Cohort Identification

A machine-processable definition of a clinical condition, expressed as a set of logical expressions and data queries, used to identify patient cohorts from electronic health records. It translates a clinical concept like 'Type 2 Diabetes' into executable code.

• Key Components: Data elements (diagnosis codes, lab results, medications), logical operators (AND, OR, NOT), and temporal constraints.
• Example: A phenotype for 'Uncontrolled Diabetes' might include HbA1c > 7.0% AND Diagnosis Code E11.x within the last 12 months.
• Purpose: Ensures consistent, reproducible cohort definitions across different studies and sites.

A standardized, machine-readable syntax for expressing the complex inclusion and exclusion rules that define a patient group for research or trial recruitment purposes. It provides a formal grammar for computable phenotypes.

• Standardization: Enables sharing of cohort definitions across different institutions and data models (e.g., OMOP CDM).
• Example: The OHDSI ATLAS tool uses a JSON-based CDL to define cohorts, specifying entry events, inclusion criteria, and censoring events.
• Benefit: Promotes transparency and reproducibility in observational research by making cohort logic explicit and portable.

A software component that runs a computable phenotype definition against a clinical data repository, resolving logical expressions and temporal constraints to return a final patient cohort. It is the runtime environment for cohort identification.

• Function: Parses the cohort definition, translates it into database-specific queries (e.g., SQL), executes them, and aggregates the results.
• Key Capability: Handles complex temporal reasoning, such as 'drug A must start at least 30 days after procedure B.'
• Output: A list of unique patient identifiers that satisfy all criteria, often with qualifying dates for each patient.

The process of converting parsed, structured eligibility criteria into executable database queries, such as SQL or FHIR API calls, to screen patient repositories. This bridges the gap between human-readable protocols and machine-actionable logic.

• Input: Structured criteria, often in a JSON or YAML format, extracted from a clinical trial protocol.
• Output: A set of parameterized queries that can be run against a data warehouse or FHIR server.
• Challenge: Accurately mapping clinical concepts (e.g., 'severe hypertension') to specific value sets and code systems (e.g., SNOMED CT codes for systolic BP > 180 mmHg).

The AI capability to interpret and validate time-dependent clinical constraints, such as washout periods or disease progression timelines, against a patient's longitudinal record. It is critical for complex trial eligibility.

• Core Logic: Evaluates the sequence and timing of clinical events, not just their presence.
• Example: An inclusion criterion like 'Disease progression within 6 months of last therapy' requires sequencing a progression event after a therapy event and checking the time delta.
• Implementation: Often uses Allen's interval algebra or custom temporal query languages to define relationships like BEFORE, AFTER, OVERLAPS, and WITHIN.

An automated, privacy-preserving initial assessment of a patient's broad suitability for a clinical trial using minimal demographic and diagnostic data before full record review. It acts as a high-recall filter.

• Goal: Rapidly narrow a large patient population to a smaller, likely-eligible subset for detailed screening.
• Data Used: Typically uses structured data like age, gender, primary diagnosis codes, and recent encounter dates.
• Architecture: Often deployed as a Trial Pre-Screening API that returns a list of potential trial matches for a given patient, without exposing the patient's full record to the trial system.