[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-128301-en":3,"doc-seo-128301-105":31,"detail-sidebar-cat-0-en-105":92},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":28,"seo_description":14,"update_tm":29,"read_time":30},128301,962085570644,"Evangeline","https://ap-avatar.wpscdn.com/davatar_994ba38a5ba835b3df7d355c54d3ed8d",8,"Research & Report","Interpretable Machine Learning-Based Detection of Coeliac Disease","Interpretable AI for coeliac disease diagnosis addresses limited interpathologist agreement and the need for transparent decision support in digital pathology. The approach trains semantic segmentation on H&E-stained duodenal biopsies to generate explainable masks of villi, crypts, intraepithelial lymphocytes, and enterocytes, then derives IEL-to-enterocyte and villus-to-crypt ratios. Logistic regression uses these ratios to diagnose coeliac disease. Evaluation on an independent set of whole slide images shows high segmentation performance and strong diagnostic accuracy, precision, and negative predictive value.","Original research  \n Interpretable machine learning-based  \ndetection of coeliac disease  \nFlorian Jaeckle  ,1,2 Rebekah Bryant,1,2 James Denholm,1,2  \nJacobo Romero Diaz,1 Benjamin Schreiber,1,3 Vrinda Shenoy,1 David Ekundayomi,1 Shelley C Evans  ,1 Mark J Arends  ,4 Elizabeth Soilleux1,2  \nTo cite: Jaeckle F, Bryant R, Denholm J, et al. Interpretable machine learning-based detection of coeliac disease. BMJ Digital Health and AI 2025;1:e000023 . doi:10 . 1136/ bmjdhai-2025-000023  \n► Additional supplemental material is published online only. To view, please visit the journal online ([https://doi.org/10.1136/](https://doi.org/10.1136/)[ ](https://doi.org/10.1136/)[bmjdhai-2025-000023](bmjdhai-2025-000023)) .  \nFJ and RB contributed equally.  \nReceived 4 February 2025 Accepted 8 September 2025  \n© Author(s) (or their employer(s)) 2025. Re-use permitted under CC BY.  \nPublished by BMJ Group. 1Department of Pathology, University of Cambridge, Cambridge, UK  \n2Lyzeum, Cambridge, UK 3Department of Applied Mathematics and Theoretical Physics, University of Cambridge, Cambridge, UK 4Division of Pathology, University of Edinburgh, Edinburgh, UK  \nCorrespondence to  \nDr Florian Jaeckle; [fj286@cam.ac.uk](fj286@cam.ac.uk)  \nABSTRACT  \nObjective Coeliac disease, an autoimmune disorder affecting approximately 1% of the global population, is typically diagnosed on duodenal biopsy. However, interpathologist agreement on coeliac disease diagnosis is only 80% . Existing machine learning solutions designed to improve coeliac disease diagnosis often lack interpretability, which is essential for building trust and enabling widespread clinical adoption. We aim to develop an interpretable artificial intelligence (AI) model segmenting key histological structures in H&E-stained duodenal biopsies, generating explainable segmentation masks, estimating intraepithelial lymphocyte (IEL)-toenterocyte and villus-to-crypt ratios and diagnosing coeliac disease.  \nMethods and Analysis Semantic segmentation models were trained to identify villi, crypts, IELs and enterocytes using 49 annotated 2048×2048 patches at 40× magnification. Subsequently, IEL-to-enterocyte and villusto-crypt ratios were calculated from segmentation masks generated by the segmentation model from 172 whole slide images (WSIs), and a logistic regression model was trained to diagnose coeliac disease based on these ratios. Evaluation was performed on an independent test set of 613 WSIs from an independent medical institution.  \nResults The villus–crypt segmentation model achieved mean precision-recall area under the curve (AUC) of 80.5%, while the IEL–enterocyte model reached precisionrecall AUC of 82% . The diagnostic model classified WSIs with 96% accuracy, 86% positive predictive value and 98% negative predictive value on the independent test set. Conclusions Our interpretable AI models accurately segmented key histological structures and diagnosed coeliac disease in unseen WSIs, demonstrating strong generalisation performance. These models provide pathologists with reliable IEL-to-enterocyte and villusto-crypt ratio estimates, enhancing diagnostic accuracy. Interpretable AI solutions like ours are essential for fostering trust among healthcare professionals and patients, complementing existing black-box methodologies.  \nINTRODUCTION  \nWith the emergence of digital pathology, artificial intelligence (AI) has the potential to improve the accuracy and speed of diagnosis. Duodenal (small intestinal) biopsies in particular are well suited to develop novel  \n\n| WHAT IS ALREADY KNOWN ON THIS TOPIC |\n| --- |\n| ⇒ Pathologist concordance in diagnosing coeliac disease from duodenal biopsies is consistently reported to be below 80%, highlighting diagnostic variability and the need for improved methods. Several recent studies have leveraged artificial intelligence (AI) to enhance coeliac disease diagnosis. However, most of these models operate as ‘black boxes’, offering limited interpretability and t","cbCaiiP4fbG2Pdli","https://ap.wps.com/l/cbCaiiP4fbG2Pdli","pdf",9417785,2,1,10,"English","en",105,"# Abstract\n## Objective and aim\n## Methods and training approach\n## Results and diagnostic performance\n## Conclusions and implications\n# Introduction\n## Digital pathology and diagnostic variability\n# What is already known on this topic\n## Pathologist concordance and black-box AI limitations\n# What this study adds\n## Interpretable segmentation and ratio-based diagnosis\n# How this study might affect research, practice or policy\n## Workflow efficiency and trust in clinical adoption","[{\"question\":\"Why is interpretability important for coeliac disease AI tools?\",\"answer\":\"Current machine-learning methods often behave as black boxes with limited transparency. Interpretability is essential to build trust for clinicians and support wider clinical adoption.\"},{\"question\":\"How does the model diagnose coeliac disease?\",\"answer\":\"It segments key histological structures on H\\u0026E-stained duodenal biopsies, produces explainable segmentation masks, computes IEL-to-enterocyte and villus-to-crypt ratios, and then uses these ratios in logistic regression to diagnose coeliac disease.\"},{\"question\":\"What performance does the model achieve on the independent test set?\",\"answer\":\"On unseen whole slide images from an independent institution, the diagnostic model reports 96% accuracy, 86% positive predictive value, and 98% negative predictive value.\"}]","Interpretable Machine Learning-Based Detection of Coeliac Disease | PDF",1785946702,25,{"code":4,"msg":32,"data":33},"ok",{"site_id":25,"language":24,"slug":34,"title":13,"keywords":35,"description":14,"schema_data":36,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":29},"interpretable-machine-learning-based-detection-of-coeliac-disease","",{"@graph":37,"@context":86},[38,54,69],{"@type":39,"itemListElement":40},"BreadcrumbList",[41,45,48,51],{"item":42,"name":43,"@type":44,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":46,"name":47,"@type":44,"position":20},"https://docshare.wps.com/document/","Document",{"item":49,"name":12,"@type":44,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":44,"position":53},"https://docshare.wps.com/document/interpretable-machine-learning-based-detection-of-coeliac-disease/128301/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":24,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":42,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-27","2026-08-05",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"Why is interpretability important for coeliac disease AI tools?","Question",{"text":76,"@type":77},"Current machine-learning methods often behave as black boxes with limited transparency. Interpretability is essential to build trust for clinicians and support wider clinical adoption.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How does the model diagnose coeliac disease?",{"text":81,"@type":77},"It segments key histological structures on H&E-stained duodenal biopsies, produces explainable segmentation masks, computes IEL-to-enterocyte and villus-to-crypt ratios, and then uses these ratios in logistic regression to diagnose coeliac disease.",{"name":83,"@type":74,"acceptedAnswer":84},"What performance does the model achieve on the independent test set?",{"text":85,"@type":77},"On unseen whole slide images from an independent institution, the diagnostic model reports 96% accuracy, 86% positive predictive value, and 98% negative predictive value.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,135],{"id":21,"doc_module":4,"doc_module_name":47,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":47,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":47,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":47,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":47,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":47,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":47,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":47,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":47,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":22,"doc_module":4,"doc_module_name":47,"category_name":133,"show_sort_weight":22,"slug":134},"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":47,"category_name":137,"show_sort_weight":107,"slug":138},19,"General","general"]