Non-adherence to medication remains a significant global health concern, adversely affecting treatment success and contributing to avoidable healthcare expenses. Studies indicate that about half of patients in developed countries and nearly 70% in developing countries do not consistently follow their prescribed medication routines, leading to worsened health outcomes and a rise in hospital stays. Many existing reminder systems suffer from limitations such as poor scalability, disease-specific designs, and dependence on smartphones and internet connectivity, restricting their usability in varied environments. This paper introduces a Mental Health Chatbot API developed with FastAPI that incorporates precise emotion detection via DeepFace, supports multilingual translation using IndicTransToolkit and transformer models, and provides context-sensitive responses through a large language model (OllamaLLM). The platform enables real-time chat streaming, emotion analysis from images, and multilingual speech synthesis optimized for Indian languages. Its scalable asynchronous backend, cloudhosted models, and SMS integration collectively ensure broad accessibility. Early assessments demonstrate strong accuracy in emotion recognition and translation, enabling personalized and culturally sensitive mental health support. This work lays the groundwork for scalable, empathetic mental health assistance that overcomes the constraints of current methods.
S et al. (Sat,) studied this question.