Learning at the Speed of AI: Using Platform Trials to Test AI Innovation

Artificial intelligence has the potential to transform healthcare, but to see this upside we need to be able to know which products work best. Clinical trials were built to evaluate technologies that are stable over time. AI updates frequently, meaning that if we rely on old evaluation approaches, evidence will be outdated before a study is complete. Head-to-head clinical trials that continuously run in multiple countries would allow us to generate evidence that keep up with the pace of innovation. This needs to be combined with an approach that makes it easier to collect evidence because if standards are too high, promising technologies will never reach patients. This paper argues that standing research infrastructure through platform trials or clinical trial networks would greatly reduce both the cost and time needed to evaluate AI, making it easier to run comprehensive studies. Further work is now needed to determine where these models add most value, how they should be funded and governed, and whether they are acceptable to regulators, developers, and health systems.