Human Phenotype Ontology (HPO) Mapper: Semantic Mapping of Clinical Findings to the Human Phenotype Ontology Using AI-Powered Embeddings and LLM-Based Quality Control

Background Structured phenotypic annotations linked to genetic data can drive diagnostic insight and therapeutic discovery in complex diseases. However, poor research access to the rich clinical data trapped in unstructured clinical records remains a significant barrier to phenotype-genotype integration. Here, we present Human Phenotype Ontology (HPO) Mapper, a scalable AI-assisted tool designed to ingest semantically structured clinical findings paired with anatomical region and accurately map them to HPO terms and associated genes.

We applied HPO Mapper to two forms of standardised clinical input extracted from inflammatory bowel disease (IBD) patient records. The first data type consisted of paired ‘clinical findings + anatomical regions’ derived from unstructured clinical reports and the second was standardised ICD-10 code-derived phenotypes. HPO Mapper achieved high semantic alignment and mapping accuracy for both data types (F1 = 0.85 ± 0.05 and 0.84 ± 0.03, respectively). Additionally, HPO Mapper was able to convert 62.3% of previously unusable free-text entries into HPO terms. Utility was demonstrated at cohort scale, whereby resultant HPO sets projected onto gene space recovered well-established IBD drivers including NOD2, IL6, STAT3, IL10RA, and CTLA4.

Our publicly available tool is suitable for converting clinical findings and regions into gene-linked HPO terms for precision medicine. This enables real-time HPO mappings, providing a foundation for scalable AI-driven phenotyping across diseases. More broadly, HPO Mapper provides a generalisable infrastructure for unlocking the latent value of clinical narrative data and bridging the gap between clinical records and genomic diagnostic discovery for targeted therapies. Competing Interest Statement JJA is a SAB member for Orchard Therapeutics, has participated in a personalisation of IBD think-tank project funded by Takeda and an IBD Delphi consensus process funded by Pfizer. Funding Statement This study was supported by the Institute for Life Sciences, University of Southampton, the NIHR Southampton BRC, and EPSRC (EP/Y01720X/1). JJA is funded by an NIHR advanced fellowship. ZG is funded by a CICRA research training fellowship. This work is supported by an SDE driver grant. Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: Ethics committee/IRB of University Hospital Southampton gave ethical approval for this work (REC 09/H0504/125). I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes Footnotes This version of the manuscript has been revised to include additional benchmarking and an additional figure demonstrating additional usage of HPO Mapper-derived HPO terms. Data Availability Due to patient privacy restrictions, underlying clinical data cannot be publicly shared. Therefore, we are providing a user-friendly demo available at https://huggingface.co/spaces/UoS-HGIG/HPOmapper and https://github.com/UoS-HGIG/HPO-Mapper, containing underlying code and data. Any additional information required to reanalyse the data reported in this work paper is available from the lead contact upon request. LIST OF ABBREVIATIONS - AI - artificial intelligence - HPO - Human Phenotype Ontology - IBD - inflammatory bowel disease - IBDU - inflammatory bowel disease unclassified - ICD-10 - International Statistical Classification of Diseases and Related Health Problems, 10th Revision - NHS - National Health Service - REC - Research Ethics Committee - GA4GH - Global Alliance for Genomics and Health - ISO - International Organization for Standardization - NLP - natural language processing - NCBO - National Center for Biomedical Ontology - LLM - large language model - QC - quality control - P1/P2/P3 - Protocol 1/Protocol 2/Protocol 3 - API - application programming interface - JSON - JavaScript Object Notation - PCA - principal component analysis - t-SNE - t-distributed stochastic neighbour embedding - L2 - Euclidean (L2) normalization - SEM - standard error of the mean - FDR - false discovery rate - TP/FP/FN - true positive/false positive/false negative - SNOMED CT - Systematized Nomenclature of Medicine Clinical Terms - HGNC - HUGO Gene Nomenclature Committee - KEGG - Kyoto Encyclopedia of Genes and Genomes - DisGeNET - disease-gene association database - GWAS - genome-wide association study - MRI - magnetic resonance imaging - CT - computed tomography - CPU - central processing unit - GPU - graphics processing unit - HPC - high-performance computing - kWh - kilowatt-hour - CO₂ - carbon dioxide - 2D/3D - two-dimensional/three-dimensional gene symbols - NOD2 - nucleotide-binding oligomerization domain containing 2 - IL6 - interleukin 6 - STAT3 - signal transducer and activator of transcription 3 - IL10RA - interleukin 10 receptor subunit alpha - CTLA4 - cytotoxic T-lymphocyte associated protein 4 - BACH2 - BTB domain and CNC homolog 2 - IL21 - interleukin 21 - XIAP - X-linked inhibitor of apoptosis - MEFV - MEFV innate immunity regulator (pyrin) - SYK - spleen associated tyrosine kinase.