📄 中文摘要
基于变换器的模型通过大规模自监督预训练在纵向电子健康记录的预测建模中取得了显著进展。然而，大多数电子健康记录变换器架构将每次临床就诊视为无序的代码集合，这限制了它们捕捉就诊内有意义关系的能力。图形变换器方法旨在通过建模就诊级结构，同时保留学习长期时间模式的能力，来解决这一限制。GT-BEHRT作为一种图形变换器架构，在MIMIC-IV重症监护结果和“我们所有人”研究项目中的心力衰竭预测中进行了评估。对报告的性能提升进行了审查，以确定其是否真实反映了模型的有效性。
纵向电子健康记录预测中图形变换器的转化差距：GT-BEHRT的批判性评估
📄 English Summary
Translational Gaps in Graph Transformers for Longitudinal EHR Prediction: A Critical Appraisal of GT-BEHRT
Transformer-based models have significantly advanced predictive modeling on longitudinal electronic health records (EHRs) through large-scale self-supervised pretraining. However, most EHR transformer architectures treat each clinical encounter as an unordered collection of codes, which limits their ability to capture meaningful relationships within a visit. Graph-transformer approaches aim to address this limitation by modeling visit-level structure while retaining the capacity to learn long-term temporal patterns. GT-BEHRT, a graph-transformer architecture, has been evaluated on MIMIC-IV intensive care outcomes and heart failure prediction in the All of Us Research Program. A critical review of the reported performance gains is conducted to assess whether they genuinely reflect the model's effectiveness.