Domain-Specific Knowledge Graphs in RAG-Enhanced Healthcare LLMs

Large Language Models (LLMs) generate fluent answers but can struggle with trustworthy, domain-specific reasoning. We evaluate whether domain knowledge graphs (KGs) improve Retrieval-Augmented Generation (RAG) for healthcare by constructing three PubMed-derived graphs: $\mathbb{G}_{1}$ (T2DM), $\mathbb{G}_{2}$ (Alzheimer's disease), and $\mathbb{G}_{3}$ (AD+T2DM). We design two probes: Probe 1 targets merged AD-T2DM knowledge, while Probe 2 targets the intersection of $\mathbb{G}_{1}$ and $\mathbb{G}_{2}$. Seven instruction-tuned LLMs are tested across retrieval sources [NoRAG, $\left.\mathbb{G}_{1}, \mathbb{G}_{2}, \mathbb{G}_{1}+\mathbb{G}_{2}, \mathbb{G}_{3}, \mathbb{G}_{1}+\mathbb{G}_{2}+\mathbb{G}_{3}\right\}$ and three decoding temperatures. Results show that scope alignment between probe and KG is decisive: precise, scope-matched retrieval (notably $\mathbb{G}_{2}$) yields the most consistent gains, whereas indiscriminate graph unions often introduce distractors that reduce accuracy. Larger models frequently match or exceed KG-RAG with a No-RAG baseline on Probe 1, indicating strong parametric priors, whereas smaller/mid-sized models benefit more from wellscoped retrieval. Temperature plays a secondary role; higher values rarely help. We conclude that precision-first, scopematched KG-RAG is preferable to breadth-first unions, and we outline practical guidelines for graph selection, model sizing, and retrieval/reranking. Code and Data available here https://github.com/sydneyanuyah/RAGComparison