Google has developed a new form of artificial intelligence based on bioacoustics. The idea is to be able to detect disease by capturing the sounds made by the human body, using a smartphone microphone.
The sounds emitted by the human body, whether breathing, speaking or coughing, can provide a wealth of information about our health. With this in mind, it was decided to develop a solution capable of detecting and monitoring certain diseases, starting with tuberculosis and chronic obstructive pulmonary disease (COPD).
The result was Health Acoustic Representations (HeAR), a bioacoustic base model specially designed to help researchers build models capable of listening to human sounds and spotting early signs of disease. HeAR has been trained with nearly 300 million audio data sets, a third of which are cough sounds.
An Indian start-up, Salcit Technologies, specialising in respiratory healthcare, used this to develop a product capable of analysing coughing sounds. Called Swaasa, it uses Google AI to assess potential health problems. The idea is to be able to detect tuberculosis as early as possible, simply by studying the sounds of a patient's cough. A 10-second sound sample is already enough for an accurate diagnosis in over 90% of cases.
In countries like India, tuberculosis can wreak havoc, mainly due to the lack of access to healthcare for a large proportion of the population. Improved diagnosis is therefore essential in the fight to eradicate this disease.
According to the latest figures from the World Health Organization (WHO), some 1.3 million people died of tuberculosis in 2022. – AFP Relaxnews
