alzheimer-research-complete / AD_DATASET_ACCESS_GUIDE.md
Satyawan1's picture
Upload AD_DATASET_ACCESS_GUIDE.md with huggingface_hub
feeb18f verified

Alzheimer's Disease & Neuroscience Dataset Access Guide

Compiled: 2026-04-05


TABLE OF CONTENTS

  1. NACC
  2. OASIS-3
  3. Bio-Hermes-001
  4. ADSP / NIAGADS
  5. HCP-Aging
  6. PREVENT-AD
  7. ANMerge (AddNeuroMed)
  8. ROSMAP
  9. Mayo Clinic Study of Aging (MCSA)
  10. AIBL
  11. DIAN
  12. AD Workbench / AD Discovery Portal
  13. GAAIN
  14. AMP-AD (via Synapse)
  15. ADNI
  16. Allen SEA-AD
  17. UK Biobank
  18. EEG/MEG Datasets
  19. Kaggle Datasets
  20. Additional/Recent Datasets

1. NACC

National Alzheimer's Coordinating Center

Field Details
Subjects 54,000+ from 39+ Alzheimer's Disease Research Centers
Apply URL https://naccdata.org/requesting-data/nacc-data/
DUA PDF https://files.alz.washington.edu/documentation/nacc_data_use_agreement.pdf

Application Requirements

  • Sign a Data Use Agreement (DUA) -- takes ~15 minutes
  • Submit electronic data request describing your research project
  • Must search for similar existing projects to ensure no significant overlap
  • Must identify a unique research hypothesis
  • Any researcher affiliated with a scientific/educational institution

Approval Timeline

  • NACC acknowledges request within 3 business days
  • Approved researchers receive data within 48 hours (excluding weekends/holidays)
  • One of the fastest turnarounds of any AD dataset

What You Get

  • Quick Access File: UDS (Uniform Data Set) standardized longitudinal data (CSV)
  • Imaging: MRI (T1w, FLAIR, DTI, T2) and PET scans in DICOM and NIfTI (.zip archives)
  • Biomarkers: CSF biomarkers, APOE genotypes, fluid biomarker data (via NCRAD)
  • Genomics: Genetic/genomic data (via NIAGADS)
  • Disease modules: FTLD, Lewy Body Dementia, Down Syndrome modules
  • Modalities: Clinical, neuropathology, MRI/PET, biospecimen, digital, EHR/Claims

Restrictions

  • Data solely for identified individuals in the request
  • Must acknowledge NACC in publications
  • Must notify NACC before and during publication submission
  • Cannot attempt to re-identify participants

2. OASIS-3

Open Access Series of Imaging Studies - 3

Field Details
Subjects 1,378 participants (ages 42-95); 755 cognitively normal, 622 with cognitive decline
Apply URL https://www.nitrc.org/projects/oasis3/ (main)
Tau PET Access https://sites.wustl.edu/oasisbrains/home/oasis-3/request-tau-access/
Contact oasis-brains@nrg.wustl.edu

Application Requirements

  • Register through NITRC (Neuroimaging Informatics Tools and Resources Clearinghouse)
  • Click "REQUEST ACCESS TO DATASETS" on the OASIS website
  • For Tau PET data: send a detailed research statement to oasis-brains@nrg.wustl.edu
  • Must already have OASIS-3 main dataset approval before requesting Tau data

Approval Timeline

  • Not explicitly stated; typically 1-2 weeks for NITRC registration
  • Tau data requires separate approval

What You Get

  • MRI: 2,842 sessions -- T1w, T2w, FLAIR, ASL, SWI, time-of-flight, resting-state BOLD, DTI
  • PET: 2,157+ scans -- PIB, AV45 (amyloid), FDG, AV1451 (Tau; 451 baseline + 85 longitudinal)
  • CT: 1,472 sessions
  • FreeSurfer: Volumetric segmentations for many MR sessions
  • Clinical: Cognitive assessments, UDS forms
  • Formats: NIfTI, BIDS-compatible
  • Size: Multiple TB total

