New research explores how hidden patterns in the human voice could serve as early indicators of disease.
Cancer of the larynx, often called the voice box, remains a major global health concern. In 2021, about 1.1 million people were diagnosed worldwide, and roughly 100,000 died from the disease. Smoking, heavy alcohol use, and infection with human papillomavirus are key risk factors. Survival rates vary widely, ranging from 35% to 78% over five years with treatment, depending on where the tumor develops and how advanced it is at diagnosis.
Early detection plays a critical role in improving outcomes. Today, diagnosis typically relies on video nasal endoscopy and tissue biopsies, which are invasive and can be difficult to access quickly. Delays in seeing a specialist may slow diagnosis and treatment.
New research published in Frontiers in Digital Health suggests a different approach. Scientists found that subtle changes in a person’s voice can reveal abnormalities in the vocal folds. These “vocal fold lesions” may be harmless, such as nodules or polyps, but they can also signal early-stage laryngeal cancer. The findings point to a potential new use for artificial intelligence: identifying early warning signs of cancer through voice analysis.
“Here we show that with this dataset we could use vocal biomarkers to distinguish voices from patients with vocal fold lesions from those without such lesions,” said Dr Phillip Jenkins, a postdoctoral fellow in clinical informatics at Oregon Health & Science University, and the study’s corresponding author.
Voice messages
Jenkins and his team are part of the ‘Bridge2AI-Voice’ project within the US National Institute of Health’s ‘Bridge to Artificial Intelligence’ (Bridge2AI) consortium. This nationwide effort aims to apply AI to complex biomedical problems. For this study, the researchers examined tone, pitch, volume, and clarity using the first public release of the Bridge2AI-Voice dataset, which includes 12,523 recordings from 306 participants across North America.
