When coaches at FC Barcelona Femení wanted to protect their players from getting hurt, they faced a puzzle that has troubled sports medicine for years: How do you actually predict when an injury will happen before it does? Now, an international team of researchers may have found an answer — and the secret lies in understanding that the longer a player works, the higher their risk climbs.
The study, published in the journal npj Digital Medicine, was led by the Barcelona Institute for Global Health (ISGlobal) and the University of Bonn in Germany, with help from FC Barcelona's own medical team, the Barça Innovation Hub, and a company called Made of Genes. Together, they built a new computer-based system that uses artificial intelligence to estimate injury risk in elite women's soccer.
Most existing AI models for injury prediction have big problems, the researchers found. They do not account for the fact that risk grows the more a player trains or competes. They also treat a small muscle tweak the same as a serious torn ligament, which does not help doctors make good decisions. And the probability numbers these models spit out often do not match what actually happens in real life.
So instead of just building another model, the team created an entirely new framework. Their approach treats injury prediction as a