New research indicates that subtle changes in a person’s voice may signal the presence of laryngeal cancer, potentially allowing for earlier detection. While these variations are often undetectable by the human ear, scientists have developed machine learning algorithms capable of distinguishing between normal and cancer-affected voice patterns. This innovative approach could transform how laryngeal cancer is diagnosed, which currently relies on invasive procedures such as video nasal endoscopy and biopsies.
According to a report from the World Health Organization, approximately 1.1 million cases of laryngeal cancer were diagnosed globally in 2021, leading to around 100,000 deaths. The findings from this research, conducted by Oregon Health and Science University and Portland State University, suggest that digital screening tools utilizing voice recordings might enable non-specialist doctors to identify patients at risk more swiftly.
Machine Learning Identifies Vocal Changes
The researchers analyzed 12,523 voice recordings from 306 participants in North America, focusing on identifying vocal features associated with both benign and malignant vocal fold lesions in men. A key factor in their analysis was the harmonic-to-noise ratio, which measures the relationship between the tonal quality of a voice and background noise. This metric proved effective in distinguishing between male voices affected by cancer, benign lesions, and other voice disorders.
While the study successfully identified vocal characteristics in men, the researchers noted that they were unable to find statistically significant features in women’s voices. They remain optimistic that larger datasets could yield more comprehensive results for female participants in future research.
Clinical informatician Phillip Jenkins from Oregon Health and Science University emphasized the next steps in this research. “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,” he explained. “We then need to test the system to ensure it works equally well for women and men.”
Future of Voice-Based Health Tools
Voice-based health tools are already in the early stages of development, with projects being piloted in various settings. Jenkins predicts that, with the expansion of datasets and clinical validation, similar tools for detecting vocal fold lesions could enter pilot testing within the next few years. This advancement represents a significant shift in laryngeal cancer diagnosis and has the potential to enhance patient outcomes through earlier intervention.
The research findings were published in the journal Frontiers in Digital Health, marking a promising step toward integrating technology into healthcare for better cancer detection methods.