Restrictions

  • Cannot attempt to identify participants (including facial recognition/3D rendering)
  • Publications using AV45 or AV1451 PET data must be submitted to Avid Radiopharmaceuticals for review 30 days before publication/presentation
  • As of Dec 2025, Tau data integrated into main OASIS-3 project

3. Bio-Hermes-001

Global Alzheimer's Platform Foundation

Field Details
Subjects 80,000+ blood and digital test results
Apply URL https://www.alzheimersdata.org/ad-workbench (via AD Discovery Portal)
Study Page https://globalalzplatform.org/biohermesstudy/

Application Requirements

  • Create a free account on AD Workbench (alzheimersdata.org)
  • Request access through the AD Discovery Portal
  • Accept Terms of Use
  • Available since August 1, 2025

Approval Timeline

  • Varies by dataset; generally days to weeks after account approval

What You Get

  • Blood biomarkers: Comprehensive plasma/serum biomarker panels
  • Digital cognitive tests: Novel digital cognitive assessment data
  • Retinal imaging: Retinal exam data
  • Speech analysis: Voice/speech biomarker data
  • PET imaging: Traditional amyloid/tau PET
  • Deeply diverse cohort: One of the most diverse AD datasets available
  • Most comprehensive biomarker dataset in Alzheimer's research history

Restrictions

  • Must use data for Alzheimer's/dementia research purposes
  • Must accept AD Workbench Terms of Use
  • Free access, no cost

4. ADSP / NIAGADS

Alzheimer's Disease Sequencing Project / NIA Genetics of AD Data Storage Site

Field Details
Subjects 110,270+ (58,507 whole genomes + 20,503 whole exomes + more)
Apply URL https://dss.niagads.org/
Application Instructions https://dss.niagads.org/documentation/data-application-and-submission/application-instructions/
ADSP Umbrella Dataset https://dss.niagads.org/datasets/ng00067/

Application Requirements

  1. Current IRB approval and protocol for the proposed project (must have 6+ months remaining)
  2. NIA Genomic Data Sharing Plan (signed by PI and Institutional Signing Official)
  3. NIAGADS Data Distribution Agreement
  4. Derived/Secondary Data Return Plan describing what data you will return
  5. Research use statement (technical and non-technical)

Approval Timeline

  • Applications reviewed by Data Access Committee
  • Typically 4-8 weeks (depends on completeness of application)
  • Approval valid for 1 year (renewable)
  • IRB must have 6+ months validity at time of review

What You Get

  • Whole Genome Sequences: 58,507 in CRAMs, gVCFs
  • Whole Exome Sequences: 20,503 in CRAMs, gVCFs
  • Quality-controlled VCFs: Project-level variant calls
  • Harmonized phenotypes: Standardized clinical data
  • Formats: CRAM, gVCF, VCF, CSV

Restrictions

  • Must return derived/secondary data to NIAGADS upon publication or DAR expiration
  • Data only for approved research use statement
  • Local IRB approval required
  • Cannot share with unauthorized users

5. HCP-Aging

Human Connectome Project - Aging / AABC

Field Details
Subjects 1,396+ adults (ages 36-100+), 2,878 sessions
Apply URL https://nda.nih.gov/ (via NIMH Data Archive permissions dashboard)
Data Use Terms https://www.humanconnectome.org/study/hcp-lifespan-aging/data-use-terms
Instructions https://www.humanconnectome.org/study/hcp-lifespan-aging/article/instructions-accessing-hcp-aging-data-releases-nda

Application Requirements

  1. Create NDA (NIMH Data Archive) account
  2. Submit Data Use Certification (DUC) via NDA Permissions Dashboard
  3. Describe how data will be accessed, managed, and eventually deleted
  4. Institutional sign-off required
  5. Annual renewal with progress report

Approval Timeline

What You Get

  • Lifespan 2.0 Release: 725 HCP-A participants + AABC Release 2 (1,396 participants)
  • Structural MRI: T1w, T2w, high-res hippocampal T2
  • Functional MRI: Resting state fMRI, task fMRI
  • Diffusion MRI: DTI
  • ASL: Arterial spin labeling perfusion
  • Phenotypic data: Demographics, behavioral assessments
  • Total size: 22+ TB
  • Formats: NIfTI, CIFTI, CSV

