Classification and use of explainable artificial intelligence in disease diagnosis

This study aims to evaluate classification algorithms on datasets used in disease diagnosis and to increase the explainability of these classifications with the LIME method, one of the explainable artificial intelligence techniques. In the study, commonly used machine learning models such as Logistic Regression, K-Nearest Neighbor, Support Vector Machines, Decision Trees and Random Forest were used on datasets belonging to Alzheimer's, Heart Failure, Breast Cancer and Cirrhosis diseases. The performances of the models were evaluated with metrics such as accuracy, precision, recall, F1 score and ROC AU. Using the LIME method, one of the explainable artificial intelligence techniques, the factors behind each classification decision were analyzed and understandable explanations were presented to healthcare professionals. The results of the study show that models such as Random Forest and Logistic Regression exhibit high performance, but LIME method provides significant contributions to each model in terms of explainability. In future studies, it is aimed to further deepen the studies in this field by using different explainable artificial intelligence techniques on larger and more diverse data sets.