According to the WHO, around 4,500 people die of tuberculosis worldwide daily, and another 30,000 fall ill with it. Although the disease is treatable, millions of people — especially in developing countries — remain unaware of their diagnosis.
A new AI tool from Google will help identify tuberculosis and other respiratory diseases in people who don't have convenient access to medical services.
🔈 HeAR (Health Acoustic Representations) model relies on bioacoustics. The Google Research team trained it on over 300 million audio recordings, including 100 million cough sounds.
The tool is easily downloadable to a smartphone. It uses cough recordings to identify early signs of TB immediately and instruct on what to do next.
Every missed case of tuberculosis is a tragedy; every late diagnosis is a heartbreak. Acoustic biomarkers offer the potential to rewrite this narrative.
— Sujay Kakarmath, Google Research, HeAR developer
👩⚕️ In India, several hospitals are already using AI tools for the primary diagnosis of lung diseases. For example, a respiratory healthcare startup and Google collaborator, Salcit Technologies, has launched a mobile app called Swaasa. It helps doctors make primary TB diagnoses remotely. All a patient needs to do is upload a 10-second recording of his cough to the app. The service makes the right call in 94% of cases.
How it's helpful
➡️ Bioacoustic tests are much cheaper than traditional diagnostic methods like blood tests or spirometry. Yet they are almost as accurate and fast.
➡️ AI lung health screening is a way to help people living in remote or poor regions where good hospitals are inaccessible.
➡️ AI bioacoustic techniques can help identify other problems, such as assessing babies' health by the sound of their crying.
➕ The HeAR API model is in the public domain for medical startups and developers.
More on the topic:
How AI is Helping Detect Autism
#news @hiaimediaen