Dominique Duncan, Ph.D. - US grants

ABSTRACT The overarching goal of this project is to secure, link, and disseminate BRAIN Initiative data, including electrophysiology, imaging, behavioral, and clinical data with all pertinent recording and imaging parameters, coming from participating sites. Our plan for a Data Archive for the Brain Initiative (DABI) is in response to RFA-MH-17-255. The Laboratory of Neuro Imaging (LONI) at USC has established itself as a hub for delivering effective informatics and analytics solutions in the context of big data for major projects in the study of a range of neurological diseases. LONI has demonstrated and proven experience with variations in data descriptions, data incompleteness, and data harmonization; we have built data portals and query engines for efficient search and utilization, and we have, most importantly, enabled broad data (re-)use toward accelerated discovery. Just as the ADNI (http://adni.loni.usc.edu/) project has been a powerful catalyst for success in biomarker research in Alzheimer's disease, this project has the power to foster a similar and potentially greater level of success for human neurophysiological data. Furthermore, we understand the needs of investigators who have collected these data and their concerns associated with data sharing, in addition to the privacy of the subjects from whom the data are collected. We have the extensive infrastructure in place to handle such large-scale data as well as LONI's Pipeline Processing Environment for streamlined integration of analytic tools. This platform will be designed with the flexibility to integrate input from those awarded grants to address standardization of data (RFA-MH-17-256) as well as those who design tools to interface with archives (RFA-MH-17-257). We will receive and de-identify data of various modalities from the participating sites, incorporating analysis tools previously developed at LONI and elsewhere, managing the data access systems, providing user interfaces to explore, visualize, interpret, and download the data, and provide comprehensive information about the projects and corresponding data through the public website that will be developed specifically for DABI. We aim to provide a tool for investigators that will decrease the burden of archiving the data once it has been generated. We have already created a true community by having established good communication among our group of collaborators from funded BRAIN Initiative programs. Furthermore, we will be working with industry partners that are anticipated major sources of data to specifically streamline methods for data upload from those sources.

The research objective of this proposal, Multimodal Signal Analysis and Data Fusion for Post-traumatic Epilepsy Prediction, with Pl Dominique Duncan from the University of Southern California, is to predict the onset of epileptic seizures following traumatic brain injury (TBI), using innovative analytic tools from machine learning and applied mathematics to identify features of epileptiform activity, from a multimodal dataset collected from both an animal model and human patients. The proposed research will accelerate the discovery of salient and robust features of epileptogenesis following TBI from a rich dataset, collected from the Epilepsy Bioinformatics Study for Antiepileptogenic Therapy (EpiBioS4Rx), as it is being acquired by investigating state-of-the-art models, methods, and algorithms from contemporary machine learning theory. This secondary use of data to support automated discovery of reliable knowledge from aggregated records of animal model and human patient data will lead to innovative models to predict post-traumatic epilepsy (PTE). This machine learning based investigation of a rich dataset complements ongoing data acquisition and classical biostatistics-based analyses ongoing in the study and can lead to rigorous outcomes for the development of antiepileptogenic therapies, which can prevent this disease. Identifying salient features in time series and images to help design a predictor of PTE using data from two species and multiple individuals with heterogeneous TBI conditions presents significant theoretical challenges that need to be tackled. In this project, it is proposed to adopt transfer learning and domain adaptation perspectives to accomplish these goals in multimodal biomedical datasets across two populations. Specifically, techniques emerging from d,eep learning literature will be exploited to augment data, share parameters across model components to reduce the number of parameters that need to be optimized, and use state-of-the-art architectures to develop models for feature extraction. These will be compared against established pipelines of hand-crafted feature extraction in rigorous cross-validation analyses. Developed techniques for transfer learning will be able to extract features that generalize across animal and human data. Moreover, these theoretical techniques with associated models and optimization methods will be applicable to other multi-species transfer learning challenges that may arise in the context of health and medicine. Multimodal feature extraction and discriminative model learning for disease onset prediction using novel classifiers also offer insights into biomarker discovery using advanced machine learning techniques through joint multimodal data analysis.

The goal of this 12-month project is to develop a data archive for multimodal (i.e., demographic information, clinical-outcome reports, imaging scans) and longitudinal data related to COVID-19 and to provide various statistical and analytic tools for researchers. There is an immediate need to study SARS-CoV-2 and COVID-19, and this archive provides access to data along with user-friendly tools for researchers to perform analyses to better understand COVID-19 and encourage collaboration on this research. The COVID-19 pandemic is spreading rapidly across the world, and governments are imposing travel bans, quarantine laws, business and school closings, and many other restrictions in efforts to contain the virus and limit the spread. However, much is still unknown about SARS-CoV-2 and COVID-19. There is an urgent need for scientists around the world to work together to model the virus, study how the virus has changed and will change over time, understand how it spreads, and discover a vaccine. The work from this project can also prepare scientists for future pandemics by putting the infrastructure in place to enable researchers to aggregate data and perform analyses quickly in the event of an emergency. Existing resources track how many cases are tallied per region, but lack imaging and other modalities that, when combined, will facilitate the ability for researchers to truly understand COVID-19 beyond the spread of the virus, in search of potential vaccines.

The approach is to develop a platform of networked and centralized web-accessible data archives to store multimodal data related to SARS-CoV-2 and COVID-19 and make them broadly available and accessible to the world-wide scientific community to expedite research in this area due to the urgent nature of the COVID-19 pandemic. The data will include clinical-evaluation (symptoms), vitals (spirometry, temperature, respiration rate, heart rate, etc.), demographic, geolocation, electrocardiography (EKG), computed tomography (CT), X-rays, position emission tomography (PET) and magnetic resonance imaging (MRI) data as well as other data that may be collecting in the coming months. By leveraging previous work in developing data repositories and archival capabilities at the Laboratory of Neuro Imaging at the University of Southern California, COVID-ARC (COVID-19 Data Archive) aims to provide an efficient and secure data-repository platform that facilitates data access and analysis. COVID-ARC provides tools for researchers to visualize and analyze various types of data as well as a website with tools for training, announcements, virtual information sessions, and a knowledgebase wherein researchers post questions and receive answers from the community.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.