---
title: "Privacy-preserving hybrid pipeline to extract uncoded UTI features from Catalonia primary care EHRs"
id: "medrxiv-16-uncoded-clinical-features-from-multilingual-electronic-health-records-in"
canonical_url: "https://medichelpline.com/clinical-feed/medrxiv-16-uncoded-clinical-features-from-multilingual-electronic-health-records-in"
content_type: "clinical_feed_article"
specialty: "General"
source_name: "medRxiv (Clinical Preprints)"
source_url: "https://www.medrxiv.org/content/10.64898/2026.09.16.26363030v1?rss=1"
published_at: "2026-09-22T12:00:00.000Z"
evidence_level: "Verified Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# Privacy-preserving hybrid pipeline to extract uncoded UTI features from Catalonia primary care EHRs
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/medrxiv-16-uncoded-clinical-features-from-multilingual-electronic-health-records-in
- **Specialty:** [General](https://medichelpline.com/clinical-feed/general.md)
- **Primary Source:** medRxiv (Clinical Preprints)
- **Source URL:** [Original Journal Publication](https://www.medrxiv.org/content/10.64898/2026.09.16.26363030v1?rss=1)
- **Published At:** 2026-09-22T12:00:00.000Z
- **Evidence Rating:** Verified Feed
## Executive GIST (TL;DR)
- Unstructured free-text in Catalonia primary care EHRs (MEAP format) holds clinically relevant information not captured by structured codes; the study targets six uncoded **urinary tract infection (UTI)** features: fever, nitrites, leukocytes, lumbar pain, abdominal pain, and haematuria. - A local hybrid pipeline combined a compact open-weight **small language model (Phi4-mini, 3.8B)** deployed inside institutional servers with post‑processing deterministic **regular expressions**, avoiding cloud APIs and external data transfer. - Model inference used Ollama within the institutional firewall; prompts were iteratively refined with clinician input. Regular expressions were applied after model output to standardize extraction. - Pipeline processed 15,498 MEAP narratives from 2,962 matched primary care patients and identified 4,663 occurrences of the target clinical features; patients who progressed to acute pyelonephritis had a higher features burden than controls (71.3% vs 55.8%), with notably higher rates of fever and lumbar pain. - Two validation arms were used: a double‑blind clinician gold standard review of 60 real patient records (consensus resolution) and an adversarial synthetic dataset of 720 notes augmented with linguistic noise generated using GPT‑4.1 and Grok‑4.1. - Real‑world clinician validation performance: 93.8% accuracy (95% CI 90.6%–96.1%), 83.6% sensitivity (95% CI 73.0%–91.2%), 96.6% specificity (95% CI 93.6%–98.4%), 87.1% PPV (95% CI 77.0%–93.9%), 95.5% NPV (95% CI 92.3%–97.7%). - Synthetic stress‑test performance: 87.2% accuracy (95% CI 84.6%–89.6%), 73.9% sensitivity (95% CI 68.9%–78.5%), 99.5% specificity (95% CI 98.1%–99.9%), 99.2% PPV (95% CI 97.2%–99.9%). - The hybrid approach achieved high **specificity** and **precision** while operating entirely on secure institutional infrastructure, avoiding API costs and protecting patient data; sensitivity was lower for complex narrative descriptions. - Data are not publicly available due to institutional and regulatory restrictions. Ethical approval was obtained from the IDIAP Jordi Gol Ethics Committee (22/089-P).
