Pretrained patient trajectories for adverse drug event prediction using common data model-based electronic health records
The CDM-based EHR pretraining model with DE can improve prediction performance for various ADEs and can provide proper explanation at cohort and patient level. Our model has the potential to serve as a foundation model due to its strong prediction performance, interpretability, and compatibility.
Authors
- Chaeeun Cho
- Taehyun Kim (ORCID: https://orcid.org/0000-0002-4478-667X)
- Joo Seong Kim (ORCID: https://orcid.org/0000-0002-0411-9805)
- Ji‐Hyang Lee (ORCID: https://orcid.org/0000-0003-4286-3114)
- Rae Woong Park (ORCID: https://orcid.org/0000-0003-4989-3287)
- Junmo Kim (ORCID: https://orcid.org/0000-0002-3908-0488)
- Kwangsoo Kim (ORCID: https://orcid.org/0000-0002-4586-5062)
- Mingyu Kim (ORCID: https://orcid.org/0000-0001-7571-9456)
Institutions
- Seoul National University (KR)
- Korea University (KR)
- New Generation University College (ET)
- Seoul National University Hospital (KR)
- Dongguk University Ilsan Hospital (KR)
- Hanyang University (KR)
- Ajou University (KR)
Publication Details
- Journal
- Communications Medicine
- Published
- 2025-06-13
- DOI
- https://doi.org/10.1038/s43856-025-00914-7
- Citations
- 5
- Primary Topic
- Machine Learning in Healthcare
- Type
- article
- Field-Weighted Citation Impact
- 12.40