Correlatehealth

Correlate Health (LooPT CTM) — A longitudinal ELIGIBILITY-STATE dataset rather than a static cohort snapshot — the scarce asset here. LooPT CTM continuously evaluates each patient's structured chart data and clinical notes (labs, diagnoses, medications, imaging, pathology, problem lists) against active trial criteria via FHIR R4 and CDS Hooks, and records the criterion-level state as met / unmet / unknown, with the source evidence for each. Because it re-evaluates continuously, the byproduct is a time series of per-criterion eligibility states per patient: when each criterion flipped from unm…

Correlatehealth offers (Alternative, Reference, Sentiment) — Not published. Structural estimate: volume scales with patients screened times active trials times re-evaluation cadence, and because evaluation is continuous with per-criterion state retained, a single deployment covering a system's oncology population across a handful of active trials plausibly generates 10^6-10^7 criterion-state observations per year (each patient x trial x criterion x evaluation). The valuable part is density per patient, not raw scale — the state-transition time series is what is unique..

Prospective patient-supply forecasting (when will this cohort reach eligibility, and how fast does a Near Miss pool convert) — impossible from snapshot data; training and benchmarking clinical NLP/note-understanding systems on note-to-criterion-verdict pairs with human rationale; evaluation of automated eligibility systems against clinician-adjudicated alerts (a labelled false-positive set with reasoning, which is what sponsors lack when assessing vendor accuracy claims); protocol criteria design research (which criteria are chronically unanswerable from real charts and therefore generate screen failures); clinical-decision-support alert-fatigue research (what makes a clinician accept or dismiss an alert); oncology staging- and lab-evolution trajectory modelling; point-of-care trial-matching system design.

The data is with 4 years of history.