Restrictions

  • Must adhere to consent-based data use limitations
  • Cannot attempt to re-identify participants
  • Must describe data management/deletion plan
  • Annual renewal required

6. PREVENT-AD

Pre-symptomatic Evaluation of Novel Treatments for AD

Field Details
Subjects 349 cognitively healthy at-risk participants (mean age 63)
Open Data Portal https://openpreventad.loris.ca
Registered Data Portal https://registeredpreventad.loris.ca
Publication https://doi.org/10.1016/j.nicl.2021.102733

Application Requirements

  • Open imaging data: Freely accessible at openpreventad.loris.ca -- just register
  • Sensitive data (CSF, genetics, cognition): Apply at registeredpreventad.loris.ca
  • Must agree to standard good data use practices
  • Must meet ethics requirements and keep data secure
  • Findable through Canadian Open Neuroscience Platform (CONP)

Approval Timeline

  • Open data: Immediate after registration
  • Registered data: Days to weeks for qualified researcher approval

What You Get

  • Imaging: Up to 5 years longitudinal MRI data (structural, functional)
  • Biomarkers: Cerebrospinal fluid biochemistry
  • Genetics: Genetic information
  • Cognitive: Neurocognitive assessments
  • Neurosensory: Sensory capacity measurements
  • Medical: Clinical/medical information
  • Formats: BIDS-compatible NIfTI, CSV

Restrictions

  • Must use for neuroscience research as stipulated in consent forms
  • Cannot attempt to re-identify participants
  • Must acknowledge PREVENT-AD in publications

7. ANMerge (AddNeuroMed)

Field Details
Subjects 1,702 participants
Access URL https://doi.org/10.7303/syn22252881 (via Synapse)
Publication https://doi.org/10.3233/JAD-200948

Application Requirements

  • Create a free Synapse account (accounts.synapse.org)
  • Accept data use conditions on Synapse
  • No complex approval process -- this is an open-access dataset

Approval Timeline

  • Immediate to days -- relatively straightforward once Synapse account is set up

What You Get

  • Clinical assessments: Longitudinal observational cohort data
  • MRI: Magnetic resonance imaging
  • Genotyping: Genetic variants
  • Transcriptomics: Gene expression profiling (whole-blood RNA)
  • Proteomics: Blood plasma proteomics
  • Formats: CSV, processed data tables
  • Fully interoperable between modalities with rigorous data curation

Restrictions

  • Must cite the dataset and primary publication
  • Standard Synapse data use conditions apply

8. ROSMAP

Religious Orders Study / Memory and Aging Project

Field Details
Subjects 3,600+ participants (longitudinal since early 1990s)
AD Knowledge Portal https://adknowledgeportal.synapse.org/Explore/Studies/DetailsPage/StudyDetails?Study=syn3219045
RADC Hub https://www.radc.rush.edu/
Molecular Networks https://www.radc.rush.edu/molecular_networks/datasets.html

Application Requirements

  • Omics data on Synapse: Register for free Synapse account; some datasets require a signed Data Use Certificate (DUC)
  • Clinical/demographic data: Request through Rush Alzheimer's Disease Center (RADC) Research Resource Sharing Hub
  • Additional phenotypes: Separate request through RADC

Approval Timeline

  • Synapse open data: Days (account registration)
  • DUC-protected data: 2-4 weeks for approval
  • RADC data: Variable, typically weeks

What You Get

  • Genomics: Whole-genome sequencing
  • Transcriptomics: RNA-seq, single-nucleus RNA-seq
  • Epigenomics: DNA methylation, histone modifications, chromatin accessibility
  • Proteomics: Mass spectrometry-based proteomics
  • Metabolomics: Metabolite profiling
  • Clinical: Longitudinal clinical assessments (since 1990s)
  • Neuropathology: Post-mortem brain tissue analyses
  • Formats: FASTQ, BAM, VCF, CSV, H5AD

Restrictions

  • Must acknowledge data source in publications
  • DUC-protected datasets (esp. Mayo/Broad samples from deceased individuals) have additional consent requirements
  • Must use for research purposes consistent with informed consent

