Despite promises that Artificial Intelligence (AI) will transform health care, the development and adoption of AI in health care has lagged behind other industries. Some of the causes for this lag include restrictions on the use of health care data, resistance from the clinical community, the gap between hype and reality of AI, ethical concerns, regulation of health technologies, and difficulties bridging the cultures of health care and engineering. Yet despite spectacular failures such as Watson Health, AI is slowly beginning to appear in health care settings, most often in the context of research, but increasingly in the form of FDA and Health Canada-approved products. The aim of this course is to build a critical understanding of end-to-end lifecycle of AI in health care, from working with raw health care data, to integration of AI with clinical workflow, through to regulatory approval. This course will be of particular interest to translational AI researchers looking to apply their work to health care, as well as health care practitioners and informaticians seeking to understand how to leverage AI in their industry.

Objectives: upon completion of the course, students will be able to: 1) Describe the opportunities for AI to provide value in health care and how to engage with health care providers to identify real clinical problems that AI can address. 2) Explain the regulatory and ethical frameworks that govern AI development in health care, including a working understanding of the governance of personal health information, the regulation of software as a medical device, and fairness in AI. 3) Develop roadmaps for how AI can be integrated into the health care environment, including clinical decision support design, health care interoperability paradigms, and clinically relevant model explanations. 4) Develop practical knowledge of the idiosyncrasies of health care data types, how to define AI research questions in health care, how to interface with hospital information systems, how to satisfy privacy and security requirements during AI development, and how to transition AI research projects into production.