Show simple item record

dc.contributor.authorPage, David
dc.date.accessioned2018-03-13T15:17:08Z
dc.date.available2018-03-13T15:17:08Z
dc.date.issued2018-02-27
dc.identifier.urihttp://hdl.handle.net/1853/59416
dc.descriptionPresented on February 27, 2018 at 3:00 p.m. in the Klaus Advanced Computing Building, Room 2443.en_US
dc.descriptionDavid Page is a Kellett and Vilas Distinguished Achievement Professor at the University of Wisconsin-Madison. His tenure home is in the School of Medicine and Public Health, Dept. of Biostatistics and Medical Informatics, and he also has an appointment in the Department of Computer Sciences where he supervises PhD students and teaches machine learning. David is the faculty lead of the Cancer Informatics Core of UW-Madison’s Carbone Cancer Center, is a member of the Genome Center of Wisconsin, and served on scientific advisory boards for the Observational Medical Outcomes Partnership and the Wisconsin Genomics Initiative.en_US
dc.descriptionRuntime: 59:18 minutesen_US
dc.description.abstractMuch of the world’s real data on people is irregularly-sampled, temporal, and observational (meaning we don’t get to experiment as in a randomized clinical trial). For example, customers make purchases on various dates of their choice, not necessarily once a week or once a month, and we only observe rather than intervene in their decisions. Patients visit the doctor whenever they feel the need, and we observe their doctors’ entries in the electronic health record (EHR), without the ability to randomize patient treatments. We show that despite this lack of control or sampling regularity, we can predict future events from such data with surprising accuracy, for example better than 80% on average across a variety of diagnosis codes in the EHR a month in advance. We further show that despite many types of potential confounding, we can actually discover causal factors (e.g., effect of a drug on a disease or on a measurement such as blood pressure) at similar levels of accuracy for real problems. The key to doing so is modeling person-specific, time-varying baseline levels, e.g. of a measurement such as blood pressure or a risk such as for heart attack. On the applied side this talk will focus entirely on medical applications, but the approaches developed and employed are general-purpose machine learning algorithms with broad potential applicability.en_US
dc.format.extent59:18 minutes
dc.language.isoen_USen_US
dc.relation.ispartofseriesCHAI Seminar Seriesen_US
dc.subjectElectronic health records (EHR)en_US
dc.subjectMedical applicationen_US
dc.titleMachine Learning from Irregularly-Sampled Temporal Data: A Case Study in Predicting Across Most Diseases in Electronic Health Recordsen_US
dc.typeLectureen_US
dc.typeVideoen_US
dc.contributor.corporatenameGeorgia Institute of Technology. Center for Heath Analytics and Informaticsen_US
dc.contributor.corporatenameUniversity of Wisconsin-Madison. Dept. of Biostatistics and Medical Informaticsen_US


Files in this item

This item appears in the following Collection(s)

Show simple item record