Researchers at Oregon Health & Science University have proposed a new way to detect laryngeal cancer using artificial intelligence that analyzes voice recordings. As part of the Bridge2AI-Voice project, they used more than 12,000 audio samples, including healthy participants and patients with various vocal cord disorders.
The AI identified key acoustic features, with the harmonic-to-noise ratio proving especially informative. In men with vocal cord lesions, both malignant and benign, these indicators differ significantly from those of healthy people.
The new technology may replace endoscopy and biopsy at the early stages of diagnosis, offering a noninvasive alternative. A short voice recording made on a smartphone would be enough.
