Combining Structured and Unstructured Data in Electronic Health Record for Readmission Prediction via Deep Learning
MetadataShow full item record
Publisher:The Ohio State University
Series/Report no.:The Ohio State University. Department of Computer Science and Engineering Honors Theses; 2020
With the aid of statistical learning tools nowadays, a variety of clinical prediction tasks can be examined and modeled in a quantitative way. Predicting hospital readmission probability is among one of the most significant tasks in that it provides a good indication of the healthcare cost and a patient's health condition. Thus, in this study, we strive to build up a quantitative prediction model for readmission prediction by utilizing both structured data and unstructured text data from a patient's Electronic Health Records (EHR). In the past, a variety of studies focused on using only structured categorical or numerical data such as lab tests and Heart Failure Signs to perform clinical risk prediction tasks, while recently, with the help of deep learning models, people started to use Natural Language Processing techniques to process unstructured patient's text data, as it contains richer information. However, with the belief that both structured and unstructured data can play significant role in predicting readmission, our research will focus on developing deep learning methods to combine the two types of data together in an efficient way, such that the predicting performance will exceed those of the previous models.
Academic Major: Computer Science and Engineering
Items in Knowledge Bank are protected by copyright, with all rights reserved, unless otherwise indicated.