Skip to main content
Loading navigation
Back to challenges

LlamaIndex MCED Diagnostic Utility RAG with GPT-5 Pro

Multi-cancer early detection (MCED) tests produce complex biomarker reports requiring precise patient-facing context and clinical guidelines. Implement an agentic RAG system using LlamaIndex and GPT-5 Pro to map MCED blood signals to evidence-based diagnostic pathways. Achieve zero hallucinated clinical recommendations on test cases.

Data ScienceHosted by Vera
Sign in free to start

What you are building

Build an advanced RAG and agent workflow using LlamaIndex with GPT-5 Pro to map multi-cancer early detection signals to clinical follow-up protocols.

How work is evaluated

Evaluation

Evaluates the factual fidelity and clinical precision of RAG generated diagnostic follow-up recommendations.

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

zero_hallucinated_citations

Verifies that all returned citations match indexed source guidelines.

Binary check

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

Dimension 2

guideline_compliance

Percentage of correct clinical diagnostic steps suggested. • 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

  • Construct a LlamaIndex VectorStoreIndex over MCED clinical practice guidelines

  • Integrate GPT-5 Pro model for zero-shot RAG synthesis with strict citation matching

  • Design custom LlamaIndex RouterQueryEngine to route patient inquiries to diagnostic sub-indices

  • Implement automated clinical factual consistency guardrails

Resources

Dataset notes

Collection of public MCED clinical consensus guidelines, NCCN pathways, and simulated patient test report cases.

Challenge setup

Evaluates the factual fidelity and clinical precision of RAG generated diagnostic follow-up recommendations.

Challenge input

JSON with patient_age, mced_signal_tissue_of_origin, biomarker_level

LlamaIndex

Required framework for document indexing and retrieval routing.

GPT-5 Pro

Provides state-of-the-art context processing and clinical analysis.

Evaluated output

JSON containing recommended_imaging, specialist_referral, and citations

Evaluation checks
  • Verifies that all returned citations match indexed source guidelines.
  • Percentage of correct clinical diagnostic steps suggested. • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: ClinicalRAG-Bench
  • Guideline Compliance 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
  • GPT-5 Pro · Required
  • Llama Index · Optional
Policy Serving
  • GPT-5 · Optional

Evaluation contract

  • zero_hallucinated_citations · Weight 1
  • guideline_compliance · Weight 1

Metrics

  • guideline_compliance · Target 0.95

Benchmarks

  • ClinicalRAG-Bench
  • biotech

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON with patient_age, mced_signal_tissue_of_origin, biomarker_level

    Output: JSON containing recommended_imaging, specialist_referral, and citations

    Execution harness: Python

Evaluation tools

  • LlamaIndex · RAG and agent data framework
  • GPT-5 Pro · LLM Provider

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about LlamaIndex MCED Diagnostic Utility RAG with GPT-5 Pro