发表机构
Westcliff University; Northern University Bangladesh; Stanton University; Washington University of Science and Technology（韦斯特克利夫大学; 孟加拉国北方大学; 斯坦顿大学; 华盛顿科技大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对ICU死亡率预测中不规则临床观测建模问题，提出CT-HEG架构并实例化为CHIRP-Net，经MIMIC-IV v3.1数据集实验，验证了双向连接、时间注意力边特征等架构选择的作用，为不规则电子健康记录数据建模提供了有效方案。
AI 中文摘要
准确的ICU死亡率预测需要对跨异构实体类型的不规则临床观测进行建模。现有序列模型可处理不规则采样，但忽略了类型化的关系结构；现有图模型则假设输入为固定间隔。我们提出了连续时间异构电子健康记录图（Continuous-Time Heterogeneous EHR Graph, CT-HEG）架构，并评估哪些架构选择会影响预测性能。CT-HEG将每一次ICU住院编码为带类型、带时间戳的图，包含三种节点类型（就诊、生命体征、检验事件）以及二维边属性（t_hours/48、标准化值），无需插补即可编码时间和数值信息。我们将CT-HEG实例化为CHIRP-Net，这是一个四层异构GATv2Conv网络，在MIMIC-IV v3.1数据集（包含31142次ICU住院，住院时长≥48小时，死亡率为13.4%）上进行五次随机种子实验，并采用自助法计算置信区间，与逻辑回归、mTAND、Transformer、GRU-D及消融模型进行对比。CHIRP-Net的五次随机种子平均AUROC为0.8449±0.0071（AUPRC为0.4958±0.0209）；集成模型的AUROC为0.8618（95%置信区间：0.8485-0.8745）。移除反向边会使观测节点与就诊读数断开连接，导致AUROC下降0.1968±0.0073；带时间注意力的边特征贡献了0.0247±0.0093的AUROC；将异构边类型合并为单一关系（参数减少7倍）在所有随机种子上的表现均优于完整模型；校准后的预期校准误差（ECE）为0.0307。我们探索了时间和人口统计学亚组分析，但暂未在此报告，待后续研究跟进。双向连接是模型使用输入的必要条件，且经验证拟合的温度缩放后，CT-HEG的校准效果良好。这些结果支持CT-HEG适用于不规则电子健康记录数据，但在声称其稳健性之前，仍需进行外部验证、预先指定的时间评估以及人口统计学公平性审核。代码：此https URL。
英文摘要
Accurate ICU mortality prediction requires modeling irregular clinical observations across heterogeneous entity types. Existing sequence models handle irregular sampling but ignore typed relational structure; existing graph models assume fixed-interval inputs. We introduce the Continuous-Time Heterogeneous EHR Graph (CT-HEG) schema and evaluate which architectural choices drive predictive performance. CT-HEG encodes each ICU stay as a typed, timestamped graph with three node types (visit, vital, lab_event) and 2D edge attributes (t_hours/48, value_norm) encoding timing and value without imputation. We instantiate CT-HEG as CHIRP-Net, a four-layer heterogeneous GATv2Conv network, evaluated on MIMIC-IV v3.1 (31,142 ICU stays, LOS>=48h, 13.4% mortality) with five seeds and bootstrapped confidence intervals, against logistic regression, mTAND, a Transformer, and GRU-D, plus an ablation study. CHIRP-Net achieved 5-seed mean AUROC 0.8449+/-0.0071 (AUPRC 0.4958+/-0.0209); the ensemble achieved AUROC 0.8618 (95% CI: 0.8485-0.8745). Removing reverse edges disconnected observation nodes from the visit readout, cutting AUROC by 0.1968+/-0.0073. Time-attentive edge features contributed 0.0247+/-0.0093 AUROC. Collapsing heterogeneous edge types into one relation (7x fewer parameters) outperformed the full model on all seeds. Post-calibration ECE was 0.0307. Temporal and demographic subgroup analyses were explored but not reported here, pending follow-up work. Bidirectional connectivity was necessary for the model to use its inputs at all, and CT-HEG was reasonably well calibrated after validation-fitted temperature scaling. These results support CT-HEG for irregular EHR data, while external validation, a pre-specified temporal evaluation, and a demographic fairness audit remain necessary before any claim of robustness. Code: https://github.com/nasiruddinstudents-ctrl/chirp-net-mimic-iv.
