A novel non-contact screening tool based on Vibraimage technology for detecting depressive disorder in psychiatric outpatients: A diagnostic accuracy study.

BACKGROUND

Depressive disorder is a common condition; its standard diagnostic process remains time-consuming due to reliance on mental examinations. This study aimed to validate a novel, non-contact screening tool based on Vibraimage technology and machine learning for rapidly detecting depressive disorder in psychiatric outpatients.

METHODS

In this study, psychiatric outpatients were recruited, underwent a 30-second video capture, which was analyzed using the Mental Health Assessment System. This system quantifies head-neck micro-movements and provides a preliminary depression diagnosis via a machine learning algorithm. The psychiatrists, blinded to the system-generated diagnosis, established the clinical diagnosis of depressive disorder. Diagnostic accuracy was evaluated by calculating the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and predictive values.

RESULTS

Of 1062 patients screened, 601 were enrolled and completed both the video capture and clinical evaluation. Among these, 483 (80.4 %) were clinically diagnosed with depressive disorder. The system achieved a sensitivity of 99.8 % and a specificity of 78.8 %, with positive and negative predictive values of 95.1 % and 98.9 %, respectively, and an overall accuracy of 95.7 %. The AUC was 89.3 % (95 % CI: 85.6-93.0 %). Calibration was excellent (Spiegelhalter Z test, P = 1.000; Brier score = 0.04), and decision curve analysis indicated a net clinical benefit.

CONCLUSION

The Vibraimage-based Mental Health Assessment System demonstrates high diagnostic accuracy for detecting depressive disorder in psychiatric outpatients, offering a rapid, non-invasive screening alternative. Further multicenter studies are warranted to confirm these findings and explore its clinical utility and evaluate the generalizability of these findings to other population.

TRIAL REGISTRATION: This study was registered at ClinicalTrials.gov (NCT06793176) on January 20, 2025, prior to participant recruitment.