What the programme covers
Longitudinal cohorts support questions about progression, response, recurrence and future outcomes. They also create risks: inconsistent follow-up, care received outside the source system, duplicated examinations and future-information leakage can distort a model if not made explicit.
What a useful specification includes
A defensible request defines the clinical task, source evidence and acceptance criteria before patient-level data moves. The exact fields and thresholds depend on the intended model claim.
- Index event, baseline window and follow-up period
- Required number and type of repeat examinations
- Treatment, procedure or outcome events to link
- Progression, response or recurrence definition
- Censoring, loss-to-follow-up and missing-data rules
- Patient-safe training and validation splits
Quality and validation controls
Longitudinal curation verifies examination order, clinical windows and source-system coverage. Observed absence is not confused with proof that an event did not occur.
- Patient and encounter resolution
- Examination sequence and timestamp consistency
- Duplicate, imported and outside-study handling
- Follow-up completeness and censoring
- Future-information and patient-leakage controls
Availability, rights and delivery
Approved releases can include pseudonymous timelines, imaging manifests and linked evidence with explicit observation windows and documented missingness.
Public pages describe a sourcing and engineering capability, not guaranteed ready inventory. Each release remains subject to verified programme inventory, programme-specific authorization, privacy review, technical acceptance and buyer licence terms.
Questions, answered directly.
Does longitudinal mean every patient has several scans?+
Not automatically. The required examination count and interval must be specified and verified.
Can treatment response be included?+
Potentially, when treatment events and response evidence are available, linkable and covered by programme authority.
Can repeat scans leak patients across dataset splits?+
Yes. All examinations for a patient should remain within the appropriate split unless the study protocol expressly requires another design.
Are outcomes always complete?+
No. Source coverage and follow-up limitations must be measured and disclosed.