Machine Learning–Based Techniques to Handle Missing Data in Meta-Regression

• In Event: Advances in Meta-Analytic and Systematic Review Approaches

Abstract

In this article, we investigated the effectiveness of model-based machine learning approaches, specifically Random Forest (RF) and LightGBM (LG), for handling missing data, juxtaposed against conventional methods. Through a simulation study, we assessed the performance of these methods by measuring bias and precision in scenarios with varying degrees of missingness (5%, 15%, 30%) and different missing data mechanisms. The findings reveal that while multiple imputation methods can provide accurate estimates in meta-regression, their efficacy varies with higher rates of missingness and when missingness is correlated with effect sizes. The results underscore the superiority of LG and RF over traditional imputation methods in meta-analytic contexts, highlighting their potential to enhance the accuracy and reliability of systematic reviews plagued by missing data.

Authors

• Kamal Chawla, University of Maine

• Henry May, University of Delaware