Development of an Intelligent Online Response System (Chatbot) for Mental Illnesses and Severe Pathologies Based on Artificial Intelligence Models

Abstract

Mental health issues such as depression and anxiety are affecting more people worldwide, with over 970 million cases reported. Many individuals do not receive proper care due to stigma, lack of professionals, and limited access to services.This thesis proposes a chat- bot system to help detect and respond to mental health conditions. It uses a multimodal approach that combines text input and visual features like facial expressions and eye movement to better understand the user’s emotional state.The system integrates Clini- calBERT for text classification, Flan-T5 for generating responses, and Ft_Transformer for visual analysis. These outputs are fused using an XGBoost model for final classifi- cation.The proposed model achieves a classification accuracy of 95%, which surpasses current state-of-the-art results in mental health detection tasks.This work offers a prac- tical and scalable tool to support mental health, especially in areas with limited access to professional care.