Outcome prediction using electronic medical record data with missing entries
Abstract/Contents
- Abstract
- The United States has been ranked 37th for overall health system performance and scores poorly in terms of patient safety, coordination and efficiency. Furthermore, the current healthcare system is costly as well as inefficient, consuming a large and increasing proportion of the gross domestic product. The use of technology and big data methodologies to improve performance and lower cost has already been demonstrated in non-healthcare fields and early studies have further demonstrated its potential in the clinical setting; however, implementation is challenging because most predictive models rely on complete data, which is costly and unrealistic to obtain. In this dissertation, we present a suite of innovative online matrix completion methodologies tailored to high amounts of missingness, which decrease the feature space and improve computation time over gold standard methods. We demonstrate their predictive performance using simulated and real electronic medical record data. Finally, we apply these methods to predictions using waveform data and in the development of a data driven system for predicting patient deterioration in a children's hospital. The use of these innovative matrix completion methods enables the implementation of sophisticated machine learning methodologies and thus allows for improved accuracy and efficiency when performing event prediction using EMR data.
Description
Type of resource | text |
---|---|
Form | electronic; electronic resource; remote |
Extent | 1 online resource. |
Publication date | 2017 |
Issuance | monographic |
Language | English |
Creators/Contributors
Associated with | Strandberg, Erika |
---|---|
Associated with | Stanford University, Program in Biomedical Informatics. |
Primary advisor | Bayati, Mohsen |
Thesis advisor | Bayati, Mohsen |
Thesis advisor | Horvitz, Eric J. (Eric Joel) |
Thesis advisor | Owens, Douglas K |
Advisor | Horvitz, Eric J. (Eric Joel) |
Advisor | Owens, Douglas K |
Subjects
Genre | Theses |
---|
Bibliographic information
Statement of responsibility | Erika Strandberg. |
---|---|
Note | Submitted to the Program in Biomedical Informatics. |
Thesis | Thesis (Ph.D.)--Stanford University, 2017. |
Location | electronic resource |
Access conditions
- Copyright
- © 2017 by Erika Strandberg
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...