Reducing vulnerabilities in the areas of suicide and self-harm in military environments through the analysis of vocal markers, visual markers, and non-invasive biomarkers

Suicide is one of the leading causes of mortality, requiring interdisciplinary research efforts to develop and implement suicide risk screening tools. With advances in artificial intelligence, these tools can predict suicidal intent and identify individuals at high risk. Considering the need to keep such individuals away from weapons and heights, AI-based systems can be utilized to detect those at risk of suicide in military settings. The current project aims to implement a smart facility leveraging various sensors to achieve high accuracy in detecting suicidal intent by integrating data collected from auditory markers, visual cues, and biomarkers. The proposed system simultaneously employs pattern-based detection (comparing an individual's feature profile with existing samples) using the XGBoost classification method and anomaly-based detection (comparing an individual's current features with their past data to assess changes) using LSTM neural networks.

کلیدواژه ها:

Suicide ، Prediction ، Vocal and Visual Markers ، Biomarkers ، Machine Learning