Daijiworld Media Network - New Delhi
New Delhi, Aug 14: In a significant medical breakthrough, a team of US scientists has demonstrated that Artificial Intelligence (AI) can potentially detect early signs of laryngeal (voice box) cancer by analyzing a person’s voice recordings.
The study, conducted by researchers at Oregon Health & Science University (OHSU) and published in Frontiers in Digital Health, explored how subtle abnormalities in the vocal folds—detectable through voice—could serve as early indicators of cancer.
Laryngeal cancer is a growing public health concern. In 2021 alone, it was estimated to affect 1.1 million people globally, with 100,000 deaths attributed to the disease. Common risk factors include smoking, alcohol consumption, and HPV infection.
While survival rates range from 35% to 78% over five years depending on cancer stage and location, early detection is crucial for improving outcomes.
The OHSU research team found that AI algorithms could distinguish between healthy individuals and those with vocal fold lesions, which may be benign (like nodules or polyps) or represent early-stage cancer.
“These findings show we can use vocal biomarkers to differentiate voices with lesions from those without,” said Dr. Phillip Jenkins, postdoctoral fellow in clinical informatics at OHSU.
In the study, researchers analyzed 12,523 voice recordings from 306 participants across North America. These included individuals with laryngeal cancer, benign lesions, and other conditions like spasmodic dysphonia and vocal fold paralysis.
AI assessed acoustic voice features such as:
• Fundamental frequency (pitch),
• Jitter (pitch variation),
• Shimmer (amplitude variation), and
• Harmonic-to-noise ratio (a balance of vocal tone vs. background noise).
Results showed notable differences in the harmonic-to-noise ratio and pitch frequency in men with vocal fold lesions and laryngeal cancer, compared to men without voice disorders. However, similar patterns were not observed in women, though the researchers believe a larger dataset could yield different results.
The study suggests that AI-based voice analysis could become a non-invasive, low-cost tool to aid in early detection and monitoring of laryngeal cancer—particularly in men.
“Tracking changes in the harmonic-to-noise ratio may help detect early-stage cancer or monitor lesion progression,” the researchers noted.
This technology opens promising possibilities for AI-driven diagnostics, where simple voice recordings could serve as the first step toward identifying serious health issues before they progress.