Automated Hierarchical Graph Construction for Multi-source Electronic Health Records
Yinjie Wang, Doudou Zhou, Yue Liu, Junwei Lu, Tianxi Cai · 2025
Read Automated Hierarchical Graph Construction for Multi-source Electronic Health Records by Yinjie Wang, Doudou Zhou, Yue Liu, Junwei Lu, Tianxi Cai (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-06576.txt.
Description
Messy hospital codes from many institutions are organized into clean hierarchies by Yinjie Wang and four co-authors in a 2025 health informatics paper. Electronic health records mix diagnoses, medicines, and lab results, but each site uses its own codes and terms. That mess blocks large studies that combine records across hospitals.
Their framework, MASH, works automatically. Neural optimal transport aligns codes between institutions, and learned embeddings in hyperbolic space, a geometry suited to trees, build the hierarchy. Training draws on pretrained language models, patterns of codes appearing together, written descriptions, and known labels. Applied to real records, MASH produced readable hierarchies for diagnosis, medication, and lab codes. Notably, it built the first automatic hierarchies for local lab codes that previously had no standard structure at all.
Catalog record
| CREATOR | Yinjie Wang, Doudou Zhou, Yue Liu, Junwei Lu, Tianxi Cai |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 66 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-06576