Skip to main content
Loading navigation
Back to challenges

Exercise Mimetic Biomarker Analysis Workflow with Mastra AI and GPT-5 Pro

Exercise-mimicking therapeutics require rigorous tracking of metabolic and muscle-retention biomarkers across trial arms. You will build a TypeScript agent using Mastra AI and GPT-5 Pro to extract clinical metrics from trial documents and forecast weight loss vs muscle mass preservation with 85% accuracy.

Data ScienceHosted by Vera
Sign in free to start

What you are building

Develop a TypeScript agent using Mastra AI and GPT-5 Pro to parse clinical trial logs and model metabolic retention profiles.

How work is evaluated

Evaluation

Evaluates TypeScript Mastra agent extraction and predictive accuracy on metabolic trial clinical logs.

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

metric_extraction_exactness

Check if extracted numbers match expected parameters

Binary check

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

Dimension 2

mae_fat_loss

Mean absolute error on fat mass change extraction • target: 0.05 • 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 TypeScript agentic workflows using Mastra AI's core framework

  • Integrate GPT-5 Pro model endpoints for advanced clinical telemetry extraction

  • Calculate lean muscle mass retention metrics from unstructured clinical logs

Resources

Dataset notes

Clinical trial notes dataset covering 200 patients in Phase 1 exercise-mimetic drug trials.

Challenge setup

Evaluates TypeScript Mastra agent extraction and predictive accuracy on metabolic trial clinical logs.

Challenge input

TypeScript string containing clinical progress notes

Mastra AI

Mandatory framework for agent orchestration in Node/TS.

GPT-5 Pro

High-capacity reasoning engine for clinical log parsing.

Evaluated output

JSON object with extracted numerical metrics and predictions

Evaluation checks
  • Check if extracted numbers match expected parameters
  • Mean absolute error on fat mass change extraction • target: 0.05 • range: 0-1
Configured evaluation evidence
  • Benchmark: ClinExtraction-Bench
  • Mae Fat Loss target: 0.05
  • 1 public reference case
Configured runtime
  • JavaScript execution harness
  • javascript 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
  • Mastra AI · Optional
Policy Serving
  • GPT-5 · Optional
Orchestration
  • Mastra AI · Optional

Evaluation contract

  • metric_extraction_exactness · Weight 1
  • mae_fat_loss · Weight 1

Metrics

  • mae_fat_loss · Target 0.05

Benchmarks

  • ClinExtraction-Bench
  • biotech

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: TypeScript string containing clinical progress notes

    Output: JSON object with extracted numerical metrics and predictions

    Execution harness: JavaScript

Evaluation tools

  • Mastra AI · TypeScript Agent Framework
  • GPT-5 Pro · LLM Inference Engine

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Exercise Mimetic Biomarker Analysis Workflow with Mastra AI and GPT-5 Pro