Population neuroscience requires investigation of brain-behavioral relationships within epidemiological samples studied in longitudinal designs, including a large number of assessment modalities and with subjects enrolled at many data collection sites to achieve a large sample and broad representation of the population. Of note, the Adolescent Brain Cognitive Development (ABCD) Study and the Human Connectome Project Development (HCP-D and HCP-A), have, in aggregate, enrolled thousands of individuals, many as children and adolescents, and are expected to follow them for up to ten years. Foreseeing the need for an accessible analysis platform to foster rigorous modeling, the ABCD Data Analysis and Informatics Center (DAIC) has developed and deployed the web-based Data Exploration and Analysis Portal (DEAP). Despite its nascent stage, the present version of DEAP already provides a flexible tool that meets many needs of both data scientists and applied investigators. Here, we propose to produce a more versatile extension of DEAP, the Data Exploration and Analysis Portal for Brain Research through Advanced Imaging Neuroscience (DEAP- BRAIN). We will implement important new features by incorporating approaches for: i) including HCP data to assess generalizability; ii) newly developed poly-vertex score as a method to summarize generalizability of whole brain patterns; and iii) bias assessment based on causal inference methods. By the completion of this proposal, we will facilitate reproducible and valid scientific discovery through providing the research community with an easy to use but sophisticated informatics tool.
The Specific Aims of the proposal are:
Specific Aim 1 : Integrating HCP data into DEAP-BRAIN. We will integrate ABCD and HCP data through homogenizing the imaging processing and variable conventions. The HCP adult cohort and HCP lifespan development cohort will be integrated in DEAP, allowing a formal assessment of generalizability.
Specific Aim 2 : Implementing poly-vertex scores (PVS) as the framework for assessing generalizability of brain mapping. We will implement our newly developed PVS method, an innovative approach to summarize the reproducibility of association patterns across brain.
Specific Aim 3 : Implementing bias assessment modules. We will implement the bias assessment module for DEAP users based on causal inference principles, using propensity scores, principal component methods, and synthetic controls.
We propose to develop a web-based informatics tool, Data Exploration and Analysis Portal for Brain Research through Advanced Imaging Neuroscience (DEAP-BRAIN), that is capable of integrating data from both ABCD and HCP. The proposed project will deploy analytic modules that maximize the probability of reproducible and valid multi-modal imaging analyses, by integrating HCP data as a replication cohort, incorporating a novel tool (polyvertex score, PVS) as a tool to assess generalizability, and incorporating a sophisticated bias assessment module using causal inference methods. DEAP-BRAIN will significantly facilitate the investigation of the complex process impacting neurodevelopment, providing research communities a platform for robust, valid and reproducible analyses. 1
