Can a large language model tell whether a clinical trial was reported properly? Across 113 sports medicine trials, GPT-4 got it right about 90% of the time. Published in BMJ Open, and presented at Stanford’s METRICS International Forum.