Drybench Patent.

What the app shows, how it works and how it was built, in the words of its own write-up, followed by the parameter defaults and the data sources with their licences.

What it shows

Type a brand, an ingredient or an application number to explore the captured FDA patent and exclusivity rows for that US product, each with its date, the rule it rests on and the row it came from. The answer is a range of listed protection: the latest substance patent, the latest listed patent of any kind and the latest non-patent exclusivity, never one date. A date window groups listed dates by product and rule; these are not entry dates.

Beside the trace sit candidate ANDA products grouped by ingredient and dosage form, with their listed therapeutic-equivalence codes and unconfirmed reference-product matching; the Paragraph IV certifications list's row for the product as FDA publishes it; the Purple Book's exclusivity dates for biologics; and what changed in FDA's files between two captures.

How it works

A scheduled job saves the FDA Orange Book data files (products, patents and exclusivities), the Purple Book download and the Paragraph IV certifications list every day and keeps each file only when it changes, because the FDA publishes current files only and keeps no public history. The app reads one static store built from those files; the store is served from this site, with the capture date on every surface, and nothing is fetched from the FDA in your browser.

The trace identifies the supported rules and their source rows. Tentative-approval data is not included in this store. NCE-1 is the only computed date. A listed patent's expiry is shown as the application holder submitted it, with its substance, product and use flags. The pediatric row FDA lists for the same patent is shown as FDA lists it; the app never adds six months itself. Each non-patent exclusivity is shown with its code and FDA's definition of the code. NCE-1 is the NCE date minus one year. The Paragraph IV row is shown as listed, and an absent row is never read as no certification. Patents with a delist request or a use code stay in the table, flagged.

Every table and export carries, for each date, the source file, the row, the file's own date when known, the capture date and the file's checksum, so a reader can check any value against the FDA's own file.

How it was built

The parsers are written against the FDA files' published layouts and say what they did not find rather than guess. The rule engine is pure TypeScript with a second implementation in Python that the tests compare it with over a fixture, and every number on screen is traced to a row. The app serves the pinned FDA capture with mechanically transcribed source fixtures and implementation checks. Independent regulatory review remains pending (SCI-124/Q-19); the original F1–F7 requirements are not all complete.

The seven rules

R1, listed patent expiry

The Orange Book patent row's expiry as the application holder submitted it, including any extension, with its drug-substance, drug-product and use flags, its use code, its delist flag and its submission date. Nothing is verified against the patent itself.

R2, pediatric exclusivity on a patent

The second row the FDA lists for the same patent, marked *PED, with the amended date, shown as listed and labelled as pediatric exclusivity. The app never adds six months: a product with a PED exclusivity row but no *PED patent row shows no extension.

R3, non-patent exclusivity

Each exclusivity row's code and date, with the FDA's definition of the code; a code outside the definitions list is shown as such, never guessed.

R4, NCE-1

The NCE exclusivity date minus one calendar year, shown as the earliest ANDA submission with a Paragraph IV certification; the one computed date, marked as computed.

R5, Paragraph IV and 180-day

The Paragraph IV list's row as the FDA publishes it: first submission date, number of ANDAs, 180-day status, decision date, first approval, first marketing and the last qualifying patent's expiry. An absent row is never read as no certification.

R6, tentative approvals unavailable

Tentative-approval data is not included. Separately, candidate ANDA products are grouped by ingredient and dosage form with listed therapeutic-equivalence codes and approval dates; this does not confirm a reference-product relationship.

R7, delisted and use-code patents

Kept in the table with their flags: a delist request keeps the patent listed while a first applicant may keep 180-day eligibility, and a use-code patent is shown as blocking the coded use.

The summary

A set, never a point: the latest substance patent (with its *PED row where listed), the latest listed patent of any kind, the latest non-patent exclusivity, and the earliest and latest of every rule's dates.

Join

By application type and number only, with rows shown per product number because patents and exclusivities are listed per strength; the Purple Book joins by BLA number; the Paragraph IV list by one isolated 4–6 digit NDA token, padded to six digits and matched to NDA product rows only. Unresolved rows are retained separately with reasons.

The store

The capture's normalised text per file (header first, rows sorted) with the capture's record of each download, served as content-hashed metadata, a names/dates index and application shards. Initial capture and transformation-only corrections remain archived but are excluded from FDA listing changes.

Data sources, licences and versions

Times on screen are local time with the UTC offset; times in exports are UTC. Dates with no time are shown at the precision the source gave them.

Each view exports as CSV with a citation per row (the source file, the row, the file's date, the capture date and the checksum), the capture date of every source and the limits in a README beside the data. The link in the address reopens the same product, view and window.

Release gate: before the app leaves the preview, its exports and printed reports will carry the data versions, parameters, limits and a timestamp needed to reproduce a result, as the regulatory position says. This section will describe each format as it ships.

Each module below is a set of automated tests in the app's source code, with the expected results written into the tests. They check the app's logic against recorded inputs and stored reference records; none of them calls a live service.

Unit tests

src/lib/exclusivity/reference-capture.test.ts

439 mechanically transcribed source rows and 30 application summaries from the pinned 17 October capture, with original coordinates and hashes. F4/F6/F7 source prerequisites remain unavailable. Independent regulatory review is pending (SCI-124/Q-19).

src/lib/exclusivity/parse.test.ts

The five parsers against the synthetic fixture: headers found and missing, dates in the FDA's forms and in words, the *PED marker, flags, quoted CSV and a changes section before the Purple Book's header.

src/lib/exclusivity/trace.test.ts

Implemented rule traces, including the negative control (a PED exclusivity without a *PED patent row shows no extension), the summary set, NCE-1 and the date window.

src/lib/exclusivity/reference.test.ts

The engine's trace over the synthetic store against the Python second implementation's expected values, field by field.

pipeline/patent-exclusivity/capture.test.ts

The capture script's built-in test: normalised layout, keyed diffs, the heartbeat, a failed source's exit code and the watchdog's allowances.

Release gate: before the app leaves the preview, each release will publish a verification dataset with expected outputs on this page, so you can run your own checks.

Scientific computation runs in your browser, and individual apps need no account. Optional Bench accounts and backups use the site server and configured providers. The privacy statement, the hosts it contacts and the headers its workspace is served with are generated from its registry entry, the same record that sets the browser's policy.

What leaves your browser

Hosts the app contacts

The app contacts no host but its own.

Headers the workspace is served with