ADaM YAML Documentation

Overview

This project provides a Python module for handling YAML-based specifications for ADaM (Analysis Data Model) datasets in clinical trials, following CDISC standards.

Features

• YAML-based Specifications: Define ADaM datasets using human-readable YAML files
• SQL-First Derivations: Most transformations executed via Polars SQL for optimal performance
• Dynamic Function Loading: Extensible function system without code modifications
• CDISC Compliant: Follows ADaM implementation guidelines

Quick Start

Installation

uv pip install -e .

Basic Usage

from adamyaml.adam_derivation import AdamDerivation

# Load specification and build dataset
engine = AdamDerivation("spec/study1/adsl_study1.yaml")
df = engine.build()

# Save to file
engine.save()

Architecture

The module uses a minimal 3-class design:

1. BaseDerivation: Abstract base class with context storage
2. SQLDerivation: Handles 90% of patterns via SQL
3. FunctionDerivation: Dynamic function loading for custom logic

Examples

• ADSL Example - Complete example of generating an ADSL dataset
• API Reference - Detailed API documentation

Key Technologies

• Python 3.10+: Modern Python with type hints
• Polars: High-performance DataFrame library
• PyYAML: YAML parsing and generation
• Quarto: Documentation generation