Google scientists have developed a revolutionary AI model to diagnose diseases from coughing and breathing sounds.
Known as Health Acoustic Representations (HeAR), the bioacoustic foundation model introduced by the tech giant earlier this year will help researchers build AI tools capable of analyzing sound patterns to produce health insights critical for diagnosing various illnesses.
The HeAR model has been trained on vast data, including 300 million pieces of audio-based data drawn from a diverse and de-identified dataset and about 100 million cough sounds.
According to Google scientists, the model, currently available for researchers, captures significant patterns in health acoustic data and creates a great foundation for medical-related audio analysis, giving it a higher rank in the field of diagnostic AI models.
Salcit Technologies, a respiratory healthcare company, is among the companies looking to integrate the new AI model into their systems to boost efficiency and accuracy in diagnosing common respiratory illnesses such as tuberculosis (TB).
Founded by Narayana Rao Sripada in 2018, Salcit Technologies has developed Swassa, an AI-powered tool for analyzing cough sounds to detect respiratory illnesses for early treatment and better patient outcomes.
Despite TB being curable, late diagnosis can result in a complex medical scenario and may turn fatal in some cases. Therefore, every missed case is a tragedy.
In some instances, the cases usually go undiagnosed due to limited access to comprehensive healthcare. However, this new AI model could change in the future, allowing patients to access diagnostic tests at affordable prices.
By integrating Salcit Technologies’ Swassa with Google’s HeAR, the company sees an opportunity to expand AI-assisted TB screening across India.
