Deep representation learning of electronic health records to unlock patient stratification at scale

Abstract

Deriving disease subtypes from electronic health records (EHRs) can guide next-generation personalized medicine. However, challenges in summarizing and representing patient data prevent widespread practice of scalable EHR-based stratification analysis. Here we present an unsupervised framework based on deep learning to process heterogeneous EHRs and derive patient representations that can efficiently and effectively enable patient stratification at scale. We considered EHRs of 1,608,741 patients from a diverse hospital cohort comprising a total of 57,464 clinical concepts. We introduce a representation learning model based on word embeddings, convolutional neural networks, and autoencoders (i.e., ConvAE) to transform patient trajectories into low-dimensional latent vectors. We evaluated these representations as broadly enabling patient stratification by applying hierarchical clustering to different multi-disease and disease-specific patient cohorts. ConvAE significantly outperformed several baselines in a clustering task to identify patients with different complex conditions, with 2.61 entropy and 0.31 purity average scores. When applied to stratify patients within a certain condition, ConvAE led to various clinically relevant subtypes for different disorders, including type 2 diabetes, Parkinson’s disease, and Alzheimer’s disease, largely related to comorbidities, disease progression, and symptom severity. With these results, we demonstrate that ConvAE can generate patient representations that lead to clinically meaningful insights. This scalable framework can help better understand varying etiologies in heterogeneous sub-populations and unlock patterns for EHR-based research in the realm of personalized medicine.

Full text

Its licence does not allow us to show the full text. Read it at PubMed Central:

References 64

1. 1.

   Jensen PB, Jensen LJ, Brunak S. Mining electronic health records: towards better research applications and clinical care. Nat. Rev. Genet. 2012;13:395.

   Not in our library10.1038/nrg3208

2. 2.

   Cutting GR. Cystic fibrosis genetics: from molecular understanding to clinical application. Nat. Rev. Genet. 2014;16:45-56.

   Not in our library10.1038/nrg3849

3. 3.

   Alexandrov V, et al. Large-scale phenome analysis defines a behavioral signature for Huntington’s disease genotype in mice. Nat. Biotechnol. 2016;34:838-44.

   Not in our library10.1038/nbt.3587

4. 4.

   Langston JW. The Parkinson’s complex: Parkinsonism is just the tip of the iceberg. Ann. Neurol. 2006;59:591-596.

   Not in our library10.1002/ana.20834

5. 5.

   de Mel, S., Lim, S. H., Tung, M. L. & Chng, W. J. Implications of heterogeneity in multiple myeloma. BioMed Res. Int. 1–12, 10.1155/2014/232546 (2014).

   Not in our library10.1155/2014/232546

6. 6.
   Type 2 diabetes: a multifaceted disease · 2019

   No notice
7. 7.

   Dugger SA, Platt A, Goldstein DB. Drug development in the era of precision medicine. Nat. Rev. Drug Discov. 2017;17:183-196.

   Not in our library10.1038/nrd.2017.226

8. 8.

   Baytas, I. M. et al. Patient subtyping via time-aware LSTM Networks. In Proc. 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (eds Matwin S, S., Yu, S. & Farooq, F.) 65–74 (ACM, New York, 2017).

   Not in our library

9. 9.

   Doshi-Velez F, Ge Y, Kohane I. Comorbidity clusters in autism spectrum disorders: an electronic health record time-series analysis. Pediatrics. 2013;133:e54-e63.

   Not in our library10.1542/peds.2013-0819

10. 10.

    Li L, et al. Identification of type 2 diabetes subgroups through topological analysis of patient similarity. Sci. Transl. Med. 2015;7:311ra174.

    Not in our library10.1126/scitranslmed.aaa9364

11. 11.

    Zhang X, et al. Data-driven subtyping of Parkinson’s disease using longitudinal clinical records: a cohort study. Scientific Rep. 2019;9:797.

    Not in our library10.1038/s41598-018-37545-z

12. 12.
    Deep learning and alternative learning strategies for retrospective real-world clinical data · 2019

    No notice
13. 13.

    Bengio Y, Courville A, Vincent P. Representation learning: a review and new perspectives. IEEE Trans. Pattern Anal. Mach. Intell. 2013;35:1798-1828.

    Not in our library10.1109/tpami.2013.50

14. 14.
    Deep learning · 2015

    No notice
15. 15.

    Miotto R, Wang F, Wang S, Jiang X, Dudley JT. Deep learning for healthcare: review, opportunities and challenges. Brief Bioinform. 2017;19:1236-1246.

    Not in our library10.1093/bib/bbx044

16. 16.

    Xiao C, Choi E, Sun J. Opportunities and challenges in developing deep learning models using electronic health records data: a systematic review. J. Am. Med. Inform. Assoc. 2018;25:1419-1428.

    Not in our library10.1093/jamia/ocy068

17. 17.
    The Impact of Phenotypic and Genetic Heterogeneity on Results of Genome Wide Association Studies of Complex Diseases · 2013

    No notice
18. 18.

    Banda JM, Seneviratne M, Hernandez-Boussard T, Shah NH. Advances in electronic phenotyping: from rule-based definitions to machine learning models. Annu. Rev. Biomed. Data Sci. 2018;1:53-68.

