Authors :
Presenting Author: Craig Press, MD, PhD, FCNS – Children's Hospital of Philadelphia
Benedetti Giulia, MD – University of Michigan School of Medicine
Dana Harrar, MD, PhD – Center for Neuroscience and Behavioral Health, Children's National Hospital, George Washington University
Raquel Farias-Moeller, MD – Medical College Wisconsin
Stuart Tomko, MD – Washington University
Anuj Jayakar, MD – Nicklaus Children's Hospital
Brian Appavu, MD – Barrow Neurological Institute at Phoenix Children's Hospital
Lindsey Morgan, MD – Seattle Childrens Hospital
Carlos Castillo-Pinto, MD – Seattle Children's Hospital
Ajay Thomas, MD, PhD – Baylor College of Medicine/Texas Children's Hospital
Stephanie Rau, B.S., C.C.R.P. – C.S. Mott Children’s Hospital, University of Michigan, Ann Arbor
Maelyn Fulton, BS – C.S. Mott Children’s Hospital, University of Michigan, Ann Arbor
Darshana Parikh, BS – University of Pennsylvania - Children's Hospital of Philadelphia
Nicholas Abend, MD – Children's Hospital of Philadelphia and University of Pennsylvania
Rishi Ganesan, MBBS MD DM FACNS – Children's Hospital - London Health Sciences Centre
Darrell De Freitas, BS – Department of Biostatistics, Epidemiology and Informatics, University of Pennsylvania;
Yomi Dare, BS – Department of Biostatistics, Epidemiology and Informatics, University of Pennsylvania;
James Riviello, MD – Baylor College of Medicine/Texas Children's Hospital
Cecil Hahn, MD – The Hospital for Sick Children, and Department of Paediatrics University of Toronto
Joost Wagenaar, PhD – University of Pennsylvania
Rationale:
Electroencephalography (EEG) is the most common tool for acute brain monitoring. Quantitative EEG (qEEG) allows quantitative analysis, visualization, and bedside display of EEG data. Obstacles to EEG research include inability to efficiently share and deidentify EEG files, lack of secure data platforms for waveform and clinical data, and non-standardized analytic pipelines.Methods:
This is an infrastructure development project for multicenter EEG requiring: (1) regulatory infrastructure as part of a research consortium, the Pediatric Quantitative EEG Strategic Taskforce (PedQuEST); (2) a data extraction system automating EEG data collection, deidentification, and transfer; and (3) a state-of-the-art research platform for EEG storage and analysis linked to clinical data.Results:
Regulatory: For the consortium we developed a data sharing agreement and centralized IRB protocol that created the legal and regulatory structure for data sharing between 10 sites with a data coordinating center and research platform (Fig 1). The agreement outlines data governance for current/future projects. It defines a Scientific Committee for evaluating and approving studies using consortium data. It creates pathways for adding new sites and future studies, ensuring data safety, and high-quality research. Our Scientific Committee also supports scientific project proposals with mentorship and advisory support for investigators embarking on pediatric qEEG research.
Data Extraction: We developed a novel program using Persyst (15b, Research) and Python programming that is installed on a virtual or local server. We are able to index all clinical EEGs with associated EEG metadata. EEGs meeting study criteria are identified, and stored in Persyst format as date shifted, deidentified files. Files are stored with a key including days from birth for each EEG allowing data to be aligned in time with other clincial data without HIPAA identifiers. Persyst file format can be accessed in both research and clinical platforms. Files are uploaded with verified data integrity to the Pennsieve platform.
Data Platform: Pennsieve is a cloud-based multi-tenant platform for scientific data collaboration, integration, and analysis created by the Wagenaar lab (Fig 2). The data management tools allow investigators to access their own data securely, while multisite data can be aggregated, analyzed, or packaged for approved external analysis. Data can be analyzed using centralized pipelines, including custom programming or use of centralized commercial platform processing to allow more rapid translation of research to beside clinical care.
Conclusions:
The development of a multicenter EEG data platform with data sharing and IRB regulatory approval for pilot and future studies eliminates many of the obstacles that have hindered pediatric EEG research while maintaining data safety and integrity. The structure and governance enables research that was not feasible without adequate subject and EEG volume which can only be achieved efficiently with multi-center collaboration.Funding:
This work was funded by The Pediatric Epilepsy Research Foundation.