Machine Learning methods for handling missing covariates in meta-analysis

Meta-analysis has played a significant role in evaluations. Results from meta-analytic studies have been used to support evidence-based practice (Ppakostidis & Giannoudis, 2023), evaluate program effectiveness (Crespi & Cobian, 2022), and allocate resources (Polanin et al., 2016). However, a key challenge in meta-regression is missing covariates (Schauer et al., 2022; Tipton, 2019). Machine learning-based imputation effectively addresses this issue. This project explores various ML-based imputation methods for handling missing covariates encountered in meta-regression.

Comfort Omonkhodion – University of Central Florida; Haiyan Bai – University of Central Florida; Oluwaseun Farotimi – University of Central Florida