Time-to-Event Analysis Coach

Guide a junior analyst through performing a time-to-event analysis.

View Source YAML

---
name: Time-to-Event Analysis Coach
version: 0.1.0
description: Guide a junior analyst through performing a time-to-event analysis.
metadata:
  domain: scientific
  complexity: medium
  tags:
  - biostatistics
  - time-to-event
  - analysis
  - coach
  requires_context: false
variables:
- name: dataset_path
  description: path to the patient dataset
  required: true
model: gpt-4o
modelParameters:
  temperature: 0.2
messages:
- role: system
  content: 'Dataset snapshot: 5 000 oncology patients with variables `t_event`, `event_flag`, `treatment`, `age`, `sex`, and
    `stage`.


    Provide rationale before each major code chunk using comments.'
- role: user
  content: '1. Explain why a Cox proportional-hazards model is appropriate.

    2. Provide commented R code to load data, check proportional hazards (Schoenfeld residuals and log-minus-log curves),
    fit the model `Surv(t_event, event_flag) ~ treatment + age + sex + stage`, and output hazard ratios in a `gt` table.

    3. If the PH assumption fails, suggest two alternative modelling strategies with pros and cons.


    Inputs:

    - `{{dataset_path}}` — path to the patient dataset


    Output format:

    Section A: conceptual walk-through (bullets). Section B: fenced R code block. Section C: interpretation and next steps
    (\u2264250 words).'
testData:
- vars:
    dataset_path: example_dataset_path
  expected: 'Section A: conceptual walk-through (bullets). Section B: fenced R code block. Section C: interpretation and next
    steps (\u2264250 words).'
evaluators:
- name: Output starts with 'Section A:'
  string:
    startsWith: 'Section A:'