Methodology
How this dataset was built, sourced, and verified
What this is
med-vc is an open, structured directory of the organizations that invest in medicine and biomedicine worldwide — venture capital firms, pharma / medtech / payer / provider corporate venture arms, crossover funds, bio accelerators and incubators, university and hospital funds, disease-foundation venture philanthropy, government programs, venture studios, angel networks and family offices — together with the companies they fund.
Two halves, and the links between them
The directory has two halves. One catalogues investors; the other catalogues the medical and biomedical companies they fund — 1,279 so far, against 1,680 investors. They are linked by 1,497 investor–company edges, 286 of which both sides independently assert. Those links are resolved by name at build time rather than hand-written, and only on an unambiguous match: a wrong link reads as a fact, while a missing one is visibly missing, so ambiguity is left unresolved and published as a backlog instead of guessed. The two halves share one vocabulary for sector, modality and indication, which is what lets a filter carry across from a fund to the companies it funds. Coverage of the company half is being filled in region by region; 937 companies named by listed investors are not profiled yet, and the Companies page says so on its face.
Sourced per institution
Every entry carries at least one real source URL, and every quantitative claim — fund size, AUM, check size, founding year, portfolio count — is tied to a verbatim quote from a source. Where a figure could not be verified, the field is left empty rather than guessed. Amounts are kept in their reported currency and string form to avoid fabricated conversions.
Confidence ratings
Each institution is rated high (official / primary source), medium (reputable secondary source), or low (a single weak or dated source). The split across the dataset is high 823 · medium 766 · low 91. Confidence is shown on every card and in the detail view so you can weigh each entry yourself.
How it was assembled
Research was fanned out across 188 slices (organization type × health subsector, per region) so coverage is complementary and overlap is minimal. Each slice was researched independently against live web sources in English and the local language, then everything was merged and de-duplicated — matching on normalized name, website domain, and slug, with accent-folding — into one validated dataset with a controlled vocabulary for type, sector, modality, indication, stage and region.
Who is behind the money
Beyond what an institution invests in, every entry can record who stands behind it — the corporate parent, anchor LPs, or sponsoring institution — as structured `backing.backers[]` rows carrying the backer's name, kind (Big Tech, frontier AI lab, pharma, medtech, payer, sovereign fund, university, foundation and nine more), and the nature of the relationship (wholly-owned venture arm vs balance-sheet fund vs anchor LP). Each row is marked verified when a source states the relationship outright, or inferred when it was derived from the institution's own name or description — a corporate venture arm is almost always named after its parent, which makes that inference reliable but not equivalent to a citation. Use the "Backed by" filter to see, for instance, every fund in the directory running on Big Tech money.
Two-layer quality check
Because the corpus was built over many runs, it went through two checks. A structural pass validated every record against the schema and flagged duplicates, missing sources and off-vocabulary values (0 critical issues remained). Then an agent fact-check audited a 120-institution sample — weighted toward lower-confidence rows — against the live web: it found 0 fabricated organizations and no systematic degradation, and the 16 minor factual corrections it surfaced (a founding year, a stale status, a fund figure) were applied and noted in each record.
Disclaimer
This is an unofficial research compilation provided as-is. Figures on fund size, equity, acceptance rates and the like should be confirmed against each institution's official disclosures via the sources listed on every entry before you rely on them.