PhysioNet 2012 challenge

该数据集来自2012年PhysioNet挑战赛，包含了至少住院48小时内的4,000名重症患者数据。数据集包括35种不同的时间序列测量指标和5个一般描述符。大约80.5%的数据存在缺失值，其中死亡标签的分布大约为1:6，具体为554例阳性病例和3,443例阴性病例。该数据集的规模涉及4,000名患者和35项时间序列测量，任务是对院内死亡率进行预测。

This dataset is derived from the 2012 PhysioNet Challenge, containing data from 4,000 critically ill patients who had been hospitalized for at least 48 hours. It includes 35 distinct time-series measurement metrics and 5 general descriptive features. Approximately 80.5% of the dataset contains missing values. The distribution of the in-hospital mortality label is approximately 1:6, with 554 positive cases and 3,443 negative cases respectively. Covering 4,000 patients and 35 time-series measurement metrics, the primary task of this dataset is in-hospital mortality prediction.

搜集汇总

背景与挑战

背景概述

PhysioNet 2012挑战赛数据集包含12,000名ICU患者的记录，旨在预测院内死亡率。数据集提供了42个生理和临床变量，包括一般描述符和时间序列数据，分为训练集和测试集，用于开发患者特异性预测方法。