This study presents an enhanced semantic knowledge graph (KG)– based framework designed to support diagnostic reasoning through improved knowledge representation and inference. Unlike traditional KGs that rely on static, binary relations, our approach incorporates probabilistic edge weights derived from Likelihood Ratios (LR+ and LR−), enabling a more nuanced and evidence-based representation of symptom–disease relationships. These weighted edges are further combined with conditional relations reflecting patient-specific attributes, resulting in a personalized and dynamic KG structure. The inference algorithm is refined to leverage these enriched representations, supporting more accurate and efficient ranking of diagnostic hypotheses under time constraints. By integrating semantic reasoning with fuzzy and probabilistic representations, this work advances the state of knowledge modeling for clinical contexts. It contributes to the development of explainable and patient-centered inference mechanisms aligned with the principles of Healthcare 4.0.