UCI dataset

The Cuff-Less Blood Pressure Estimation Dataset [2] from the UCI Machine Learning Repository. It is a subset of the MIMIC-II Waveform Dataset that contains 12000 records of simultaneous PPG and ABP from 942 patients with a sampling rate of 125 Hz. The 12000 records were uniformly split into four parts with 3000 records each. However, as the subject information is lacking, the Hold-one-out strategy was utilized to generate training, validation, and test sets once the data was preprocessed. In the end, the UCI dataset had 291,078 segments, which was around 404 hours of recording, making it substantially the biggest data set with a considerably higher ratio of continuous segments per record (32.15). [2] Kachuee, M., Kiani, M. M., Mohammadzade, H. & Shabany, M. Cuff-less blood pressure estimation data set (2015). UCI repository https://archive.ics.uci.edu/ml/datasets/Cuff-Less+Blood+Pressure+Estimation.

来自UCI机器学习库（UCI Machine Learning Repository）的无袖带血压估计数据集（Cuff-Less Blood Pressure Estimation Dataset）[2]，是MIMIC-II波形数据集（MIMIC-II Waveform Dataset）的子集。该数据集包含来自942名患者的12000条同步光电容积描记（Photoplethysmography，简称PPG）与有创动脉血压（Arterial Blood Pressure，简称ABP）记录，采样率为125 Hz。该12000条记录最初被均匀划分为4个部分，每部分包含3000条记录。但由于缺失受试者相关信息，在完成数据预处理后，研究人员采用留一法（Hold-one-out）生成训练集、验证集与测试集。最终，该UCI数据集共包含291078个数据段，总记录时长约404小时，是目前规模最大的同类数据集，且单条记录对应的连续数据段比例显著更高（32.15）。[2] Kachuee, M., Kiani, M. M., Mohammadzade, H. & Shabany, M. 无袖带血压估计数据集（2015）. UCI机器学习库 https://archive.ics.uci.edu/ml/datasets/Cuff-Less+Blood+Pressure+Estimation.