Artificial intelligence in healthcare: past, present and future

This paper surveys the state of artificial intelligence in healthcare, framing AI as technology that aims to mimic human cognitive functions and is driving a paradigm shift in medicine thanks to increasingly available healthcare data and rapid advances in analytics. It describes how AI can be applied to both structured and unstructured healthcare data. For structured data, popular techniques include classical machine learning methods such as support vector machines and neural networks, along with modern deep learning; for unstructured data, natural language processing is used. The survey notes that major disease areas employing AI tools include cancer, neurology, and cardiology.

The authors then review AI applications in stroke in more depth, spanning three areas: early detection and diagnosis, treatment, and outcome prediction and prognosis evaluation. They conclude by discussing pioneering AI systems such as IBM Watson and the hurdles that stand in the way of deploying AI in real clinical practice. As a broad, accessible overview, the paper became a widely cited reference point for understanding how AI techniques map onto healthcare data types and clinical use cases.