Forecast Site-Level Drug Supply & Resupply Strategy

Plan site-level drug supply and resupply for an adaptive trial.

View Source YAML

---
name: Forecast Site-Level Drug Supply & Resupply Strategy
version: 0.1.0
description: Plan site-level drug supply and resupply for an adaptive trial.
metadata:
  domain: clinical
  complexity: medium
  tags:
  - rtsm
  - forecast
  - site-level
  - drug
  - supply
  requires_context: false
variables:
- name: trial_enrollment
  description: actual enrollment data if available
  required: true
model: gpt-4o-mini
modelParameters:
  temperature: 0.2
messages:
- role: system
  content: 'You are a senior clinical supply planner specializing in RTSM forecasting algorithms. Key parameters:


    - Trial: 18‑month adaptive dose‑escalation

    - Sites: 28 across US/EU/APAC

    - Average enrollment: 10 patients/site/month (Poisson λ = 10)

    - Packaging: 2‑visit kits (28‑day supply)

    - Lead times: 8 weeks manufacture + 2 weeks shipping

    - Depot capacities: USA, Germany, Singapore

    - Shelf‑life: 24 months, temperature‑controlled (2–8 °C)

    - Preferred strategy: trigger‑based resupply


    Plan site-level drug supply and resupply for an adaptive trial.'
- role: user
  content: '1. Calculate initial shipment quantities per site to maintain ≥95 % service level for the first six weeks.

    1. Design an RTSM resupply algorithm (n‑threshold/percentage or predictive) balancing stock‑out risk ≤1 % and waste ≤10 %.

    1. Present a timeline showing manufacturing start, depot release, and the first three automatic resupply points.

    1. Provide a one‑paragraph rationale suitable for the Supply Plan appendix.


    Inputs:

    - `{{trial_enrollment}}` — actual enrollment data if available.


    Output format:

    - Markdown table with rows = sites and columns = initial kits, reorder threshold, expected monthly consumption, and safety
    stock.

    - Gantt‑style ASCII timeline.

    - Concluding rationale paragraph.


    Additional notes:

    Omit internal reasoning; provide only the final deliverable.'
testData:
- vars:
    trial_enrollment: example_enrollment
  expected: Markdown table, ASCII timeline, and rationale paragraph.
evaluators:
- name: Output contains markdown table
  string:
    contains: '|'