Off-campus WSU users: To download campus access dissertations, please use the following link to log into our proxy server with your WSU access ID and password, then click the "Off-campus Download" button below.

Non-WSU users: Please talk to your librarian about requesting this thesis through interlibrary loan.

Access Type

WSU Access

Date of Award

January 2020

Degree Type

Thesis

Degree Name

M.S.

Department

Industrial and Manufacturing Engineering

First Advisor

Ratna Chinnam

Abstract

Heterogeneous data including patient demographics, clinical notes, vitals, lab test

results, and visit metadata accumulate in the Electronic Health Record (EHR) during the

course of a patient encounter. Having advanced notification of a patient’s care pathway

can facilitate patient flow leading to increased patient satisfaction, more effective resource

utilization, and lower operating cost. The objective of this study is to use EHR data to

predict emergency department (ED) disposition decision.

The role of distributed representation, including pre-trained word embeddings and

embeddings learned jointly with the network parameters during training, are identified as

critical components to successful deep learning models. Based on empirical results a

prototype clinician interface for interpreting prediction results is developed and pipelines

for dynamically predicting ED disposition decision throughout the course of the patient

encounter in real time are detailed. The data processing pipelines for heterogeneous

models are contrasted between the areas of traditional classification methods vs. deep

neural network-based classification.

Off-campus Download

Share

COinS