Bridging the EHR Divide: Asymmetric Contrastive Learning for Cross-National Medical Representation Transfer

2026-10-04

AI Summary

This study addresses the challenge of transferring longitudinal representations from cross-national electronic health records (EHRs), where differences in coding systems, patient populations, and clinical workflows hinder cross-domain generalization. We propose an asymmetric supervised contrastive learning pre-training objective that uniquely optimizes for clinical outcome heterogeneity without explicitly attracting negative sample trajectories. By integrating a temporal Transformer encoder with a hybrid semantic mapping pipeline, our method enables efficient representation transfer from Taiwanese to U.S. healthcare systems. Evaluated on the MIMIC-IV and EHRSHOT benchmarks, the proposed approach significantly outperforms random initialization and substantially narrows the performance gap with in-domain pre-training. Furthermore, it achieves higher average AUPRC than standard contrastive learning and direct supervised transfer, effectively enhancing cross-domain few-shot predictive capabilities.

Abstract

Cross-system transfer of longitudinal Electronic Health Record (EHR) representations is challenging because clinical coding, patient populations, and healthcare workflows differ substantially across institutions and countries. We introduce Asymmetric Supervised Contrastive Learning (Asymmetric SupCon), a task-specific pre-training objective motivated by the heterogeneity of negative clinical outcomes. The objective clusters patients sharing a target positive outcome without explicitly attracting negative trajectories toward one another. We pre-train temporal Transformer encoders on longitudinal records from 3.98 million patients in the Taiwanese National Health Insurance Research Database (NHIRD) and transfer them to two U.S. EHR datasets, MIMIC-IV and EHRSHOT. A hybrid semantic mapping pipeline combining direct mappings with embedding-based retrieval enables transfer across heterogeneous clinical vocabularies. On MIMIC-IV, NHIRD pre-training consistently improves over random initialization while substantially narrowing the performance gap to task-specific in-domain pre-training. On EHRSHOT, the transferred models show particularly strong few-shot performance for incident disease prediction. A controlled objective ablation shows that Asymmetric SupCon achieves the best AUPRC on three of four evaluated tasks and is 0.003 AUPRC below Standard SupCon on the fourth. These results support asymmetric contrastive pre-training as an effective approach for task-specific cross-national EHR representation transfer. Code is available at https://github.com/qingYzhang/Asymmetric_SupCon.

Problem

• Electronic Health Record
• Cross-national transfer
• Representation learning
• Clinical heterogeneity

Innovation

• Asymmetric Contrastive Learning
• Electronic Health Record
• Cross-National Transfer
• Temporal Transformer
• Semantic Mapping