ClarusC64/clinical-perfusion-instability-v0.1

该数据集用于评估模型是否能检测组织灌注的不稳定性。每个数据行代表一个简化的循环灌注场景，观察三个时间点。任务目标是判断组织灌注是保持稳定还是趋向于不稳定。数据集包含多个信号轨迹，如平均动脉压轨迹、毛细血管再充盈代理轨迹、灌注指数代理轨迹、乳酸轨迹、代谢需求代理和干预延迟等变量。预测目标为二元分类，1表示灌注不稳定，0表示稳定。此外，还包含了一些干扰变量如实验室噪声和图表噪声。数据集是Clarus稳定性推理基准的一部分。

This dataset evaluates whether models can detect instability in tissue perfusion. Each row represents a simplified circulatory perfusion scenario observed across three time points. The task is to determine whether tissue perfusion remains stable or is moving toward perfusion instability. The dataset includes multiple signal trajectories such as mean arterial pressure trajectory, capillary refill proxy trajectory, perfusion index proxy trajectory, lactate trajectory, metabolic demand proxy, and intervention delay. The prediction target is binary classification, where 1 indicates perfusion instability and 0 indicates stable perfusion. Additionally, it includes decoy variables such as lab noise and chart noise. This dataset is part of the Clarus Stability Reasoning Benchmark.