What Is Multimodal ADRD (Alzheimer's Disease and Related Dementias) Data?
Multimodal ADRD (Alzheimer's disease and related dementias) data refers to research data that combines two or more distinct data types collected from the same participants or cohorts from multiple biological and clinical angles at once. These data types include neuroimaging, genomic and proteomic profiles, cognitive assessments, biofluid biomarkers, and clinical records. The AD Data Initiative supports multimodal analysis through AD Workbench, which hosts data from international consortia.
ADRD is driven by overlapping biological processes, including protein misfolding, vascular changes, and neuroinflammation, that no single data type can capture in isolation. Researchers use multimodal datasets to connect biomarker changes to cognitive decline, link genetic risk factors to imaging findings, and distinguish overlapping conditions such as Alzheimer's disease and vascular dementia. As ADRD research shifts toward precision medicine, multimodal data has become central to identifying disease subtypes and predicting individual disease trajectories.
Why Does Multimodal Data Matter for ADRD Research?
Multimodal data lets researchers test whether a genetic variant, a proteomic signature, and an imaging finding are connected within the same individuals, rather than inferring connections across separate studies. This approach has been used to identify ADRD subtypes that present similarly on cognitive tests but differ substantially in underlying biology, a distinction with direct implications for treatment selection and clinical trial design.
What Types of Data Make Up a Multimodal ADRD Dataset?
A multimodal ADRD dataset typically draws on some combination of the following: structural and functional neuroimaging (MRI, PET), genomic and transcriptomic data, proteomic and metabolomic biomarker panels, cerebrospinal fluid and blood-based biomarkers, cognitive and neuropsychological assessments, and longitudinal clinical records. Datasets accessible through AD Workbench vary in which of these modalities they include, and researchers can filter the AD Discovery Portal catalog by data type to identify datasets that match a specific research question.
How Does Federated Data Analysis Support Multimodal Research?
Much of the data needed for multimodal ADRD research is held by different institutions under different governance and access requirements, which has historically limited researchers to analyzing one data source at a time. Federated data analysis, supported through the AD Data Initiative's Federated and Distributed Data Sharing Appliance (FDSA), allows permissioned researchers to query remotely hosted datasets without requiring all data to be centralized in one location. This makes it possible to combine modalities across institutions while each contributing organization retains control over its own data.
Bringing multimodal ADRD data into a single analytical environment is one of the core problems AD Workbench was built to solve. Researchers can discover multimodal datasets through the AD Discovery Portal, request access, and analyze combined data types within secure AD Workbench workspaces, available at no cost to qualified researchers.
Frequently Asked Questions
What is multimodal ADRD data?
Multimodal ADRD data combines two or more distinct data types, such as imaging, genomic, and clinical data, collected from the same participants to study Alzheimer's disease and related dementias from multiple angles at once.
How is multimodal ADRD data different from a single dataset?
A single dataset typically captures one data type, such as imaging or genetics, from a group of participants. Multimodal ADRD data links two or more data types for the same participants, allowing researchers to study how findings in one modality relate to findings in another.
Can researchers combine multimodal ADRD data from more than one institution?
Yes. Through the Federated and Distributed data Sharing Appliance (FDSA), researchers can query and combine remotely hosted datasets from multiple contributing institutions without requiring the data to be centralized in one location.
How is multimodal ADRD data different from the FDSA?
Multimodal ADRD data describes a type of data, information that combines two or more distinct data types on the same participants. The (FDSA is a product, a tool that enables researchers to access and query remotely hosted datasets across institutions.
Where can researchers access multimodal ADRD datasets?
Researchers can discover multimodal ADRD datasets through the AD Discovery Portal and analyze them by registering for AD Workbench, available at no cost to qualified researchers.