Classification of Social Anxiety Disorder Using Explainable Machine Learning and Pearson's Correlation Technique

Social Anxiety Disorder (SAD) is a widespread mental health issue marked by significant fear or discomfort during social interactions. It can greatly affect a person's life and overall happiness, causing problems like emotional distress, low self-esteem and depression. Thorough examination by a mental health expert is often required to diagnose SAD. The diagnostic and statistical manual of mental disorders' particular criteria are used to diagnose SAD that includes clinical interviews, self-report questionnaires, and a thorough evaluation of the individual's symptoms like severe anxiety while engaging or conversing with strangers. In this work, we employ explainable artificial intelligence (XAI) and machine learning techniques to identify SAD in individuals. Critical attributes were identified using Pearson's correlation technique. Random Forest yields optimal outcomes with 88% accuracy, 93 % precision, 83 % recall, 84 % F1-score, and 94% Area Under Curve (AUC). Furthermore, XAI methods such as Shapley Additive Values (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) have been applied to improve the models' accuracy, comprehensibility, and precision. Automated SAD diagnosis helps in early detection and increased accessibility that allows for timely intervention, treatment, and facilites access to social anxiety testing and screening for an individual.