Related Experiment Video
Updated: Aug 6, 2026
Establishment and Evaluation of a Risk Prediction Model for Pathological Escalation of Gastric Low-Grade Intraepithelial Neoplasia
Published on: February 16, 2024
Predicting Malignant Transformation in Oral Epithelial Dysplasia: A Systematic Comparison of Artificial
Diele Carine Barreto Arantes1,2, Lucas Monteiro Delgado3, Marília de Marco Pinto Paiva Dos Santos1
1Oral Medicine and Pathology, Faculdade São Leopoldo Mandic, Research Institute, Rua Dr. José Rocha Junqueira, Campinas, São Paulo, Brazil.
Artificial intelligence (AI) models show superior accuracy in predicting malignant transformation from oral epithelial dysplasia (OED) compared to human analysis. While AI improves decision precision, further validation is needed for clinical use in oral potentially malignant disorders (OPMDs).
Area of Science:
- Oral pathology
- Medical artificial intelligence
- Histopathology
Background:
- Oral potentially malignant disorders (OPMDs) require accurate risk stratification for personalized management.
- Histopathological analysis of oral epithelial dysplasia (OED) is crucial for predicting malignant transformation (MT).
- Risk prediction models (RPMs) aid in stratifying OPMD patients.
Purpose of the Study:
- To systematically review and compare the performance of AI-based RPMs versus conventional human microscopy in predicting MT in OED.
- To evaluate the discriminative ability of AI in OPMD risk assessment.
Main Methods:
- Systematic review and meta-analysis of 32 studies (1998-2026).
- Included studies focused on OED in oral leukoplakia, erythroleukoplakia, and erythroplakia.
- AI models utilized machine learning (ML), deep learning (DL), or hybrid methods; human assessment involved microscopy.
Main Results:
- AI models demonstrated strong discriminative performance (OR=12.64) with higher specificity and consistency.
- Human evaluation showed a significant association between OED and MT (OR=3.42) but with substantial heterogeneity.
- AI models improved decision precision for MT prediction in OED.
Conclusions:
- AI-based RPMs show promising and more reliable discriminative ability for MT prediction in OED.
- AI primarily enhances decision precision rather than detection rates.
- Further standardization, external validation, and multicentric datasets are necessary for clinical implementation of AI models.
