Hybrid Manhattan Distance-Certainty Factor for Early Cardiovascular Diagnosis in a Hospital

Abstract

Cardiovascular disease drives high mortality and operational strain in public hospitals, underscoring the need for tools that standardize rapid early decisions. We evaluated a hybrid expert system that integrates Manhattan Distance (MD) for case-based similarity with a Certainty Factor (CF) framework for rule-based evidence aggregation. Using a locally curated knowledge base (110 cases, 69 symptoms, 13 conditions) and a 26-case hold-out against specialist references, the system retrieves nearest cases via MD on symptom vectors and then computes per-diagnosis confidence with CF. The system achieved 23/26 exact matches (accuracy 88.46%), with confidence values spanning 71.90–99.99% higher when nearest-case patterns and rules converged and moderated in ambiguous presentations (e.g., suspected aneurysm). Outputs were interpretable and suitable for 15–30-minute consultations, supporting consistent triage where specialist capacity is limited. These findings suggest a practical pathway to improve timeliness and reduce variability. Future work should pursue multi-site validation, knowledge-based expansion for atypical phenotypes, and governance for safe, equitable deployment.