Ichlibitiche's picture
Point explorer link to combined Sample Explorers Space
57f4778 verified
|
Raw
History Blame Contribute Delete
3.46 kB
metadata
license: cc-by-nc-4.0
pretty_name: SuppDB  Supplements & Nootropics Dataset (Free Sample)
tags:
  - health
  - chemistry
  - supplements
  - nootropics
  - nih-dsld
  - pubchem
  - recommender-systems
size_categories:
  - 1K<n<10K

💊 SuppDB — Supplements & Nootropics Dataset (Free Sample)

A free sample of SuppDB: a structured dataset of real supplement & nootropic products built exclusively from the public NIH Dietary Supplement Label Database (DSLD) — every active ingredient normalized to milligrams, proprietary blends flagged where the dose is undisclosed, and compounds enriched with NIH PubChem chemical identity. Think INCIDecoder for supplements: one row per active ingredient, with dose, form, safety reference, and molecular identity.

This sample contains 2,249 ingredient records across 300 real products from 218 brands.

The full dataset covers 17,000+ products, 2,000+ brands, 115,000+ active-ingredient records, and 40,000+ proprietary-blend flags (SQLite · CSV · JSON).

Get the Full Dataset

Key Columns

Column Description
brand, product_name, upc_barcode, form_type Product identity as printed on the label
ingredient, ingredient_form, ingredient_category Active ingredient per row
amount_per_serving_mg Dose normalized to mg (mcg, g, substance-specific IU handled correctly)
is_proprietary_blend 1 where the dose is hidden in a proprietary blend (amount = 0) — transparency, not omission
recommended_daily_mg, upper_safety_limit_mg NIH DRI reference intakes; NULL where no official value exists
pubchem_cid, molecular_formula, molecular_weight, canonical_smiles, inchikey PubChem chemistry — InChIKey canonicalizes the same molecule across label names
dsld_label_id, source_url Exact NIH DSLD label page — every record re-verifiable

Quick Start

import pandas as pd

df = pd.read_csv("hf://datasets/Ichlibitiche/suppdb-supplements-sample/suppdb_sample.csv")
print(len(df), "ingredient records,", df["product_id"].nunique(), "products,", df["brand"].nunique(), "brands")
hidden = df[df["is_proprietary_blend"] == 1]
print(len(hidden), "ingredients with doses hidden in proprietary blends")

Use Cases

  • AI health co-pilots & supplement recommendation apps (structured dose + chemistry data)
  • Ingredient/dose comparison and proprietary-blend transparency tools
  • ML / RAG corpora over supplement labels
  • Formulation, market, and assortment research across brands and ingredient categories

License

Sample data: CC BY-NC 4.0 — attribution, non-commercial. Full dataset commercially licensed at suppdb.net; underlying facts are public-domain U.S. Government data (NIH DSLD + PubChem) — the license covers SuppDB's curated, normalized compilation. Not medical advice — always verify against the current physical label. Contact: suppdb.doorframe589@simplelogin.com.