Sleep in patients with disorders of consciousness characterized by means of machine learning DATA

Data underlying findings described in the manuscript: Wielek et al., 'Sleep in patients with disorders of consciousness characterized by means of machine learning'.Subfolders:=> DOC_classif: .csv files for each patient with epoch-wise: classifier predictions, eyes state and day/night period information. Python script perfMetrics4DOC_pl.py computes average classification performance and reproduces Fig.4=> DOC_clustering: .RData files with sampled and everaged epochs (sampling from each subject and averaging separatelly across UWS and MCS). R script hierarchClust_DOC_pl.R computes cluster analysis and plots heatmaps, used to generate Fig.2=> healthy_classif: .txt files containing classifier prediction and ground truth for each healthy. f1weighted4healthyPred_pl.py reproduces Fig.3 based on data: *_nn ending: neuronal networks classification (e.g: 1trueVSpred_nn.txt) *_dummy ending: dummy classification, chance estimate: (e.g.: 1trueVSpred_dummy.txt) *no additional ending: random forest classification: (e.g.: 1trueVSpred.txt) *subfolders; 14electrodes and 26electrodes: classification based on diff. #channels *subfolders; tau1 and tau3: PE computed with different tau parameter

本数据集为Wielek等人发表的题为《采用机器学习方法表征意识障碍患者睡眠》的手稿中的各项研究发现提供数据支撑。附属子文件夹说明如下：
=> DOC_classif：存放每位患者的.csv格式文件，内含逐睡眠分段（epoch-wise）的分类器预测结果、眼动状态以及昼夜时段信息。Python脚本perfMetrics4DOC_pl.py可用于计算平均分类性能，并复现图4。
=> DOC_clustering：包含采样与平均后的睡眠分段数据的.RData格式文件（对每位受试者进行采样后，分别针对无反应觉醒综合征（Unresponsive Wakefulness Syndrome, UWS）与最低意识状态（Minimally Conscious State, MCS）两组群体进行单独平均）。R脚本hierarchClust_DOC_pl.R可执行聚类分析并绘制热图，用于生成图2。
=> healthy_classif：存放每位健康受试者的分类器预测结果与真实标签的.txt格式文件。脚本f1weighted4healthyPred_pl.py可基于以下数据复现图3：
- 后缀为_nn的文件：对应神经网络分类结果（例如：1trueVSpred_nn.txt）
- 后缀为_dummy的文件：对应虚拟分类（即随机猜测的性能估计值，例如：1trueVSpred_dummy.txt）
- 无额外后缀的文件：对应随机森林分类结果（例如：1trueVSpred.txt）
该子文件夹还包含两级子目录：
1. 14electrodes与26electrodes：基于不同电极数量的分类结果数据集
2. tau1与tau3：采用不同tau参数计算得到的PE数据集