6. DATA SCIENCE FOR WEARABLES:
PERCEPTION & PERSONALIZATION
• Hardware is maturing
• Sensor data is growing exponentially
• Unlocking potential requires deriving meaning from data
7.
8. BEAUTY + ENGINEERING IN SERVICE OF A BETTER LIFE
Measure steps, sleep states, workouts, heart rate
21. • How can I understand this data?
• How should I feel about what it tells me?
• What action should I take in response?
22. Smart Coach Remembers
Remember how you took 45,365 steps on July 4?
Smart Coach remembers! On your health journey,
don't forget to stop and celebrate.
Step Update
Smart Coach noticed a surge in activity. In fact,
you surpassed 9,690 steps, your typical
5:00pm average.
Last night you had 35m of REM sleep, less than
the 1h9m that is typical for your age group. One
way to improve your chances for more REM is to
try an earlier bedtime than last night's 12:35am.
You can set a bedtime Reminder for 11:35pm to
help.
REM TimeLong Journey?
Looks like you've been traveling recently, which
can throw off your routine. Try setting a bedtime
reminder for tonight to help you adjust.
Your daily average of 17,543 steps places you
in the top 3% of UP females in their 30s.
Bravo, Angela.
Welcome to the 3%
29. This morning’s resting heart rate was higher
than 61bpm, your 30-day average.
Dehydration may be the cause. If you think
you were dehydrated last night, make up for
it today with 8 glasses of water.
Start with Hydration
33. Commitment and
Consistency
Foot In The Door
Technique
Source: Freedman, J.L. & Fraser, S.C. (1966). Compliance
without pressure: The foot-in-the-door technique. Journal of
Personality and Social Psychology, 4, 195-202.
BEHAVIOR CHANGE
34. Commitment and
Consistency
Foot In The Door
Technique
Goldilocks Tasks
Source: Pink, Daniel (2009). Drive: The Surprising Truth
About What Motivates Us. New York, NY: Riverhead Books.
BEHAVIOR CHANGE
35. Commitment and
Consistency
Foot In The Door
Technique
Goldilocks Tasks
Source: Carpenter, Chris. (2013) A meta-analysis of the
effectiveness of the "but you are free" compliance-gaining
technique
Reactance
BEHAVIOR CHANGE
36. 72%Increased likelihood to go to
bed early enough to hit their
sleep goal
23mMinutes earlier to bed,
compared to if they didn’t
receive a TIW
BEHAVIOR CHANGE
43. “The fact that the tracker
measured my sleep and my
activity level was a big part of
my recovery. I had this way to
‘metric’ my body as I went
through this. Sleep is so
important in brain function
anyway, and when you're
recovering from a brain injury,
it's even more important.”
44.
45.
46.
47. PARTING THOUGHTS…
• Wearables will help us live healthier
• Health data at unprecedented scale and granularity
• Data Science can play a critical role in unlocking their potential
by deriving meaning from this sensor data
48.
49.
50.
51.
52. • Sensor and Accelerometer data
• > 1GB/sec aggregated across users
• Compacted on band into code-words
UP Band Phone
{
steps: 12,
hr: 78,
ts: 1455741797
…..
….
}
53. • Phone adds context to band signals
• Collects eventing and logging data from app
• Eventing/Logging passes through Kinesis
• User data is stored in the appropriate DB
Kinesis
Eventing Data
JB server
User Data
Phone Server
Platform
54. • Run batch ETL jobs using
Elastic MapReduce to
clean and process data.
• Choose the appropriate
processing framework
depending on type of job
(Hadoop/Spark)
• Store cleaned and
anonymized data in
Redshift.
Server Warehouse
Kinesis
Platform
Redshift
Processing
Aggregations
New tables of
interest
57. WEARABLES FOR BETTER HEALTH
Chronic disease care is 86% of US healthcare cost
• Diabetes affects 12.3% population, costs $245B
• Obesity affects 36% population, costs