Rare Disease Research Atlas

Of 7,126 diseases with usable interventional-trial matching in this build (8,171 total records; incomplete trial fetches and uncapped scans excluded from this percentage)

5,501

have no registered interventional clinical trial matched directly to their specific disease name — 77.2% . Including broader parent-disease registrations reduces this estimate to about 62.7%.

Data as of 9 September 2026. Current-state claim — present tense.

Using exact disease-name matching, 77.2% (5,501 of 7,126) have no matched interventional trial. When trials registered under broader parent conditions are counted as evidence, the estimate falls to 62.7% (4,466 of 7,126). We report the conservative figure and show parent-category trials on each disease page. Prior matching choices in this project landed in the mid-50s to mid-70s — that spread is itself a finding about how poorly disease naming maps between literature and trial registries. Against a 33-disease gold set (dual-model adjudication with light human fix of disagreements), trial matching recall is 96% and precision 86%. Precision is among NCT IDs already labelled in that gold set. Full unaided human validation is not yet complete. Interventional trials are studies in which participants receive an intervention (drug, device, procedure, etc.). Observational and natural-history studies are shown separately on disease pages because they are meaningful research and may be open to families, but they are not counted as interventional trials. Publication name-collision flags do not remove diseases from this trials denominator.

A fixed historical sample measurement (past tense) is on Methods & findings; the live site may differ.

Broader comparison: 4,984 of 7,126 (69.9%) have no matched registered study of any type, including observational studies. The difference between this and the headline is an editorial definition, not a correction to the data. Both figures exclude pan-disease registries.

Among those with no specific-condition trial, 1,243 still have substantial published literature — a name that demonstrably matches papers, making the trial zero far likelier to be real than a search artifact.

Separately, 989 of 8,171 records in this build returned nothing on either database and are excluded from every percentage as probable broken queries, not measured absence.

Secondary finding: 183 of 3,951 (4.6%) have no publication in the last ten years and no interventional trial — intersection of the publications and trials denominators (3,967 and 7,126 respectively). Obsolete and “non rare in Europe” Orphanet entries are removed before sampling.

Corpus: full · excluded 801 obsolete/non-rare