## Clinical Analysis & Structured Key Points
Uncoded Clinical Features from Multilingual Electronic Health Records in Catalonia: Development and Validation Study | medRxiv Skip to main content Uncoded Clinical Features from Multilingual Electronic Health Records in Catalonia: Development and Validation Study View ORCID Profile Dan Ouchi , Silvia Riudavets , View ORCID Profile Alvaro Franquet , View ORCID Profile Silvia Fernández-García , View ORCID Profile Carl Llor , View ORCID Profile Ana Moragas Moreno , View ORCID Profile Maria Giner-Soriano , View ORCID Profile Ferran Torres , View ORCID Profile Rosa Morros doi: https://doi.org/10.64898/2026.09.16.26363030 Dan Ouchi 1 IDIAP Jordi Gol; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Dan Ouchi For correspondence: dov.bcn{at}gmail.com Silvia Riudavets 2 IDIAP Jordi Gol i Gurina; Find this author on Google Scholar Find this author on PubMed Search for this author on this site Alvaro Franquet 3 Institut Català de la Salut; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Alvaro Franquet Silvia Fernández-García 4 Gerència d'Atenció Primària i a la Comunitat (GAPiC) Baix Llobregat; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Silvia Fernández-García Carl Llor 2 IDIAP Jordi Gol i Gurina; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Carl Llor Ana Moragas Moreno 5 Universitat Rovira i Virgili; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Ana Moragas Moreno Maria Giner-Soriano 2 IDIAP Jordi Gol i Gurina; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Maria Giner-Soriano Ferran Torres 6 Universitat Autònoma de Barcelona Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Ferran Torres Rosa Morros 2 IDIAP Jordi Gol i Gurina; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Rosa Morros Abstract Info/History Metrics Supplementary material Data/Code Preview PDF Abstract Background: Unstructured free‑text narratives in electronic health records (EHRs) contain critical clinical information that is not captured by structured standard codes. In Catalonia, primary care notes follow a semi‑structured format known as MEAP (in Catalan). However, extracting structured variables using cloud‑hosted commercial large language models (LLMs) raises substantial data privacy concerns and incurs high operational costs. Objective: To develop and validate a privacy-preserving local hybrid pipeline combining a lightweight small language model (SLM) with deterministic regular expressions for extracting six uncoded urinary tract infection (UTI) clinical features (fever, nitrites, leukocytes, lumbar pain, abdominal pain, and haematuria) from primary care EHR narratives. Methods: We deployed the open‑weight model (Phi4-mini, 3.8B parameters) within the institutional firewall using Ollama. Prompts were iteratively refined in collaboration with clinicians, and regular expressions were applied post‑inference. Validation was conducted across two independent arms: (1) a double‑blind clinician gold standard review of 60 real patient records (consensus‑resolved cases), and (2) an adversarial synthetic dataset of 720 notes enriched with linguistic noise, generated using GPT‑4.1 and Grok‑4.1. Point estimates and 95% confidence intervals (CIs) were computed for key performance metrics. Results: The pipeline processed 15,498 MEAP narratives from a matched cohort of 2,962 primary care patients and identified 4,663 clinical feature occurrences. Patients progressing to acute pyelonephritis (cases) presented a higher features burden than non-progressing controls (71.3% vs. 55.8%; standardized mean difference [SMD] = 0.326), particularly for fever (33.0% vs. 9.0%; SMD = 0.616) and lumbar pain (29.0% vs. 9.8%; SMD = 0.502). In real-world clinician validation, overall performance yielded 93.8% accuracy (95% CI 90.6%-96.1%), 83.6% sensitivity (95% CI 73.0%-91.2%), 96.6% specificity (95% CI 93.6%-98.4%), 87.1% positive predictive value (PPV; 95% CI 77.0%-93.9%), and 95.5% negative predictive value (NPV; 95% CI 92.3%-97.7%). In synthetic stress-testing, the pipeline demonstrated 87.2% accuracy (95% CI 84.6%-89.6%), 73.9% sensitivity (95% CI 68.9%-78.5%), near-perfect specificity (99.5%, 95% CI 98.1%-99.9%), and 99.2% PPV (95% CI 97.2%-99.9%). Conclusions: A privacy-preserving local hybrid framework combining a compact open-weight SLM with regular expression rules achieves high specificity and precision for extracting six uncoded clinical features from multilingual primary care narratives. Running entirely within institutional servers, this approach keeps patient data secure, meets privacy standards, and avoids Application Programming Interface (API) costs, making it a practical tool for EHR research, though with lower sensitivity for narratively complex descriptions. Competing Interest Statement The authors have declared no competing interest. Clinical Protocols https://www.researchprotocols.org/2023/1/e44244 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: The study was conducted in accordance with the Declaration of Helsinki, Good Research Practice principles and guidelines, and the Real Decreto 957/2020, of November 3, which regulates observational studies of medicines for human use. The study protocol was approved by the Ethics Committee of IDIAP Jordi Gol with ethical approval code 22/089-P (protocol code IJG-ITUCAT-2022, and date of approval: 27 July 2022). All data was stored, processed and analysed in our secure institutional server. 