Diagnostic Performance of an AI-Assisted Mobile Screening App for Oral Potentially Malignant Disorders: A Comparative Study in Rural and Urban Pakistan

Abstract

Oral Potentially Malignant Disorders (OPMDs) are still a major problem in the public health of Pakistan, especially in the rural population that lacks access to early detection services. Mobile applications aided by artificial intelligence (AI) have become an inexpensive and quick method of early screening and risk classification of oral lesions. The purpose of this research was to assess and compare the diagnostic capabilities of an AI-based mobile screening application, i.e., sensitivity, specificity and general accuracy, in identifying OPMDs among adults living in rural and urban areas of Pakistan. A comparative cross-sectional study design was used. A random sample size of 50 subjects was used in the study, and these participants aged 18 to 65 years were sampled through a stratified random sampling method that incorporated both rural and urban populations into the sample. The study was done in the primary healthcare facilities and community health campaigns within the rural districts and dental outpatient units of tertiary-care hospitals within urban settings. (1) The AI-Assisted Mobile Screening App was used to collect the data in which intraoral photographs were analysed to identify possible OPMDs, and (2) the Clinical Oral Examination Form was used as the reference gold standard in which data were collected by trained dental surgeons. The variables that were used to evaluate AI performance in the two settings were sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and diagnostic accuracy. The pilot data showed that the AI-based app was more sensitive among rural populations than urban ones, and specificity was a bit better in urban areas. The overall diagnostic accuracy was good in both regions, indicating that the mobile app is usable in the early detection of OPMD. The AI-assisted mobile screening application had great potential to be used as a useful, accessible application to assist in the early detection of OPMDs in rural and urban Pakistan. Its use can contribute to the fact that the diagnostic delays decrease, and more preventive oral healthcare is provided, especially in underserved populations.