BPG is committed to discovery and dissemination of knowledge
Minireviews
Copyright ©The Author(s) 2025.
World J Gastroenterol. Oct 14, 2025; 31(38): 110999
Published online Oct 14, 2025. doi: 10.3748/wjg.v31.i38.110999
Table 1 Common clinical features and diagnostic criteria of eosinophilic esophagitis[1,65]
Feature
Description
Relevance
Definitive diagnosisHistological identification of 15 or more eosinophils per HPF[65]Gold standard for diagnosis
Endoscopic findingsRings, linear furrows, exudates, edema[65]Key visual indicators
Associated atopic conditionsAsthma, eczema, allergic rhinitis, food allergies[1]Common co-morbidities, important for management
SymptomsDysphagia, food impaction, chest pain, reflux-like symptoms[1]Impact on quality of life
Risk of complicationsIncreased risk of esophageal strictures and remodeling[1]Importance of early diagnosis
Table 2 Summary of artificial intelligence methodologies applicable to eosinophilic esophagitis diagnosis
AI methodology
Description
Applications in EoE
Supervised MLAlgorithms that rely on labeled training data to learn and predict outcomesClassifying endoscopic images, predicting eosinophil counts
DLSubset of ML utilizing neural networks for pattern recognitionAnalyzing endoscopic and histopathological data
Convolutional neural networksSpecial type of DL particularly adept at image recognitionEnhancing diagnostic accuracy in endoscopic images
Random decision forestML algorithm combining multiple data sources for improved decision makingIntegrating clinical and endoscopic data for diagnosis
Natural language processingAI field focused on interaction between computers and human languageAnalyzing electronic health records for diagnostic insights
Table 3 Comparative table of the main artificial intelligence models applied to endoscopic diagnosis of esophagitis diagnosis
Model
Training dataset
Population size
Validation status
Key limitations
Methodological rigor
AI-EoE-EREFS (DL/CNN)[48]Endoscopic images, EREFS-labeled, multi-center484 images from 134 patientsRetrospective, some external validationRetrospective data, limited diversity, mainly binary classification, limited real-world scenariosHigh accuracy (AUC = 0.992), but generalizability and real-world robustness uncertain
CNN[47]Endoscopic images (EoE vs controls)1192 characteristic endoscopic images of 108 patientsTested against patients and controlsDataset size not reported, unclear diversity, retrospective designHigh accuracy (95%), but limited transparency and unclear reproducibility
CNN with EREFS integration[49]Endoscopic images annotated with EREFS200 WLIs, including 100 WLIs from EoE patients and 100 WLIs of normal esophagusCompared to human expertsTraining/test data overlap possible, unclear external validationSensitivity improved to 85%, specificity 95% with EREFS
Random decision forest[50]Clinical + endoscopic dataNot specifiedReal-world data, AUC up to 0.94Complexity, integration challenges, population diversity not detailedRobust with multi-source inputs, but needs more transparent validation
General endoscopic AI (DL/CNN)Mixed datasets (WLI, NBI, CT images)Varies; often smallMostly retrospective, single/multi-centerData heterogeneity, lack of multi-center, prospective validationHigh reported accuracy, but shallow validation, risk of bias due to dataset curation