What Clinical Trial Data Is, Exactly

This file describes the categories exposed at 25-0912b, because the corpus has no comparable data class.

Not a diagnosis, not a treatment record — a set of values describing physiology at a point in time. Cholesterol, inflammatory markers, glucose response, genetic indicators depending on the study.

They are closer to biometric data at 26-0324 than to a medical record: intrinsic, unchangeable in the relevant sense, and describing the person rather than an interaction with a healthcare system.

Whether a participant’s immune system responded to a compound, and how. It is study-specific, clinically meaningful, and not something that appears in any ordinary medical file.

This desk cannot say what harm follows from its exposure, and neither can anyone else — which is itself the point. The corpus filed at 25-1219 that the link between exposure and harm is unproven generally; for a field like this it is not even hypothesised.

Smoking, alcohol, diet, exercise, sometimes sexual history depending on the protocol. Disclosed by the participant to a researcher, in confidence, as a condition of enrolling.

It is the closest thing in this corpus to the free-text problem at 25-1006 and 25-1024 — information volunteered in one context, uninventoried, and meaningful to anybody who reads it.

The other three require expertise to read. Participation in a trial for a named condition does not.

The corpus’s repeated finding — that sensitivity lies in the inference rather than the field, at 25-0603, 26-0326 and 25-1130 — reaches its clearest case here. Graded medium: this describes the category and is not a claim about any specific study.