摘要目的 采用视频分析方法分析2～3岁孤独症谱系障碍(ASD)患儿和健康儿童的早期症状差异,为2～3岁ASD患儿的早期筛查提供参考.方法 采用视频分析方法,研究中山大学附属第三医院儿童发育行为中心根据《美国精神障碍诊断与统计手册(第5版)》(DSM-5)确诊的25例ASD患儿(ASD组)和社区招募的21例健康儿童(正常组)的行为,2组儿童分别进行3分钟标准流程的互动,并采集录像信息.采用盲法由2位专业评估人员根据5个项目对录像内容进行评估打分,比较2组儿童项目评分的差异性,并根据评分结果对儿童进行诊断,比较3分钟视频评分诊断和临床诊断的敏感性、特异性.结果 ASD组叫名反应得分[2(2)分]高于正常组[0(0)分],对发声物体的反应得分[0(1)分]高于正常组[0(0)分],社会性微笑得分[1(1)分]高于正常组[0(1)分],总分[4(2)分]高于正常组[2(0)分],差异均有统计学意义(Z=2.272,P=0.000;Z =0.976,P=0.010;Z=1.763,P=0.001;Z =2.355,P=0.000).2组间语言表达和手指指物得分的差异均无统计学意义(均P＞0.05).3分钟视频评估诊断ASD的特异性为85.7％,敏感性为80.0％,3分钟视频评分与DSM-5的诊断结果一致性为0.652(P=0.000).结论 2～3岁ASD儿童在叫名反应、对发声物体的反应和社会性微笑较健康儿童差,通过对3分钟标准化视频分析可以帮助筛查ASD儿童的早期症状,可以用于2～3岁儿童中早期发现ASD儿童.

abstractsObjective To explore the difference in early symptoms between 2-3 years old autism spectrum disorder (ASD) children and healthy children through video analysis,in order to provide evidence for the identification and screening of ASD children.Methods The study involved 25 cases of ASD children who were admitted to Children Development and Behavior Center,the Third Affiliated Hospital,Sun Y at-Sen University,and confirmed with diagnosis standards of Diagnosis and Statistical Manual of Medical Disorder 5th edition (DSM-5) and 21 healthy children recruited in community.Three-minute videos of children in standard procedure were collected.Mter that,all the videos were scored on 5 items by 2 professional evaluators,unaware of diagnostic status.Comparison analysis of video scores between ASD and the healthy controls was made,and the sensitivity and specificity of video analysis were evaluated.Results Response to roll-call score was higher in ASD children [2 (2) scores] than that of the healthy children [0 (0) scores],response to audible object score was higher in ASD children [0(1) scores] than that of the healthy children [0(0) scores],social smiling score was higher in ASD children [1 (1) scores] than that of the healthy children [0 (1) scores],and total score was higher in ASD children [4 (2) scores] than that of the healthy children [2 (0) scores],and the differences were statistically significant (Z =2.272,P =0.000;Z =0.976,P =0.010;Z =1.763,P =0.001;Z =2.355,P =0.000).But,no difference was found in speech and finger pointing (all P ＞ 0.05).The consistency between three-minute video analysis and standard diagnosis was 0.652 (P =0.000),with 80.0％ of sensitivity and 85.7％ of specificity.Conclusions The ASD children perform worse than the healthy children in response to roll-call,response to audible object and social smiling.Three-minute standard video analysis can help to detect the early symptoms of ASD children.This result also demonstrates the potential of video-based analysis used as a ASD screening instrument in 2-3 years old children.