Benchmarking machine learning missing data imputation methods in large-scale mental health survey databases

Databases tied to mental and behavioral health surveys suffer from the issue of missing data when participants skip the entire survey, which affects the data quality and sample size. These missing data patterns were investigated and the imputation performance was evaluated in Simons Foundations Powering Autism Research for Knowledge, a large-scale autism cohort consists of over 117,000 participants. Four common methods were assessed – Multiple imputation by chained equations (MICE), K-nearest neighbors (KNN), MissForest, and multiple imputation with denoising autoencoders (MIDAS). In a complete subset of 15,196 autism participants, three types of missingness patterns were simulated. We observed that MIDAS and KNN performed the best as the random missingness rate increased and when blockwise missingness was simulated. The average computational times were each 10 min for MIDAS and KNN, 35 min for MissForest, and 290 min for MICE. MIDAS and KNN both provide promising imputation performance in mental and behavioral health survey data that exhibit blockwise missingness patterns.