基于体质健康的多指标联合ROC曲线法建立儿童青少年24 h活动行为推荐量的探索性研究  

An Exploratory Research on Establishing a 24—Hour Recommended Amount of Movement Behaviors for Children and Adolescents by Multiple-Index Joint ROC Curve Based on Physical Health

作　　者：孙毅 刘媛[2] 尹小俭 吴慧攀[4] 李明[5] 张婷 张凤[7] 郭亚茹 孙鹏伟 洪俊[7] SUN Yi;LIU Yuan;YIN Xiaojian;WU Huipan;LI Ming;ZHANG Ting;ZHANG Feng;GUO Yaru;SUN Pengwei;HONG Jun(Ludong University,Yantai 264025,China;Shanghai University,Shanghai 200444,China;Shanghai Institute of Technology,Shanghai 201418,China;Taiyuan Institute of Technology,Taiyuan 030008,China;Xizang Minzu University,Xianyang 712082,China;Suzhou University,Suzhou 234111,China;East China Normal University,Shanghai 200062,China)

机构地区：[1]鲁东大学,山东烟台264025 [2]上海大学,上海200444 [3]上海应用技术大学,上海201418 [4]太原工业学院,山西太原030008 [5]西藏民族大学,陕西咸阳712082 [6]宿州学院,安徽宿州234111 [7]华东师范大学,上海200062

出　　处：《中国体育科技》2024年第4期10-18,28,共10页China Sport Science and Technology

基　　金：国家社会科学基金一般项目(21BTY121)。

摘　　要：目的:采用多指标联合ROC曲线方法对24 h活动行为的推荐量进行探索性研究,并与其他方法的推荐量进行比较。方法:在全国7个地区选取435名儿童青少年为研究对象,采用加速度计测量身体活动,问卷调查视屏时间(screen time,ST)和睡眠(sleep,SLP),从身体成分、心肺耐力、柔韧性、力量、速度和协调灵敏性等方面评价体质健康水平。使用成分数据的分析方法对24 h活动行为进行描述和变换。采用线性回归、二元逻辑回归和ROC曲线法探索24 h活动行为推荐量。采用准确率和Kappa值检验准确性和一致性,采用协方差分析比较满足不同项目推荐量的体质健康水平差异。结果:1)控制年龄、性别和家庭社会经济状况后,非视屏静坐(non-screen-time sedentary behavior,Non-ST SB)与体质健康呈负相关,低强度身体活动(light-intensityphysicalactivity,LPA)和中高强度身体活动(moderate to vigorous-intensity physical activity,MVPA)与体质健康呈正相关;2)多指标联合的预测模型ROC曲线下面积为0.710,高于所有单项指标且有统计学意义。联合指标预测值最佳切点值为0.38,对应的SLP为568.49 min/d,ST为130.00 min/d,Non-ST SB为584.92 min/d,LPA为99.92 min/d,MVPA为56.67 min/d;3)多指标联合预测的效果最好,准确率为88.72%,精确率为78.43%,召回率为90.91%,Kappa一致性系数为0.76。满足多指标联合推荐量的被试体质健康评分最高,且呈现随着满足项目数的增多,体质健康评分逐渐上升的趋势(Ptrend<0.001)。结论:研究采用的成分数据结合多指标联合ROC曲线法准确性和一致性较高,满足该方法的推荐量与较高的体质健康水平有关。Objective:To conduct an exploratory research of the recommended amount of 24-hour movement behaviors using multiple-index joint ROC curve and comparing it with the amount of traditional methods.Methods:A total of 435 children and adolescents in seven regions of China were recruited to measure physical activity with accelerometer.Questionnaires were used to measure screen time(ST)and sleep time(SLP).Physical health level was evaluated in terms of body composition,cardiorespiratory endurance,flexibility,strength,speed,and agility.The 24-hour movement behaviors were described and transformed using the analysis of component data.Linear regression,logistic regression and ROC curve was used to explore the recommended amount of 24-hour movement behaviors.Accuracy and Kappa value were used to test for accuracy and consistency,and analysis of covariance was used to compare differences in physical health levels meeting the recommended amounts of different items.Results:1)After controlling for age,gender,and socioeconomic status,non-screen-time sedentary behavior(Non-ST SB)was negatively correlated with physical health,and low intensity physical activity(LPA)and moderate to vigorous-intensity physical activity(MVPA)were positively correlated with physical health;2)the area under the ROC curve of the prediction model combined with multiple-index was 0.710,which was higher than all single indices and statistically significant.The best cut-off value of the prediction value of multiple-index was 0.38,corresponding to 568.49 min/d for SLP,130.00 min/d for ST,584.92 min/d for Non-ST SB,99.92 min/d for LPA and 56.67 min/d for MVPA;3)the multiple-index joint prediction was the best:accuracy was 88.72%,precision was 78.43%,recall was 90.91%,and Kappa consistency coefficient was 0.76.Subjects who met the recommended amount of multiple-index had the highest physical health scores and showed a gradual increase in scores as the number of meeting items increased(Ptrend<0.001).Conclusions:Multiple-index joint ROC curve method had high acc

关 键 词：多指标联合 儿童青少年 24 h活动行为 推荐量 

分 类 号：G804.49[文化科学—运动人体科学]