Personalized Treatment Approaches for Social Phobia: An ML-Based Decision Support System

Despite extensive research on social phobia and its treatments, there is a notable gap in the literature concerning a comprehensive framework for determining the most suitable treatment for individuals with social phobia, particularly in the context of Sri Lanka. Current treatments lack customization, prompting the need for a support system that aids psychiatrists in selecting appropriate treatment types or combinations for their patients. This research addresses the absence of tailored treatments by considering various parameters such as symptoms, user availability, preferences, and patient history. With a significant number of individuals in Sri Lanka who have social phobia, awareness of available treatments remains low. The proposed framework aims to bridge this gap by identifying the most effective treatment type for each patient based on their unique parameters. The research involves data collection from social phobia patients to discern the most suitable treatment types. The findings will contribute to the development of a model capable of identifying optimal therapies based on individual severity of symptoms and other relevant parameters. To assess the accuracy of the developed model confusion matrix will be employed and using metrices like micro and macro averaging for precision and recall, accuracy will be measured. Validation will be performed using both the trained model, a new dataset and manual annotator. Demonstrating higher accuracy in the validation process will affirm the model's efficacy, indicating it as a superior solution to address current shortcomings. This model will serve as a valuable tool for psychiatrists, offering crucial guidance in the decision-making process and enhancing the customization of treatments for individuals with social phobia.