Diagnosing autism severity associated with physical fitness and gray matter volume in children with autism spectrum disorder: Explainable machine learning method.

PURPOSE: This study aimed to investigate the relationship between physical fitness, gray matter volume (GMV), and autism severity in children with autism spectrum disorder (ASD). Besides, we sought to diagnose autism severity associated with physical fitness and GMV using machine learning methods. METHODS: Ninety children diagnosed with ASD underwent physical fitness tests, magnetic resonance imaging scans, and autism severity assessments. Diagnosis models were established using extreme gradient boosting (XGB), random forest (RF), support vector machine (SVM), and decision tree (DT) algorithms. Hyperparameters were optimized through the grid search cross-validation method. The shapley additive explanation (SHAP) method was employed to explain the diagnosis results. RESULTS: Our study revealed associations between muscular strength in physical fitness and GMV in specific brain regions (left paracentral lobule, bilateral thalamus, left inferior temporal gyrus, and cerebellar vermis I-II) with autism severity in children with ASD. The accuracy (95 % confidence interval) of the XGB, RF, SVM, and DT models were 77.9 % (77.3, 78.6 %), 72.4 % (71.7, 73.2 %), 71.9 % (71.1, 72.6 %), and 66.9 % (66.2, 67.7 %), respectively. SHAP analysis revealed that muscular strength and thalamic GMV significantly influenced the decision-making process of the XGB model. CONCLUSION: Machine learning methods can effectively diagnose autism severity associated with physical fitness and GMV in children with ASD. In this respect, the XGB model demonstrated excellent performance across various indicators, suggesting its potential for diagnosing autism severity.

Predicting academic performance associated with physical fitness of primary school students using machine learning methods.

OBJECTIVES: This study aimed to explore the relationship between physical fitness and the academic performance of primary school students and to predict the academic performance associated with physical fitness using machine learning methods. The results provide new evidence confirming the relationship between physical fitness and the academic performance of primary school students. This study provides a practical foundation for early intervention methods to improve the physical fitness and academic performance of primary school students via physical exercise. METHODS: A total of 432 fifth-grade students from five primary schools in Huai'an, China, were selected using the cluster sampling method. Their physical fitness was evaluated in terms of their body mass index, muscle strength, flexibility, speed, and aerobic endurance. The final exam scores in Chinese, mathematics, and foreign language were used to quantify their academic performance. The Mann-Whitney U test was used to investigate the differences in physical fitness between academic performance groups. The Spearman correlation analysis was used to quantify the relationship between physical fitness and academic performance. Machine learning models based on random forest (RF), support vector machine (SVM), and K-nearest neighbor (KNN) algorithms were used to predict the academic performance of primary school students. The respective prediction performances of machine learning models were evaluated using the accuracy and validated in the test sample. RESULTS: The body mass index (z = -2.046, p < 0.05) of high-score (HS) primary school students was lower than non-high-score (NHS) students, and the upper limb (z = -2.143, p < 0.05), trunk (z = -3.399, p < 0.05), and lower limb strength (z = -2.525, p < 0.05) and aerobic endurance (z = -2.105, p < 0.05) of HS students were better than NHS students. The academic performance of primary school students was negatively correlated with body mass index (r = -0.105, p < 0.05) and positively correlated with upper limb (r = 0.11, p < 0.05), trunk (r = 0.175, p < 0.05), and lower limb strength (r = 0.13, p < 0.05) and aerobic endurance (r = -0.108, p < 0.05). The average accuracy of RF, SVM, and KNN models in predicting the academic performance of primary school students in training samples were 59.4% ± 5.16%, 56.41% ± 3.81% and 57.89% ± 4.98%, respectively, which were found to be higher than baseline accuracy, as validated in the test sample. CONCLUSION: The body mass index, muscle strength, and aerobic endurance of primary school students are significantly different between academic performance groups and are correlated with their academic performance. Machine learning methods can effectively predict academic performance associated with the physical fitness of primary school students.