30294
Loris Platform for Automating Clinical Workflows and Multimodal Data Management
The growing quantity and diversity of data collections in autism research require robust data systems for harmonizing multi-modal cohorts across large-scale distributed collaborations. Enabling researchers to query and manage diverse brain, behaviour and genotypic data types in one comprehensive platform is pivotal to the future of neurodevelopmental research. By leveraging the computational customizability of specialized neuroinformatics platforms, research teams can optimize labour-intensive clinical workflows and accelerate the return of results.
Objectives:
LORIS is an open-source web-based data management software developed at the Montreal Neurological Institute (MNI). Its modular design aims to automate high-value processes and tasks from data capture, quality control, visualization and analysis workflows. Embedded tools for normalized scoring and data verification provide computational validation for labour-intensive tasks otherwise requiring clinical expertise.
Native clinical and imaging workflows are customizable and extensible across other data modalities supported by LORIS, including electrophysiological (EEG), biobanking and summary genetic data types. Coupled with the CBRAIN processing platform and embedded BrainBrowser visualization tool, this combined neuroinformatics environment endeavors to address the challenges of large-scale, multi-modal data collection analysis.
Methods:
Complex instrument scoring using norms tables and scales can be computed automatically in LORIS during data entry, reducing error and time-cost in manual scoring. Clinical algorithms can be updated and re-calculated to validate accuracy and reproducibility of results.
Multi-modal querying of longitudinal cohorts using the Data Querying Tool enables users to build, save and export queries across timepoints. At-a-glance data summaries are visualized in interactive Dashboard graphs, and detailed in the Statistics module.
Salient demographic variables are easily accessed and queried along with longitudinal data. The Candidate Profile module transparently captures top-level enrolment, demographic, and consent information at every stage of a study. Related participants are cross-linked through the Family Information sub-module.
Results:
In autism and neurodevelopmental research, projects such as the IBIS network, the Baby Connectome Project, and the NIHPD database use LORIS as a combined gene-brain-behaviour longitudinal data platform. In such instances, clinical norm algorithms are embedded in behavioural forms, providing instant return of results to clinicians. No external calculations or manual checks are needed to validate complex scores. These initiatives have also contributed significantly to the data standards and interoperability of the NIMH Data Archive (NDA).
For large, multidisciplinary data collections, this native ability to link and query participant data across modalities in one system conserves time and expertise that would otherwise be spent coordinating and maintaining multiple data systems in parallel.
Recent integrations of EEG data and interoperability with the BIDS data structure increase the global utility of LORIS. Interoperability with processing platforms such as CBRAIN facilitates the flow of data from storage platform to analysis toolkits.
Conclusions:
Customizable for large-scale multi-modal collections, the LORIS platform allows investigators to amplify the translational impact of data assets in autism and neurodevelopmental research. Equipped with powerful tools for cross-modal querying of longitudinal datasets, verifiable scoring of complex algorithms, and multi-level data validation and review, research teams can reduce error and efficiently utilize clinical expertise via a single neuroinformatics platform.