Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Using latent class mixture models to define sepsis endotypes

Taylor, Samantha (2017) Using latent class mixture models to define sepsis endotypes. Master's Thesis, University of Pittsburgh. (Unpublished)

Submitted Version

Download (1MB) | Preview


Severe sepsis is associated with high mortality and is a common problem in the United States. Recently, studies have shown that efforts focused on lowering cytokine levels improve survival. The aim of this work is to define sepsis endotypes using longitudinal cytokine measurements.
Sepsis endotypes were defined using latent class mixture models. Latent class mixture models were modeled using a natural log transformation of the actual time measurements. The outcome was the natural log of the cytokine value. No other covariates were modeled and a parameterized link function using a basis of I-splines was chosen over a linear transformation to increase flexibility in the latent class trajectories. The number of latent classes were determined by a combination of the lowest BIC and clinical significance.
After creating models for a variety of subsets derived from the source population, it was determined that mortality within a particular trajectory class is not only dependent upon the baseline cytokine value, but also dependent upon the rate of decent after baseline. A class with high baseline cytokine values that decrease quickly has lower mortality rates than classes who do not decline quickly. It was also determined that those who have increasing cytokine values from baseline to 6 hours have worse outcomes than those who decrease in the same time frame.
Public Health Significance: Given the public health significance of sepsis, understanding prognosis is extremely important. Previously, having a high IL6 measurement implied a poor prognosis. Our results show that many factors play into the determination of prognosis and patients can be treated accordingly.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Taylor, Samanthasjt42@pitt.edusjt42
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Thesis AdvisorChang,
Committee MemberClermont,
Committee MemberYouk,
Date: 24 February 2017
Date Type: Publication
Defense Date: 19 December 2016
Approval Date: 24 February 2017
Submission Date: 22 November 2016
Access Restriction: 1 year -- Restrict access to University of Pittsburgh for a period of 1 year.
Number of Pages: 52
Institution: University of Pittsburgh
Schools and Programs: School of Public Health > Biostatistics
Degree: MS - Master of Science
Thesis Type: Master's Thesis
Refereed: Yes
Uncontrolled Keywords: Sepsis
Date Deposited: 24 Feb 2017 19:07
Last Modified: 01 Jan 2018 06:15


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item