Generating Multi-label Discrete Patient Records using Generative Adversarial Networks

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 67
  • Download : 0
DC FieldValueLanguage
dc.contributor.authorChoi, Yoonjaeko
dc.contributor.authorBiswal, Siddharthko
dc.contributor.authorMalin, Bradleyko
dc.contributor.authorDuke, Jonko
dc.contributor.authorStewart, Walter Fko
dc.contributor.authorSun, Jimengko
dc.date.accessioned2021-03-15T00:31:23Z-
dc.date.available2021-03-15T00:31:23Z-
dc.date.created2020-12-03-
dc.date.issued2017-08-
dc.identifier.citationMachine Learning for Healthcare Conference 2017-
dc.identifier.urihttp://hdl.handle.net/10203/281519-
dc.description.abstractAccess to electronic health record (EHR) data has motivated computational advances in medical research. However, various concerns, particularly over privacy, can limit access to and collaborative use of EHR data. Sharing synthetic EHR data could mitigate risk. In this paper, we propose a new approach, medical Generative Adversarial Network (medGAN), to generate realistic synthetic patient records. Based on input real patient records, medGAN can generate high-dimensional discrete variables (e.g., binary and count features) via a combination of an autoencoder and generative adversarial networks. We also propose minibatch averaging to efficiently avoid mode collapse, and increase the learning efficiency with batch normalization and shortcut connections. To demonstrate feasibility, we showed that medGAN generates synthetic patient records that achieve comparable performance to real data on many experiments including distribution statistics, predictive modeling tasks and a medical expert review. We also empirically observe a limited privacy risk in both identity and attribute disclosure using medGAN.-
dc.languageEnglish-
dc.publisherMLHC-
dc.titleGenerating Multi-label Discrete Patient Records using Generative Adversarial Networks-
dc.typeConference-
dc.type.rimsCONF-
dc.citation.publicationnameMachine Learning for Healthcare Conference 2017-
dc.identifier.conferencecountryUS-
dc.identifier.conferencelocationNortheastern University-
dc.contributor.localauthorChoi, Yoonjae-
dc.contributor.nonIdAuthorBiswal, Siddharth-
dc.contributor.nonIdAuthorMalin, Bradley-
dc.contributor.nonIdAuthorDuke, Jon-
dc.contributor.nonIdAuthorStewart, Walter F-
dc.contributor.nonIdAuthorSun, Jimeng-
Appears in Collection
RIMS Conference Papers
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0