EMAILED ON February 8, 2018 BY Conor Grant

Your watch knows you’ll get sick, but it doesn’t know why

San Francisco-based health-tracking company, Cardiogram, has released clinical studies showing its machine-learning model, DeepHeart, can use Apple Watch data to forecast potential health ailments. 

Among the watch’s purported powers: the ability to predict abnormal heart rhythms with 97% accuracy, sleep apnea with 90% accuracy, diabetes with 85% accuracy, and hypertension with 82% accuracy.

Diagnosing with data (not medicine)

To put health information generated by wearable devices like Apple Watch or Fitbit into operation, engineers at Cardiogram (mostly ex-Googlers) turned not to medicine, but to big data. 

By adapting speech-recognition software to synthesize health data instead of words, Cardiogram stumbled upon a way to accurately predict a disease without knowing anything about its cause.

So, DeepHeart will tell you that you have diabetes — but not how it knows. 

Health insurance and big tech stole each other’s hearts

The FDA makes it illegal for wearables to make true medical diagnoses, but the savings made possible by fitness-tracking and diagnostic “recommendations” still have health insurance companies drooling.

Aetna, John Hancock, and UnitedHealthcare all have programs that reward members for using wearable devices to increase fitness and improve health.

Big tech spent about 10 times as much on health-related deals in 2017 (at least $2.7B) as it did in 2012 ($277m) — and if the money continues to flow, expect to see companies like Cardiogram pumping it through the veins of a new data-driven healthcare system.