目的 探索基于图像认知的心理测评技术的复合指标早期变化在抑郁症急性期治疗中的 疗效预测作用。方法 选取 2019 年 1 月至 2020 年 12 月在首都医科大学附属北京安定医院门诊就诊的 135 例抑郁症患者为研究对象。所有患者均接受抗抑郁剂治疗，于基线期、第 8 周末采用 17 项汉密尔顿 抑郁量表（HAMD-17）评估患者的治疗疗效，根据 HAMD-17 减分率将患者分为治疗有效组和治疗无效 组。于基线期、第2周末完成基于图像认知的心理测评，通过测评过程获取眼动轨迹和反应时两类指标， 经过数据预处理提取出可用于分析的特征，采用多因素 Logistic 回归分析构建预测模型，采用受试者工 作特征（ROC）曲线分析复合指标对抑郁症患者抗抑郁剂治疗第 8 周末的疗效预测能力。结果 复合指 标包括认知速度 PN、认知速度 PP。多因素 Logistic 回归分析结果显示，复合指标预测抑郁症患者第 8 周 末治疗有效的 ROC 曲线下面积为 0.628，敏感度为 65.9%，特异度为 55.8%。结论 基于图像认知的心 理测评技术的复合指标可以预测抑郁症患者的急性期治疗结局，但仍有待大规模独立样本验证。
Objective To explore the effect of early changes of composite indicators of image cognition-based psychometric techniques in the treatment of acute depression. Methods From January 2019 to December 2020， a total of 135 patients with depression who were treated in the outpatient clinic of Beijing Anding Hospital Affiliated to Capital Medical University were selected. All patients received antidepressants and assessed for treatment efficacy by Hamilton Depression Scale-17 （HAMD-17） at baseline and the end of 8 weeks. According to the HAMD-17 reduction rate， patients were divided into treatment effective group and treatment ineffective group. The psychological assessment based on image cognition was completed at baseline and the end of 2 weeks. Eye movement trajectory and reaction time were obtained by assessment. The features that could be used for analysis were extracted through data pre-processing， and a prediction model was constructed using multifactorial Logistic regression analysis. It was evaluated that the prediction ability of composite indexes on antidepressant treatment in patients with depression after 8 weeks in receiver operating characteristic （ROC） curve. Results The composite index includes cognitive speed PN and cognitive speed PP. Multifactorial Logistic regression analysis showed that the area under the ROC curve for the composite index to predict effective treatment at the end of week 8 in depressed patients was 0.628， with a sensitivity of 65.9% and a specificity of 55.8%. Conclusions The composite index of psychological assessment technology based on image cognition can be used to predict the treatment outcome of patients with depression in the acute phase， but it still needs to be verified by large-scale independent samples.
DOI :10.3969/j. issn.1009-6574.2023.04.003.