Clinical RCT Analysis — just your Excel

中文 · English

A reusable Codex skill for converting clinical randomized controlled trial (RCT) spreadsheets into an audit-ready, reproducible analysis workflow: from data checks to statistical code, tables, figures, and manuscript-ready results.

What it supports

• Data inventory, validation, ID/visit reconciliation, and missing-data checks
• FAS, ITT, PPS, and safety-set definitions with participant-flow reporting
• Model selection for continuous, binary, count, ordinal, and time-to-event outcomes
• Repeated-measures analysis using mixed-effects models or an appropriate alternative
• Missing-data, multiplicity, subgroup, and sensitivity-analysis requirements
• Reproducible R-oriented code generation with explicit quality checks
• Publication-ready tables, figures, and results narratives aligned to verified output

Repository structure

SKILL.md                         # Main Codex skill instructions
agents/openai.yaml               # Skill display metadata
references/analysis-requirements.md
references/reporting-guide.md

Install and use

Copy this repository folder into your Codex skills directory, retaining the folder name clinical-rct-analysis. Then invoke:

$clinical-rct-analysis

Provide the protocol/SAP or study question, data file, variable definitions, primary endpoint and time point, and planned analysis-set rules. The skill treats any data-derived assumption as provisional until it is verified against those study materials.

Scope and privacy

This skill supports statistical analysis and reporting. It does not replace a protocol/SAP, qualified biostatistical review, ethics oversight, clinical judgment, or regulatory review.

No clinical study data, participant information, project-specific code, or study results are included in this repository. Do not upload identifiable or otherwise sensitive trial data to a public repository.

---

中文说明

这是一个可复用的 Codex Skill，用于将临床随机对照试验（RCT）的 Excel/CSV 数据转化为可审计、可复现的统计分析流程。它覆盖数据核查、FAS/ITT/PPS 分析集、纵向随访模型、缺失数据、多重比较、R 代码生成、结果表图与论文结果段落。

使用时请提供研究方案或统计分析计划、数据文件、变量说明、主要结局与主要时间点，以及分析集的既定规则。Skill 不会把当前项目的结局指标或临时规则强加于新的研究。

该仓库不包含任何临床原始数据、受试者信息、项目代码或项目结果。公开发布时请勿上传可识别或其他敏感试验数据。

English

This repository contains a reusable Codex skill for clinical RCT data analysis. It supports transparent, SAP-aligned workflows from spreadsheet validation through reproducible code, statistical reporting, tables, figures, and manuscript-ready results. It is intentionally endpoint-agnostic and does not reuse assumptions from a prior trial without confirmation from the new study materials.