Cancers ordered by how many trial records in the corpus name them.
Loading
14 league tables, each computed from counts and dates already in the corpus: no weighting, no composite score, no number that cannot be traced to a record. Every table says in one sentence what it counts, shows the top 50, and discloses how completely the counted field is filled. Ties are broken by name.
These tables rank what the corpus records, not clinical merit. A cancer with many trials is well studied, not well treated; a company with many approval records has many labels, not better drugs. Each table's page carries a "How this is counted" note and a JSON companion.
Cancers ordered by how many trial records in the corpus name them.
Cancers ordered by how many products with status approved list them.
Cancers with at least ten trials, excluding umbrella records that other cancers name as their parent, ordered by trials divided by approved products, with cancers that have no approved product first.
Targets ordered by how many product records in the corpus aim at them, any phase.
Targets ordered by how many cancers their record lists as places the target is found or acted on.
Approved products ordered by how many approval records they carry, one per regulator and indication.
Products ordered by how many trial records are linked to them from either side.
Approved products ordered by the number of years between their earliest approval record and the current year.
Companies ordered by how many trial records in the corpus link to them.
Companies ordered by the number of approval records across the products that name them.
People ordered by how many papers their record lists.
Glossary terms in the Biomarkers category ordered by how many cancers are linked to them from either side.
Trials ordered by the number of people their record says were enrolled, randomised, analysed or treated.
Roadmaps ordered by how many distinct eras their steps span.
A row's value is a count of linked records, a ratio of two such counts, or the difference between two years on the record. The registry lives in src/lib/rankings.ts; a table is only registered when the field it counts is filled on at least 60 percent of the records it draws from, and that share is printed under every table.
Tables considered and not built, because the corpus does not yet carry the field widely enough: institutions by trials (4 percent of trials name an institution), cancers by ideas (38 percent of ideas name a cancer), time from first trial to first approval (trials carry a readout year, not a start date, on 18 percent of records), and countries by approvals or trial sites (approvals carry a regulator, not a country, and trials carry no site list). Insurer and country tables built from published external figures are on Coverage rankings and Countries.