Laryngeal cancer, also known as voice box cancer, affects millions worldwide. With an estimated 1.1 million cases in 2021 and 100,000 deaths, it’s a significant public health burden. Current diagnosis methods involve invasive procedures like video nasal endoscopy and biopsies, often leading to delays.
Researchers have now identified a new method using AI to detect early laryngeal cancer from voice recordings. The study analyzed voice data from over 12,500 participants and found significant differences in tone, pitch, volume, and clarity between those with benign vocal fold lesions, laryngeal cancer, or healthy individuals.
The discovery focuses on variations in acoustic features of the voice, such as fundamental frequency, jitter, shimmer, and harmonic-to-noise ratio. Men with laryngeal cancer showed distinct differences from those without any disorders, while women did not display similar patterns.
While further testing is needed to confirm these findings, the study suggests that AI-powered voice analysis could become a valuable tool in detecting laryngeal cancer early. The next step involves training models on larger datasets and validating their performance in clinical settings.
This breakthrough has the potential to revolutionize voice-based health tools, enabling earlier diagnosis and treatment of laryngeal cancer.
Source: https://medicalxpress.com/news/2025-08-ai-early-voice-cancer.html