An Exploration and Analysis of Generative AI in Enhancing Personalized Treatment Plans for Chronic Disease Management

Generative AI (GenAI) shows a growing potential for personalized treatment planning in chronic disease management. Although Large Language Models (LLMs) are capable of processing clinical text and generating treatment-related recommendations, their adoption in real clinical practice remains limited due to concerns around trust, safety, bias, privacy, and accountability. From a Human–Computer Interaction (HCI) perspective, these concerns directly shape whether GenAI systems can be trusted, and used responsibly in high-stakes medical environments. In this study we verify the specific trust-related concerns that shape clinicians’ willingness to engage with GenAI. Using a mixed-methods approach, we first analyzed public discourse on GenAI and healthcare taken from social media platforms, professional forums, news outlets, and academic databases. Topic modeling and sentiment analysis were applied to identify recurring patterns, resulting in six core categories of concern. These concerns were subsequently validated through a survey with practicing clinicians to assess their relevance in real clinical contexts. The results subsequently indicated low clinician comfort with GenAI systems, reinforcing the need for cautious, transparent, and human-in-the-loop system behavior. This study contributes an empirically validated framework that establishes a foundation for future design and evaluation of safer, more inclusive GenAI systems for chronic disease management.