Learning a Health Knowledge Graph from Electronic Medical Records

Maya Rotmensch, Yoni Halpern, Abdulhakim Tlimat, Steven Horng, David Sontag

Research output: Contribution to journalArticle

Abstract

Demand for clinical decision support systems in medicine and self-diagnostic symptom checkers has substantially increased in recent years. Existing platforms rely on knowledge bases manually compiled through a labor-intensive process or automatically derived using simple pairwise statistics. This study explored an automated process to learn high quality knowledge bases linking diseases and symptoms directly from electronic medical records. Medical concepts were extracted from 273,174 de-identified patient records and maximum likelihood estimation of three probabilistic models was used to automatically construct knowledge graphs: logistic regression, naive Bayes classifier and a Bayesian network using noisy OR gates. A graph of disease-symptom relationships was elicited from the learned parameters and the constructed knowledge graphs were evaluated and validated, with permission, against Google's manually-constructed knowledge graph and against expert physician opinions. Our study shows that direct and automated construction of high quality health knowledge graphs from medical records using rudimentary concept extraction is feasible. The noisy OR model produces a high quality knowledge graph reaching precision of 0.85 for a recall of 0.6 in the clinical evaluation. Noisy OR significantly outperforms all tested models across evaluation frameworks (p < 0.01).

Original languageEnglish (US)
Article number5994
JournalScientific Reports
Volume7
Issue number1
DOIs
StatePublished - Dec 1 2017

Fingerprint

Electronic medical equipment
Health
Maximum likelihood estimation
Bayesian networks
Decision support systems
Medicine
Logistics
Classifiers
Statistics
Personnel
Statistical Models

ASJC Scopus subject areas

  • General

Cite this

Rotmensch, M., Halpern, Y., Tlimat, A., Horng, S., & Sontag, D. (2017). Learning a Health Knowledge Graph from Electronic Medical Records. Scientific Reports, 7(1), [5994]. https://doi.org/10.1038/s41598-017-05778-z

Learning a Health Knowledge Graph from Electronic Medical Records. / Rotmensch, Maya; Halpern, Yoni; Tlimat, Abdulhakim; Horng, Steven; Sontag, David.

In: Scientific Reports, Vol. 7, No. 1, 5994, 01.12.2017.

Research output: Contribution to journalArticle

Rotmensch, M, Halpern, Y, Tlimat, A, Horng, S & Sontag, D 2017, 'Learning a Health Knowledge Graph from Electronic Medical Records', Scientific Reports, vol. 7, no. 1, 5994. https://doi.org/10.1038/s41598-017-05778-z
Rotmensch, Maya ; Halpern, Yoni ; Tlimat, Abdulhakim ; Horng, Steven ; Sontag, David. / Learning a Health Knowledge Graph from Electronic Medical Records. In: Scientific Reports. 2017 ; Vol. 7, No. 1.
@article{e887ecc1949e4860af089ec386f88c18,
title = "Learning a Health Knowledge Graph from Electronic Medical Records",
abstract = "Demand for clinical decision support systems in medicine and self-diagnostic symptom checkers has substantially increased in recent years. Existing platforms rely on knowledge bases manually compiled through a labor-intensive process or automatically derived using simple pairwise statistics. This study explored an automated process to learn high quality knowledge bases linking diseases and symptoms directly from electronic medical records. Medical concepts were extracted from 273,174 de-identified patient records and maximum likelihood estimation of three probabilistic models was used to automatically construct knowledge graphs: logistic regression, naive Bayes classifier and a Bayesian network using noisy OR gates. A graph of disease-symptom relationships was elicited from the learned parameters and the constructed knowledge graphs were evaluated and validated, with permission, against Google's manually-constructed knowledge graph and against expert physician opinions. Our study shows that direct and automated construction of high quality health knowledge graphs from medical records using rudimentary concept extraction is feasible. The noisy OR model produces a high quality knowledge graph reaching precision of 0.85 for a recall of 0.6 in the clinical evaluation. Noisy OR significantly outperforms all tested models across evaluation frameworks (p < 0.01).",
author = "Maya Rotmensch and Yoni Halpern and Abdulhakim Tlimat and Steven Horng and David Sontag",
year = "2017",
month = "12",
day = "1",
doi = "10.1038/s41598-017-05778-z",
language = "English (US)",
volume = "7",
journal = "Scientific Reports",
issn = "2045-2322",
publisher = "Nature Publishing Group",
number = "1",

}

TY - JOUR

T1 - Learning a Health Knowledge Graph from Electronic Medical Records

AU - Rotmensch, Maya

AU - Halpern, Yoni

AU - Tlimat, Abdulhakim

AU - Horng, Steven

AU - Sontag, David

PY - 2017/12/1

Y1 - 2017/12/1

N2 - Demand for clinical decision support systems in medicine and self-diagnostic symptom checkers has substantially increased in recent years. Existing platforms rely on knowledge bases manually compiled through a labor-intensive process or automatically derived using simple pairwise statistics. This study explored an automated process to learn high quality knowledge bases linking diseases and symptoms directly from electronic medical records. Medical concepts were extracted from 273,174 de-identified patient records and maximum likelihood estimation of three probabilistic models was used to automatically construct knowledge graphs: logistic regression, naive Bayes classifier and a Bayesian network using noisy OR gates. A graph of disease-symptom relationships was elicited from the learned parameters and the constructed knowledge graphs were evaluated and validated, with permission, against Google's manually-constructed knowledge graph and against expert physician opinions. Our study shows that direct and automated construction of high quality health knowledge graphs from medical records using rudimentary concept extraction is feasible. The noisy OR model produces a high quality knowledge graph reaching precision of 0.85 for a recall of 0.6 in the clinical evaluation. Noisy OR significantly outperforms all tested models across evaluation frameworks (p < 0.01).

AB - Demand for clinical decision support systems in medicine and self-diagnostic symptom checkers has substantially increased in recent years. Existing platforms rely on knowledge bases manually compiled through a labor-intensive process or automatically derived using simple pairwise statistics. This study explored an automated process to learn high quality knowledge bases linking diseases and symptoms directly from electronic medical records. Medical concepts were extracted from 273,174 de-identified patient records and maximum likelihood estimation of three probabilistic models was used to automatically construct knowledge graphs: logistic regression, naive Bayes classifier and a Bayesian network using noisy OR gates. A graph of disease-symptom relationships was elicited from the learned parameters and the constructed knowledge graphs were evaluated and validated, with permission, against Google's manually-constructed knowledge graph and against expert physician opinions. Our study shows that direct and automated construction of high quality health knowledge graphs from medical records using rudimentary concept extraction is feasible. The noisy OR model produces a high quality knowledge graph reaching precision of 0.85 for a recall of 0.6 in the clinical evaluation. Noisy OR significantly outperforms all tested models across evaluation frameworks (p < 0.01).

UR - http://www.scopus.com/inward/record.url?scp=85025466587&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85025466587&partnerID=8YFLogxK

U2 - 10.1038/s41598-017-05778-z

DO - 10.1038/s41598-017-05778-z

M3 - Article

VL - 7

JO - Scientific Reports

JF - Scientific Reports

SN - 2045-2322

IS - 1

M1 - 5994

ER -