Using Case-Based Causal Reasoning to Provide Explainable Counterfactual Diagnosis in Personalized Sprint Training

Intelligent sport training (IST) is an urgent need for both professional athletes and ordinary people, but personalized intelligent diagnosing is still lacking. In this paper we proposed a personalized sports training diagnosis framework called CBCR, which combines CBR with causal inference in 2 ways: 1) In the case selection stage, the traditional distance metric is replaced by the weighted distance based on causal effect; 2) In the counterfactual diagnosis stage, the solution is the counterfactual training effects estimated from the individual causal model. We developed a set of sprint diagnosis algorithms on a very small case base, and evaluated it on data from both Olympic candidates and college students, and by an Olympic final case study as well.