Dr. Google really exists! AI tool listens to your cough and determines whether you have a certain disease

Have you ever received a diagnosis from Dr. Google? Since the “office” is open 24 hours a day, 365 days a year, all you have to do is type your symptoms into the Google search bar and you'll get a diagnosis that may or may not apply to your actual condition. So if you talk to a real doctor and explain why you think you're terminally ill, they'll probably ask you, “Who gave you that diagnosis, Dr. Google?”

While Dr. Google is considered a quack, Google itself is working on an AI model called “HeAR” that uses acoustic data and AI to detect the early signs of certain diseases. One of the diseases that can be detected early using this method is tuberculosis. For most “deadly” diseases, speed of diagnosis is the key to survival. And that's more than just a hope for the future. A respiratory health company in India is now using HeAR to improve its own bioacoustic AI models.

HeAR is short for Health Acoustic Representations, a tool that allows researchers to build AI models that, as Google says, “listen to human sounds and detect early signs of disease.” Google trained HeAR on 300 million audio samples, and 100 million, a third of them, were coughing sounds. In a new high-tech version of “turn right and cough,” a doctor (maybe even Dr. Google) could ask you to cough and use AI to diagnose you based on the AI's analysis of the cough.

Nowadays, Dr. Google is only known for self-diagnosing concerned individuals. | Image credit: PhonbeArena

According to Google, HeAR outperforms its competitors with less training data than other AI models. And what makes it all so exciting is that this technology fits into a mobile phone. The potential is enormous. Imagine taking a smartphone to a remote area, far from hospitals and medical care, and being able to perform screening tests using a smartphone microphone instead of an expensive imaging device like an X-ray machine, CT scanner or MRI.

Google has partnered with Indian company Salcit Technologies, which specializes in respiratory diseases and has its own AI bioacoustics model called Swaasa. The latter uses cough sounds to determine the condition of a patient's lungs. Salcit also uses patient cough sounds to improve Swaasa's early detection of tuberculosis.

Convincing doctors that they can diagnose patients using technologies like HeAR will be a challenge. But it certainly helps when a recognized organization with a great reputation like the United Nations StopTB Partnership supports HeAR. Maybe one day you will actually be diagnosed and cured by Dr. Google.