Google’s HeAR model for tuberculosis detection helps researchers in India detect life-threatening diseases through audio analysis. A cough often indicates a health issue, but how much can we learn from the sound alone? According to a recent Google Research report, a team in India is using Google’s AI model to “hear” specific respiratory diseases. These include tuberculosis (TB) and chronic obstructive pulmonary disease (COPD). The HeAR model enhances tuberculosis detection by analyzing cough sounds and other health-related data.

Launch of the HeAR Model for Tuberculosis Detection

Google introduced the HeAR model for tuberculosis detection in March. The research team trained HeAR on 300 million audio data points from a diverse and de-identified dataset. They also trained the cough model using approximately 100 million cough sounds. This model specifically focuses on identifying health patterns related to respiratory diseases such as tuberculosis.

AI Analyzes Speech Patterns and Coughs

The AI can analyze how a person speaks, including tone, pitch, and tempo. This analysis can detect potential health issues like dementia. Shravya Shetty, Director of Engineering at Google Research, explained that the HeAR model recognizes patterns in health-related sounds. This capability makes HeAR a powerful tool for medical audio analysis. She also noted that HeAR outperforms other models across various tasks. It generalizes well across different microphones, capturing meaningful patterns in health-related acoustic data.

Real-World Applications in Tuberculosis Detection

Salcit Technologies, a respiratory health company based in India, uses the HeAR model in its Swaasa app. The app analyzes cough sounds to detect tuberculosis. TB is treatable, but millions of cases go undiagnosed each year. Often, this happens because people lack easy access to healthcare services. Improving diagnosis is crucial for eradicating TB. AI can enhance detection, making care more accessible and affordable worldwide.

For more details, check out Google Research on HeAR and Salcit Technologies’ Swaasa App which leverages HeAR for respiratory health solutions.