The researchers used the ECG detector on the Apple Watch Series 6 to establish a correlation between ECG data, including heart rate acceleration and capacity decline, and the actors' reported stress levels at the exact moment the readings were taken. They also used this information to develop machine learning algorithms that created a prognostic model.
They found that the stress model had a "high level of accuracy," but lower recall. The study suggests that the Apple Watch has promising potential for detecting stress levels, and proposes that incorporating additional data points, such as sleep and fitness information, could improve the predictive accuracy of the stress models.
The researchers hypothesize that the Apple Watch could be used to support personal health care by providing conditioning techniques, such as breathing exercises, to neutralize stress signals and respond to changes in internal health in advance. Other competitors such as Samsung, Fitbit, and Garmin have already offered a stress score feature, but it is yet to be implemented by Apple in its Health app.