• Computer Science > Artificial Intelligence [Submitted on 23 Feb 2026] Title:An artificial intelligence framework for end-to-end rare disease phenotyping from clinical notes using large language models View PDF HTML (experimental)Abstract:Phenotyping is fundamental to rare disease diagnosis, but manual curation of structured phenotypes from clinical notes is labor-intensive and difficult to scale. • Existing artificial intelligence approaches typically optimize individual components of phenotyping but do not operationalize the full clinical workflow of extracting features from clinical text, standardizing them to Human Phenotype Ontology (HPO) terms, and prioritizing diagnostically informative HPO terms. • We developed RARE-PHENIX, an end-to-end AI framework for rare disease phenotyping that integrates large language model-based phenotype extraction, ontology-grounded standardization to HPO terms, and supervised ranking of diagnostically informative phenotypes. • We trained RARE-PHENIX using data from 2,671 patients across 11 Undiagnosed Diseases Network clinical sites, and externally validated it on 16,357 real-world clinical notes from Vanderbilt University Medical Center. • Using clinician-curated HPO terms as the gold standard, RARE-PHENIX consistently outperformed a state-of-the-art deep learning baseline (PhenoBERT) across ontology-based similarity and precision-recall-F1 metrics in end-to-end evaluation (i.e., ontology-based similarity of 0.70 vs. • Ablation analyses demon

Article Summaries:

  • A new AI system, RARE‑PHENIX, was introduced to streamline rare‑disease diagnosis by automatically extracting, standardizing, and ranking phenotypic features from clinical notes. The framework combines large‑language‑model extraction, mapping to Human Phenotype Ontology (HPO) terms, and supervised prioritization of diagnostically relevant HPO codes. Trained on 2,671 patients from 11 Undiagnosed Diseases Network sites and validated on 16,357 notes from Vanderbilt, RARE‑PHENIX achieved an ontology‑based similarity of 0.70 versus 0.58 for the PhenoBERT baseline. Ablation studies confirmed incremental gains from each module, suggesting the system could support clinicians in real‑world rare‑disease workflows.

Sources: