Abstract
Artificial Intelligence (AI) is transforming healthcare by enabling advanced solutions in diagnostics, treatment planning, and healthcare management. This systematic review, conducted using the PRISMA framework, analyses existing studies to examine AI applications in healthcare, focusing on methodologies, datasets, performance, and challenges. Deep learning techniques such as convolutional neural networks (CNNs) for medical imaging and recurrent neural networks (RNNs) for disease progression prediction have shown strong results in radiology, oncology, clinical decision support, and drug discovery, improving diagnostic accuracy and personalized care. Supervised learning methods, including support vector machines (SVMs) and random forests, are commonly used for diagnostic tasks with labeled data, while unsupervised and semi-supervised approaches are applied when annotated datasets are limited. Hybrid models combining structured and unstructured data are increasingly adopted. Despite promising results, challenges remain, such as the lack of large, high-quality annotated datasets, data imbalance, privacy concerns, limited model generalization, and the black-box nature of deep learning models. Ethical issues and integration into clinical workflows further hinder adoption. Addressing these gaps is essential to enable trustworthy AI systems and improve real-world healthcare outcomes.
Keywords: Artificial Intelligence, Clinical Decision Support Systems, Explainable AI, Medical Imaging, Natural Language Processing.