Caries research · 2026
Artificial Intelligence for Detecting Secondary Caries: A Systematic Review and Meta-analysis of Diagnostic Test Accuracy.
СтатьяОбзорНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Sachdeva Akshat, Kumar Yata Prashanth, Chawla Amrita, Kumar Vijay, Sharma Sidhartha, Upadhyay Ashish Datt, Ballal Vasudev, Logani Ajay
Аннотация
INTRODUCTION: Secondary caries (SC) is one of the untoward consequences of dental restorations, which if detected timely can decrease the expense of frequent reinterventions. The aim of this systematic review is to assess the diagnostic accuracy of artificial intelligence (AI) in the detection of secondary caries (SC).
METHODS: The protocol was registered in PROSPERO (CRD420251014888). PubMed, EBSCOhost, Scopus, Embase, IEEE Xplore and Web of Science databases were searched up to 11th June, 2026. Data extraction focused on different AI models used, dataset characteristics, and diagnostic performance metrics. Risk of bias assessment was done using the quality assessment of diagnostic accuracy studies (QUADAS-2) tool. A bivariate random-effects model was used to pool sensitivity and specificity. Heterogeneity, threshold effect, and publication bias were evaluated.
RESULTS: After screening 116 articles, nine studies fulfilled the inclusion criteria and were included in the review. The pooled sensitivity and specificity of AI for SC detection were 76.83% (95% CI: 51.70-91.13) and 84.43% (95% CI: 29.06-98.63), respectively. The HSROC analysis yielded an AUC of 0.85 (95% CI: 0.81-0.88) and heterogeneity of 59.47%, indicating moderate diagnostic ability. Three studies demonstrated a high risk of bias in the index test domain and one in the patient selection domain.
CONCLUSION: Pooled results suggest that AI demonstrated reasonable diagnostic accuracy for detecting SC. However, given the limited and heterogeneous evidence base, its clinical use remains promising yet insufficiently validated.
METHODS: The protocol was registered in PROSPERO (CRD420251014888). PubMed, EBSCOhost, Scopus, Embase, IEEE Xplore and Web of Science databases were searched up to 11th June, 2026. Data extraction focused on different AI models used, dataset characteristics, and diagnostic performance metrics. Risk of bias assessment was done using the quality assessment of diagnostic accuracy studies (QUADAS-2) tool. A bivariate random-effects model was used to pool sensitivity and specificity. Heterogeneity, threshold effect, and publication bias were evaluated.
RESULTS: After screening 116 articles, nine studies fulfilled the inclusion criteria and were included in the review. The pooled sensitivity and specificity of AI for SC detection were 76.83% (95% CI: 51.70-91.13) and 84.43% (95% CI: 29.06-98.63), respectively. The HSROC analysis yielded an AUC of 0.85 (95% CI: 0.81-0.88) and heterogeneity of 59.47%, indicating moderate diagnostic ability. Three studies demonstrated a high risk of bias in the index test domain and one in the patient selection domain.
CONCLUSION: Pooled results suggest that AI demonstrated reasonable diagnostic accuracy for detecting SC. However, given the limited and heterogeneous evidence base, its clinical use remains promising yet insufficiently validated.