This project aims to establish a centralised, open-source repository that provides comprehensive access to legal, policy, and strategy documents on Artificial Intelligence for Health, Health Data, and Digital Health, with a strong focus on supporting Low- and Middle-Income Countries (LMICs). The platform will serve as a one-stop platform, allowing policymakers and decision-makers to easily access official documents, enable cross-country and cross-regional comparisons, and support informed analysis for evidence-based decision-making, particularly in LMICs where resources and governance frameworks may be limited. In addition to providing policies and guidelines produced by governments, regional bodies, and multilateral organisations, the repository will leverage AI-powered tools to assess alignment with global equity frameworks and evolving Health Data Governance Principles, ensuring that positioning and principles remain central to its design.
Built as a living and agile platform, the repository will be continuously and updated to include both newly published and revised legal, policy, and strategy documents as they emerge. To ensure its relevance and reliability, the initiative is coordinated by United Nations University through its International Institute for Global Health (UNU-IIGH) and Campus Computing Centre, and will invite voluntary contributions from experts and non-experts worldwide in multiple languages, with UNU-IIGH curating and validating entries.
AI Approach:
Leveraging a hybrid of Generative AI and NLP, the platform transforms static policy documents into actionable insights by mapping them to recognised governance and equity frameworks. Uniquely, the system employs a strength-based evaluation model: rather than solely identifying gaps, it highlights where policies successfully reflect global principles. This positive reinforcement approach is designed to catalyse knowledge exchange and capacity building, specifically enabling policymakers in LMICs to recognise and scale their existing governance assets. The result is an objective, transparent analysis that promotes best practices while remaining sensitive to local contexts.
Technology Platform:
Our platform leverages a flexible Generative AI ecosystem, orchestrating top-tier providers, such as OpenAI, Azure, Google Gemini, Anthropic Claude, Perplexity, and DeepSeek, alongside local models for full control and privacy. This multi-model architecture balances cost and accuracy by deploying lightweight models for metadata extraction and frontier reasoning engines for complex policy analysis. The Streamlit-based web interface enables iterative refinement, facilitating prompt calibration and output comparison to identify the optimal LLM provider. Airtable integration enables structured, shareable data, with flexible exports (JSON, DOCX, PDF, Markdown) for easy analysis and transparent results. The backend is built on Python, laying the foundation for a planned batch processing phase. This upcoming capability will enable scalable operations by targeting specific directories of policy documents for automated, high-volume execution with minimal manual intervention.
