Barth · PloS one 2020 · Matched case-control study using supervised machine learning · n=166

The path to international medals: A supervised machine learning approach to explore the impact of coach-led sport-specific and non-specific practice.

Cited 18 times in the scientific literature.

Level 4 - case-series / case-control

Retrospective matched case-control study

PubMed 32976547 · doi:10.1371/journal.pone.0239378 · record verified 2026-08-26

What was done

Supervised machine learning classification algorithms (non-ensemble and tree-based ensemble methods) were applied to analyze multivariate, non-linear effects of coach-led practice in main sports versus other sports on international athletic success. The sample comprised 166 adult athletes analyzed as matched pairs (matched on sport, sex, and age) consisting of international medallists and non-medallists.

What was found

eXtreme Gradient Boosting (XGBoost) was identified as the best-performing classification model and showed fair discrimination between medallists and non-medallists. Coach-led practice in other sports up to age 14 was identified as the most important predictive feature. Both main-sport practice and other-sports practice showed non-linear associations with medal success: main-sport practice followed a parabolic pattern, whereas other-sports practice followed a saturation pattern. Specific numerical performance metrics (such as classification accuracy, area under the curve, or exact training hours) were not reported in the abstract.

Why it matters

The findings challenge early sports specialization by showing that multisport coach-led participation up to age 14 is a stronger predictor of senior international podium success than excessive early focus on a single sport.

Limits

The study is observational and retrospective, relying on recall of childhood training history in a modest sample (n = 166). The abstract does not provide quantitative model performance metrics, confidence intervals, or specific training hour thresholds.

Cited by