What cTAKES is doing to the note

The exact words in the note come first

The wrapper keeps the original text unchanged. cTAKES records where a phrase begins and ends as two character offsets, so another person can return to the same characters and see exactly what produced a row. The note below is the same synthetic fixture that How it works walks station by station; its codes are fake on purpose, shaped like DEMO-C001 instead of real terminology identifiers.

The row stays connected to the phrase that created it. Clinical review happens after extraction, not inside the terminology lookup.

The identifier fields on a candidate row

"Renal calculus" and "kidney stone" are different wording for one clinical idea, and the CUI, a Concept Unique Identifier, is the shared code that links both to it. A real CUI is the letter C plus seven digits, such as C0015733 for the feces concept. The note above uses fabricated DEMO- codes. The fixture rows and codes are made up, and the site does not embed a dictionary extract. The separately labeled reviewed failure on the WSD page names real CUIs. The code system behind the CUI is UMLS, the Unified Medical Language System; How it works introduces terminology, ontology, and UMLS beside the synthetic fever span.

The mapping does not decide whether the concept is true for this patient or guarantee that one CUI perfectly captures the sentence. Assertion fields and the source sentence carry that separate information, such as negation or uncertainty.

Four identifiers clinicians will see

The recipe names the engines and their order

The recipe is a small .piper file that tells cTAKES which analysis engines to load and in what order; order matters because one stage often needs annotations created by an earlier stage. The default recipe's key is star, and its full stage-order table, with what each engine adds and why the order holds, is on Dictionary and recipes. How it works walks the same order on the synthetic note.

Assertions describe how the note talks about a candidate

A CUI alone says which concept matched. Assertion fields say whether the language is negated, uncertain, conditional, generic, or about someone other than the patient. The wrapper keeps missing values blank because a fabricated default would look like a model result.

Degree and location relations are separate. A relation is a directed link between two annotations: a pain candidate can link to a right-flank anatomy candidate, while a severity cue such as "severe" can link to the concept it modifies. These links still need representative clinical review because a structurally valid relation can point in the wrong direction or attach to the wrong nearby site.

WSD handles one phrase with more than one candidate

Word-sense disambiguation (WSD) runs when one text span carries multiple distinct candidate meanings, the way "mass" at span 170–174 carries a body-structure candidate and a quantitative one. It keeps every candidate row, and it writes one Y, one or more N values, and scores only when a passing decision is possible; otherwise it abstains and both WSD columns stay blank. The score compares candidates inside that one ambiguity group and is not the probability that the concept is clinically true. The WSD page carries the full method, the decision path, the thresholds, the wrapper 2.1.1 controlled 858-note comparison, and the not-evaluated clinical status.

Federated learning keeps the source notes at each site

In a federated study, Site A and Site B do not pool their raw notes into one central folder: each site runs the same extraction definition locally, and only protocol-approved results cross the boundary. The wrapper supplies the first requirement, because both sites can prove they used the same wrapper source, pipeline, dictionary bytes, SQLite schema, and WSD model by comparing analysis_id; it does not implement the federated trainer, authentication, secure aggregation, coordinator, or privacy accounting. Multi-site use carries the full workflow and data boundary.

The site-wide glossary

Every page on this site links here. Terms are defined the way this wrapper uses them, not as general textbook entries.

Continue to the exact terminology selection, or see how these terms and identities appear in the delivered files.