Decaying relevance of clinical data towards future decisions in data-driven inpatient clinical order sets

International Journal of Medical Informatics
Jonathan H ChenRuss B Altman

Abstract

Determine how varying longitudinal historical training data can impact prediction of future clinical decisions. Estimate the "decay rate" of clinical data source relevance. We trained a clinical order recommender system, analogous to Netflix or Amazon's "Customers who bought A also bought B..." product recommenders, based on a tertiary academic hospital's structured electronic health record data. We used this system to predict future (2013) admission orders based on different subsets of historical training data (2009 through 2012), relative to existing human-authored order sets. Predicting future (2013) inpatient orders is more accurate with models trained on just one month of recent (2012) data than with 12 months of older (2009) data (ROC AUC 0.91 vs. 0.88, precision 27% vs. 22%, recall 52% vs. 43%, all P<10-10). Algorithmically learned models from even the older (2009) data was still more effective than existing human-authored order sets (ROC AUC 0.81, precision 16% recall 35%). Training with more longitudinal data (2009-2012) was no better than using only the most recent (2012) data, unless applying a decaying weighting scheme with a "half-life" of data relevance about 4 months. Clinical practice patterns (automatically) le...Continue Reading

Citations

Jun 29, 2017·The New England Journal of Medicine·Jonathan H Chen, Steven M Asch
Jul 19, 2018·Clinical Orthopaedics and Related Research·Michala Skovlund SørensenMichael Mørk Petersen
Aug 8, 2019·NPJ Digital Medicine·Marc Dewey, Uta Wilkens
Feb 7, 2020·The Journal of Bone and Joint Surgery. American Volume·Thomas G MyersConstantinos Ketonis
Jun 2, 2020·Annals of Internal Medicine·Alison CallahanJonathan H Chen
Aug 3, 2019·Journal of the American Medical Informatics Association : JAMIA·Ben Van CalsterGary S Collins
Oct 27, 2018·World Journal of Gastrointestinal Endoscopy·Muthuraman AlagappanTyler M Berzin
Feb 6, 2020·Scientific Reports·Piotr DworzynskiTune H Pers
Mar 7, 2020·Clinical Pharmacology and Therapeutics·Kathryn RoughAlvin Rajkomar
May 8, 2020·The Journal of Bone and Joint Surgery. American Volume·Thomas G MyersConstantinos Ketonis
Oct 28, 2020·Journal of the American Medical Informatics Association : JAMIA·Andre KumarJonathan H Chen
May 7, 2021·Journal of the American Medical Informatics Association : JAMIA·Sophie-Camille HogueMaxime Thibault
Jul 21, 2021·Methods of Information in Medicine·John H HolmesJason H Moore
Jan 26, 2021·Archives of Pathology & Laboratory Medicine·James H HarrisonMichelle N Stram
Dec 28, 2021·The American Surgeon·Mert Marcel DagliCharles E Butler

❮ Previous
Next ❯

Related Concepts

Related Feeds

BioHub - Researcher Network

The Chan-Zuckerberg Biohub aims to support the fundamental research and develop the technologies that will enable physicians to cure, prevent, or manage all diseases in our childrens' lifetimes. The CZ Biohub brings together researchers from UC Berkeley, Stanford, and UCSF. Find the latest research from the CZ Biohub researcher network here.

Related Papers

Journal of the American Medical Informatics Association : JAMIA
Jonathan H ChenRuss B Altman
Journal of the American Medical Informatics Association : JAMIA
Jonathan H ChenRuss B Altman
AMIA Joint Summits on Translational Science Proceedings
Jonathan H Chen, Russ B Altman
© 2022 Meta ULC. All rights reserved