Thrust 1 — Imaging Biomarkers
Automated DaT-SPECT and T1 MRI pipelines generate harmonized biomarker matrices that sharpen CSFSAA stratification and feed downstream agents.
Progressive AI for Parkinson's care
We build actionable, multimodal agents that differentially diagnose Parkinson's mechanisms, surface subtype-specific biomarkers, and keep neurologists in the loop with interactive decision support.
Two coordinated pillars—domain thrusts and clinical workflows—move harmonized data from cohorts into bedside insight.
The problem we are solving
- Parkinson’s disease is heterogeneous and multisystem; current care pathways remain largely symptomatic with limited disease-modifying success.
- Clinical scales are episodic and noisy—objective imaging, diffusion, biospecimen, and wearable markers tighten risk stratification.
- Multimodal data remains siloed; coordinated inference unlocks individualized trajectories and intervention planning.
Our architecture
We harmonize PPMI and allied cohort assets—DaT-SPECT, T1 and diffusion MRI, biospecimens, genetics, gait sensors, and rich clinical batteries—into progressive agents that reason over time. Modality-specific thrusts extract stable features, while workflows coordinate generative modeling and clinician delivery.
The result: multimodal latent spaces that expose severity-aligned phenotypes, plus ActionIntel tooling that grounds recommendations in real-world practice.
Four coordinated thrusts
Thrust 2 — Diffusion + Clinical Co-Clustering
Free-water–corrected diffusion metrics align with clinical scales inside a compositional Bayesian co-clustering framework, revealing severity-aligned patient subtypes.
Thrust 3 — Wearable & Cognitive Phenotyping
Wearable IMU data, cognitive batteries, and demographics combine to predict CSFSAA status and uncover multimodal phenotypes.
Thrust 4 — Mechanism Inference
Large-scale statistical testing links multimodal biomarkers to putative biological pathways, grounding discoveries in mechanism-centric hypotheses.
See all thrusts on the Domain Thrusts page.
Two translating workflows
Workflow 1 — Generative Latent-Space Modeling
Harmonizes imaging, diffusion, and clinical data into interpretable latent spaces that seed state estimators and policy agents.
Workflow 2 — Clinician-Centered Decision Pipeline
Delivers multimodal inferences to neurologists via interactive visualization, case review, and cohort monitoring tools.
Learn how modeling meets care on the Workflows page.
Recent findings
- Diffusion metrics fused with clinical scales uncover severity-aligned latent subtypes with interpretable diffusion asymmetries.
- Automated DaT-SPECT and T1 MRI processing differentiates CSFSAA positive vs. negative Parkinsonisms with AUC 0.93 ± 0.04 when fused with clinical and biologic anchors.
- Wearable-derived gait and arm-swing metrics partner with cognition to predict CSFSAA status non-invasively (AUC 0.93 ± 0.07).
- Mechanism inference links multimodal signatures to pathways such as LRRK2 kinase hyperactivity and brainstem α-synucleinopathy.
Status & near-term priorities
- Core imaging pipelines, diffusion–clinical co-clustering, and wearable phenotyping are trained on harmonized cohorts.
- Clinician-facing prototypes integrate patient timelines, motion analysis, and policy recommendations for case review.
- Upcoming work: prospective validation across sites, device-drift audits, subgroup fairness reporting, and SOP finalization for longitudinal follow-up.
Partner with us
Strategic philanthropy, industry collaborations, and clinical trial partnerships accelerate progressive AI deployment. Visit the partners page to learn how funding and clinical collaborators engage, or connect through the resources portal.