Prediction of Marathon Performance using Artificial Intelligence.

Although studies used machine learning algorithms to predict performances in sports activities, none, to the best of our knowledge, have used and validated two artificial intelligence techniques: artificial neural network (ANN) and k-nearest neighbor (KNN) in the running discipline of marathon and compared the accuracy or precision of the predicted performances. Official French rankings for the 10-km road and marathon events in 2019 were scrutinized over a dataset of 820 athletes (aged 21, having run 10 km and a marathon in the same year that was run slower, etc.). For the KNN and ANN the same inputs (10-km race time, body mass index, age and sex) were used to solve a linear regression problem to estimate the marathon race time. No difference was found between the actual and predicted marathon performances for either method (p>0,05). All predicted performances were significantly correlated with the actual ones, with very high correlation coefficients (r>0,90; p<0,001). KNN outperformed ANN with a mean absolute error of 2,4 vs 5,6%. The study confirms the validity of both algorithms, with better accuracy for KNN in predicting marathon performance. Consequently, the predictions from these artificial intelligence methods may be used in training programs and competitions.

Predictions of the Distance Running Performances of Female Runners Using Different Tools.

This study examined the validity and compared the precision and accuracy of a distance-time linear model (DTLM), a power law and a nomogram to predict the distance running performances of female runners. Official rankings of French women ("senior" category: between 23 and 39 years old) for the 3000-m, 5000-m, and 10,000-m track-running events from 2005 to 2019 were examined. Performances of runners who competed in the three distances during the same year were noted (n=158). Mean values and standard deviation (SD) of actual performances were 11.28±1.33, 19.49±2.34 and 41.03±5.12 for the 3000-m, 5000-m, and 10,000-m respectively. Each performance was predicted from two other performances. Between the actual and predicted performances, only DTLM showed a difference (p<0.05). The magnitude of the differences in these predicted performances was small if not trivial. All predicted performances were significantly correlated with the actual ones, with a very high correlation coefficient (p<0.001; r>0.90), except for DTLM in the 3000-m, which showed a high correlation coefficient (p<0.001; r>0.895). Bias and 95% limits of agreement were acceptable because, whatever the method, they were≤-3.7±10.8% on the 3000-m, 1.4±4.3% on the 5000-m, and -2.5±7.4% on the 10,000-m. The study confirms the validity of the three methods to predict track-running performance and suggests that the most accurate and precise model was the nomogram followed by the power law, with the DTLM being the least accurate.

Prediction of Distance Running Performances of Female Runners Using Nomograms.

This study examined the validity, precision and accuracy of the predictions of distance running performances in female runners from three nomograms. Official rankings of French women for the 3000-m, 5000-m, and 10 000-m track-running events from 2005 to 2019 were examined. Only female runners who performed in the three distance events within the same year were included (n=158). Each performance over any distance was predicted using the three nomograms from the two other performances. The 3000-m, 5000-m and 10 000-m performances were 11min17 s±1min20 s, 19min29 s±2min20 s, 41min18 s±5min7 s, respectively. No difference was found between the actual and predicted running performances regardless of the nomogram (p>0.05). All predicted running performances were significantly correlated with the actual ones, with a very high correlation coefficient (p<0.001; r>0.90). Bias and 95% limits of agreement were acceptable because, whatever the nomogram, they were less than or equal to - 0.0±6.2% on the 3000-m, 0.0±3.7% on the 5000-m, and 0.1±9.3% on the 10 000-m. The study confirms the validity of the three nomograms to predict track-running performance with a high level of accuracy. The predictions from these nomograms are similar and may be used in training programs and competitions.