https://github.com/cmangun/model-governance-framework

AI governance for healthcare ML. Bias detection, fairness metrics, FDA-ready model cards, responsible AI.
https://github.com/cmangun/model-governance-framework

bias-detection fairness fda healthcare ml-governance model-cards python responsible-ai

Last synced: 5 months ago
JSON representation

AI governance for healthcare ML. Bias detection, fairness metrics, FDA-ready model cards, responsible AI.

• Host: GitHub
• URL: https://github.com/cmangun/model-governance-framework
• Owner: cmangun
• Created: 2026-01-12T20:17:17.000Z (5 months ago)
• Default Branch: main
• Last Pushed: 2026-01-15T17:13:07.000Z (5 months ago)
• Last Synced: 2026-01-15T18:13:37.744Z (5 months ago)
• Topics: bias-detection, fairness, fda, healthcare, ml-governance, model-cards, python, responsible-ai
• Language: Python
• Size: 30.3 KB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 3
• Metadata Files:
  • Readme: README.md
  • Contributing: CONTRIBUTING.md

Awesome Lists containing this project

README

          
# Model Governance Framework

[![Python 3.11+](https://img.shields.io/badge/python-3.11+-blue.svg)](https://www.python.org/downloads/)

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)

Comprehensive AI governance framework for regulated industries with bias detection, fairness metrics, and model documentation.

## Business Impact

- **FDA-ready** model documentation with automated model cards

- **Bias detection** across protected demographic attributes

- **Explainability** with SHAP and LIME integration

- **Zero compliance violations** across $51M+ portfolio

## Key Features

### Fairness and Bias Detection

- **Demographic Parity**: Equal positive rates across groups

- **Equalized Odds**: Equal TPR and FPR across groups

- **Disparate Impact**: 4/5ths rule compliance

- **Intersectional Analysis**: Multi-attribute fairness

### Explainability

- SHAP value computation

- LIME explanations

- Feature importance analysis

- Decision boundary visualization

### Model Documentation

- Automated model cards (Google format)

- Intended use documentation

- Limitation and risk disclosure

- Healthcare-specific compliance notes

## Quick Start

```python

from src.fairness.bias_detector import BiasDetector, ProtectedAttribute

detector = BiasDetector()

report = detector.analyze(

    model_id="diabetes-classifier",

    model_version="1.0.0",

    predictions=[...],

    actuals=[...],

    protected_attributes={

        ProtectedAttribute.GENDER: ["male", "female", ...],

        ProtectedAttribute.AGE_GROUP: ["18-40", "41-65", "65+", ...],

    },

)

print(f"Overall Status: {report.overall_status}")

for violation in report.violations:

    print(f"  - {violation.metric}: {violation.description}")

```

## Project Structure

```

model-governance-framework/

├── src/

│   ├── fairness/

│   │   └── bias_detector.py     # Fairness metrics

│   ├── explainability/

│   │   └── explainer.py         # SHAP/LIME integration

│   ├── cards/

│   │   └── model_card.py        # Documentation generation

│   └── documentation/

│       └── report_generator.py  # Compliance reports

├── tests/

└── templates/

```

## Author

**Christopher Mangun** - [github.com/cmangun](https://github.com/cmangun)