Data Management and Statistical Core

ABSTRACT – ALLFTD2: DATA MANAGEMENT AND STATISTICAL CORE The Data Management and Statistical Core (DMSC) supports the goals of ARTFL LEFFTDS Longitudinal Frontotemporal Lobar Degeneration, Cycle 2 (ALLFTD2) by maintaining a secure state-of-the art relational database for the collection of participant data, by sharing these data with the other ALLFTD2 Cores, Projects, national data repositories, and investigators both internal and external to ALLFTD2, and by assisting with the statistical planning and analysis across ALLFTD2. Using the previously constructed 21 CFR Part 11 Compliant ALLFTD database, the DMSC will improve existing and develop new automated tools for real-time data entry and checking, reporting back to the cores and clinical sites with data issues and educating to ensure higher quality for future data entries. The DMSC will continue to work with and develop new tools for electronic data capture when feasible. The DMSC will work closely with the other cores and clinical sites to ensure efficient and accurate data entry and capture. The DMSC will assist with patient tracking and recruitment and retention of study participants and assist with exchange of participant samples and data within the ALLFTD2 consortium and with other research groups including the National Alzheimer’s Coordinating Center (NACC), National Centralized Repository for Alzheimer's Disease and Related Dementias (NCRAD), the Laboratory of Neuroimaging (LONI) and the FTD Disorders Registry (Registry). The DMSC will share data between the Cores and with the Projects as well as investigators internal and external to ALLFTD2. The DMSC will provide statistical support to all cores and projects, assisting with study design, statistical planning, analysis and interpretation of results, and work with the development and implementation of new methods when needed. The DMSC’s methods development includes new methods for the derivation of improved normative values for the psychometric measures used in ALLFTD2, Bayesian Mixed Effects Models to obtain better estimates of whole-brain structural trajectories and disease age from longitudinal data, and machine learning methods for analysis of high dimensional data.