New Delhi: Google has unveiled a groundbreaking AI system that can potentially detect early signs of disease by analysing audio signals, according to a recent Bloomberg report. This technology, part of Google’s Health Acoustic Representations (HeAR) project, could revolutionise healthcare accessibility in underserved regions.
The AI model, trained on 300 million audio samples including coughs, sniffles, and breathing patterns, aims to identify diseases like tuberculosis (TB) through subtle acoustic cues. Google has partnered with Salcit Technologies, an Indian respiratory healthcare AI startup, to integrate this technology into smartphones, making it accessible to high-risk populations in areas with limited healthcare resources.
TB, which kills nearly 4,500 people daily according to the World Health Organisation, is a primary target for this technology. In India alone, TB claims about a quarter-million lives annually, underscoring the urgent need for early detection methods.
Bloomberg reports that the HeAR AI model was trained on 100 million cough sounds, enabling it to detect TB based on minute differences in cough patterns. This smartphone-based tool could significantly enhance screening capabilities in remote areas, potentially saving countless lives through early intervention.
Salcit Technologies is leveraging Google’s AI model to improve its own machine learning system, Swaasa, which has already received approval from India’s medical device regulator. The Swaasa app allows users to upload a 10-second cough sample for disease screening, boasting a 94 per cent accuracy rate.
While the technology shows promise, challenges remain. These include gaining acceptance in clinical practices, ensuring clean audio samples, and addressing user familiarity with the technology in rural areas.
Google is also exploring other applications of this bioacoustic AI, including early breast cancer detection using ultrasound at Chang Gung Memorial Hospital in Taiwan.
