The Apple Watch could predict your health much more precisely thanks to this data

Last February, we learned that Apple would launch a particularly ambitious scientific study. Moreover, The idea was then to offer its users to entrust the data collected by their Apple Watch to find. Consequently, out if certain algorithms could provide for health. However,

To go further
Apple apple watch could predict your new Watch Series apple watch could predict your 10, SE, Ultra 2… What Apple Watch model to choose in 2025? Therefore,

It must be said that. For example, so far. For example, the algorithms used by Apple are essentially based on user movements and on data from the optical cardiac frequency sensor. However, the study aimed to know if, by integrating more data, the predictions could be more reliable. Pell-mell. the study was aimed at using the respiratory frequency at night. blood oxygen levels, body mass, calories spent, number of steps, heart rate at rest or ascension speed.

On June 30, Apple finally published the first results of its study, relayed by the site 9to5Mac.

A new AI Apple -based algorithm

In total. data of nearly 162,000 people were used, for a total of 15 billion hours of data. Logically. to analyze as much raw data. Apple has gone through an AI, in this case Mamba-2, to assess a new health algorithm, called WBM (for Wearable Behavior Modelor portable behavior model).

apple watch could predict your

To go further
Vital signs: how does the new vital application of Apple Watch. work

The idea was then to compare the analyzes of the WBM algorithm compared to the only optical. analysis of the heart rate (PPG) currently used by Apple in certain scenarios. Apple thus sought to find out if this model could better predict arterial diseases. pregnancy. urinary tract infections, anxiety, atrial fibrillation, sleep phases or sleep apnea:

WBM representations reach a high performance and constantly exceed basic medical history. Above all. the combination of WBM. the PPG does better in almost all tasks, demonstrating the complementarity of behavioral data and sensor data for health detection tasks.

For example, the new algorithm combined with the optical sensor has thus analyzed a pregnancy in 92.1 % of cases, against 87.3 % for the PPG sensor apple apple watch could predict your new watch could predict your alone. For the duration of the sleep, the accumulation of the two data made it possible to go from 59.0 to 60.1 % precision. Only counter-examples, diabetes remains better measured by the PPG alone, at 82.9 %, than by the integration of behavioral data, at 82.8 %.

For the time being. it is only a study on the part of Apple and, a fortiori, of a prepublication. If it seems more than likely that the American brand will integrate. in the more. less long term, behavioral data with its various health algorithms, it has not yet officially communicated on the integration of these measures within its Apple Health application.