---
title: "Patient-Visit-Spanned Hypergraph Learning for EHR-based Diagnosis Prediction"
authors:
  - "Ye Yuan"
  - "Haiyan Wang"
  - "Lun Hu"
  - "Xin Luo"
doi: "10.24963/ijcai.2026/780"
url: "https://ijcai.org/proceedings/2026/780"
pdf: "https://ijcai.org/proceedings/2026/0780.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "7010-7018"
track: "AI and Health"
keywords:
  - "EHR analysis: EHR analysis"
  - "Medical diagnosis: Medical diagnosis"
  - "Clinical decision support system: Clinical decision support system"
---

# Patient-Visit-Spanned Hypergraph Learning for EHR-based Diagnosis Prediction

Ye Yuan, Haiyan Wang, Lun Hu, Xin Luo

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 7010-7018. DOI: [10.24963/ijcai.2026/780](https://doi.org/10.24963/ijcai.2026/780)

## Abstract

Hypergraphs effectively model complex interactions in structured Electronic Health Records (EHR). Consequently, Hypergraph Neural Networks (HGNNs) are commonly applied to EHR-based diagnosis prediction. However, existing HGNNs struggle to capture patient-visit long-range dependencies when processing EHR-derived hypergraphs. To tackle this issue, we propose a Patient-Visit-Spanned HyperGraph Learning (PVHGL) framework specifically designed for diagnosis prediction. Concretely, PVHGL initially constructs a unified patient-visit hypergraph that integrates visit records from all patients, enabling the capture of shared healthcare patterns across the patient population. Subsequently, it incorporates a Transformer architecture enhanced by two structure-encoding matrices to facilitate one-step message propagation, which preserves both local and global hypergraph structural properties effectively. Additionally, the framework integrates a medical code co-occurrence matrix to explicitly guide the learning process by highlighting critical medical code interactions. Comprehensive experiments on three real-world datasets demonstrate that the proposed PVHGL significantly outperforms state-of-the-art baselines in diagnosis prediction.