An Explainable Artificial Intelligence Framework for Medical Diagnosis

Artificial intelligence (AI) has recently gained prominence in healthcare. It is increasingly applied in disease detection and clinical decision-making. Although the results generated by such applications are highly accurate, many of them exhibit limited transparency. Most often, they operate as a "black box," i.e., while their output invisible, no clear explanation exists for how a particular result was reached. Such opacity is counterproductive in healthcare since trust and accountability are key considerations. To tackle the problem, the concept of Explainable Artificial Intelligence (XAI) has emerged. XAI refers to research direction dedicated to increasing the transparency of models to enable users to trace decisions. In this paper, we suggest a framework in which a traditional machine learning algorithm is combined with explain ability techniques to achieve an effective balance between accuracy and interpretability. We compare conventional and explainable artificial intelligence algorithms using different evaluation criteria such as accuracy, the level of interpretability, clinician trust, and time spent validating the output. Based on the analysis of our experiments, it becomes evident that the explainable AI model remains sufficiently accurate (approximately87%) but increases the visibility of the decision-making process. Furthermore, it saves time needed for decision validation. In general, the research reveals that the integration of explain ability helps increase the practical application of AI-based solutions to healthcare practice. It becomes possible for AI-powered tools not only to present the output but also assist doctors in comprehending and relying on it.