Abstract: Objective To explore the influencing factors of program for the education & enrichment of relational skills(PEERS) training on the rehabilitation effect of children with autism spectrum disorder (ASD), to construct a Nomogram for predicting rehabilitation effect, and to test the predictive efficiency of the model. Method A total of 146 ASD children who received PEERS training in Guangzhou Women and Children Medical Center from February 2019 to September 2021 were included in the study. The participants were divided randomly into training set (102 cases) and validation set (44 cases) in a ratio of 7∶3, by using the R software. All children were evaluated by the Autism Treatment Evaluation Checklist(ATEC) before and after training, and were divided into good rehabilitation group (ATEC score decline rate ≥30%) and poor rehabilitation group (ATEC score decline rate < 30%) according to the rehabilitation effect. The t-test and χ²test were used to compare the baseline characteristics of the training set and validation set. Univariate Logistic analysis and multivariate Logistic analysis were used to determine the factors influencing the rehabilitation effect of children with ASD in training set. Nomogram was constructed by the RMS packages in R software based on the multivariate Logistic model, and the differentiation and prediction efficiency was evaluated. Results Among 102 ASD children receiving PEERS training, of whom 75 cases (73.53%) recovered well and 27 cases (26.47%) recovered poorly. Multivariate Logistic regression analysis of the training set showed that the initial age of training (OR=1.708, P=0.013), severity of disease (OR=3.040, P=0.034), and no confidence in healing of caregivers (OR=4.265, P=0.013) were risk factors for the rehabilitation effect of ASD children after PEERS training. The receiver operating characteristic (ROC) curve showed that the Nomogram model has good discrimination (AUC=0.791, 95%CI:0.685 - 0.896), and the calibration curve showed that the model has good prediction efficiency. Conclusion The Nomogram prediction model, based on four factors:the severity of the child's condition, the child's age at training onset, the caregiver's education and the caregiver's confidence in healing, has good discriminatory and predictive power for the rehabilitation outcomes of ASD children after group social training.

Key words: group social training, program for the education &enrichment of relational skills, autism spectrum disorder, rehabilitation effect, Nomogram prediction model

摘要： 目的 探讨团体社交训练对孤独症谱系障碍(ASD)儿童康复效果的影响因素,构建预测康复效果的Nomogram模型,并检验该模型的预测效能。 方法 收集2019年2月—2021年9月在广州市妇女儿童医疗中心接受团体社交训练的ASD儿童及其照护者的信息,共146例ASD儿童纳入研究,利用R软件对原始人群进行随机抽样,按7∶3的比例,将研究对象分为训练集(102例)和验证集(44例);所有儿童在接受训练前后采用孤独症治疗评价量表(ATEC)对其开展评价,根据康复效果分成良好组(ATEC评分下降率≥30%)和欠佳组(ATEC评分下降率＜30%)。 使用t检验和χ²检验比较训练集和验证集的基线特征;在训练集中使用单因素Logistic回归分析和多因素Logistic回归分析确定ASD儿童康复效果的影响因素,利用R软件中的RMS包绘制康复效果的Nomogram并评价其区分度及预测效果。 结果 训练集102例ASD儿童接受团体社交训练后,康复良好75例(73.53%),康复欠佳27例(26.47%)。训练集经多因素Logistic回归分析显示,训练起始年龄(OR=1.708,P=0.013)、儿童病情严重程度为重度(OR=3.040, P=0.034)、照护者无治愈信心(OR=4.265, P=0.013)是团体社交训练后ASD儿童康复效果的危险因素。受试者工作特征(ROC)曲线显示Nomogram模型的区分度较好(AUC=0.791,95%CI:0.685~0.896),校准曲线的结果显示模型预测效果较好。 结论 基于儿童病情严重程度、儿童训练起始年龄、照护者学历、照护者有无治愈信心等4项因素构建的Nomogram预测模型对ASD儿童接受团体社交训练后的康复效果具有较好的区分力和预测力。

关键词: 团体社交训练, 社交技能教育和促进项目, 孤独症谱系障碍, 康复效果, Nomogram预测模型