临床和认知特征指标预测注意缺陷多动障碍药物疗效  

Prediction of drug response in attention deficithy peractivity disorder based on clinical and cognitive characteristics

作　　者：苏怡[1] 王业鑫 杨莉[1] 曹庆久[1] 孔庆梅[1] 王玉凤[1] 杨智 SU Yi;WANG Yexin;YANG Li;CAO Qingjiu;KONG Qingmei;WANG Yufeng;YANG Zhi(Peking University Sixth Hospital,Peking University Institute of Mental Health,NHC Key Laboratory of Mental Health(Peking University),National Clinical Re search Center for Mental Disorders(Peking University Sixth Hospital),Beijing 100083,China;Computer Science,Peking University,Beijing 100871,China)

机构地区：[1]北京大学第六医院,北京大学精神卫生研究所,国家卫生健康委员会精神卫生学重点实验室(北京大学),国家精神心理疾病临床医学研究中心(北京大学第六医院),北京100083 [2]北京大学计算机学院,北京100871

出　　处：《中国心理卫生杂志》2022年第3期211-216,共6页Chinese Mental Health Journal

基　　金：国家自然科学基金青年科学基金项目(81701348);杨森科学研究基金(CON-I-07-CN-029-B)。

摘　　要：目的:采用机器学习模型,探讨预测哌甲酯控释片(OROS MPH)与托莫西汀(ATX)治疗注意缺陷多动障碍(ADHD)疗效的临床和认知特征指标。方法:纳入符合DSM-IV诊断标准的6~16岁ADHD患者237例,1∶1随机接受OROS MPH或ATX治疗,以缓解作为结局指标。以基线ADHD评定量表、临床总体印象量表、中国韦氏儿童智力量表、执行功能测查结果为特征,分别构建两种药物的疗效预测模型。结果:临床症状较轻、C因子智商较高、执行功能计划和抑制受损较轻的患者,OROS MPH疗效更好,疗效预测模型准确率81.3%;而对于ATX,B因子智商较高、执行功能启动和抑制受损较轻的患者,疗效更好,预测模型准确率80.7%。结论:患者基线智商水平、执行功能受损程度是OROS MPH与ATX疗效的重要预测指标,同时,临床症状严重程度也是OROS MPH疗效的预测指标。Objective:Using the machine learning model to explore the clinical and cognitive predictors for the response of osmotic release oral system methylphenidate(OROS MPH) and atomoxetine(ATX) in the treatment of attention deficit hyperactivity disorder(ADHD).Methods:Totally 237 patients with ADHD from 6 to 16 years old who met the diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition(DSM-IV) were included, and randomly divided into two groups receiving OROS MPH or ATX treatments, respectively. "Remission" was taken as the response outcome index.The ADHD Rating Scale-IV,Clinical Global Impressions-ADHD-Severity(CGI-ADHD-S),Chinese-Wechsler Intelligence Scale for Children(C-WISC),and Executive Function Test at baselines were treated as features to construct the predictive model.Results:Patients with less severe clinical symptoms, higher C factor IQ,and less impaired executive function on planning and inhibition, had better response to OROS MPH,and the accuracy of the response prediction model was 81.3%.For ATX,patients with higher B factor IQ,less impaired executive function on initiation and inhibition had better response, and the accuracy of the prediction model was 80.7%.Conclusion:The patient’ s baseline IQ level and the degree of impaired executive function are important predictors of the response of OROS MPH and ATX.In addition, the severity of clinical symptoms is also a predictor of the response of OROS MPH.

关 键 词：疗效预测 注意缺陷多动障碍 机器学习 哌甲酯 托莫西汀 

分 类 号：R749.94[医药卫生—神经病学与精神病学]