This paper proposes a medication recommendation method that utilizes the hypergraph approach to capture cross-patient relationships in Electronic Health Records.
Overall architecture of HypeMed. HypeMed comprises two stages: the Medical Entity Relevance Representation Stage (MedRep) and the Similar Visit Enhanced Medication Recommendation Stage (SimMR). MedRep focuses on encoding intra-visit set-level combinatorial semantics into a globally consistent, retrieval-friendly embedding space. SimMR integrates longitudinal history and visit-conditioned retrieved similar visits to refine latent condition estimation.
AI-driven medication recommendation aims to generate safe and effective medication combinations from patients’ electronic health records (EHRs). However, accurately recommending medications hinges on inferring a patient’s latent clinical condition from sparse and noisy observations, which requires both (i) preserving the visit-level combinatorial semantics of co-occurring diagnoses/procedures and (ii) leveraging informative historical references through effective, visit-conditioned retrieval. Most existing methods fall short in one of these aspects: graph-based modeling often fragments higher-order intra-visit patterns into pairwise relations, while inter-visit augmentation methods commonly exhibit an imbalance between learning a globally stable representation space and performing dynamic retrieval within it. To address these limitations, this paper proposes HypeMed, a two-stage hypergraph-based medication recommendation framework that unifies intra-visit coherence modeling with inter-visit reference augmentation. In the first stage, MedRep represents each visit as a hyperedge and performs knowledge-aware hypergraph contrastive pre-training to encode high-order interactions into a globally consistent, retrieval-friendly embedding space. In the second stage, SimMR conducts visit-conditioned dynamic retrieval directly in this hyperedge-aware space and fuses the retrieved references with the patient’s longitudinal history to refine condition estimation and improve medication prediction. Experiments on MIMIC-III/IV and eICU demonstrate that HypeMed consistently improves recommendation accuracy while reducing the drug–drug interaction (DDI) rate, indicating enhanced effectiveness and safety. The implementation is publicly available at https://github.com/xansar/HypeMed.
This project contains the necessary python scripts for HypeMed as well as the directions.
Considering that unauthorized public access to the MIMIC-III and MIMIC-IV databases is prohibited to distribute, we only provide the example data (first 100 entries of records of MIMIC-III and MIMIC-IV) in data. You can use our provided scripts to preprocess the raw data once you have obtained the relevant data.
Python==3.8.36
Torch==1.13.1+cu116
NumPy==1.24.4
Below is a guide on how to use the scripts. Before processing, please change the 'pathtomimic' in HypeMed.py to the real path.
Please download drug-DDI.csv, which a large file, containing the drug DDI information. Please put it in data. The file could be downloaded from https://drive.google.com/file/d/1mnPc0O0ztz0fkv3HF-dpmBb8PLWsEoDz/view?usp=sharing
# Data Processing.
python data/processing.py # MIMIC-III
python data/processing_4.py #MIMIC-IV
# Contrastive Learning Pretraing (on MIMIC-III)
python HypeMed.py --pretrain --pretrain_epoch 3 --pretrain_lr 1e-3 --pretrian_weight_decay 1e-5 --mimic 3 --name example
# debug
python HypeMed.py --debug --mimic 3 --pretrain_epoch 3 --name example
# Training
python HypeMed.py --mimic 3 --pretrain_epoch 3 --name example
# Testing
python HypeMed.py --mimic 3 --pretrain_epoch 3 --Test --name example
# ablation study
python HypeMed.py --mimic 3 --pretrain_epoch 3 --channel_ablation only_his --name exampleYou can explore all adjustable hyperparameters through the config.py file.
This project was inspired by and partially built upon several excellent open-source repositories. We sincerely thank the original authors for making their code publicly available.
Special thanks to:
Their valuable implementations greatly helped the development of this project.