TY - JOUR
T1 - Inferring multimodal latent topics from electronic health records
AU - Li, Yue
AU - Nair, Pratheeksha
AU - Lu, Xing Han
AU - Wen, Zhi
AU - Wang, Yuening
AU - Dehaghi, Amir Ardalan Kalantari
AU - Miao, Yan
AU - Liu, Weiqi
AU - Ordog, Tamas
AU - Biernacka, Joanna M.
AU - Ryu, Euijung
AU - Olson, Janet E.
AU - Frye, Mark A.
AU - Liu, Aihua
AU - Guo, Liming
AU - Marelli, Ariane
AU - Ahuja, Yuri
AU - Davila-Velderrain, Jose
AU - Kellis, Manolis
N1 - Publisher Copyright:
© 2020, The Author(s).
PY - 2020/12/1
Y1 - 2020/12/1
N2 - Electronic health records (EHR) are rich heterogeneous collections of patient health information, whose broad adoption provides clinicians and researchers unprecedented opportunities for health informatics, disease-risk prediction, actionable clinical recommendations, and precision medicine. However, EHRs present several modeling challenges, including highly sparse data matrices, noisy irregular clinical notes, arbitrary biases in billing code assignment, diagnosis-driven lab tests, and heterogeneous data types. To address these challenges, we present MixEHR, a multi-view Bayesian topic model. We demonstrate MixEHR on MIMIC-III, Mayo Clinic Bipolar Disorder, and Quebec Congenital Heart Disease EHR datasets. Qualitatively, MixEHR disease topics reveal meaningful combinations of clinical features across heterogeneous data types. Quantitatively, we observe superior prediction accuracy of diagnostic codes and lab test imputations compared to the state-of-art methods. We leverage the inferred patient topic mixtures to classify target diseases and predict mortality of patients in critical conditions. In all comparison, MixEHR confers competitive performance and reveals meaningful disease-related topics.
AB - Electronic health records (EHR) are rich heterogeneous collections of patient health information, whose broad adoption provides clinicians and researchers unprecedented opportunities for health informatics, disease-risk prediction, actionable clinical recommendations, and precision medicine. However, EHRs present several modeling challenges, including highly sparse data matrices, noisy irregular clinical notes, arbitrary biases in billing code assignment, diagnosis-driven lab tests, and heterogeneous data types. To address these challenges, we present MixEHR, a multi-view Bayesian topic model. We demonstrate MixEHR on MIMIC-III, Mayo Clinic Bipolar Disorder, and Quebec Congenital Heart Disease EHR datasets. Qualitatively, MixEHR disease topics reveal meaningful combinations of clinical features across heterogeneous data types. Quantitatively, we observe superior prediction accuracy of diagnostic codes and lab test imputations compared to the state-of-art methods. We leverage the inferred patient topic mixtures to classify target diseases and predict mortality of patients in critical conditions. In all comparison, MixEHR confers competitive performance and reveals meaningful disease-related topics.
UR - http://www.scopus.com/inward/record.url?scp=85085156889&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85085156889&partnerID=8YFLogxK
U2 - 10.1038/s41467-020-16378-3
DO - 10.1038/s41467-020-16378-3
M3 - Article
C2 - 32439869
AN - SCOPUS:85085156889
SN - 2041-1723
VL - 11
JO - Nature communications
JF - Nature communications
IS - 1
M1 - 2536
ER -