff 43 (16)
KNOWME:Conclusions and Next Steps
• Research & Technology: Tortoise and Hare?• Hopeful pilot results for a tough chore• “Its like having a Doctor in your pocket!”• Did SMS prompts lead to physical activity
behavior change responses ? How long did it it take? How long did it last?
• Next steps: Redevelop system with new hardware, new software, geared for long-term wear
• Full clinical trial
• Participants• Research Team
• Funders Qualcomm NIMHD P60
002564
THANKS TO
Eat Well and Be Active!
Thank you! [email protected]