Erin Rose

Founder, Verified Supplement Data

I build tools that make hidden health data legible. I'm the founder of CareCost Estimate, where I parsed tens of terabytes of federally mandated — but practically unreadable — healthcare pricing files into a tool anyone can use. Verified Supplement Data applies the same approach to supplements: take the primary sources most people never read, and turn them into numbers you can check.

What I am, and what I'm not

I am not a doctor, dietitian, or pharmacist, and nothing on this site is medical advice. I'm a data engineer. That's exactly why every page here is built so you don't have to take my word for anything:

• Every clinical claim links to its PubMed source (PMID)
• Product data comes from the NIH Dietary Supplement Label Database
• Safety-signal data comes from the FDA's FAERS database, deficiency data from CDC NHANES
• The math (cost per clinically effective dose, elemental content) is shown on the page
• When I get something wrong, it goes on the public corrections log

Sites in this space usually rent a clinician's photo for the byline. I'd rather show you the sources. The accountability here is the methodology, the editorial standards, and data you can download and audit yourself.

Verifiable work

The dataset behind this site is published openly, under my name, in places I can't quietly edit:

• Figshare (DOI): 10.6084/m9.figshare.32917331 — the Verified Supplement Evidence dataset, CC BY 4.0
• Hugging Face (DOI): 10.57967/hf/9356 — canonical dataset mirror
• Kaggle: verified-supplement-evidence-database
• GitHub: erinheit451/verified-supplement-evidence — open CSVs, CC BY 4.0
• npm / MCP Registry: supplement-advisor-mcp — an open-source supplement evidence server for AI assistants, listed in the official Model Context Protocol Registry
• Free API: the recommendation API (OpenAPI spec) — the same data this site uses, free for anyone

Why I built this

The supplement industry runs on confusion: compound weight passed off as elemental dose, prices that vary 10x for equivalent products, and evidence locked in journals nobody reads. The information to cut through all of it is free and public — NIH, PubMed, FDA, CDC — it's just not legible. Making it legible is a data problem, and data problems are what I do.

Questions, corrections, or just want to check my work? [email protected] — corrections get logged publicly.