Comment on "Black Box Prediction Methods in Sports Medicine Deserve a Red Card for Reckless Practice: A Change of Tactics is Needed to Advance Athlete Care"

Abstract

We read with interest the article by Bullock et. al. [1] on the applicability of Machine Learning (ML) approaches in sports medicine settings. Overall, we agree with the central thrust of this paper-without sufficient interpretability, ML models are not useful in the sports injury domain. As the authors correctly state, black box models can cause problems. If we cannot establish how a model is making its decisions, and the predictive power is low, there is no value to using these models.However, we feel there are areas of research within ML that have been overlooked which strongly temper the conclusions that the authors make. In the following, we look at how these are being addressed by the ML community.