Presenter Information

Layaal KhellahFollow

Presentation Type

Oral Presentation

Start Date

5-8-2024 9:00 AM

End Date

5-8-2024 11:00 AM

Subjects

Translational bioinformatics, Bioinformatics

Advisor

Mohammad Adibuzzaman

Student Level

Undergraduate

Abstract

Introduction:

Propensity Score Matching (PSM) and Structural Causal Models (SCMs) are key approaches in causal inference for treatment evaluation and clinical decisions, both rooted in the potential outcome framework but differing in their foundations—PSM follows Rubin's Causal Model, while SCMs adhere to the Structural Theory of Causation.

Methods:

Our study compares PSM and SCMs in clinical contexts, focusing on albumin as an AKI treatment. We examine each method's steps, highlighting differences. Using data from AKI patients with cirrhosis, we assess albumin’s treatment efficacy with both PSM and SCMs, offering insights into their performance in estimating causal effects..

Results:

PSM and SCM showed similar findings regarding the effect of treatment on AKI recovery and in-hospital survival, indicating no significant association with improved outcomes. PSM found no significant effect on AKI recovery (OR 0.70, 95% CI: 0.59-1.07, P = .130) or in-hospital survival (OR 0.76 [95% CI: 0.46-1.25], P = .280). SCM also showed no statistically significant average treatment effect on primary (-0.07487) or secondary (-0.09710) outcomes (p > 0.005), suggesting no discernible causal effect of albumin use.

Conclusion:

Our study provides insights into PSM and SCMs' practical application in clinical research. By demonstrating their effectiveness in a real-world scenario, we underscore their relevance and utility in informing treatment decisions and improving patient outcomes.

Creative Commons License or Rights Statement

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.

Persistent Identifier

https://archives.pdx.edu/ds/psu/41934

Share

COinS
 
May 8th, 9:00 AM May 8th, 11:00 AM

Systematic Comparison of Propensity Score Matching and Structural Causal Modeling for Clinical Applications, with a Case Study of Albumin Treatment for Acute Kidney Injury (AKI) Patients: Propensity Score Matching vs. Structural Causal Models

Introduction:

Propensity Score Matching (PSM) and Structural Causal Models (SCMs) are key approaches in causal inference for treatment evaluation and clinical decisions, both rooted in the potential outcome framework but differing in their foundations—PSM follows Rubin's Causal Model, while SCMs adhere to the Structural Theory of Causation.

Methods:

Our study compares PSM and SCMs in clinical contexts, focusing on albumin as an AKI treatment. We examine each method's steps, highlighting differences. Using data from AKI patients with cirrhosis, we assess albumin’s treatment efficacy with both PSM and SCMs, offering insights into their performance in estimating causal effects..

Results:

PSM and SCM showed similar findings regarding the effect of treatment on AKI recovery and in-hospital survival, indicating no significant association with improved outcomes. PSM found no significant effect on AKI recovery (OR 0.70, 95% CI: 0.59-1.07, P = .130) or in-hospital survival (OR 0.76 [95% CI: 0.46-1.25], P = .280). SCM also showed no statistically significant average treatment effect on primary (-0.07487) or secondary (-0.09710) outcomes (p > 0.005), suggesting no discernible causal effect of albumin use.

Conclusion:

Our study provides insights into PSM and SCMs' practical application in clinical research. By demonstrating their effectiveness in a real-world scenario, we underscore their relevance and utility in informing treatment decisions and improving patient outcomes.