ClarusC64/clinical-respiratory-drive-instability-v0.1

该数据集用于评估模型是否能检测呼吸控制中的不稳定性。每一行代表一个简化的呼吸调节场景，包含三个时间点的数据。任务是根据呼吸率轨迹、CO₂代理轨迹、呼吸驱动代理轨迹、氧饱和度轨迹、代谢需求代理和干预延迟等变量，判断呼吸调节是否稳定（标签0表示稳定，1表示不稳定）。数据集还包括干扰变量如实验室噪声和图表噪声。评估指标包括准确率、精确率、召回率、F1分数、混淆矩阵和数据集完整性诊断。数据集反映了通过可观察代理变量体现的潜在稳定性几何结构，是Clarus稳定性推理基准的一部分。

This dataset evaluates whether models can detect instability in respiratory control. Each row represents a simplified respiratory regulation scenario observed across three time points. The task is to determine whether ventilatory regulation remains stable or is moving toward respiratory drive instability (label 0 for stable, 1 for unstable) based on variables such as respiratory rate trajectory, CO₂ proxy trajectory, respiratory drive proxy trajectory, oxygen saturation trajectory, metabolic demand proxy, and intervention delay. The dataset also includes decoy variables like lab_noise and chart_noise. Evaluation metrics include accuracy, precision, recall, f1, confusion matrix, and dataset integrity diagnostics. The dataset reflects latent stability geometry through observable proxies and is part of the Clarus Stability Reasoning Benchmark.