Multi-stage prognosis of COVID-19 using a clinical event-based stratification of disease severity

Chen, Haotian

Permalink

https://hdl.handle.net/2142/113924 Copy

Description

Title

Multi-stage prognosis of COVID-19 using a clinical event-based stratification of disease severity

Author(s)

Chen, Haotian

Issue Date

2021-12-08

Director of Research (if dissertation) or Advisor (if thesis)

Iyer, Ravishankar K.

Department of Study

Electrical & Computer Eng

Discipline

Electrical & Computer Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Prognosis
• Machine Learning
• COVID-19
• Health Informatics
• Artificial Intelligence

Abstract

The COVID-19 disease has shown remarkable diversity in its manifestation. Precise anticipation of these manifestations is important to enable earlier intervention for high-risk patients and efficient deployment of medical resources. In this thesis, a multi-stage prognostic framework is developed for assessing COVID-19 patients at hospital admission and during disease progression. The analysis is conducted upon 10,123 COVID-19 patients treated at Rush University Medical Center at Chicago between 03/17/2020 and 08/07/2020. In order to characterize the patients with different severity, a stratification scheme is first established to assign patients to different stages of disease severity based on discrete clinical events (i.e., admission to hospital, admission to ICU, mechanical ventilation, and death). Then two prognostic frameworks were developed to predict the progression of COVID-19 through these stages: 1) a baseline model which uses the measurements collected at hospital admission to predict disease escalation to severe stages; 2) a progressive model which uses the measurements collected at the patient’s latest stage to predict further escalation. It is found that future clinical stages can be predicted using baseline measurements with clinically significant accuracy. Finally, key risk factors are identified using Least Absolute Shrinkage and Selection Operator (LASSO) and decision tree algorithms. The developed multi-stage framework can be used to anticipate COVID-19 disease progression, allowing earlier interventions as well as better management of hospital resources.

Graduation Semester

2021-12

Type of Resource

Thesis

Permalink

http://hdl.handle.net/2142/113924

Copyright and License Information

Copyright 2021 Haotian Chen

Owning Collections