ClarusC64/clinical-microcirculation-instability-v0.1

该数据集用于评估模型是否能够从简短的临床代理轨迹中检测组织灌注的不稳定性。每一行代表一个简化的临床场景，描述了三个时间点的宏观循环信号和组织灌注指标。任务是判断系统是否保持微循环稳定或趋向于组织灌注衰竭。数据集的核心思想是微循环不稳定性可能在传统宏观血流动力学指标看似正常的情况下发展。数据集包含多个代理变量（如MAP轨迹、乳酸轨迹、毛细血管再充盈轨迹等）和干扰变量（如监测噪声、图表噪声）。预测目标是判断微循环不稳定性是否正在形成（label=1）或组织灌注稳定（label=0）。数据集是Clarus稳定性推理基准的一部分，支持微循环功能障碍、组织氧输送衰竭、轨迹临床推理等研究领域。

This dataset evaluates whether models can detect instability in tissue-level perfusion from short clinical proxy trajectories. Each row represents a simplified clinical scenario observed across three time points describing macro-circulatory signals together with indicators of tissue perfusion. The task is to determine whether the system remains microcirculatory stable or is moving toward tissue-level perfusion failure. The core stability idea is that microcirculatory instability often develops even when traditional macro-hemodynamic indicators appear acceptable. The dataset includes proxy variables (e.g., MAP trajectory, lactate trajectory, capillary refill trajectory) and decoy variables (e.g., monitor noise, chart noise). The prediction target is to determine whether microcirculatory instability is emerging (label=1) or tissue perfusion is stable (label=0). This dataset is part of the Clarus Stability Reasoning Benchmark and supports research into microcirculatory dysfunction, tissue oxygen delivery failure, trajectory-based clinical reasoning, etc.