ClarusC64/clinical-drug-interaction-instability-v0.1

该数据集用于评估模型是否能检测由药物相互作用引起的不稳定性。每个数据行代表三个时间点的简化药理学轨迹，任务是确定药物组合是否保持稳定或趋向于相互作用驱动的不稳定性。数据集测试了药物A、B、C的轨迹、肝功能和肾清除代理、代谢能力、相互作用指数和纠正延迟等因素。预测目标是标记药物相互作用不稳定性（1）或稳定的药理学轨迹（0）。数据集中还包含了一些干扰变量，如监控噪声和图表噪声，这些变量看起来有意义但不单独决定标签。评估方法包括准确率、精确率、召回率、F1分数和混淆矩阵。数据集是Clarus稳定性推理基准的一部分。This dataset evaluates whether models can detect instability caused by interacting medications. Each row represents a simplified pharmacologic trajectory across three time points. The task is to determine whether the drug combination remains stable or is moving toward interaction-driven instability. The dataset tests interaction reasoning across drug A, B, C trajectories, liver function proxy, renal clearance proxy, metabolism capacity, interaction index, and correction delay. The prediction target is label = 1 for drug interaction instability or label = 0 for stable pharmacologic trajectory. Decoy variables like monitor_noise and chart_noise are included but do not determine the label alone. Evaluation metrics include accuracy, precision, recall, f1, and confusion matrix. This dataset is part of the Clarus Stability Reasoning Benchmark.