Abstract:
Objective To establish a multi-index method for the diagnosis of exercise intensity, and provide a methodological reference for accurately diagnosing exercise intensity of obese children and adolescents.
Methods 160 obese children and adolescents aged 10~17 were randomly divided into training group (n=120) and validation group (n=40). The METs, %HRR and vector magnitudes (VM3) were measured for all participants during free walking and running at 3~7 km/h. Logistic regression was used to establish the multi-index joint diagnosis model, and the ROC curve was conducted with the prediction probability values (P3METs and P6METs), and the cut-off point of intensity was determined by the maximum Youden index. The area under the ROC curve was evaluated for diagnostic value, accuracy and weighted Kappa values were tested for accuracy and consistency.
Results The 3METs multi-index joint diagnosis model was Logist(P3METs)=10.666-9.760×10-2×%HRR-1.035×10-3×VM3-0.290×Age, while the 6METs was Logist(P6METs)=-15.34+0.127×%HRR+6.316×10-4×VM3+0.176×Age. The cut-off point of 3METs for P3METswas 0.315, and the area under curve of ROC was 0.891.The cut-off point of 6METs for P6METswas 0.248, and the area under curve of ROC was 0.956. The accuracy of multiple-index joint diagosis was 79.62%, and the weighted Kappa value was 0.686.
Conclusion Multiple-index joint diagnosis model can be used to diagnose the exercise intensity of obese children and adolescents, which has high diagnostic value.The diagnostic criteria of light-intensity was P3METs>0.315, moderate-intensity P3METs ≤ 0.315 and P6METs < 0.248, and high-intensity P6METs ≥ 0.248.