PhysioNet 2019

该数据集包含了来自三个不同医院系统的重症监护室（ICU）患者的临床数据，旨在利用这些数据对脓毒症进行早期检测。该数据集已经过处理，以固定的小时为单位进行离散化，并且大约有80%的数据缺失。数据集的规模涉及40,336名患者，任务是预测在临床预测前6小时内脓毒症的发作。

This dataset collects clinical data of intensive care unit (ICU) patients from three distinct hospital systems, and is aimed at leveraging these data for early sepsis detection. The dataset has been preprocessed and discretized at fixed hourly intervals, with approximately 80% of the data missing. It includes a total of 40,336 patients, and the core predictive task is to forecast the onset of sepsis within 6 hours prior to the clinical prediction timestamp.

搜集汇总

背景与挑战

背景概述

PhysioNet 2019数据集是一个用于早期预测败血症的临床数据集，包含来自三个医院系统的40,336名ICU患者的多项生理和实验室测量数据。该数据集的目标是通过算法分析这些数据，提前6小时预测败血症的发生，从而改善患者预后。