9. Mayo Clinic Study of Aging (MCSA)

Application Requirements

  • Apply via LONI IDA or GAAIN
  • Agree to Data Use Agreement
  • Describe research project and collaborators
  • Must be a qualified academic or industry researcher

Approval Timeline

  • 2-4 weeks for LONI IDA review
  • GAAIN requests reviewed individually

What You Get

  • Clinical data: Longitudinal data, 1-12 visits at ~15-month intervals
  • MRI: T1w, T2w-FLAIR, diffusion MRI from 1,802 participants
  • Future releases: De-faced amyloid PET images planned
  • Molecular data: Whole-genome genotype + gene expression from 2,655 individuals (842M+ datapoints)
  • Interactive tool: Multiomic Atlas of AD Brain Endophenotypes (free web app)
  • Formats: NIfTI, CSV

Restrictions

  • Standard DUA restrictions apply
  • Cannot re-identify participants
  • Must acknowledge Mayo Clinic in publications

10. AIBL

Australian Imaging, Biomarkers and Lifestyle

Application Requirements

  • Apply via LONI IDA online form
  • Read and agree to AIBL project Terms of Use (carefully)
  • Describe your research and list collaborators
  • Can apply jointly for AIBL + ADNI data

Approval Timeline

  • 2-4 weeks (similar to ADNI process via LONI)

What You Get

  • Neuroimaging: Amyloid PET (PiB, flutemetamol), FDG PET, structural MRI
  • Blood biomarkers: Amyloid-beta, tau, inflammatory markers
  • CSF biomarkers: Cerebrospinal fluid analytes
  • Cognitive assessments: Battery of neuropsychological tests
  • Genetics: APOE genotyping and broader genetic data
  • Lifestyle: Diet, exercise, sleep, social engagement data
  • Longitudinal: Assessments every 18 months
  • Formats: DICOM/NIfTI (imaging), CSV (clinical)

Restrictions

  • Must comply with AIBL Terms of Use
  • Acknowledge AIBL in publications
  • Cannot re-identify participants

11. DIAN

Dominantly Inherited Alzheimer Network

Field Details
Subjects 533 individuals across 206 families (autosomal dominant AD)
Data Request Form https://dian.wustl.edu/dian-observational-data-request-form/
Website https://dian.wustl.edu
Investigator Resources https://dian.wustl.edu/for-investigators/

Application Requirements

  • Submit DIAN Observational Data Request Form
  • Research proposal reviewed by DIAN Obs Resource Committee
  • Must be a qualified researcher
  • Must accept and comply with DIAN data sharing/publication policies
  • Strict publication policy: all publications using DIAN data must follow their guidelines

Approval Timeline

  • Weeks to months -- committee review required
  • More restrictive than most datasets

What You Get

  • MRI: Structural and functional brain imaging
  • PET: Amyloid and tau PET scans
  • Clinical: Longitudinal cognitive/clinical assessments
  • Biofluid: CSF and blood biomarkers
  • Genetics: Deep genetic phenotyping (PSEN1, PSEN2, APP mutations)
  • 15+ years of longitudinal data on autosomal dominant AD
  • Unique value: Only large-scale dataset on dominantly inherited (genetic) AD

Restrictions

  • Strict publication and authorship policies
  • Violations can result in being barred from future data/biospecimen requests
  • Potential institutional involvement or legal action for policy deviations
  • Must comply with DIAN-TU Data and Biospecimen Sharing Policy

12. AD Workbench / AD Discovery Portal

Field Details
Datasets 100+ novel datasets across multiple modalities
Portal URL https://discover.alzheimersdata.org
Workbench URL https://www.alzheimersdata.org/ad-workbench
How-To Guide https://www.alzheimersdata.org/how-to-use-the-ad-workbench

Application Requirements

  • Create a free user account on AD Workbench
  • Account creation is automatic; data permissions require review
  • Use FAIR Search for data discovery
  • Request workspace for analysis
  • Accept Terms of Use

Approval Timeline

  • Account creation: Immediate
  • Data access permissions: Variable per dataset (days to weeks)

What You Get

  • 100+ datasets: Imaging, omics, clinical, multi-modal
  • Cloud workspaces: Secure, private analysis environments
  • Tools: Data visualization, curation, combination tools
  • Bio-Hermes-001: Available through this portal
  • Free: No cost for any tool or data access

