The primary goal of next-generation healthcare systems is to deliver preventive, predictive, precise, participatory, and personalized health, thereby improving the quality of patient care. The objective of this thesis is to advance cutting-edge AI-enabled clinical decision support systems (CDSS), including large language models, to facilitate translational personalized medicine. Firstly, I investigate data-centric AI methods to prepare integrative, large-scale, and high-quality patient data to address data scarcity and quality issues, with a specific focus on rare diseases. Secondly, I develop an AI-enabled CDSS framework for improved patient care by solving responsible AI considerations, including model transparency, accountability, trustworthiness, and fairness. Thirdly, with the emerging few-shot capabilities of LLMs in reasoning and planning, I deploy translational and interactive LLM-based agents for improved user experience in real-world clinical practice. Specifically, clinicians can specify tasks in natural language, and the LLM agent autonomously generates and executes code to interact with EHRs for answers, eliminating the need for specialized expertise or extra effort from data engineers in conventional clinical settings. LLM-based agents enable simple and efficient interactions among clinicians, EHR systems, and external tools, simplifying clinical workload and reducing additional burdens when introducing new tools. By advancing emerging AI in personalized medicine, the proposed framework aims to bridge the gap between biomedical research and clinical practice, thereby promoting the adoption of AI in real-world clinical practice.