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 Data Availability The data produced in the present study are not available due to institutional and regulatory restrictions on the sharing of patient‑level clinical data. Copyright The copyright holder for this preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Back to top Previous Next Posted September 22, 2026. Download PDF Supplementary Material Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Uncoded Clinical Features from Multilingual Electronic Health Records in Catalonia: Development and Validation Study Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Uncoded Clinical Features from Multilingual Electronic Health Records in Catalonia: Development and Validation Study Dan Ouchi , Silvia Riudavets , Alvaro Franquet , Silvia Fernández-García , Carl Llor , Ana Moragas Moreno , Maria Giner-Soriano , Ferran Torres , Rosa Morros medRxiv 2026.09.16.26363030; doi: https://doi.org/10.64898/2026.09.16.26363030 Share This Article: Copy Citation Tools Uncoded Clinical Features from Multilingual Electronic Health Records in Catalonia: Development and Validation Study Dan Ouchi , Silvia Riudavets , Alvaro Franquet , Silvia Fernández-García , Carl Llor , Ana Moragas Moreno , Maria Giner-Soriano , Ferran Torres , Rosa Morros medRxiv 2026.09.16.26363030; doi: https://doi.org/10.64898/2026.09.16.26363030 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Areas All Articles Addiction Medicine (619) Allergy and Immunology (903) Anesthesia (335) Cardiovascular Medicine (4835) Dentistry and Oral Medicine (479) Dermatology (419) Emergency Medicine (651) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1634) Epidemiology (15982) Forensic Medicine (33) Gastroenterology (1206) Genetic and Genomic Medicine (7070) Geriatric Medicine (734) Health Economics (1067) Health Informatics (5070) Health Policy (1438) Health Systems and Quality Improvement (1776) Hematology (587) HIV/AIDS (1347) Infectious Diseases (except HIV/AIDS) (16313) Intensive Care and Critical Care Medicine (1177) Medical Education (673) Medical Ethics (155) Nephrology (727) Neurology (7308) Nursing (370) Nutrition (1087) Obstetrics and Gynecology (1245) Occupational and Environmental Health (1007) Oncology (3622) Ophthalmology (1058) Orthopedics (398) Otolaryngology (458) Pain Medicine (476) Palliative Medicine (140) Pathology (709) Pediatrics (1805) Pharmacology and Therapeutics (739) Primary Care Research (769) Psychiatry and Clinical Psychology (5923) Public and Global Health (9778) Radiology and Imaging (2437) Rehabilitation Medicine and Physical Therapy (1456) Respiratory Medicine (1250) Rheumatology (643) Sexual and Reproductive Health (773) Sports Medicine (582) Surgery (783) Toxicology (107) Transplantation (306) Urology (293)
## Related Clinical Research

- [First‑trimester glycolipid and inflammatory markers linked to gestational diabetes risk in rural S](https://medichelpline.com/clinical-feed/bmj-open-11-evaluation-of-the-first-trimester-maternal-glycolipid-profile-on-gestational.md)
- [Caesarean Deliveries in Tamil Nadu: Prevalence, Costs, and Postpartum Quality of Life](https://medichelpline.com/clinical-feed/plos-one-13-caesarean-deliveries-and-their-impact-on-financial-burden-and-quality-of-life.md)
- [Key factors for equitable UBI-based health interventions for low-income Black men with chronic ill](https://medichelpline.com/clinical-feed/plos-one-10-key-factors-for-an-equitable-health-intervention-for-black-men-with-low-income.md)
- [Usability and Impact of the WHO TAP eHealth Module on Assistive Product Provision](https://medichelpline.com/clinical-feed/medrxiv-13-evaluation-of-an-ehealth-assistive-device-provision-training-tool-a-quasi.md)
- [Patient Safety Culture at an Indian Tertiary Hospital Using I-HSOPSC 2.0](https://medichelpline.com/clinical-feed/medrxiv-3-assessing-patient-safety-culture-among-doctors-nurses-and-allied-health.md)

## Navigation
- [← Back to General Feed](https://medichelpline.com/clinical-feed/general.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.