Restrictions

  • Must be approved by ADDI
  • Data use per individual dataset terms
  • Research purposes only

13. GAAIN

Global Alzheimer's Association Interactive Network

Field Details
Subjects ~500,000 from nearly 50 institutions worldwide
Portal URL https://www.gaaindata.org
Website https://gaain.org

Application Requirements

  • Register on gaaindata.org
  • Free access for researchers worldwide
  • Use the Interrogator tool for cohort discovery and analysis

Approval Timeline

  • Immediate to days for most federated queries
  • Individual partner datasets may have their own access requirements

What You Get

  • Federated data platform: Query across multiple cohorts simultaneously
  • Clinical data: Cognitive scores, demographics, diagnoses
  • Imaging data: Various imaging modalities from partner studies
  • Genomics: Genetic data from contributing cohorts
  • Analytics tools: Built-in analytics and visualization
  • Partner datasets: MCSA, ADNI, and many international cohorts

Restrictions

  • Individual partner datasets retain their own data use policies
  • Cannot download all raw data -- federated query model
  • Must acknowledge GAAIN and contributing studies

14. AMP-AD (via Synapse)

Accelerating Medicines Partnership - Alzheimer's Disease

Application Requirements

  • Register for free Synapse account
  • Browse public content freely
  • Download data requires Synapse login
  • Some datasets require signed Data Use Certificate (DUC)
  • DUC datasets (esp. from Mayo Clinic/Broad Institute with deceased donor samples) have additional review

Approval Timeline

  • Open data: Immediate after registration
  • DUC-protected data: 1-4 weeks

What You Get

  • Multi-omics: Genomics, transcriptomics, epigenomics, proteomics, metabolomics
  • Studies include: ROSMAP, MayoRNAseq, MSBB (Mount Sinai Brain Bank), many more
  • Clinical: Longitudinal clinical and neuropathological data
  • Tools: Analysis pipelines, pre-computed results
  • Formats: FASTQ, BAM, VCF, CSV, H5AD, AnnData

Restrictions

  • Data use conditions per informed consent of each study
  • Must acknowledge AMP-AD and NIA
  • Some datasets have embargo periods for new data

15. ADNI - BONUS ESSENTIAL DATASET

Alzheimer's Disease Neuroimaging Initiative

Field Details
Subjects 2,000+ across ADNI-1, ADNI-2, ADNI-GO, ADNI-3, ADNI-4
Apply URL https://ida.loni.usc.edu/collaboration/access/appApply.jsp
Website https://adni.loni.usc.edu/

Application Requirements

  • Review and agree to ADNI Data Use Agreement
  • Application reviewed by Data Sharing and Publications Committee (DPC)
  • Must be affiliated with a scientific or educational institution
  • Describe proposed research or data use

Approval Timeline

  • ~2 weeks for DPC review

What You Get

  • MRI: Structural, functional, DTI
  • PET: Amyloid (AV45/PiB), Tau (AV1451), FDG
  • Biomarkers: CSF (amyloid-beta, tau, p-tau), blood biomarkers
  • Genomics: GWAS, WGS, WES
  • Clinical: Longitudinal cognitive/clinical assessments
  • Formats: DICOM, NIfTI, CSV

Restrictions (IMPORTANT - 2025 UPDATE)

  • New AI restriction: ADNI DUA now explicitly forbids use of external AI tools on the data
  • AI tools restricted to within university/company (no external release allowed)
  • Cannot share data with others
  • Must acknowledge ADNI in all publications
  • Must follow ADNI publication policies

16. Allen SEA-AD

Seattle Alzheimer's Disease Brain Cell Atlas

Field Details
Subjects 84 donors spanning AD pathology spectrum
Open Data https://portal.brain-map.org/explore/seattle-alzheimers-disease/seattle-alzheimers-disease-brain-cell-atlas-download
AWS Registry https://registry.opendata.aws/allen-sea-ad-atlas/
Controlled Data Via AD Knowledge Portal (Synapse)

Application Requirements

  • Open/processed data: No application needed -- freely downloadable
  • Raw sequencing data: Apply through Synapse AD Knowledge Portal

