{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/learning-tasks-for-multitask-learning","title":"Learning Tasks for Multitask Learning: Heterogenous Patient Populations in the ICU","arxiv_id":"1806.02878","date":"2018-06-07","proceeding":null,"authors":["Harini Suresh","Jen J. Gong","John Guttag"],"abstract":"Machine learning approaches have been effective in predicting adverse\noutcomes in different clinical settings. These models are often developed and\nevaluated on datasets with heterogeneous patient populations. However, good\npredictive performance on the aggregate population does not imply good\nperformance for specific groups.\n  In this work, we present a two-step framework to 1) learn relevant patient\nsubgroups, and 2) predict an outcome for separate patient populations in a\nmulti-task framework, where each population is a separate task. We demonstrate\nhow to discover relevant groups in an unsupervised way with a\nsequence-to-sequence autoencoder. We show that using these groups in a\nmulti-task framework leads to better predictive performance of in-hospital\nmortality both across groups and overall. We also highlight the need for more\ngranular evaluation of performance when dealing with heterogeneous populations.","url_abs":"http://arxiv.org/abs/1806.02878v1","url_pdf":"http://arxiv.org/pdf/1806.02878v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"learning-tasks-for-multitask-learning","repo_url":"https://github.com/mit-ddig/multitask-patients","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.02878","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}