ECNETNews reports that a cutting-edge AI initiative is exploring the realm of “bioacoustics,” a fascinating intersection of biology and sound that enables researchers to understand how the presence of pathogens can influence human vocalizations. Research indicates that the sounds we produce can provide crucial insights into our health.
Recent developments reveal that a new AI model has been established to utilize sound signals for “predicting early signs of disease.” This technology could serve as a vital tool in areas with limited access to quality healthcare, requiring only a smartphone’s microphone to operate.
How the Bioacoustics AI Operates
The AI model, known as HeAR (Health Acoustic Representations), has been trained on an extensive dataset of 300 million audio samples, each lasting two seconds, encompassing coughs, sniffles, sneezes, and breathing patterns. These samples have been gathered from non-copyrighted, publicly available audio sources.
For instance, the training data includes recordings from a hospital in Zambia where patients were screened for tuberculosis. Notably, HeAR has analyzed around 100 million cough sounds specifically aimed at enhancing tuberculosis detection.
Insights suggest that bioacoustics can reveal “near-imperceptible clues” that provide subtle indications of illness, aiding healthcare professionals in patient diagnostics. Furthermore, the AI model can discern minute variations in cough patterns, enabling it to detect early signs of either improvement or worsening of a patient’s condition.
In an innovative collaboration, a partnership has been established with an AI healthcare startup focused on enhancing HeAR’s accuracy for tuberculosis and lung health screening. This collaboration leverages an existing mobile application that allows users to submit brief cough samples for analysis. Reports indicate that this application can accurately determine the presence of a disease with a 94 percent accuracy rate.
This auditory-based testing is significantly more affordable than traditional methods, with a cost of $2.40 compared to around $35 for a spirometry test in local clinics.
Despite its potential, challenges remain. Developers are currently addressing issues associated with users submitting audio samples that contain excessive background noise.
While the bioacoustics-based AI model is not yet at a commercial stage, the integration of AI technology into the medical field through sound analysis presents a promising and innovative frontier.
Topics
Artificial Intelligence
