QIME: Constructing Interpretable Medical Text Embeddings via Ontology-Grounded Questions

Researchers have developed QIME, a new framework for creating interpretable medical text embeddings that uses ontology-grounded questions to represent biomedical text. Unlike black-box AI models, QIME provides clinically meaningful explanations while achieving performance close to traditional dense embeddings in medical text analysis tasks.

• →QIME creates interpretable medical embeddings where each dimension corresponds to a clinically meaningful yes/no question.
• →The framework uses medical ontologies and concept signatures to generate semantically atomic questions for fine-grained biomedical text analysis.
• →QIME offers a training-free embedding construction strategy that eliminates the need for per-question classifier training.
• →The method outperforms prior interpretable embedding approaches while narrowing the performance gap to black-box biomedical encoders.
• →The framework provides concise and clinically informative explanations for medical AI decision-making.