EPI-Eval/flusight-forecast-hub

FluSight预测中心流感住院目标数据集是一个用于流感预测模型评分基准的真实数据集合。该数据集记录了美国从2022年2月5日到2026年4月18日期间每周的流感相关住院情况，更新频率为每周一次。数据覆盖国家层面和州层面（共53个地理位置），包含两个核心指标：每周确认的流感住院人数（作为FluSight评分目标）和每10万人口每周住院率。该数据集派生自美国国家医疗安全网络（NHSN）的医疗相关感染数据（nhsn-hrd），但可能比原始NHSN数据延迟一个周期，因为预测中心需要固定真实值用于评分。数据集主要用于评估流感预测模型的准确性，是FluSight预测中心生态系统的组成部分。

The FluSight Forecast Hub — flu hospital admission target dataset is a ground truth collection used as a scoring benchmark for influenza forecasting models. It records weekly influenza-related hospital admissions in the United States from February 5, 2022 to April 18, 2026, with a weekly cadence. The data covers national and state levels (53 unique geographic locations) and includes two core metrics: confirmed flu hospital admissions per week (serving as the FluSight scoring target) and weekly admission rate per 100,000 population. Derived from the National Healthcare Safety Network (NHSN) Healthcare-associated Infections data (nhsn-hrd), this dataset may lag NHSN by one cycle as the hub freezes truth values for scoring purposes. Primarily used to evaluate the accuracy of influenza forecasting models, it forms an integral part of the FluSight Forecast Hub ecosystem.