Artificial Intelligence for the Diagnosis of Pediatric Appendicitis: A Systematic Review

Abstract

BACKGROUND: While acute appendicitis is the most frequent surgical emergency in children, its diagnosis remains complex. Artificial intelligence (AI) and machine learning (ML) tools have been employed to improve the accuracy of various diagnoses, including appendicitis. The purpose of this study was to systematically review the current body of evidence regarding the efficacy of AL and ML approaches for the diagnosis of acute pediatric appendicitis. METHODS: This systematic review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to identify articles from Pubmed, Scopus, and iEEE Xplore. Eligible articles included full text, English-language articles assessing the use of AI technologies for the diagnosis of acute pediatric appendicitis. Study quality of reporting was appraised using The Transparent Reporting of a multivariable prediction model of Individual Prognosis Or Diagnosis (TRIPOD) statement. RESULTS: A total of fourteen studies were included in the final analysis of which ten were published after 2019. Two studies originated in the United States while half were carried out in Europe. Artificial Neural Network and Random Forest AI methods were the most commonly used modeling approaches. Commonly used predictors were pain and laboratory blood findings. The average area under the curve that was reported among the fourteen studies was greater than 80%. CONCLUSIONS: AI and ML technologies have the potential to improve the accuracy of acute appendicitis diagnosis in pediatric patients. Further investigation is needed to identify barriers to adoption of these technologies and to assess their efficacy in real world usage once integrated into clinical workflows.