Holland Bloorview Kids Rehabilitation Hospital's Bloorview Research Institute, together with DNAstack and other partners today launched Neuroscience AI, a new federated platform for collaborative neuroscience research. By enabling privacy-preserving AI-powered insights across distributed networks of data, Neuroscience AI will enable researchers to make discoveries from larger and more diverse datasets than ever before. The platform is being used to advance collaborative research in autism by connecting global members of the Autism Sharing Initiative.

Towards a national strategy for autism

Autism, or sometimes called autism spectrum disorder, refers to a broad range of conditions characterized by differences with social skills, repetitive behaviours, speech and nonverbal communication. It can be understood as a lifelong neurodevelopmental difference in which autistic people perceive the world, think, and interact with others in unique ways. There are a number of conditions that may co-occur with autism such as epilepsy, anxiety, sleep disorders, learning disabilities, and digestive hypersensitivities that can significantly affect quality of life. Each person on the autism spectrum has a distinct set of strengths and differences. There is not one autism but many subtypes, influenced by a combination of genetic and environmental factors.

A recent CDC report published in March 2023 suggests that autism is even more prevalent than previously thought, with 1 in 36 children (2.7%) being diagnosed in the USA. Canadian data from 2019 indicated a prevalence of 1 in 50 (2%). Experts say that estimates of autism prevalence are rising due to increased understanding, better diagnostics, genetics, and other factors.

On March 30th 2023, the Canadian House of Commons passed the Federal Framework on Autism Spectrum Disorder Act. The bill mandates the development of a national strategy to support autistic Canadians, their families, and their caregivers, and calls for the establishment of a national research network to promote research and improve data collection.

“The Federal Framework on Autism Spectrum Disorder Act is a critical step towards providing better support and services for autistic individuals and their families. By developing a framework that will include a national research network, we can improve our understanding of autism and its co-occurring conditions, and ultimately work towards delivering more personalized and effective care,” said Jill Farber, Executive Director at Autism Speaks Canada. “Neuroscience AI will help unlock the potential of big data and artificial intelligence to help optimize outcomes for autistic people.”

Enabling a new era of precision health

Precision health refers to a new kind of healthcare that tailors services and treatments to individuals based on in-depth analysis of their genes, biology, demographics, and environment. In the context of autism, precision health creates the opportunity to better understand subgroups of autism, leading to more individualized care with the end goal of creating better outcomes for autistic individuals.

A key challenge to enabling precision health in autism is that the core attributes and co-occurring conditions are broad and complex. Researchers therefore need to collect and analyze large, diverse, and multifaceted datasets, which are not readily available. Many research studies focus on one condition at a time, with narrowly defined cohorts. To complicate matters, these datasets are siloed across many different institutions, databases, cloud providers, and regions. Being able to combine data across all these systems and geographies, while covering a range of related conditions, has the potential to dramatically increase the power of the global research enterprise to make data-driven discoveries.

Neuroscience AI: Increasing the power of research through federated learning

Neuroscience AI increases the power of global research by streamlining the ability to perform federated learning, a new technique that makes it possible to derive insights across distributed networks of data in a privacy-preserving way. Federated learning works by aggregating the results of analyses that run within controlled environments where data are stored. Compared to the alternative approach of centralized “data sharing”, federated systems are faster, more efficient, secure, scalable, and compliant, and allow data stewards to retain control, measure impact, and automate attribution. The federated design of Neuroscience AI came out of extensive engagement with stakeholders, including participant advocacy groups, families, and individuals, who expressed the need for new techniques that promote more diversity, inclusion, privacy, and autonomy in research.

“The ability to explore multiple datasets from different jurisdictions will give us both the volume and diversity of data to have the power to make transformative discoveries for the lives of kids, youth, and adults with developmental differences and disabilities,” said Dr. Evdokia Anagnostou, vice president of research and director of the Bloorview Research Institute at Holland Bloorview Kids Rehabilitation Hospital. “Crucially, doing so without having to move the data from local environments will enable us to respect the requests of various communities to protect their privacy and confidentiality.”

Data stewards can use Neuroscience AI to increase the impact of datasets by registering them with the network. DNAstack provides software for stewards to easily integrate, harmonize, organize, protect, and register new datasets. Data consumers can, in turn, accelerate their research by discovering, accessing, and analyzing data connected to the network. DNAstack also provides software for researchers to analyze the large and growing universe of data on the network through simple user and programmatic interfaces within popular data science frameworks.

“Neuroscience AI has created the opportunity to conduct secure, inter-institutional, multivariate and integrative data analyses,” said Kristina Calli, Project Manager for the iTARGET Autism Initiative at the University of British Columbia. “This means that our one-of-a-kind multi omic data resource called ASDbase, which includes whole genome, exome, metabolome and meticulously captured mind and body whole phenome, clinome and family history data, will remain securely stored in our own research facility and the de-identified data is made accessible to approved collaborators for use in other studies. This will not only enable novel genome-wide and phenome-wide association studies but will also benefit critical independent validation of each of our own study findings within an expanded cohort of data.”

Neuroscience AI is unique in that it can perform federated learning in compliance with open standards developed by the Global Alliance for Genomics & Health (GA4GH), the de facto international standards body for genomics and related data. The system aligns with FAIR Principles by making scientific data more findable, accessible, interoperable, and reusable, while enabling data stewards and consumers to adhere to international privacy regulations like PIPEDA, HIPAA, and GDPR.

A blueprint for the future of collaborative discovery

Researchers can already start using Neuroscience AI to analyze over 50 public and controlled access datasets related to autism, Alzhiemer’s, Parkinson’s, and other areas of research. Organizations such as Autism Speaks, The Centre for Applied Genomics (TCAG) at The Hospital for Sick Children (SickKids), and University of British Columbia have connected data to the network, with Hoffmann-La Roche Limited (Roche Canada), Ontario Brain Institute, Molecular You, and others coming online soon. The ecosystem will become more powerful over time, as each new dataset and researcher that joins the network creates opportunities for new cross-analyses and associations to be made.

DNAstack collaborated with Autism Speaks Canada, the Pacific Autism Family Network, and AIMS-2-TRIALS in the development of Neuroscience AI to gather and implement feedback from members of the autism community. Including the views of autistic adults, family members, and carers will continue as the platform develops.

Neuroscience AI builds on the award-winning effort by DNAstack and partners from previous projects to deliver Viral AI, the first federated network for international genomic surveillance of infectious diseases like COVID, and demonstrates continuity of the company’s vision of making global research more diverse, equitable, inclusive, sovereign, sustainable, and impactful.

“Neuroscience AI follows our blueprint for creating the future of global collaborative research,” said Dr. Marc Fiume, CEO at DNAstack. “We believe the next generation of discoveries, clinical trials, and healthcare will be enabled by building federated ecosystems that can work together to deliver deeper insights and better outcomes for individuals and families faster than ever before.”

Read full media release here.