Artificial Intelligence for Screening Voice Disorders: Aspects of Risk Factors

Early detection of voice disorders significantly enhances diagnostic accuracy and treatment outcomes. The objective of this paper is to emphasize the existing lack of evidence regarding the clinical application of artificial intelligence (AI) in verbal communication disorders. A literature search conducted through the Royal Society of Medicine, UK, on AI and voice disorders identified 24 AI-related articles, with Parkinson's Disease being the most frequently studied condition. However, only a limited number of AI applications provided clinically useful results. The underlying challenges pertain to data measurement, data detection, software training and testing, and inadequate specificity, sensitivity, and accuracy. The necessity of clinically validated AI models is crucial, also in addressing neurological and genetic disorders, which affect 6% and 15% of the population, respectively, aside from primary laryngeal disorders. Transparent AI software is essential for future applications in foundational software models.