Summary
The OncoTRAD translation agent is an AI-powered translation platform developed to transform highly technical, medically validated oncology texts into clear, plain Spanish for patients, caregivers, and the general public. By combining a rigorous, rule-based linguistic methodology with large language models, the solution drastically reduces translation time and cost while preserving medical accuracy and linguistic quality.Description
The project originated within a university linguistics research group as part of the oncoTRAD research initiative. Historically, cancer-related content from trusted medical sources was manually rewritten into plain Spanish—a process that was slow, expensive, and difficult to scale.
Nieve collaborated with the research team to build an AI-assisted translation system that follows the same explicit linguistic rules used by human translators. Rather than “free-form” generation, the AI agent operates within a strict framework covering tone, vocabulary, structure, and explanation of complex concepts, ensuring consistency, clarity, and trustworthiness.
The platform integrates an existing, client-owned medical terminology dictionary and supports a human-in-the-loop review workflow with both linguistic and medical validation. The architecture is modular and future-proof, allowing model upgrades, fine-tuning, and reuse across other domains with complex technical language.
Key Objectives
- Automate the transformation of technical medical texts into plain, accessible language
- Preserve full medical accuracy and linguistic rigor
- Significantly reduce translation time and operational costs
- Enable scalable content production for public-facing health communication
- Maintain human oversight through structured linguistic and medical review
- Build a reusable architecture adaptable to other knowledge-heavy domains
Expected Outcomes
- Translation time reduced from hours or days to minutes per document
- Very low cost per translated document with parallel processing capabilities
- A scalable, high-quality library of patient-friendly oncology content
- Improved accessibility of medical information for non-expert audiences
- A robust foundation for future extensions, including automatic publishing and conversational AI
- A reusable system architecture already validated for use beyond healthcare