The era of cyber health checkups has arrived.

Google trained a health foundation model, SensorFM, and also received recognition from professional doctors.
It can simultaneously analyze more than 30 signals from smart watch records, including heart rate, blood oxygen, sleep, exercise, and body temperature, to predict cardiovascular health, metabolic risk, sleep status, and metrics related to depression and anxiety.
Google then put it through an exam using data from nearly 14k people:
Looking only at watch records to judge these people’s health status, then comparing against the real outcomes, across 35 tests, SensorFM was more accurate than conventional methods in 34 of them.
SensorFM’s training data comes from 5 million people, totaling more than 1 trillion minutes. Even if there was a period in the middle without wearing the watch, it can estimate the missing data based on the records before and after.