Approval Timeline

  • Open data: Immediate
  • Controlled raw data: 1-4 weeks via Synapse

What You Get

  • snRNA-seq: Single-nucleus RNA sequencing
  • snATAC-seq: Single-nucleus chromatin accessibility
  • Multiome: Combined RNA + ATAC from same nuclei
  • Neuropathology: Quantitative pathology data
  • Spatial transcriptomics: MERFISH data
  • Formats: H5AD, AnnData, CSV, FASTQ (raw)
  • Already in your project: You have SEA-AD metadata in /data/allen_sea_ad/

Restrictions

  • Must cite per Allen Institute Citation Policy
  • Cite both primary publication and specific dataset

17. UK Biobank

Field Details
Subjects 500,000+ (26,000+ with brain MRI)
Apply URL https://www.ukbiobank.ac.uk/enable-your-research/register
Website https://www.ukbiobank.ac.uk

Application Requirements

  • Register as a researcher
  • Submit research application describing study
  • Institutional affiliation required
  • Application fee: ~2,000-5,000 GBP (varies by data type)
  • Ethics approval may be required

Approval Timeline

  • Several weeks to months for full approval

What You Get

  • Brain MRI: 26,000+ participants with structural/functional imaging
  • 4,000+ imaging-derived phenotypes: Pre-computed brain measures
  • Genetics: Genome-wide genotyping, exome sequencing, WGS
  • Clinical: GP records, hospital admissions, cognitive tests
  • Lifestyle: Diet, exercise, socioeconomic data
  • Longitudinal: Repeat visits over 15+ years
  • Formats: Various (bulk data downloads)

Restrictions

  • Application fee required
  • Strict data security requirements
  • Must return results/findings
  • UK-based ethics oversight
  • Not AD-specific but massive AD-relevant subset

18. EEG/MEG Datasets for AD

18a. OpenNeuro EEG AD Dataset (ds004504)

Field Details
Subjects 88 (36 AD, 23 FTD, 29 healthy)
URL https://openneuro.org/datasets/ds004504
Access Free, immediate download
Format BIDS-compliant EEG (EDF)
License CC0 (public domain)
Content Resting state EEG (eyes closed), raw + preprocessed

18b. Complementary Photic Stimulation EEG Dataset (2025)

Field Details
Subjects Same 88 participants as ds004504
Content Eyes-open photic stimulation recordings
Format BIDS-compliant
Published April 2025

18c. PEARL-Neuro Database

Field Details
Subjects 192 middle-aged (50-63) at-risk participants
URL https://openneuro.org/datasets/ds004796
Content EEG + fMRI + APOE/PICALM genetics + psychometric tests + blood tests
Access Open access via OpenNeuro
Format BIDS-compliant
Unique value Multi-modal (EEG + fMRI) with genetic risk factors

18d. LEAD Corpus (Research Reference)

Field Details
Subjects 813 across 9 combined datasets (330 public, 483 private)
Publication https://arxiv.org/html/2502.01678v1
Note World's largest EEG-AD corpus; public subset downloadable, private portion restricted

19. Kaggle AD Datasets

19a. Alzheimer MRI 4-Class Dataset

Field Details
URL https://www.kaggle.com/datasets/sachinkumar413/alzheimer-mri-dataset
Classes Non-Demented, Very Mild, Mild, Moderate Demented
Size ~6,400 images
Format JPG/PNG
Access Free, immediate download

19b. Augmented Alzheimer MRI Dataset

Field Details
URL https://www.kaggle.com/datasets/uraninjo/augmented-alzheimer-mri-dataset
Content Augmented MRI slices for classification
Access Free

19c. Alzheimer's Disease Clinical Dataset

Field Details
URL https://www.kaggle.com/datasets/rabieelkharoua/alzheimers-disease-dataset
Content Clinical features, demographics, lifestyle, cognitive scores
Access Free

19d. OASIS-derived Kaggle Dataset

Field Details
URL https://www.kaggle.com/datasets/jboysen/mri-and-alzheimers
Content OASIS cross-sectional and longitudinal MRI data
Access Free

Note: Kaggle datasets are great for prototyping and model development but are NOT suitable for clinical validation or publications requiring primary data.


