The integration of Artificial Intelligence (AI) in healthcare has revolutionized medical diagnostics by improving accuracy, efficiency, and clinical decision- making. This paper presents a comprehensive review of AI- based medical diagnosis systems, emphasizing their end-to- end workflow—from user input interfaces that collect and preprocess patient data, to AI models that extract features and select optimal architectures for diagnosis, and finally to output modules that deliver interpretable predictions and actionable recommendations. Deep Learning, an integral part of these systems, excels at analyzing complex medical data (e.g., imaging, electronic health records, and genomic profiles), while Predictive Analytics enables forecasting of disease progression and treatment outcomes. We evaluate these technologies across medical domains such as oncology, cardiology, and radiology, highlighting their clinical applicability and performance. The review also addresses ethical challenges, including data privacy, model interpretability, and the importance of user feedback loops for continuous model improvement. By synthesizing recent advancements and identifying research gaps, we propose future directions for developing robust, transparent, and ethically responsible AI frameworks that integrate seamlessly into healthcare workflows—ensuring alignment with the iterative, patient centric process depicted in modern system architectures.
Sharvary et al. (Tue,) studied this question.