Apple Watch: An AI model reveals hidden health conditions.

A recent study published by Apple- supported researchers revealed that a new AI model trained on behavioral data from Apple Watch can now predict a wide range of health conditions more accurately than traditional sensor-based approaches.

Apple Watch: New AI model can predict health conditions more accurately

The study, titled “Beyond Sensor Data: Foundation Models of Behavioral Data from Wearables Improve Health Predictions,” introduces the Wearable Behavior Model (WBM) . This foundation model uses behavioral data collected by the Apple Watch, such as step count, sleep duration, heart rate variability, and mobility, to predict underlying health problems.

Unlike traditional approaches, which rely on instantaneous measurements of heart rate, oxygenation, etc., this AI model analyzes user behavior over the long term.

The result is greater accuracy in identifying persistent or fluctuating health conditions, such as sleep quality, beta-blocker use, respiratory infections, and pregnancy.

To train the model, the Cupertino giant used data from over 160,000 users collected through the Heart and Movement study, with more than 2.5 billion hours of data analyzed.

The model is a time-series machine learning architecture designed to identify changes in behavior over the course of days or weeks, allowing it to identify health conditions that develop over time rather than instantaneously.

The researchers argue that wearable devices have now evolved to the point where they can support this type of AI-based analysis on a large scale. It's currently unknown whether Apple will actually integrate this model in the future, but it certainly demonstrates that the current Apple Watch hardware can go much further in terms of health analytics.