Artificial Intelligence in Healthcare: A Comprehensive Review of Data-Driven Approaches and Public Datasets

Abstract

Artificial intelligence (AI) is steadily growing in importance as an essential part of the health industry because it helps make it more data-driven and supports more informed and personalized healthcare practices. The following paper will discuss the history of the use of artificial intelligence technologies in healthcare, focusing on computational techniques, application fields, and the use of public datasets for conducting research. It will provide a brief introduction into AI technology, namely, into the main approaches such as machine learning, deep learning, reinforcement learning, and hybrid AI technologies and show how they are used in medicine. The use of large public healthcare data sets like MIMIC-IV, PhysioNet, AmsterdamUMCdb, UK Biobank, and mental health data is also reviewed. Moreover, some major obstacles to the adoption process include data heterogeneity, incomplete data, biases, lack of interpretability, privacy problems, and difficulties in deploying ML models into clinical settings. Finally, ethical and legal aspects, such as fairness, transparency, and data privacy, are also covered. Lastly, the review provides future research directions and focuses on explainable AI, federated learning, multimodal data integration, real-time AI systems, and human-centered design. Overall, the article shows that despite the enormous potential of AI to transform healthcare, its effectiveness in the long run is possible to be attained with the help of trustworthy models, high-quality data, accountable governance, and adequate integration into real healthcare practices.