Causal Inference for ICU Patients

Presenter(s)

Jack O'Connor

Abstract

Sepsis is an infection that can lead to vital organ dysfunction, and it is estimated to occur in 30% of ICU patients. This research aims to assess whether receiving a Transthoracic Echocardiogram (TTE) has an effect on the 28-day mortality rate of ICU patients with Sepsis, as TTEs are currently widely used in medical treatment. Our data source containing ICU patients comes from the Electronic Health Records (EHR) database MIMIC-III. This is an observational data source, so we attempted to establish causality by addressing assumptions and using various causal effect estimators. Additionally, we estimated heterogeneity in treatment effects across different groups of sepsis patients. Ultimately, we established a causal effect and identified which covariates are associated with large treatment effects. This research can be used to better inform clinical decisions when providing care for ICU patients with Sepsis.

College

College of Science & Engineering

Department

Mathematics & Statistics

Campus

Winona

First Advisor/Mentor

Silas Bergen

Start Date

4-24-2025 10:00 AM

End Date

4-24-2025 11:00 AM

Presentation Type

Poster Session

Format of Presentation or Performance

In-Person

Session

1b=10am-11am

Poster Number

48

Comments

WSU Review Needed - Check advisors are different in the program

Faculty Mentors: Rahul Ladhania and Snigdha Panigrahi

Share

COinS
 
Apr 24th, 10:00 AM Apr 24th, 11:00 AM

Causal Inference for ICU Patients

Sepsis is an infection that can lead to vital organ dysfunction, and it is estimated to occur in 30% of ICU patients. This research aims to assess whether receiving a Transthoracic Echocardiogram (TTE) has an effect on the 28-day mortality rate of ICU patients with Sepsis, as TTEs are currently widely used in medical treatment. Our data source containing ICU patients comes from the Electronic Health Records (EHR) database MIMIC-III. This is an observational data source, so we attempted to establish causality by addressing assumptions and using various causal effect estimators. Additionally, we estimated heterogeneity in treatment effects across different groups of sepsis patients. Ultimately, we established a causal effect and identified which covariates are associated with large treatment effects. This research can be used to better inform clinical decisions when providing care for ICU patients with Sepsis.