Google Develops AI to Detect Illness Through Sound
The new model, HeAR, uses bioacoustics to identify early signs of diseases like tuberculosis from audio samples.
Overview
- HeAR was trained on 300 million audio samples, including 100 million cough sounds.
- The AI can detect subtle changes in coughs, sniffles, and breathing patterns.
- Google is collaborating with India's Salcit Technologies to enhance the model's accuracy.
- Salcit's app, Swaasa, can diagnose diseases with 94% accuracy from a 10-second cough sample.
- The technology aims to improve healthcare access in regions with limited medical resources.