Published Paper Details:

Mental health has emerged as a pressing global challenge, with escalating cases of stress, anxiety, depression, and other psychological disorders affecting individuals across all age groups. Traditional mental health support systems depend largely on human therapists, counselors, and clinical infrastructure. While effective, these systems face significant bottlenecks such as limited accessibility in remote areas, high consultation costs, long waiting times, and persistent social stigma that prevents many from seeking timely help. To overcome these challenges, ManasAI is developed as an AI-powered mental health chatbot designed to provide real-time emotional support, personalized mood tracking, and preliminary well-being guidance. Leveraging advanced Natural Language Processing (NLP) and sentiment analysis, the chatbot identifies underlying emotions in conversations and generates empathetic responses tailored to user needs. In addition, ManasAI integrates handwriting-based psychological pattern recognition, which enables it to analyze handwriting variations associated with different emotional states, thereby enhancing multimodal emotion detection. The technical framework of the system employs FastAPI as the backend for efficient and scalable deployment, MongoDB for secure and structured data storage, and AI/ML models including Transformer-based architectures for text analysis, Convolutional Neural Networks (CNNs) for handwriting recognition, and anomaly detection models to identify inconsistencies or potential cases of user deception. A fusion model is incorporated to combine multimodal signals, leading to more robust and accurate emotion recognition. Beyond simple conversational responses, ManasAI offers features such as mood trend visualization, risk scoring mechanisms, and safety escalation protocols to detect critical conditions like suicidal ideation or severe distress proactively.