20. Additional & Recent Datasets (2025-2026)

20a. OASIS-4 (NEW)

  • Latest release in the OASIS series
  • MR, clinical, cognitive, and biomarker data for individuals with memory complaints
  • Access: Same as OASIS-3 via sites.wustl.edu/oasisbrains

20b. CLARiTI (via NACC - NEW)

20c. Allen Brain Cell Atlas (Broader)

20d. ADNI-4 (Latest Phase)

  • URL: https://adni.loni.usc.edu/
  • Newest ADNI phase with updated protocols
  • Blood-based biomarker focus
  • NOTE: New AI restrictions in DUA

20e. Bio-Hermes-002 (Upcoming)

  • GAP Foundation + Alamar Biosciences collaboration announced January 2026
  • Next-generation biomarker study building on Bio-Hermes-001
  • Watch: https://globalalzplatform.org/

SYNTHETIC / AUGMENTED AD DATA APPROACHES

For when real data is insufficient or for pre-training:

Approach Description Reference
CycleGAN MRI augmentation Generate synthetic MRI scans; achieved 95% F1 (vs 89% without) Frontiers in Medicine 2025
SMOTE for tabular data Oversample minority AD classes in clinical datasets Multiple papers 2025
Diffusion model MRI generation Generate 2D slice projections of 3D MRI scans ScienceDirect 2026
3D CNN with data augmentation Standard augmentation (flip, rotate, scale) on 3D MRI arxiv 2505.04097
AdaBoost synthetic generation Boost training data diversity for clinical features PMC 2025

PRIORITY APPLICATION ORDER (Recommended)

Based on ease of access, data richness, and relevance to multi-modal AD research:

Tier 1 -- Apply Immediately (Fast Approval, High Value)

  1. NACC -- 48hr turnaround, 54K subjects, multi-modal
  2. AD Workbench / Bio-Hermes-001 -- Free account, 80K+ results, diverse cohort
  3. ANMerge -- Open on Synapse, 1,702 subjects, multi-modal
  4. Allen SEA-AD -- Open download, single-cell multi-omics (you already have metadata)
  5. OpenNeuro EEG datasets -- Immediate free download, BIDS format
  6. Kaggle datasets -- Immediate, good for prototyping

Tier 2 -- Apply This Week (1-4 Week Approval)

  1. OASIS-3 -- NITRC registration, rich imaging data
  2. AMP-AD / Synapse -- Free account, massive multi-omics
  3. ROSMAP -- Via Synapse + RADC, deep longitudinal omics
  4. AIBL -- Via LONI IDA, good imaging + lifestyle data
  5. ADNI -- ~2 week review, gold standard (watch AI restrictions)
  6. MCSA -- Via GAAIN/LONI, large clinical + imaging release

Tier 3 -- Apply When Ready (Longer Approval, More Requirements)

  1. HCP-Aging -- NDA DUC required, 22+ TB connectome data
  2. ADSP/NIAGADS -- IRB required, massive genomics
  3. DIAN -- Committee review, unique genetic AD data
  4. UK Biobank -- Fee required, months for approval, massive scale
  5. PREVENT-AD registered -- For pre-symptomatic biomarkers
  6. GAAIN -- Federated queries across 500K subjects

QUICK REFERENCE: ALL APPLICATION URLS


ESTIMATED TOTAL DATA AVAILABLE

Category Approximate Scale
Total unique subjects across all datasets ~700,000+
Genomics subjects ~150,000+ (ADSP, ROSMAP, UK Biobank)
Neuroimaging subjects ~50,000+ (ADNI, OASIS, HCP, NACC, AIBL, UK Biobank)
Clinical/cognitive subjects ~600,000+ (NACC, UK Biobank, GAAIN)
Single-cell omics ~84 donors, millions of cells (SEA-AD)
EEG subjects ~1,000+ (OpenNeuro, PEARL-Neuro, LEAD corpus)
Blood biomarkers ~80,000+ results (Bio-Hermes-001)
Pre-symptomatic/at-risk ~2,000+ (PREVENT-AD, DIAN, HCP-Aging)