Cancerous vocal fold lesions could exert a subtle influence over the sound of a person’s voice, which could help catch laryngeal cancer earlier than current methods.
While the distinction is impossible to detect with the human ear, scientists have found that machine learning algorithms can tell them apart.
Across the world, around 1.1 million cases of laryngeal or ‘voice box’ cancer were diagnosed in 2021, and approximately 100,000 people died from it. Currently, this cancer is diagnosed by specialists using invasive procedures like video nasal endoscopy and biopsies.
Related: Your Voice Could Hide The Tell-Tale Signs of Parkinson’s Disease
Digital screening tools that detect the early warning signs of laryngeal cancer using voice recordings could help non-specialist doctors identify patients at risk, and help them receive a diagnosis sooner than they might otherwise.
By analyzing 12,523 voice recordings from 306 North American participants, researchers from Oregon Health and Science University and Portland State University identified in men the telltale vocal features of vocal fold lesions that were either benign or cancerous.
In particular, the harmonic-to-noise ratio (the relationship between tone and noise) helped to differentiate between male voices with cancer, benign lesions, and voice disorders.
The researchers were unable to find statistically significant identifying features in women’s voices in this study, but are hopeful that a wider dataset could offer better results for female voices in future.
“To move from this study to an AI tool that recognizes vocal fold lesions, we would train models using an even larger dataset of voice recordings, labeled by professionals. We then need to test the system to make sure it works equally well for women and men,” says clinical informatician Phillip Jenkins, from Oregon Health and Science University.
“Voice-based health tools are already being piloted. Building on our findings, I estimate that with larger datasets and clinical validation, similar tools to detect vocal fold lesions might enter pilot testing in the next couple of years.”
This research was published in Frontiers in Digital Health.
Source link
