📄 中文摘要
中风每年影响数百万人的健康，但症状识别不佳常导致就医延误。为了解决风险识别的缺口，开发了一种被动监测系统，旨在通过糖尿病患者自报的症状实现早期中风风险检测。构建了基于患者自身语言的症状分类法，并采用双重机器学习流程（异构图神经网络和弹性网/LASSO），识别与后续中风相关的症状模式。将研究结果转化为一种混合风险筛查系统，整合了症状的相关性和时间接近性，通过基于电子健康记录的模拟在3至90天的窗口内进行评估。在故意设计的保守阈值下，筛查系统成功实现了减少误报的目标。
以患者为中心的图增强人工智能被动监测系统用于高风险个体早期中风风险检测
📄 English Summary
Patient-Centered, Graph-Augmented Artificial Intelligence-Enabled Passive Surveillance for Early Stroke Risk Detection in High-Risk Individuals
Stroke affects millions annually, yet poor symptom recognition often leads to delays in seeking care. To bridge the risk recognition gap, a passive surveillance system was developed for early stroke risk detection using patient-reported symptoms among individuals with diabetes. A symptom taxonomy grounded in patients' own language was constructed, alongside a dual machine learning pipeline (heterogeneous GNN and EN/LASSO), to identify symptom patterns associated with subsequent strokes. Findings were translated into a hybrid risk screening system that integrates symptom relevance and temporal proximity, evaluated through EHR-based simulations across 3-90 day windows. Under conservative thresholds, intentionally designed to minimize false alerts, the screening system achieved significant results.