Constructing a clinical knowledge graph from electronic health records for enhanced decision-making and disease diagnosis

The increasing complexity of clinical data presents both challenges and opportunities for modern healthcare. This study proposes a robust framework for building a Clinical Knowledge Graph (KG) by leveraging unstruc tured Electronic Health Records (EHRs) and clinical notes. Using state-of-the-art natural language processing tools such as MetaMap and the Unified Medical Language System (UMLS), the proposed system structures het erogeneous medical data into a unified format. By analyzing demographic, symptomatic, and laboratory data, this framework enables enhanced decision-making and insights into disease correlations. Demonstrated using the MIMIC-III database, the system achieves high granularity, providing actionable intelligence for personalized recommendations and supporting predictive diagnostic models.