Personalized Prediction of rTMS Clinical Response in Medication-Refractory Depression Data

This article describes a dataset that was generated as part of the article: Personalized prediction of transcranial magnetic stimulation clinical response in patients with treatment-refractory depression using neuroimaging biomarkers and machine learning (DOI: 10.1016/j.jad.2021.04.081). We collected resting-state functional Magnetic Resonance Imaging data from 70 medication-refractory depressed subjects before undergoing four weeks of repetitive transcranial magnetic stimulation targeting the left dorsolateral prefrontal cortex. The data presented here include information about the seed-based analyses to identify connectivity biomarkers of treatment response, such as regions of interest/seeds, individual/group functional connectivity maps and second-level analysis contrast maps. The contrast maps are controlled for age, gender, duration of the current depressive episode, duration since the first depressive episode, and symptom scores. Demographics, clinical characteristics, and categorical treatment response variables are reported as well. Further, the individual connectivity values of the identified neuroimaging biomarkers of long-term clinical response were used as features in the support vector machine models are presented in combination with the trained classifiers of the support vector machine models. Post hoc analyses that were not published in the original analyses are presented as well. Finally, the R or MATLAB code scripts for all figures published in the co-submitted paper are included.本文介绍了一项数据集，该数据集依托于论文《利用神经影像生物标志物与机器学习对难治性抑郁症患者经颅磁刺激临床应答进行个性化预测》（DOI: 10.1016/j.jad.2021.04.081）生成。本研究从70名药物难治性抑郁症受试者中，在其接受为期四周、靶向左侧背外侧前额叶皮层的重复经颅磁刺激前，采集了静息态功能磁共振成像（resting-state functional Magnetic Resonance Imaging）数据。本次公开的数据包含用于识别治疗应答相关连接生物标志物的基于种子点的分析信息，具体包括感兴趣区/种子点、个体/组水平功能连接图以及二级分析对比图。上述对比图已针对年龄、性别、当前抑郁发作时长、首次抑郁发作至今时长以及症状评分进行了协变量校正。此外还公开了受试者的人口学特征、临床特征以及分类变量形式的治疗应答相关变量。进一步地，本研究还公开了经识别的长期临床应答相关神经影像生物标志物的个体连接数值，该数值曾作为支持向量机（support vector machine）模型的特征，并与训练完成的支持向量机分类器一同呈现。同时还公开了原始分析中未发表的事后检验分析结果。最后，本数据集还包含与共同投稿论文中所有图表对应的R语言或MATLAB代码脚本。