Early Infection Alert on an Apple Watch based on resting heart rate elevation.
Early detection is crucial in stopping the spread of infection. We should use all the tools at our disposal.
Infections elevate your resting heart rate, sometimes even before you can feel the symptoms or the temperature is elevated. 1 2
Wearables provide an accurate way to measure resting heart rate during sleep by wearing the watch during the night.
We first create a personal baseline over a couple of nights, then every morning compare the recording with the previous baseline.
Elevation of the resting heart rate over the baseline could indicate an infection.
We display a green/✅ (not elevated) red/❗ (elevated) symbol on the apple watch face and anonymously collect the data into a public database.
Other factors could elevate the resting heart rate, such as:
- high stress
- Alcohol
- lack of sleep.
- ... But the user is usually aware of them.
I expect this to have a fairly low false-negative rate (high sensitivity) but a pretty high false-positive rate (low specificity).
Apple watch collects heart rate data continuously. The users would ideally wear the watch during the night.
Data is extracted from the Apple HealtKit API in the morning, and resting heart rate elevation calculated.
Resting Heart Rate can be measured with a smartphone app Instant Heart Rate4 (clinically validated) and anonymously transferred to the Infection Alert app using the Azumio Connect SDK.3
Care must be taken that all resting heart rate measurements are done at approximately the same time daily and in sitting resting state.
Let's anonymously collect all resting heart rate data from users who want to participate. With enough coverage, we can create an infection spread map.
What to collect (per user per day):
- daily heart rate measurements (close to raw data for tracking back and recalculation with a better algorithm)
- heart rate
- timestamp
- activity level
- daily resting heart rate and elevation above the resting heart rate
- location of the user (could randomly blur the exact location to protect privacy)
- unique id of the user (randomly generated on the device at install time)
All anonymized data is stored in a publicly accessible repository, so any researcher can have free access to it.
Dashboards can be built...
Google BigQuery for storing, heart rate and location data.
HeartRate:
- userid
- location
- timestamp
- day
- baseline
- heart rate
- raw data records
- timestamp
- heart rate
- motion
I'm Peter Kuhar, I've developed the most successful heart rate measurement app for iOS and Android. With more than 50M Downloads.
I've also participated in research regarding heart pulse signals, wearables... and lead internal research efforts.
This is a personal project outside my work at Azumio Inc.
[1] Harnessing wearable device data to improve state-level real-time surveillance of influenza-like illness in the USA: a population-based study - https://www.thelancet.com/journals/landig/article/PIIS2589-7500(19)30222-5/fulltext
[2] Digital Health: Tracking Physiomes and Activity Using Wearable Biosensors Reveals Useful Health-Related Information https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5230763/