# Robustly Extracting Medical Knowledge from EHRs: A Case Study of Learning a Health Knowledge Graph

arXiv: https://arxiv.org/abs/1910.01116 · HTML: https://mlsanity.com/paper/1910.01116/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 Oct 2019.

## Built on by (most cited first, 12 of 12)

- [Graph Representation Learning in Biomedicine](https://mlsanity.com/paper/2104.04883/) (2021)
- [GraphCare: Enhancing Healthcare Predictions with Personalized Knowledge Graphs](https://mlsanity.com/paper/2305.12788/) (2023)
- [A Semantic Web Technology Index](https://mlsanity.com/paper/2201.07034/) (2022)
- [Patient-Centric Knowledge Graphs: A Survey of Current Methods, Challenges, and Applications](https://mlsanity.com/paper/2402.12608/) (2024)
- [Inference of Dependency Knowledge Graph for Electronic Health Records](https://mlsanity.com/paper/2312.15611/) (2023)
- [Refining Diagnosis Paths for Medical Diagnosis based on an Augmented Knowledge Graph](https://mlsanity.com/paper/2204.13329/) (2022)
- [Review of Natural Language Processing in Pharmacology](https://mlsanity.com/paper/2208.10228/) (2022)
- [Healthcare Knowledge Graph Construction: State-of-the-art, open issues, and opportunities](https://mlsanity.com/paper/2207.03771/) (2022)
- [Semantic web technologies in sensor-based personal health monitoring systems: A systematic mapping study](https://mlsanity.com/paper/2306.04335/) (2023)
- [Liver Cancer Knowledge Graph Construction based on dynamic entity replacement and masking strategies RoBERTa-BiLSTM-CRF model](https://mlsanity.com/paper/2410.18090/) (2024)
- [From Patient Consultations to Graphs: Leveraging LLMs for Patient Journey Knowledge Graph Construction](https://mlsanity.com/paper/2503.16533/) (2025)
- [Semantic Reasoning in Medicine: The Role of Knowledge Graphs Across Five Key Domains](https://mlsanity.com/paper/2606.15155/) (2026)

## Abstract (arXiv)

Increasingly large electronic health records (EHRs) provide an opportunity to algorithmically learn medical knowledge. In one prominent example, a causal health knowledge graph could learn relationships between diseases and symptoms and then serve as a diagnostic tool to be refined with additional clinical input. Prior research has demonstrated the ability to construct such a graph from over 270,000 emergency department patient visits. In this work, we describe methods to evaluate a health knowledge graph for robustness. Moving beyond precision and recall, we analyze for which diseases and for which patients the graph is most accurate. We identify sample size and unmeasured confounders as major sources of error in the health knowledge graph. We introduce a method to leverage non-linear functions in building the causal graph to better understand existing model assumptions. Finally, to assess model generalizability, we extend to a larger set of complete patient visits within a hospital system. We conclude with a discussion on how to robustly extract medical knowledge from EHRs.