Google AI Could Soon Use a Person’s Cough to Diagnose Disease

A team of Google scientists has developed an incredible machine-learning tool to help detect health conditions by analyzing noises like coughing and breathing. The system is called Health Acoustic Representations, or HeAR. It uses a massive data set trained on millions of audio recordings and pairs it with health information about the person who made the sounds. HeAR has been trained to detect COVID-19, tuberculosis, and whether a person smokes.

Using audio as a biomarker for disease gained traction during the height of the COVID-19 pandemic. Scientists realized it was possible to determine a respiratory disease by listening to a person’s cough. The team hopes to eventually use machine learning to diagnose diseases like COVID-19 and tuberculosis through audio input. Although acoustic science has existed for decades, AI and machine learning can lead to breakthrough technologies that improve health outcomes.

Source