IntroductionAnxiety disorder is one of the most prevalent mental disorders in China. However, there are obvious subjective factors in the current assessment of anxiety disorders, which may lead to certain diagnostic errors. The identification and diagnosis of anxiety disorders can be further improved if objective biological indicators are added in the assessment process. The current research validates facial expression recognition as a screening tool to assist in detecting generalized anxiety disorder. MethodsBased on the International Affective Picture System, we constructed an aided diagnostic experimental paradigm and recorded their facial expression. The split-half reliability was displayed by the Pearson correlation heatmap. The paradigm, GAD-7 and HAMA scales were administered to 60 generalized anxiety disorder patients and 60 matched healthy controls to evaluate the criterion-related validity. Additionally, we conducted a diagnostic study by using MINI as a gold standard and calculated ROC analysis to examine the screening performance of the facial expressions. ResultsThe heatmap showed very high correlations (r > 0.60, PS < 0.05) along the diagonal of the square heatmap (from the bottom left corner to the top right). The Pearson correlation coefficients between the GAD-7, HAMA and seven facial expressions ranged from −0.35(neutral, P < 0.01) to 0.34(angry, P < 0.01). The intergroup effects of neutral, anger and fear emotions were statistically significant (F = 18.893, P < 0.001; F = 20.535, P < 0.001; F = 9.091, P = 0.003). ROC analysis showed AUC for neutral, angry and scared facial expressions were 0.723, 0.792 and 0.727 respectively. ConclusionThis study constructed a tool for auxiliary screening of GAD patients and provided an objective automatic facial expression recognition method to assist psychological diagnosis.
Read full abstract