Analyze Adjudication KPIs
Calculate adjudication performance metrics and recommend improvements.
---
name: Analyze Adjudication KPIs
version: 0.1.0
description: Calculate adjudication performance metrics and recommend improvements.
metadata:
domain: clinical
complexity: medium
tags:
- adjudication
- analyze
- kpis
requires_context: false
variables:
- name: adjudication_log.csv
description: event log export
required: true
model: gpt-4
modelParameters:
temperature: 0.2
messages:
- role: system
content: '- CSV file `adjudication_log.csv` lists all events in an oncology trial.
- Leadership expects a plan to reduce median cycle time by 20%.
Request a data dictionary if any column in the CSV is ambiguous before starting the analysis.'
- role: user
content: "1. Load the CSV and compute:\n - median and 90th percentile cycle time from event trigger to final decision\n\
\ - reviewer disagreement rate\n - top three root causes of delays inferred from status fields\n2. Create bar charts\
\ for each metric and save them as PNGs.\n3. Recommend at least five concrete process changes tied to these metrics that\
\ would achieve the target reduction.\n\nInputs:\n- `{{adjudication_log.csv}}` – event log export\n\nOutput format:\n\
- **Metrics Summary Table**\n- Embedded charts or download links for each PNG\n- Bullet list of recommendations"
testData:
- vars:
adjudication_log.csv: example_adjudication_log.csv
expected: '- **Metrics Summary Table**
- Embedded charts or download links for each PNG
- Bullet list of recommendations'
evaluators:
- name: Output includes 'Metrics Summary Table'
string:
contains: Metrics Summary Table