AI Clinical Trial Eligibility Agent

How Cassidy automates this using AI

Step 1: Ingest trial protocol and structure criteria

The Workflow triggers when a new protocol is addedwhether from ClinicalTrials.gov, a sponsor PDF, or your CTMS. Cassidy parses the eligibility section, extracts inclusion/exclusion criteria, and converts them into computable logic with normalized concept mappings and version-controlled audit trails.

Step 2: Connect to EHR and harmonize patient data

Cassidy integrates with your EHR via SMART on FHIR to pull structured data (demographics, problem lists, labs, medications, procedures) and uses NLP to extract key signals from clinical notes, imaging reports, and molecular pathologynormalizing units and handling temporal indexing automatically.

Step 3: Run criterion-level eligibility reasoning

For each patient, Cassidy evaluates every criterion using deterministic rules for numeric thresholds and temporal windows, combined with AI-powered adjudication for nuanced clinical language. The result: a criterion-by-criterion assessment with confidence scores and direct links to the source EHR evidence.

Step 4: Generate ranked candidate lists with explanations

Cassidy produces a prioritized pre-screen list for coordinators, showing eligibility scores, unmet criteria, and clickable evidence trails. Each decision includes the reasoning behind itsupporting both clinical trust and regulatory auditability.

Step 5: Human-in-the-Loop review and workflow handoff

Coordinators review candidates in a research dashboard or via CDS Hooks alerts in the EHR. They can confirm eligibility, order missing tests, or flag patients as ineligiblewith all actions logged automatically to screening logs and your CTMS for sponsor reporting and 21 CFR Part 11 compliance.