The rapid integration of Large Language Models and Artificial Intelligence in the healthcare sector has opened up innovative frontiers for the monitoring of public health and psychiatric practice, but key issues related to the reliability, equity, and human-centric integration of AI in the healthcare sector are still pending. This paper provides a complete synthesis of the state of the art in the field, including the application of advanced models in the formulation of public health policies, the application of conversational agents for personalized psychiatric communication, and the development of automated medical information extraction systems. Although the existing literature indicates high rates of diagnostic accuracy through the application of specialized models for social media analysis and the development of “Digital Twin” solutions for the early detection of mental illness, key issues are still pending. These include the high risk of AI hallucinations in clinical decision-making, the potential for human counselors to be crowded out in online health forums, and the absence of multilingual support in mental health applications. The current state of research tends to disregard the racial disparities in the mental health treatment outcomes and lacks standardized certification processes for ensuring the safety of mobile health technologies. The proposed paper will be able to address these problems by providing a strong, interpretable, and multilingual AI framework that highlights the significance of fair design and emotional intelligence. By strictly following a comprehensive nine-domain Health Technology Assessment framework, this research fills the existing knowledge gap between technical feasibility and safety, ensuring that AI technology is used as a transparent, inclusive, and collaborative tool that supports rather than replaces human expertise in the global mental health context.
Paper
Full text
Carebot: Smart Conversational Healthcare System
Semantic Scholar · 2026
Abstract
The rapid integration of Large Language Models and Artificial Intelligence in the healthcare sector has opened up innovative frontiers for the monitoring of public health and psychiatric practice, but key issues related to the reliability, equity, and human-centric integration of AI in the healthcare sector are still pending. This paper provides a complete synthesis of the state of the art in the field, including the application of advanced models in the formulation of public health policies, the application of conversational agents for personalized psychiatric communication, and the development of automated medical information extraction systems. Although the existing literature indicates high rates of diagnostic accuracy through the application of specialized models for social media analysis and the development of “Digital Twin” solutions for the early detection of mental illness, key issues are still pending. These include the high risk of AI hallucinations in clinical decision-making, the potential for human counselors to be crowded out in online health forums, and the absence of multilingual support in mental health applications. The current state of research tends to disregard the racial disparities in the mental health treatment outcomes and lacks standardized certification processes for ensuring the safety of mobile health technologies. The proposed paper will be able to address these problems by providing a strong, interpretable, and multilingual AI framework that highlights the significance of fair design and emotional intelligence. By strictly following a comprehensive nine-domain Health Technology Assessment framework, this research fills the existing knowledge gap between technical feasibility and safety, ensuring that AI technology is used as a transparent, inclusive, and collaborative tool that supports rather than replaces human expertise in the global mental health context.