    Not in our library10.1146/annurev-biodatasci-080917-013315

19. 19.

    Cote RA, Robboy S. Progress in medical information management: systematized nomenclature of medicine (snomed). JAMA. 1980;243:756-762.

    Not in our library10.1001/jama.1980.03300340032015

20. 20.
    Deep Patient: An Unsupervised Representation to Predict the Future of Patients from the Electronic Health Records · 2016

    No notice
21. 21.

    Dougherty ER, Brun M. A probabilistic theory of clustering. Pattern Recogn. 2004;37:917-925.

    Not in our library

22. 22.

    Dalton, L. A., Benalcázar, M. E. & Dougherty, E. R. Optimal clustering under uncertainty. PLoS ONE13, 10.1371/journal.pone.0204627 (2018).

    Not in our library10.1371/journal.pone.0204627

23. 23.

    Brun M, et al. Model-based evaluation of clustering validation measures. Pattern Recogn. 2007;40:807-824.

    Not in our library

24. 24.

    Amigó E, Gonzalo J, Artiles J, Verdejo F. A comparison of extrinsic clustering evaluation metrics based on formal constraints. Inform. Retrieval. 2009;12:461-486.

    Not in our library

25. 25.

    McInnes, L., Healy, J., Saul N., & Grossberger, L. UMAP: Uniform Manifold Approximation and Projection for dimension reduction. J Open Source Softw3, 861, 10.21105/joss.00861 (The Open Journal, 2018).

    Not in our library

26. 26.

    Cowie, C. C., Casagrande, S. S. & Geiss, L. S. Prevalence and incidence of type 2 diabetes and prediabetes. In Diabetes in America 3rd edn (eds Barrett-Connor, E. et al.) 3–1 (National Institutes of Health, Bethesda, 2018).

    Not in our library

27. 27.

    de Lau LML, Breteler MMB. Epidemiology of Parkinson’s disease. Lancet Neurol. 2006;5:525-535.

    Not in our library10.1016/s1474-4422(06)70471-9

28. 28.

    Qiu C, Kivipelto M, von Strauss E. Epidemiology of alzheimeras disease: occurrence, determinants, and strategies toward intervention. Dialog. Clin. Neurosci. 2009;11:111.

    Not in our library10.31887/dcns.2009.11.2/cqiu

29. 29.

    Kazandjian, D. Multiple myeloma epidemiology and survival: a unique malignancy. In Seminars in Oncology, Vol. 43 (eds Ahn I. E. & Mailankody, S.) 676–681 (Elsevier, 2016).

    Not in our library10.1053/j.seminoncol.2016.11.004

30. 30.

    Cancer Stat Facts: Prostate Cancer. https://seer.cancer.gov/statfacts/html/prost.html (2019).

    Not in our library

31. 31.

    Cancer Stat Facts: Female Breast Cancer. https://seer.cancer.gov/statfacts/html/breast.html (2019).

    Not in our library

32. 32.

    Vallon V, Komers R. Pathophysiology of the diabetic kidney. Compr. Physiol. 2011;1:1175-1232.

    Not in our library10.1002/cphy.c100049

33. 33.

    Malaguarnera L, Cristaldi E, Malaguarnera M. The role of immunity in elderly cancer. Crit. Rev. Oncol. Hematol. 2010;74:40-60.

    Not in our library10.1016/j.critrevonc.2009.06.002

34. 34.

    Delamaire M, et al. Impaired leucocyte functions in diabetic patients. Diabetic Med. 1997;14:29-34.

    Not in our library10.1002/(sici)1096-9136(199701)14:1<29::aid-dia300>3.0.co;2-v

35. 35.

    Jain S, Lo SE, Louis ED. Common misdiagnosis of a common neurological disorder. Arch. Neurol. 2006;63:1100-1104.

    Not in our library10.1001/archneur.63.8.1100

36. 36.

    Alves G, Wentzel-Larsen T, Larsen JP. Is fatigue an independent and persistent symptom in patients with Parkinson disease? Neurology. 2004;63:1908-1911.

    Not in our library10.1212/01.wnl.0000144277.06917.cc

37. 37.

    Siciliano M, et al. Fatigue in Parkinson’s disease: a systematic review and meta-analysis. Mov. Disord. 2018;33:1712-1723.

    Not in our library10.1002/mds.27461

38. 38.

    Alzheimer’s association. Younger/Early-Onset Alzheimer’s. https://www.alz.org/alzheimers-dementia/what-is-alzheimers/younger-early-onset (2019).

    Not in our library

39. 39.

    Manji H, Jäger HR, Winston A. HIV, dementia and antiretroviral drugs: 30 years of an epidemic. J. Neurol. Neurosurg. Psychiatry. 2013;84:1126-1137.

    Not in our library10.1136/jnnp-2012-304022

40. 40.

    Lyketsos CG, et al. Prevalence of neuropsychiatric symptoms in dementia and mild cognitive impairment. JAMA. 2002;288:1475-1483.

    Not in our library10.1001/jama.288.12.1475

Reads

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.