SAGE generates medical training data using semantic anchors locally
SAGE: Semantic Anchor-Guided Evolution for Grounded Medical QA Data Synthesis
Computation and LanguageArtificial Intelligence
Summary
Training computer models to answer medical questions is hard because there isn’t enough expert-labeled data and privacy rules limit sharing information. The authors developed a method called SAGE that creates high-quality medical question-answer data using small models and public medical knowledge bases without relying on big datasets or outside services. SAGE builds up training examples step-by-step, guided by medical concepts, which helps the computer learn better. Tests show models trained on SAGE-made data perform better than other ways of generating training examples.
What this means in practice
- •For hospital data teams: Create high-quality training data for medical question-answering models without relying on external data or cloud APIs, working within hospital privacy rules.
- •For healthcare ai developers: Build more effective medical language models in resource-limited settings by synthesizing grounded training data from minimal seeds and publicly available taxonomies.
Authors
Chuan Li, Chengyu Wang, Cen Chen, Ye Lyu, Mingyuan Fan, Ming Gao
Abstract
Developing reliable models for clinical tasks, such as Medical Question Answering (QA), is severely constrained by the limited availability of high-quality, expert-annotated training data. This challenge is exacerbated by stringent privacy requirements and the impracticality of utilizing large open-source corpora or proprietary cloud APIs within resource-limited clinical settings. To address these obstacles, we introduce SAGE (\textit{Semantic Anchor-Guided Evolution}), a novel data synthesis framework that enables small, locally deployed models to generate high-quality medical training data. SAGE leverages lightweight, publicly available taxonomies such as MeSH as semantic anchors, imposing a structured prior to effectively guide and ground the data generation process. At its core, SAGE iteratively interleaves atomic (individual concept-based) and associative (relation-based) synthesis, bootstrapping training data from minimal seeds. This approach eliminates the need for large collections of medical documents or reliance on external APIs, providing a practical solution for on-premises data creation. Extensive experiments across multiple medical question-answering benchmarks demonstrate that models fine-tuned with SAGE-synthesized data consistently outperform those trained using self-derived or conventional document-based paradigms, highlighting tangible improvements in data efficiency and resource utilization for medical LLM development. Code is available at https://github.com/DIaacKr/SAGE.