WH2014 Session: Smartphone platform survey-scale heart rate collection – a performance evaluation in ethiopia

WLSACONVERGENCE SUMMIT

“SMARTPHONE PLATFORM SURVEY-SCALE HEART RATE COLLECTION – A PERFORMANCE EVALUATION IN ETHIOPIA”

JAMES A. J. HEATHERSPH.D., UNIVERSITY OF SYDNEY

Ph.D., University of SydneyJAMES A. J. HEATHERS

“Smartphone Platform Survey-Scale Heart Rate Collection – a Performance Evaluation in Ethiopia”

3

WHAT IS HRV?

4

PROBLEMS WITH HRV

5

PROBLEMS WITH HRV

HRV NOW:

• POORLY CONTROLLED• SMALL SAMPLE SIZE

6

TWO BETTER MODELS:

1. CAREFUL EXPERIMENTAL CONTROL

2. MASSIVELY INCREASED SAMPLE SIZE

PROBLEMS WITH HRV

7

TWO BETTER MODELS:

1. CAREFUL EXPERIMENTAL CONTROL

2. MASSIVELY INCREASED SAMPLE SIZE

PROBLEMS WITH HRV

8

IDEAL MASS COLLECTION:

• Accurate• Robust• Cheap• Components

• Off shelf• Swappable

THE NEW SYSTEM

9

THE NEW SYSTEM

• iOS device (iPod, iPad etc.)

• Hardware switching box

• Optical finger sensor (PPG)

• www.myithlete.com

A/DBandpassfilter

d/dtPeak

detectTime

stamp

Sensor

iPod/iPhone

Logfile

HRV

10

HEATHERS et al (2012); HEATHERS (2014b)

Is pulse rate accurate enough to measure HRV?

Yes.

PROBLEMS WITH HRV

11

HEATHERS (2013);HEATHERS (2014a)

Is this system accurate enough to support HRV?

Yes.

PROBLEMS WITH HRV

12

“The International Food Policy Research Institute (IFPRI) seeks sustainable solutions

for ending hunger and poverty.”

THE IFPRI

13

HRV in Development Economics

• HRV as stress correlate (in realistic sample)

• HRV as predictor of program success or covariate

• HRV as control against response bias

THE IFPRI

14

TWO PURPOSES

1. FIELD TRIAL, IN CONTEXT

2. PROPERTIES OF MASSIVELY INCREASING SAMPLE SIZE

EXPERIMENT

15

PARAMETERS• 17 enumerators (locals, no research experience)

EXPERIMENT

16

PARAMETERS• 20 iPod Touch + extra sensors

EXPERIMENT

17

PARAMETERS• 6 supervisors

EXPERIMENT

18

PARAMETERS• Animated indicators on phone

EXPERIMENT

19

SAMPLE

• N=679• n=1205 recordings• 91.7% usability (93.3% in new sample)

RESULTS

20

OUTCOMES

• Age/gender relationships as expected

• Relationship with alcohol consumption (analogue of self-control?)

RESULTS

21

ONGOING:

• Correlate of food-aid program (n=1142), still analysing

• Largest HRV study ever… (presently n=3408)

RESULTS

22

PAPER

WLSACONVERGENCE SUMMIT

www.wirelesshealth2014.org