Bridging Big Data: Procedures for Combining Non-equivalent Cognitive Measures from the ENIGMA Consortium

AbstractInvestigators in the cognitive neurosciences have turned to Big Data to address persistent replication and reliability issues by increasing sample sizes, statistical power, and representativeness of data. While there is tremendous potential to advance science through open data sharing, these efforts unveil a host of new questions about how to integrate data arising from distinct sources and instruments. We focus on the most frequently assessed area of cognition - memory testing - and demonstrate a process for reliable data harmonization across three common measures. We aggregated raw data from 53 studies from around the world which measured at least one of three distinct verbal learning tasks, totaling N = 10,505 healthy and brain-injured individuals. A mega analysis was conducted using empirical bayes harmonization to isolate and remove site effects, followed by linear models which adjusted for common covariates. After corrections, a continuous item response theory (IRT) model estimated each individual subject’s latent verbal learning ability while accounting for item difficulties. Harmonization significantly reduced inter-site variance by 37% while preserving covariate effects. The effects of age, sex, and education on scores were found to be highly consistent across memory tests. IRT methods for equating scores across AVLTs agreed with held-out data of dually-administered tests, and these tools are made available for free online. This work demonstrates that large-scale data sharing and harmonization initiatives can offer opportunities to address reproducibility and integration challenges across the behavioral sciences.Significance StatementData sharing can increase the quality and rigor of scientific claims, but with large scale data sharing comes an increased need to combine non-equivalent measurements across studies. Auditory verbal learning tasks (AVLTs) are one of the most common research and clinical tools to evaluate memory constructs, but numerous distinct AVLTs are in common use, which creates challenges for data aggregation. We report methods to convert raw scores across common verbal learning instruments, constructed using harmonizing data from 53 studies from around the world. This approach can be replicated in other domains to address long standing data compatibility issues for researchers and clinicians.