Artificial intelligence continues to amaze with innovative applications in healthcare. Google is taking important steps in this direction by developing an AI model called Health Acoustic Representations (HeAR).
It can detect early signs of serious diseases such as tuberculosis by ‘listening’ and analyzing sounds such as labored breathing, coughing or sneezing.
Model trained with over 300 million sounds
HeAR represents an innovation in the field of bioacoustics. It is trained with an impressive amount of audio data: over 300 million samples. Of these, 100 million are cough sounds.
The massive database allows the model to identify subtle signs in the sounds emitted by the human body, signs that may indicate the presence of respiratory conditions.
94% accuracy
Google’s collaboration with Indian startup Salcit Technologies brings this technology closer to users in areas with limited access to advanced medical services.
Through the Swaasa app, developed by Salcit, HeAR is used to analyze cough sounds and assess lung health. The 94% accuracy recorded in TB screenings is impressive.
Without treatment, the death rate from tuberculosis exceeds 50%. Sujay Kakarmath, product manager at Google Research, points out the transformative potential of acoustic biomarkers.
Use the phones’ microphones
Using the microphones of smartphones, devices that around 60% of the world’s population own, HeAR can bring diagnostic solutions to remote areas where advanced medical equipment is scarce or non-existent.
The incredible potential of HeAR
Google doesn’t stop there. The company is exploring the use of the new model in detecting other conditions, including the early stages of breast cancer.
It is also being studied to combine it with ultrasound technology to provide cost-effective diagnostic options.
It will not replace regular medicine, but will complement it
However, there are challenges that need to be addressed. Maintaining the quality of audio samples, especially in noisy environments, is essential for optimal system performance. It is also important that this technology is introduced carefully and does not replace existing medical practices.
Google hopes to advance the development of diagnostic tools and monitoring solutions for various lung conditions and other medical fields.
Source: Quartz
Source: www.go4it.ro