Google AI Might Be Able To Detect Your TB First Just By Hearing The Way You Cough & Breathe

Tuberculosis (TB), which causes approximately 4,500 deaths daily as reported by the World Health Organization, is a key focus of this initiative.

Tuberculosis is responsible for around a quarter-million deaths every single year in India.

Source : Unsplash/Alban

Google has been making advancements in healthcare quite rapidly. The tech giant has recently unveiled a new AI system which can possibly detect early signs of diseases just by analysing audio signals. According to a Bloomberg report, this groundbreaking technology is a part of Google's Health Acoustic Representations (HeAR) project, and it might hold the power to revolutionise healthcare accessibility in underserved regions. If the model achieves success as envisioned, you might be able to early detect tuberculosis (TB)

This new AI model has been trained on 300 million audio samples, including cough, sniffle, and breathing pattern. It primarily aims to identify diseases such as TB through subtle acoustic cues.

How Is This Happening?

Salcit Technologies is an India startup that specialises in respiratory healthcare AI and Google has teamed up with it to incorporate this technology into smartphones. The tech giant hopes to reach high-risk populations in regions with limited healthcare access. TB, which causes approximately 4,500 deaths daily as reported by the World Health Organisation, is a key focus of this initiative.

If we just talk about India, TB is responsible for around a quarter-million deaths every single year. Such statistics highlight the critical need for effective early detection. As per Bloomberg, the HeAR AI model was trained on 100 million cough sounds to identify TB through subtle variations in cough patterns. This AI tool holds the potential of improving screening in remote areas and can ultimately save many lives through early detection.

With this partnership in place, Salcit Technologies is also enhancing its machine learning system, Swaasa, with the help of Google’s AI model. The Swaasa app has been approved already by India’s medical device regulator. This app lets users submit a 10-second cough sample for screening and as per reports it has an accuracy rate of 94 per cent.

Despite such potential, it faces several challenges, such as integrating it into clinical practice, ensuring high-quality audio samples, and familiarising users with the technology in rural areas.

Apart from focusing on TB, Google is also looking into other uses for this bioacoustic AI, such as its initiative at Chang Gung Memorial Hospital in Taiwan to detect breast cancer early with the help of ultrasound.