Study: Apple Watch is said to reveal hidden diseases with AI

With the help of the sensor data of an Apple Watch, various diseases may be recognized at an early stage. This emerges from a study that Apple has published. Researchers from the company have developed a new basic model for this, the Wearable Behavior Model (WBM), as in your paper Write that was published at the Preprint website Arxiv.org at the end of June.

Data from 162,000 people

For the development of the WBM, data from 162,000 volunteers to Apple’s heart and movement study was evaluated over a total of 2.5 billion hours. A total of 57 tasks relevant to health were then tested with the model. The evaluated sensor data included the areas of heart health (via pulse measurement), physical activity (via acceleration sensor), data entered by the user such as BMI, the so -called vital, which now also include changes to body temperature when the Watch is worn at night, as well as mobility / gait behavior.

Above all, it is useful to the new basic model that it records health changes over a time course – and not only can analyze individual values as they are normally presented to the user. The detection works well, for example, in the case of infectious diseases of the respiratory tract or heart diseases that require the use of beta blockers. Even the detection of pregnancy worked amazingly well. It was also checked whether a diabetes status or a mobility violation could be determined. The WBM was also able to reconstruct input signals.

Potential for a product?

So far, Apple has not been given any information on whether the WBM is made a product. In the AI area, the group is now prone to publishing a few whole series of papers that initially describe mere research projects. The problem of early detection via Apple Watch is still the accuracy of the devices – and the admission problem.

For example, Apple has only titled the blood oxygen saturet meter in its computer clock as a “wellness” device, but not as a medical sensor. So it will probably take a while for Wearables to be prescribed as a cash benefit. However, the fact that the data is available and are recorded in the long term results in completely new possibilities, as the WBM shows.