Wearable tech with sensors that monitor and report biometric data can help assess your current health, but devices with algorithms that detect life-threatening conditions could potentially save your life. The iBeat smartwatch app monitors heart activity to alert to dangerous conditions and researchers at North Carolina State University developed a wearable tech system called the Health and Environment Tracker that helps predict asthma attacks.

A team of University of Washington (UM) engineers and researchers developed a support vector machine (SVM) that detects cardiac arrest with no physical contact. An SVM is a machine learning algorithm that classifies data into one of two classes based on a value called a hyperplane that maximally differentiates the classes. In other words, if you score above the value you are most likely in one class and if you score below that value you’re most likely in the other class. The UM SVM classifies agonal breathing to identify cardiac arrest.

Agonal breathing occurs when the body is deprived of adequate oxygen. Therefore, agonal breathing can be a diagnostic of cardiac arrest. The University of Washington group trained the SVM with recorded 9-1-1 calls from people in cardiac arrest. They tested the machine against recorded sleep audio that included snoring, hypopnea, and sleep apnea and in home sleep environments. The SVM accuracy determining agonal breathing was in the high 90% range, with false positives in just 0 to 0.22% of the samples.

The University of Washington SVM can be adapted as a smartphone app or a smart device skill. In the study’s discussion, the researchers suggest that, due to battery power usage and potential data fees, wired smart device skills for devices such as Amazon Alexa smart speakers and smart displays that listen by default could be more practical than smartphones. People concerned about smart devices and privacy might object to allowing Alexa or Google Assistant, for example, listen to them at night. However, the potential life-saving aspect of detecting and sending real-time alerts of agonal breathing for people who might never make it to a phone to call 911 is undeniable.