HD Research Hub

Collaborate on open HD R&D

This is an open research project. Everything, the experiments, the code, the data pipeline, the hypotheses, is public and forkable. Here are concrete ways to work together, and how to start.

Ways to collaborate

Reproduce or critique an experiment

Every experiment ships its code, prompts, and data. Re-run it, challenge the method, or flag where the AI got something wrong. Skeptical review is the highest-value contribution.

Test a falsifiable hypothesis

We turn AI ideas into concrete, testable predictions (e.g. the chorea-footwear margin-of-stability claim). If you have a lab or cohort, try to break one.

Contribute a data source or an agent

Add a feed (a registry, a dataset, a scanner) or a new autonomous agent to the pipeline. Everything that lands feeds the shared knowledge base.

Co-design an experiment

Bring a question. We'll help scope it into a reproducible run (search → analyze → score → honest write-up), and publish it with your framing.

Validate in a lab or cohort

Take a computational hypothesis into the real world, wet-lab, gait lab, or an HD cohort (e.g. via Enroll-HD / CHDI / an HDSA Center of Excellence). Pre-registration welcome.

Lend lived experience

Patients, families, and caregivers: tell us what actually matters day to day. It should shape what we prioritize. For care, always start with your clinician and HDSA.

How it works

  1. 1Start a conversation. Open a GitHub Discussion or reach out on LinkedIn with what you'd like to do, no formal proposal needed.
  2. 2Scope it together. We turn it into a small, concrete, reproducible piece of work with a clear question and an honest success criterion.
  3. 3Build in the open. Code and data are public. You keep authorship of your framing; the AI's role is always labeled.
  4. 4Publish honestly. Every result says what worked and what didn't, with limitations and how to reproduce it. No hype.

What we bring to the table

• An open agent pipeline (PubMed, ClinicalTrials.gov, HDBuzz, Open Targets, preprints, NIH funding)
• A chatbot grounded in real papers, section by section
• Reproducible experiments with honest write-ups and citation-integrity checks
• A drug-hypothesis tracker cross-referenced to real trials
• Parametric design + 3D-CAD workflows for hardware concepts
• Everything MIT-licensed and forkable, zero cost to run

Ready when you are

Pick any of the above, or bring your own. The fastest start is a GitHub Discussion, we'll take it from there.

Open research collaboration for education and discovery, not a company, a partnership offer, or medical advice. AI-generated outputs are unvalidated hypotheses for expert review. We are data scientists, not doctors. For HD care, contact HDSA (hdsa.org) or your clinician.