MIT-BIH Arrhythmia Database

该数据集是对MIT-BIH数据库的一个平衡子集，它包含了健康和异位心跳的数据，专为心电图信号分类的训练和测试而设计。为确保数据的有效性和公正性，该数据集被划分为训练集、验证集和测试集，确保了来自同一位患者的样本不会在训练集和测试集之间混合。具体规模为：训练集包含8520个样本，测试集包含6440个信号。该数据集的任务是进行心电图信号的分类。This dataset is a balanced subset of the MIT-BIH database, which contains data of both healthy and ectopic heartbeats, and is specifically designed for the training and testing of electrocardiogram (ECG) signal classification tasks. To guarantee data validity and impartiality, this dataset is split into training, validation, and test sets, with the strict constraint that samples from the same patient will not be mixed between the training and test sets. The specific scale of the dataset is as follows: the training set includes 8,520 samples, while the test set contains 6,440 ECG signals. The core task of this dataset is ECG signal classification.

搜集汇总

背景与挑战

背景概述

MIT-BIH Arrhythmia Database是一个经典的心律失常心电图数据库，包含48个半小时的双通道动态心电图记录，来自47名受试者，数据采集于1975年至1979年，采样率为360 Hz，具有11位分辨率和约110,000次心跳的专家注释。该数据库自1980年发布以来，广泛用于心律失常检测算法的评估和心脏动力学研究，是生物医学信号处理领域的标准基准数据集。