JMIR Publications

ABSTRACT

Background:

Artificial intelligence (AI)-based tools for oral cancer screening have shown promising performance in curated or retrospective image datasets, but prospective evidence from mobile community screening workflows remains limited. In remote or resource-limited settings, intraoral images are often acquired by non-specialist personnel under variable field conditions, making on-site image quality assurance, risk stratification, and expert oversight essential for responsible AI-assisted screening

Objective:

This study aimed to evaluate the prospective field performance of a parameter-locked, mobile AI-assisted oral lesion triage system embedded within routine community oral cancer screening.

Methods:

We conducted a prospective community-based field evaluation in eastern Taiwan from June 17 to December 31, 2025. Trained non-specialist personnel acquired standard white-light intraoral images using handheld mobile devices during routine oral cancer screening activities. The AI system incorporated on-site image quality assessment and lesion-level risk stratification into green, yellow, and red triage categories. Images that remained technically inadequate after repeated acquisition attempts were classified as ungradable. Three board-certified oral and maxillofacial specialists established an operational clinical reference standard through structured expert adjudication. The primary outcome was lesion-level identification of high-risk lesions requiring specialist referral, defined as red versus non-red triage. Diagnostic performance was estimated with Wilson 95% confidence intervals.

Results:

Among 602 screened participants, 4283 interpretable intraoral images were included in expert adjudication and lesion-level analysis. The AI system flagged 68 red lesions; 12 were confirmed as high risk by expert adjudication, and 4 additional high-risk lesions were identified during expert review. For high-risk lesion identification, sensitivity was 75.0% (95% CI 50.5%-89.8%), specificity was 98.6% (95% CI 98.2%-98.9%), positive predictive value was 17.6% (95% CI 10.4%-28.4%), negative predictive value was 99.9% (95% CI 99.7%-100.0%), and accuracy was 98.5% (95% CI 98.1%-98.8%). Discriminative performance was consistent across triage categories, with area under the curve values of 0.957 for red, 0.931 for yellow, and 0.964 for green triage.

Conclusions:

A parameter-locked mobile AI-assisted oral lesion triage system was feasibly integrated into a real-world community oral cancer screening workflow operated by trained non-specialist personnel. The system maintained high specificity and very high negative predictive value for high-risk red triage under field conditions, supporting its potential role as human-supervised triage support for referral prioritization rather than autonomous diagnosis. Further multicenter studies with blinded assessment, participant-level outcomes, workflow-efficiency measures, and longitudinal follow-up are needed to evaluate clinical impact and scalability.