The digital transformation of healthcare has generated a wealth of new information, ranging from electronic health records to patient-generated data. The availability of these data has led to the rapid adoption of artificial intelligence in medicine. Although these algorithms have the potential to guide precision therapies, improve efficiency, and achieve better outcomes, there has been limited regulatory oversight to assess the quality and efficacy of predictive algorithms in the clinical setting. Recent articles by Parikh et al. in Science and Collins and Moons in The Lancet and draft guidance from the Food and Drug Administration have started to explore reporting and regulatory frameworks for predictive algorithms (1–3). Recent draft guidance from the Food and Drug Administration on clinical decision support systems and artificial intelligence in software as a medical device has started …
Schulz et al. (Mon,) studied this question.