Honest composite scoring
The composite is computed only over dimensions that actually carry varying data for a given fiscal year, with weights renormalized to sum to 100% — so a year missing award dollars never gets a hidden, constant score floor.
An open, transparent view of public NIH RePORTER data — and an institution funding ranking, in development, designed so every score can be audited.
Built on one principle: a ranking is only useful if you can see exactly how it was computed and how current the data is. OpenNIH is being built to disclose both, in full. This site is a preview while the interactive app is in active development.
Why OpenNIH is different
OpenNIH is designed to be checkable end to end — the scoring rules, the excluded rows, the data's as-of date. These are the things that make a public ranking trustworthy.
The composite is computed only over dimensions that actually carry varying data for a given fiscal year, with weights renormalized to sum to 100% — so a year missing award dollars never gets a hidden, constant score floor.
Each view will report the data's as-of date. NIH's RePORTER database refreshes weekly; OpenNIH is built to state how current its copy is and flag when it may be lagging — no fake "live" timestamps.
A documented, versioned methodology and a citation file, with every figure traceable to public NIH data through a reproducible pipeline. The code is permissively licensed and will be opened on release.
Institutions are resolved by stable identifiers, and intramural NIH activity is kept out of the institution ranking — so totals aren't fragmented across name variants or double-counted.
FY1985 by the numbers
These figures come straight from the public NIH ExPORTER FY1985 snapshot. They are project counts — award dollars are not recorded in the FY1985 snapshot, so no funding amounts are shown.
FY1985 · loading…
Couldn't load the dataset — please refresh.
By FY1985 NIH project count
By FY1985 NIH project count
By FY1985 NIH project count (R01, M01, P01…)
By FY1985 NIH project count (institution location)
Ask the data
Ask questions in plain English. Answers are generated by an AI model (Kimi K2.6) and grounded in the same FY1985 project-count data charted above — it will tell you when something isn't in the data. AI can still make mistakes; verify against the numbers.
Grounded in the public NIH ExPORTER FY1985 snapshot · figures are project counts, not dollars.
How it works
A small, legible pipeline. No proprietary inputs, no manual nudging of results.
Project records are pulled from NIH RePORTER and ExPORTER — the same public sources anyone can download.
Records are normalized to a consistent schema and institutions are resolved to canonical entities.
Weighted dimensions are combined into a composite over the dimensions that carry data, with the rule disclosed on every page.
Results will ship with the data's as-of date, the active/excluded dimensions, and a methodology you can read.
Data & currency
OpenNIH only shows numbers it can trace to a public source, and it tells you how fresh they are.
The weekly-refreshed source of project records. OpenNIH tracks newly-added awards by their RePORTER date_added for incremental currency.
Bulk annual extracts used as a periodic baseline. Treated as annual, not as a weekly freshness source.
Publication and citation metrics (e.g. iCite, OpenAlex) are planned enrichments — listed here as planned, not shipped.
How current is the data? NIH's RePORTER database updates weekly. The interactive app will report the exact as-of date of its loaded dataset and warn when that copy may be lagging NIH. This preview page intentionally shows no funding figures — those will be served by the app once it is live, computed from a dated dataset.
Preview site. The interactive ranking and search app is in active development. This page describes the methodology and approach. It deliberately does not display any sample or placeholder funding numbers.
See how the ranking is computed and how the data's currency is disclosed, or follow along as the interactive app comes online.