Acoustic and Machine Learning Methods for Speech-Based Suicide Risk Assessment: A Systematic Review (2505.18195v1)

Abstract: Suicide remains a public health challenge, necessitating improved detection methods to facilitate timely intervention and treatment. This systematic review evaluates the role of AI and Machine Learning (ML) in assessing suicide risk through acoustic analysis of speech. Following PRISMA guidelines, we analyzed 33 articles selected from PubMed, Cochrane, Scopus, and Web of Science databases. These studies primarily explored acoustic differences between individuals at risk of suicide (RS) and those not at risk (NRS), and evaluated ML classifier performance. Findings consistently showed significant acoustic feature variations between RS and NRS populations, particularly involving jitter, fundamental frequency (F0), Mel-frequency cepstral coefficients (MFCC), and power spectral density (PSD). Classifier effectiveness varied based on algorithms, modalities, and speech elicitation methods, with multimodal approaches integrating acoustic, linguistic, and metadata features demonstrating superior performance. However, limitations such as methodological variability, small sample sizes, lack of longitudinal data, and limited linguistic and demographic diversity restrict generalizability. Future research should focus on standardizing methods, expanding multimodal analyses, and utilizing larger, diverse datasets to support AI integration in clinical suicide risk assessment.