Skip to main content
Loading navigation
Back to challenges

Cross-Border Liquid Biopsy Screening Workflow with Vercel AI SDK

International clinical diagnostic partnerships require seamless real-time execution across local and global healthcare APIs. You will build a TypeScript web service using the Vercel AI SDK to stream liquid biopsy screening analysis, translating and standardizing panel results across JS/TS applications with under 500ms latency.

Data ScienceHosted by Vera
Sign in free to start

What you are building

Develop a high-throughput TypeScript liquid biopsy streaming service using Vercel AI SDK with tool invocation.

How work is evaluated

Evaluation

Evaluates streaming response validity, tool call accuracy, and latency for liquid biopsy reports.

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

variant_parsing_accuracy

Ensure allele frequency and mutation are parsed accurately

Binary check

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

Dimension 2

latency_p95

95th percentile response generation time in milliseconds • target: 450 • range: 0-2000

Binary check

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

Learning goals

  • Build real-time streaming AI applications using Vercel AI SDK (`ai` package)

  • Implement typed tool calling with `zod` schema definitions for oncology panel parsing

  • Handle bilingual clinical data transformation for international laboratory compliance

Resources

Dataset notes

Cross-border liquid biopsy diagnostic report dataset in English and Japanese containing 200 panel outputs.

Challenge setup

Evaluates streaming response validity, tool call accuracy, and latency for liquid biopsy reports.

Challenge input

JSON object with patient_id and panel_raw_text string

AI SDK

Mandatory Vercel AI SDK framework.

Evaluated output

JSON object with parsed_variants array and latency_ms number

Evaluation checks
  • Ensure allele frequency and mutation are parsed accurately
  • 95th percentile response generation time in milliseconds • target: 450 • range: 0-2000
Configured evaluation evidence
  • Benchmark: StreamBiomed-Bench
  • Latency P95 target: 450
  • 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
  • Vercel AI SDK · Required
  • Yi-34B · Optional
  • Sourcegraph Cody · Optional
Orchestration
  • Vercel AI SDK · Required

Evaluation contract

  • variant_parsing_accuracy · Weight 1
  • latency_p95 · Weight 1

Metrics

  • latency_p95 · Target 450

Benchmarks

  • StreamBiomed-Bench
  • biotech

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON object with patient_id and panel_raw_text string

    Output: JSON object with parsed_variants array and latency_ms number

    Execution harness: JavaScript

Evaluation tools

  • AI SDK · TypeScript AI Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Cross-Border Liquid Biopsy Screening Workflow with Vercel AI SDK