“A doctor in your pocket”: new AI assistant reads your tests and warns about health risks

On March 25, a major technology company unveiled a prototype of a personal medical assistant powered by generative AI that can analyze wearable device data, uploaded lab tests and a user’s medical history, and produce preliminary health recommendations. The assistant tracks heart rate, heart rate variability, sleep quality and activity levels in real time, comparing them with individual baselines rather than just population averages. According to the developers, the goal is not to replace doctors, but to nudge users when it is time to book an appointment, order an extra test or reduce physical load.

The system runs on a model trained on anonymized medical data and consensus guidelines from international professional organizations. To reduce hallucinations, the assistant’s responses are accompanied by links to clinical guidance and an explanation of its confidence level. Users can forbid storing some data locally or in the cloud and can also opt out of having their data used for further model training. At the same time, the company announced partnerships with several clinics: doctors will get a separate interface where they can review AI outputs and leave comments, turning the assistant into a joint tool for patient and physician.

Regulators have already stated that they will treat the service not as a full‑fledged medical device, but as an “enhanced informational tool,” while emphasizing that the company will still be responsible for algorithm transparency and data protection. Experts expect that in the coming years such AI assistants will become standard features of wearables, from watches and fitness bands to home blood pressure monitors and glucometers. For healthcare, this could mean a shift from occasional doctor visits to continuous preventive “background monitoring,” where risky trends are detected weeks or months before symptoms appear.