Reproducibility Problems Found in Cancer Research
December 10, 2021
In a new study, researchers repeated 50 early-stage cancer experiments reported in major journals a decade ago, but they were unable in roughly half the cases to reproduce the initial scientific claims. Indeed, irreproducibility is a problem across biomedical research, Shannon Brownlee and Bibiana Bielekova write in Issues. As a remedy, the authors propose a tool that uses artificial intelligence to assess the methodological rigor, reproducibility, and utility of studies. Its widespread use, they argue, “can help transform the scientific enterprise into a self-regulating system that increases the rate at which it produces scientific breakthroughs that have meaningful impacts for society.”