The increasing complexity of patient care demands a shift toward personalized clinical decision-making. Current systems rely on correlation-based methods and generalized, one-size-fits-all guidelines that fail to address individual patient characteristics, limiting their effectiveness and trustworthiness. Furthermore, clinicians are burdened with unstructured data like clinical notes, which remain underutilized due to inefficiencies in extracting actionable insights. This lack of precision, combined with the growing workload of clinicians, hinders optimal patient outcomes and contributes to systemic inefficiencies. This research advances personalized decision support through three key aims: (1) advancing treatment strategies from generalized to individualized care via causal inference, (2) enhancing personalized diagnosis through data-driven insights (3) integrating LLMs with causal reasoning to generate interpretable, data-driven insights. By combining these approaches, this work empowers clinicians with precise recommendations, reduces cognitive burden, and optimizes healthcare workflows, driving innovation in evidence-based medicine. Merging causal reasoning and LLM capabilities ensures not only the ability to process complex unstructured data but also to provide context-aware, interpretable insights essential for personalized and evidence-based care. This dual capability enhances decision-making in resource-constrained settings, structures critical causal information for improved clinical documentation, and supports treatment efficacy and surgical outcome analysis.