AI + patient safety

This chapter discusses the AI transformation of patient safety within modern healthcare systems, particularly those digitally extended with telehealth. It carefully details the definitions and debates in patient safety (including the major actors and researchers in this community who share a general critique about the slow to absent sustained, substantive, and equally shared improvements in healthcare systems' safe delivery of patient care). The chapter provides the standard (including WHO) and innovative though still practical conceptualizations of patient safety and then progresses to recent more promising recent advances in human-centered, standardized, and AI-enabled patient safety (including safety as design thinking and system strategy). The chapter illustrates these developments with concrete use cases (in AI-enabled drug safety, clinical reports, and alarms) and augmentation with automation (including embedded, ambient, and command center safety intelligence). The chapter concludes by considering new and growing challenges in AI-driven patient safety (including data security, privacy, bias, and inconsistency) and emerging solutions (including blockchain, bias reduction, reproducibility, explainability, effectiveness, and safety in embedded design).