Skip to main content
Loading navigation
Back to challenges

Build an OpenAI Agents SDK Trial Matching Agent for IL-6 Inflammation Targets

Novartis and other pharmaceutical firms require automated participant screening to identify candidates eligible for IL-6 inflammatory target trials. Build an agent using OpenAI Agents SDK to parse electronic health records against complex inclusion/exclusion criteria and calculate trial eligibility.

Data ScienceHosted by Vera
Sign in free to start

What you are building

Develop an OpenAI Agents SDK workflow to screen cardiovascular trial candidate EHR records based on IL-6 inflammatory biomarkers.

How work is evaluated

Evaluation

Evaluates agent decision accuracy when matching patient FHIR bundles to trial inclusion parameters.

Datasets

Public datasets, plus the private datasets you have access to.

How submissions are scored

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

eligibility_classification_match

Verifies that eligible patient is correctly marked true.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

screening_f1_score

F1 score across candidate patient cohort trial matching evaluations. • target: 0.95 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

  • Implement structured autonomous agents with the OpenAI Agents SDK

  • Extract lab values (e.g., hsCRP, IL-6 levels) and diagnosis codes from FHIR JSON bundles

  • Evaluate eligibility logic against clinical trial protocol inclusion/exclusion constraints

  • Generate explicit reasoning logs explaining why patient records passed or failed screening

Resources

Dataset notes

Cohort of 30 synthetic FHIR patient records with varying inflammatory biomarker profiles and cardiovascular history.

Challenge setup

Evaluates agent decision accuracy when matching patient FHIR bundles to trial inclusion parameters.

Challenge input

JSON FHIR patient resource bundle containing Observation and Condition resources.

OpenAI Agents SDK

Official framework for autonomous agent development.

Evaluated output

JSON containing is_eligible (boolean), matched_criteria (list), and exclusion_reasons (list).

Evaluation checks
  • Verifies that eligible patient is correctly marked true.
  • F1 score across candidate patient cohort trial matching evaluations. • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: FHIR Trial Screening Benchmark
  • Screening F1 Score target: 0.95
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Shows the first task, up to three tools, and up to three evidence items.

View technical recipe

Configured tools

Action Space
  • OpenAI · Required
  • OpenVINO Toolkit · Optional
  • Lepton AI · Optional

Evaluation contract

  • eligibility_classification_match · Weight 1
  • screening_f1_score · Weight 1

Metrics

  • screening_f1_score · Target 0.95

Benchmarks

  • FHIR Trial Screening Benchmark
  • biotech

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON FHIR patient resource bundle containing Observation and Condition resources.

    Output: JSON containing is_eligible (boolean), matched_criteria (list), and exclusion_reasons (list).

    Execution harness: Python

Evaluation tools

  • OpenAI Agents SDK · Agent Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build an OpenAI Agents SDK Trial Matching Agent for IL-6 Inflammation Targets