2019/11/26 by Sandeep Nallan Chakravarthula, Md Nasir, Chakravarthula, Sandeep Nallan +15 · 2 citations
Psychology · #Audio and Speech Processing (eess.AS) #FOS: Electrical engineering #Mental Health via Writing #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1911.11927
openalex publication_date 2019/11/26 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Suicide is a major societal challenge globally, with a wide range of risk\nfactors, from individual health, psychological and behavioral elements to\nsocio-economic aspects. Military personnel, in particular, are at especially\nhigh risk. Crisis resources, while helpful, are often constrained by access to\nclinical visits or therapist availability, especially when needed in a timely\nmanner. There have hence been efforts on identifying whether communication\npatterns between couples at home can provide preliminary information about\npotential suicidal behaviors, prior to intervention. In this work, we\ninvestigate whether acoustic, lexical, behavior and turn-taking cues from\nmilitary couples' conversations can provide meaningful markers of suicidal\nrisk. We test their effectiveness in real-world noisy conditions by extracting\nthese cues through an automatic diarization and speech recognition front-end.\nEvaluation is performed by classifying 3 degrees of suicidal risk: none,\nideation, attempt. Our automatic system performs significantly better than\nchance in all classification scenarios and we find that behavior and\nturn-taking cues are the most informative ones. We also observe that\nconditioning on factors such as speaker gender and topic of discussion tends to\nimprove classification performance.\n