Objective:To compare knowledge graphs (KGs) constructed from standardized clinical pathways and actual examination records within 24 hours of emergency care for acute gastrointestinal hemorrhage (AGH), acute myocardial infarction (AMI), and intracerebral hemorrhage (ICH), and to visually analyze discrepancies between guideline recommendations and real-world practice, thereby exploring a novel methodology for clinical pathway optimization.Methods:KGs were developed using clinical pathway standards and actual examination data collected within the first 24 hours of emergency treatment for AGH, AMI, and ICH. Entity attributes were weighted to visually represent the frequency and extent of examination usage through variable node sizes in the KG. The constructed KGs were used to compare and analyze the differences in type and frequency of examinations performed relative to pathway standards.Results:The proportion of examination items with >50% adherence to clinical pathway standards within 24 hours was 76.92% for AGH, 44.44% for AMI, and 78.57% for ICH. Items from the clinical pathways that were not performed in over 50% of patients accounted for 15.38%, 27.78%, and 21.43% of cases, respectively. Non-pathway examinations increased by 9, 7, and 4 items for each condition, of which 17 items (85%) were performed at least once in more than half of the patients. Visualization via KGs revealed a reduction in redundant examinations by 38.64% between AGH and AMI, 35.00% between AGH and ICH, and 37.50% between AMI and ICH. Overall, a 54.84% reduction in redundant examinations was achieved across all three critical conditions.Conclusions:The visual KG approach effectively integrates both guideline-recommended and experience-driven examinations, serving as a correlational analysis tool to assess deviations between actual clinical practice and standardized pathways. It provides a quantitative foundation for optimizing clinical pathways, with potential for greater efficiency gains as more critical conditions are incorporated into the graph.