数字化、人工智能技术辅助早期筛查和早期诊断儿童孤独症谱系障碍

Research on the application of digitalization and artificial intelligence technologies to assist in the early screening and diagnosis for autism spectrum disorder in children

目的探索使用数字化、人工智能预诊系统辅助早期筛查、早期诊断儿童孤独症谱系障碍(autism spectrum disorder,ASD)的可行性,建立ASD“社区初筛-区妇幼复筛-三甲医院诊断”的三级筛查诊断模式。方法采用分层随机抽样的方法,从广州市白云区全区23个街镇按经济水平高中低随机抽取3个街镇为调查点,于2022年1月1日至2022年12月31日,在儿童保健/预防接种服务时,使用数字化、人工智能预诊系统对调查点内所有18~30月龄儿童进行ASD网络初级筛查;初级筛查阳性者转诊至广州市白云区妇幼保健院,使用CHAT⁃23量表B部分、儿心量表-Ⅱ和S-S语言发育迟缓评价法进行面对面复筛;复筛阳性病例转诊至三甲医院使用DSM⁃V进行确诊。结果共纳入3175例儿童,初筛阳性率为12.15%(386/3175);复筛阳性率为1.45%(46/3175);最终确诊ASD 39人,其中男童28人,女童11人;ASD整体患病率为1.23%。男童现患率为1.58%,女童现患率0.78%,差异有统计学意义(χ2=4.15,P=0.04)。结论应用数字化、人工智能预诊系统辅助早期筛查、早期诊断ASD儿童具有良好的临床应用价值;应推动建立ASD“社区初筛-区妇幼复筛-三甲医院诊断”的三级筛查诊断模式。

Objective To explore the feasibility of employing a digital,artificial intelligence preliminary diagnosis system to aid in early screening and diagnosis of autism spectrum disorder in children,and establishing a three⁃tier screening and diagnostic process:primary screening in community,re⁃screening in maternity and child health hospital of district,and diagnosis for ASD in top three hospitals.Methods A stratified random sampling method was used to randomly select 3 streets from 23 streets of Baiyun district based on economic levels.A digital artificial intelligence preliminary diagnosis system was employed during child healthcare/preventive vaccination services to conduct initial ASD network screening for all children aged 18⁃30 months in the selected survey points from January 1,2022,to December 31,2022.Children who tested positive in the primary screening were referred to the Child Mental Health Department of Baiyun District Maternal and Child Health Hospital for re⁃screening with face⁃to⁃face behavior observation,CHAT⁃23 Scale Part B,Developmental scale for children aged 0⁃6 years and S⁃S method.Positive cases in the re⁃screening phase were referred to top three hospitals for diagnosis based on the criterion of Diagnostic and Statistical Manual of Mental Disorders,Fifth Edition(DSM⁃V).Results A total of 3175 cases of children were investigated.The primary screening positive rate was 12.15%and 46(1.45%)were positive after being re⁃screened.Totally 39 autism cases(28 boys and 11 girls)were diagnosed and the overall autism prevalence rate was 1.23%;the prevalence rate among the boys was significantly higher than that among the girls(1.58%vs 0.78%,χ²=4.15;P=0.04).Conclusion The prevalence rate of children autism in Guangzhou Baiyun District is in the higher level.The application of digital and artificial intelligence preliminary diagnosis system has been proved to have significant clinical utility in assisting early screening and diagnosis of ASD in children.It is advisable to promote the establishment of a three⁃tier screening and diagnostic process for ASD.