Abstract
AIM: Suicide is the most severe consequence of Major Depressive Disorder (MDD). Current risk assessments rely heavily on subjective self-reports, which lack reliability. Emerging technologies, such as facial and behavioral recognition devices, are being explored to improve suicide risk evaluation. This study aimed to examine the potential of 3D facial features in identifying suicide risk and uncovering sex-specific characteristics in patients with MDD. METHODS: We conducted a cross-sectional study involving 222 MDD patients. Suicide-related information was collected from caregivers, while independent raters assessed depressive symptoms and recorded sociodemographic data. Three-dimensional facial scans were acquired using the 3dMDface System, followed by preprocessing to extract key facial landmarks. Sex-stratified subgroup analyses were performed to identify suicide risk-associated facial features. Logistic regression analysis was used to evaluate predictors, including demographic data, clinical characteristics, and the identified facial markers. RESULTS: Data from 203 patients were analyzed, including 110 in the suicide-risk group and 93 in the non-suicidal group. The suicidal group exhibited significantly shorter philtrum length (t = 2.137, p < 0.05). Analyses revealed sex-specific facial patterns, with males demonstrating suicide risk association with philtrum depth (t=2.389, p < 0.05) and females showing nose-eye distance variations (U = 1121, p < 0.05). Logistic regression identified female (OR = 2.055, 95% CI: 1.107-3.873, p < 0.05) and shallow philtrum (OR = 0.644, 95% CI: 0.419-0.952, p < 0.05) as potential factors, with a significant interaction effect (OR = 1.963, 95% CI: 0.419-0.952, p < 0.05). CONCLUSION: This study identified sex-specific facial features associated with suicide risk in MDD, with reduced philtrum depth in females emerging as a correlate. These objective measures could complement current clinical risk assessments, though further longitudinal validation is required. CLINICAL TRIAL REGISTRATION: https://www.chictr.org.cn, identifier ChiCTR2400090458.