an overview of the myheart counts app
TRANSCRIPT
Acknowledgements
Aleksandra Pavlovic, Michael V. McConnell, Martin Landray,
Elizabeth Orchard, Lionel Tarassenko, Carmelo Velardo
Introduction
We know that being active helps to prevent heart disease
However, this knowledge is largely based on survey data
Thanks to smartphones and wearable devices we can now measure activity in greater detail
Can we use smartphones to identify what type and what amount of activity are best for the heart to stay healthy?
MyHeart Counts
• One of the 5 apps launched with Apple’s Research Kit
• Aim: to study activity, fitness, and cardiovascular health in the general population
• Collaboration between the University of Oxford and Stanford University
https://med.stanford.edu/myheartcounts.html
Study description
• Participants:• Adults, 18 and over
• Protocol:• Answer validated questionnaires about diet, cardiovascular health, sleep
quality, physical activity, wellbeing and risk perception, “heart age”• Perform 6-minute walk test• Measure daily activity for a week• Use the app for a week every three months
• Goals:• Compare measured activity to markers of cardiovascular health (fitness, risk
factors, disease status)• Follow for changes in cardiovascular health status to understand the types of
activity and other factors most predictive of improvement• Study behavioral interventions for modifying activity and ultimately
cardiovascular health (TBD)
Collaboration between Oxford and Stanford• A first version of the app was launched by Stanford
University for iPhones only
• University of Oxford is developing a version for Android phones
(Partially) based on Google Fit
Based on Apple HealthKit and Research Kit
Functionalities: consent and registration• Year-long discussions between ethicists and lawyers to
develop and approve a consent process on a smartphone
• Offers an introduction to the tasks to be performed by the participant and the legal and ethical consequences of the study
Functionalities: questionnaires
• Diet, Physical Activity Readiness, Activity and Sleep, Well-Being & Risk Perception, General Blood Pressure, Cardiovascular Health• Most sources are derived from AHA’s “MyLifeCheck”
toolkit
Functionalities: education
• Available educational content:• Information about the study
• Symptoms
• Treatments
• Online resources and news feed
Functionalities: 7-day activity assessment
• HealthKit and Google Fit provide activity detection based on accelerometry, wearable devices and manual input from the user
• The app sends a report about the activities performed in a week
Functionalities: heart age
• Uses the 2013 AHA/ACC ASCVD risk score system to provide 10-year and lifetime risks of having a heart attack or stroke
• Requires blood pressure and cholesterol data (manually input)
Functionalities: 6MWT
• Uses the pedometer and GPS to calculate distance
• HR is also captured if the appropriate device is connected
Issues about 6MWT
• Outdoor:• GPS signal is not always
available or reliable
• Indoor:• Counting steps can only
roughly approximate distance
• University of Oxford is developing an improved version of both measurements (higher accuracy)
Improved 6MWT: Indoor
• Counts lapses around two points at known distance
• Uses compass to detect U-turns
• Additional movement sensors and step counters are used to improve accuracy
• Error: 3% (5-20 meters)
Improved 6MWT: Outdoor
• Uses all means of localisation (GPS, WiFi, cell towers)
• Filters out unstable signal
• Additional movement sensor and step counter for increased accuracy
• Error: 4% (10-20 meters)
Android version
• Mostly similar, but some design aspects are adapted to Google’s recommendations
• Uses Google Fit instead of HealthKit for activity detection
• Still under development
Backend
• Managed by Sage Bionetworks
• Synapse: a tool to collect genomics, molecular, and imaging data.• Used for annonating data, tracking the relationship between
digital assets and sharing data. Integrates with R, Python and Linux shell.
• BRIDGE: a web-based, open-source tool that collects data from patients’ phones and feeds them into Synapse
Research Kit, experience so far
• We have used it only partially at University of Oxford• Offers a set of use cases already implemented *:
• Surveys• Consent• Gait and Balance• Tapping speed• Fitness (e.g. 6MWT)• Spatial memory• Sustained phonation
• Uses HealthKit as a personal health record and for activity recognition
• Open Source, SW framework, used by developers
* http://researchkit.org/docs/docs/ActiveTasks/ActiveTasks.html
Results
• The study was launched in March 2015 and is still ongoing
• More than 47000 participants joined
• Oxford improved 6MWT accuracy
Open questions
• How reliable are Apple Health Kit and Google Fit?
• How measurements vary among different devices?
• How to engage users in the long term?