Application of the certainty factor method in an expert system for diagnosing diabetes mellitus based on clinical symptoms

Abstract

Diabetes Mellitus is a chronic metabolic disease characterized by elevated blood glucose levels due to impaired insulin secretion or function, which can lead to serious complications such as cardiovascular disease, kidney failure, and neuropathy if not detected early. Early detection remains a significant challenge, particularly in regions with limited access to laboratory-based diagnostic facilities, where many cases are only identified after the disease has progressed to a more serious stage. This study applies the Certainty Factor method as an expert system approach to diagnose Diabetes Mellitus based on patients' clinical symptoms. The dataset used was the Early Stage Diabetes Risk Prediction Dataset from the UCI Machine Learning Repository, consisting of 520 patient records with 14 binary clinical symptom attributes. The CF Expert value for each symptom was derived from the proportion of symptom occurrence among positive diabetes cases, and the combined certainty scores were used to classify each patient as diabetic or non-diabetic based on an optimized threshold value. Several threshold values ranging from 0.50 to 0.95 with an interval of 0.05 were systematically tested to determine the optimal decision boundary. The experimental results demonstrate that a threshold of 0.95 yielded the best classification performance, achieving an Accuracy of 79.81%, Precision of 79.61%, Recall of 90.31%, and F1-Score of 84.63%. The high Recall value of 90.31% indicates that the method is highly effective in minimizing undetected positive cases, which is critically important in a medical screening context. Furthermore, the Certainty Factor method demonstrated advantages in interpretability and computational efficiency, as diagnostic decisions can be directly traced to individual symptom contributions without requiring a model training process, making it suitable for implementation in clinical decision support systems for early diagnosis of Diabetes Mellitus.