Tuberculosis or TB is a treatable disease, but millions of cases go undiagnosed. Consequently, Google harnessed the power of artificial intelligence to help detect this illness.
It created the Health Acoustic Representations (HeAR) to help researchers build AI models that can flag early signs of disease.
India-based respiratory healthcare firm Salcit Technologies used this technology to create Swaasa. It’s an AI that examines cough sounds to assess lung health.
How does the tuberculosis detector work?
In March 2024, Google published a study on the Cornell University website discussing its HeAR artificial intelligence. The company says they trained this foundational model on 300 million audio recordings.
Specifically, they trained the cough model with approximately 100 million cough sounds. Consequently, HeAR can identify patterns from health-related sounds for medical audio analysis.
The Health Acoustic Representations model suits a wide range of tasks and functions on various microphones. As a result, it demonstrates a superior ability to capture meaningful patterns in health-related acoustic data.
READ: Tuberculosis: Asia-Pacific gets new weapon against drug-resistant strain
Google made this health analysis AI available to researchers so that they could develop similar tools more efficiently.
One of these researchers was Salcit Technologies, which built the Swaasa AI program. It uses artificial intelligence to detect sounds related to tuberculosis to facilitate diagnoses.
Swaasa lets more people access lung healthcare by reducing costs and the need for specialized equipment.
“Every missed case of tuberculosis is a tragedy; every late diagnosis, a heartbreak,” said Sujay Kakarmarth, a product manager at Google Research working on HeAR.
READ: More young people use hearing aids
“Acoustic biomarkers offer the potential to rewrite this narrative. I am deeply grateful for the role HeAR can play in this transformative journey.”
Google reports that The StopTB Partnership, a UN-hosted org aimed at ending tuberculosis by 2030, supports this technology. Its digital health specialist, Zhi Zhen Qin, shared this statement:
“Solutions like HeAR will enable AI-powered acoustic analysis to break new ground in tuberculosis screening and detection, offering a potentially low-impact, accessible tool to those who need it most.”