Extension of Dynamic Network Biomarker using the propensity score method: Simulation of causal effects on variance and correlation coefficient

2025-05-20

AI Summary

In clinical dynamic network biomarker (DNB) modeling, confounding factors introduce causal bias into variance and Pearson correlation coefficient estimation. To address this, we propose the first systematic integration of propensity score matching (PSM) into the DNB construction pipeline, extending causal inference specifically to DNB’s core metrics—variance and pairwise Pearson correlations. Using Monte Carlo simulations, we rigorously evaluate its bias-correction efficacy. Results demonstrate that PSM substantially reduces inter-group comparison bias, markedly enhancing DNB stability and discriminative accuracy under confounding. This approach fills a critical methodological gap in estimating causal effects of variance and correlation at the DNB level, providing a statistically robust, interpretable framework for causal inference in clinical biomarker discovery.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

In clinical biomarker studies, the Dynamic Network Biomarker (DNB) is sometimes used. DNB is a composite variable derived from the variance and the Pearson correlation coefficient of biological signals. When applying DNB to clinical data, it is important to account for confounding bias. However, little attention has been paid to statistical causal inference methods for variance and correlation coefficients. This study evaluates confounding adjustment using propensity score matching (PSM) through Monte Carlo simulations. Our results support the use of PSM to reduce bias and improve group comparisons when DNB is applied to clinical data.

Problem

• Extends Dynamic Network Biomarker with propensity score method
• Evaluates confounding bias in variance and correlation coefficients
• Simulates causal effects using Monte Carlo methods

Innovation

• Uses propensity score matching for bias reduction
• Applies Monte Carlo simulations for evaluation
• Adjusts confounding in Dynamic Network Biomarker