CogniCare AI: An Intelligent Mental-Health Companion

Abstract

Mental health challenges, such as stress, anxiety, and depression, affect millions around the world, but access to timely professional support remains limited due to stigma, cost, and resource constraints. The system utilizes transformer-based NLP models for sentiment classification and linguistic feature analysis algorithms to compute cognitive load. Underpinned by a FastAPI backend with MongoDB for data persistence, it processes user messages to generate empathetic, context-sensitive responses while maintaining comprehensive records of emotional trends. The clinician dashboard offers longitudinal behavioral analytics comprising emotional heatmaps, daily metrics, and correlation patterns to enable data-driven therapeutic decisions. Testing on 500 labeled messages yielded 82% accuracy in sentiment classification with an RMSE of 1.31 for cognitive load estimation. Scalability analysis revealed an average response time of 14-25ms for concurrent user volumes. This research closes the gap between automated emotional support and professional mental healthcare by acting as both an easily accessible, non-judgmental conversation partner and a clinical monitoring tool.