ROLE OF AI IN PREDICTING CARDIOVASCULAR RISK USING ROUTINE DENTAL IMAGING: A SYSTEMATIC REVIEW

Abstract

Background: Cardiovascular disease (CVD) remains the leading global cause of mortality, often progressing asymptomatically until advanced stages. Routine dental imaging, particularly panoramic radiographs, may incidentally capture vascular calcifications indicative of subclinical atherosclerosis. While artificial intelligence (AI) has shown potential in automating such detections, current evidence is fragmented, and no prior systematic review has synthesized its diagnostic value in this context.

Objective: This systematic review aimed to evaluate the accuracy and clinical utility of AI algorithms in identifying early cardiovascular risk indicators from routine dental radiographs in adult populations.

Methods: Following PRISMA guidelines, a systematic review was conducted using PubMed, Scopus, Web of Science, and the Cochrane Library to identify relevant studies published between 2019 and 2024. Eligible studies included cross-sectional and retrospective designs using AI models to detect cardiovascular risk markers via dental imaging. Data were extracted on study characteristics, AI model performance, and risk of bias, which was assessed using the Newcastle-Ottawa Scale. A narrative synthesis was conducted due to heterogeneity in model types and outcome measures.

Results: Eight studies comprising 788 to 3,200 participants were included. All employed deep learning-based AI algorithms, primarily convolutional neural networks, to detect markers such as carotid artery calcifications. Reported accuracies ranged from 82.5% to 95.3%, with AUC values up to 0.91. Most studies demonstrated moderate-to-high methodological quality. However, variability in model training and limited external validation restricted meta-analytic pooling.

Conclusion: AI algorithms show strong potential in identifying early cardiovascular risk using dental radiographs, offering a novel, non-invasive screening opportunity within dental care. Despite promising results, further prospective studies with standardized methodologies and external validation are essential to support clinical integration.