Calibration-First Cross-Cohort Multimodal Temporal Learning for Transferable Asthma-Risk Forecasting
Calibration-First Cross-Cohort Multimodal Temporal Learning for Transferable Asthma-Risk Forecasting
首选校准的跨队列多模态时间学习：用于可迁移的哮喘风险预测
Abstract: Asthma deterioration forecasting must remain reliable when patient populations, sensor ecosystems, and available modalities change across cohorts. Existing models commonly optimize within-cohort discrimination and may produce poorly calibrated probabilities after transfer.
摘要： 当患者群体、传感器生态系统和可用模态在不同队列间发生变化时，哮喘恶化预测必须保持可靠。现有的模型通常优化队列内的判别能力，但在迁移后可能会产生校准性较差的概率结果。
We present CALIBRA, a calibration-first multimodal temporal framework for short-horizon risk prediction with incomplete data. Dedicated recurrent encoders process environmental, pulmonary, symptom, medication, wearable, and context streams; a reliability-conditioned gate suppresses stale or absent modalities, while gradient-reversal training discourages avoidable cohort signatures.
我们提出了 CALIBRA，这是一个以校准为先的多模态时间框架，用于处理不完整数据的短期风险预测。专用的循环编码器处理环境、肺功能、症状、药物、可穿戴设备和上下文数据流；可靠性条件门控机制抑制了陈旧或缺失的模态，而梯度反转训练则抑制了可避免的队列特征。
A shrinkage-based hierarchical logistic layer calibrates probabilities using a patient-disjoint target subset, and split conformal prediction provides abstention-capable prediction sets. To avoid fabricating clinical evidence, we evaluate the complete implementation on a documented three-cohort semi-synthetic benchmark with controlled distribution shift, informative missingness, and sealed target patients.
一个基于收缩的分层逻辑回归层利用患者不相交的目标子集来校准概率，而拆分共形预测则提供了具备弃权能力的预测集。为了避免捏造临床证据，我们在一个有记录的三队列半合成基准测试上评估了完整的实现，该基准测试具有受控的分布偏移、信息性缺失和密封的目标患者。
Across five configured seeds, CALIBRA achieved mean target-test AUPRC 0.224 versus 0.240 for the strongest non-ablation comparator, TemporalTransformer; mean AUROC was 0.717, and Brier score was 0.098. Experiments additionally assess complete-modality failures, calibration, conformal coverage, decision curves, subgroup behavior, ablations, runtime, and parameter count.
在五个配置种子下，CALIBRA 的平均目标测试 AUPRC 为 0.224，而最强的非消融对比模型 TemporalTransformer 为 0.240；平均 AUROC 为 0.717，Brier 分数为 0.098。实验还评估了完全模态故障、校准、共形覆盖、决策曲线、子组行为、消融研究、运行时间和参数数量。
The results verify the method and reproducible pipeline under controlled shift, but do not establish clinical effectiveness. External validation on harmonized real asthma. Overall this artifact provides evidence for carefully governed real-cohort validation.
研究结果验证了该方法和可重复流水线在受控偏移下的有效性，但尚未确立其临床有效性。目前正在进行针对统一真实哮喘数据的外部验证。总体而言，该研究成果为经过严格管理的真实队列验证提供了证据。