Artificial Intelligence for the Diagnosis and Prognosis of Oral Potentially Malignant Disorders and Oral Squamous Cell Carcinoma: A Systematic Review and Independent Quantitative Synthesis
Mahek Suryanarayan, Santha Devy Arumugam, Vezhavendhan Nagarajan, Umamaheswari GiriAbstract
Objective:
Oral potentially malignant disorders (OPMDs) are associated with an increased risk of progression to oral squamous cell carcinoma (OSCC), making early and accurate risk prediction essential for improving patient outcomes. While clinical examination and histopathological assessment remain the standard diagnostic approaches, predicting which lesions will undergo malignant transformation continues to be challenging. In recent years, artificial intelligence (AI) has emerged as a promising adjunctive tool for enhancing diagnostic precision and prognostic assessment. This systematic review and independent quantitative synthesis evaluates and compares contemporary AI-based algorithms used for early detection and prediction of OPMDs and OSCC.
Methods:
A comprehensive search of PubMed, Scopus, Google Scholar, Cochrane Central Register of Controlled Trials, and Web of Science was conducted for the studies published between 2015 and 2025. Studies assessing AI models for the diagnosis or prediction of malignant transformation were included. Of the 1909 records identified, 7 unique studies (yielding 8 independent datasets) met the eligibility criteria and were included in the qualitative and independent quantitative synthesis.
Results:
The included studies reported improvement in the diagnostic accuracy and predictive performance of AI-based tools such as random survival forest, DeepSurv, and convolutional neural network in identifying high-risk lesions. Among the evaluated models, DeepSurv demonstrated comparatively stronger prognostic capability for predicting lesion progression. However, variability in datasets, validation strategies, and outcome measures were noted across several studies.
Conclusion:
AI-based predictive models show meaningful potential in supporting early detection and risk stratification of OPMDs and OSCC. However, further standardized, large-scale studies are necessary to establish their reliability and facilitate integration into routine clinical practice.
PROSPERO registry number:
CRD420251236709.