Foundation model for multimodal volumetric brain MRI that extends the V-JEPA 2 joint-embedding predictive architecture to 3D and combines a latent predictive objective with a sparse Mixture-of-Experts backbone (3D ViT-Base-MoE, 96×108×96 voxels). Adds foreground-aware objectives and improved masking for T1-weighted, T2-weighted, and FLAIR sequences. Pretrained on ~1.55 million curated scans from 428,647 studies across 282,693 patients at NYU Langone, and evaluated across 47 downstream tasks spanning diagnosis, prognosis, regression, multimodal learning, and time-to-event prediction over 12 public cohorts plus 3 large clinical cohorts (NYU Langone, NYU Long Island, Massachusetts General Hospital). Reports average gains of 4.4–6.4% AUROC on unimodal and 5.8–7.6% AUROC on multimodal tasks over prior neuroimaging foundation models and CNN baselines. Code is MIT; model weights are gated and released under CC-BY-NC-ND-4.0 (non-commercial, no derivatives).
Assess — committed within 3 months; permissive licence (MIT).
Derived deterministically by the MACH scorer from the signals below — not written prose.
Active — last commit ~2 mo ago; no tests/CI/releases detected
★ 78 stars · 12 forks · 1 open issues
Python 90% Jupyter Notebook 8% Shell 2%
Every value records the source it was harvested from.
| Field | Value | Source |
|---|---|---|
| description | Neuro-JEPA: Learning Sparse Latent Predictive Foundation Model for Multimodal Neuroimaging | github_api |
| discovered_conda | not_found | discovery |
| discovered_cran | not_found | discovery |
| discovered_doi | not_found | discovery |
| discovered_hf | NYUMedML/Neuro-JEPA,NYUMedML/Neuro-JEPA-Large | discovery |
| discovered_npm | not_found | discovery |
| discovered_pypi | neurojepa | discovery |
| has CI | False | github_api |
| has docs | True | github_api |
| has releases | False | github_api |
| has tests | False | github_api |
| HuggingFace downloads (monthly) | 99 | huggingface |
| judgement | Assess | mach_scorer |
| judgement reason | Assess — committed within 3 months; permissive licence (MIT). | mach_scorer |
| license | MIT | github_api+license_text |
| PyPI downloads (monthly) | rate_limited | pypi |
| archived | False | github_api |
| repository created | 2026-06-13T06:04:21Z | github_api |
| GitHub forks | 12 | github_api |
| repo health | Active | mach_health |
| repo health reason | last commit ~2 mo ago; no tests/CI/releases detected | mach_health |
| homepage | ok | github_api |
| languages | Python:90,Jupyter Notebook:8,Shell:2 | github_api |
| last commit | 2026-09-16T20:45:52Z | github_api |
| open issues | 1 | github_api |
| primary language | Python | github_api |
| repo_size_kb | 20657 | github_api |
| GitHub stars | 78 | github_api |
| topics | ok | github_api |
Recorded deviations — when a harvested value changed between runs.
| When | Field | Was | Now | Source |
|---|---|---|---|---|
| 2026-11-20T21:53:46.631300 | repo_health | — | Active | mach_health |
| 2026-11-20T21:53:46.631300 | repo_health_reason | — | last commit ~2 mo ago; no tests/CI/releases detected | mach_health |
| 2026-11-20T21:53:46.631236 | judgement | — | Assess | mach_scorer |
| 2026-11-20T21:53:46.631236 | judgement_reason | — | Assess — committed within 3 months; permissive licence (MIT). | mach_scorer |
| 2026-11-20T21:52:30.313969 | hf_downloads | — | 99 | huggingface |
| 2026-11-20T21:52:30.215744 | pypi_downloads | — | — | pypi |
| 2026-11-20T21:52:30.215659 | discovered_conda | — | — | discovery |
| 2026-11-20T21:52:30.215659 | discovered_cran | — | — | discovery |
| 2026-11-20T21:52:30.215659 | discovered_doi | — | — | discovery |
| 2026-11-20T21:52:30.215659 | discovered_hf | — | NYUMedML/Neuro-JEPA,NYUMedML/Neuro-JEPA-Large | discovery |
| 2026-11-20T21:52:30.215659 | discovered_npm | — | — | discovery |
| 2026-11-20T21:52:30.215659 | discovered_pypi | — | neurojepa | discovery |
| 2026-11-20T21:52:28.482367 | has_ci | — | False | github_api |
| 2026-11-20T21:52:28.482367 | has_docs | — | True | github_api |
| 2026-11-20T21:52:28.482367 | has_releases | — | False | github_api |
| 2026-11-20T21:52:28.482367 | has_tests | — | False | github_api |
| 2026-11-20T21:52:28.482367 | repo_languages | — | Python:90,Jupyter Notebook:8,Shell:2 | github_api |
| 2026-11-20T21:52:28.315810 | description | — | Neuro-JEPA: Learning Sparse Latent Predictive Foundation Model for Multimodal Neuroimaging | github_api |
| 2026-11-20T21:52:28.315810 | license | — | MIT | github_api+license_text |
| 2026-11-20T21:52:28.315810 | repo_archived | — | False | github_api |
| 2026-11-20T21:52:28.315810 | repo_created | — | 2026-06-13T06:04:21Z | github_api |
| 2026-11-20T21:52:28.315810 | repo_forks | — | 12 | github_api |
| 2026-11-20T21:52:28.315810 | repo_homepage | — | — | github_api |
| 2026-11-20T21:52:28.315810 | repo_last_commit | — | 2026-09-16T20:45:52Z | github_api |
| 2026-11-20T21:52:28.315810 | repo_open_issues | — | 1 | github_api |
| 2026-11-20T21:52:28.315810 | repo_primary_language | — | Python | github_api |
| 2026-11-20T21:52:28.315810 | repo_size_kb | — | 20657 | github_api |
| 2026-11-20T21:52:28.315810 | repo_stars | — | 78 | github_api |
| 2026-11-20T21:52:28.315810 | repo_topics | — | — | github_api |