AI System Recommends Pathological Tests, Improving Diagnostic Efficiency

Accurate and timely medical diagnoses are crucial for patient care, yet delays in test recommendations and subjective interpretations by physicians can impede this process. This research presents an innovative pathological test recommendation system designed to streamline diagnostic test selection. The system utilizes patient symptoms to suggest appropriate tests even before a physician's initial consultation. The core of this system is a multi-label classification approach, employing the Classifier Chain (CC) technique to account for dependencies between various diagnostic tests. Researchers built a custom dataset from SOUTHERN.IML pathology data and applied several machine learning algorithms. The Logistic Regression model combined with CC achieved the highest accuracy, while an ensemble model offered the best balance of precision, recall, and F1-score. To ensure transparency and clinical trust, the system incorporates Explainable AI (XAI) using SHAP values. This allows the model to reveal how each symptom contributes to a test recommendation, providing diagnostic reasoning that aligns with established medical knowledge. This interpretability can assist physicians in making more logical and confident decisions, particularly in critical scenarios, ultimately enhancing the efficiency of the diagnostic process.