Accepted for/Published in: Journal of Medical Internet Research
Date Submitted: Feb 22, 2024
Date Accepted: Sep 20, 2024
Accuracy of Machine Learning in Discriminating Kawasaki Disease and Other Febrile Illnesses: A Systematic Review and Meta-Analysis
ABSTRACT
Background:
Currently, the differentiation of Kawasaki disease (KD) from other febrile illnesses remains a serious challenge in clinical practice, and there is still a lack of efficient predictive tools for the prediction of coronary artery lesions in patients with KD. With the development and application of artificial intelligence in the field of medicine, some researchers have tried to explore the performance of machine learning (ML) for diagnosing KD and predicting coronary artery lesions in patients with KD. However, their findings appear somewhat controversial.
Objective:
To summarize the accuracy of ML in differentiating KD from other febrile illnesses and predicting coronary artery lesions in patients with KD.
Methods:
PubMed, Cochrane Library, Embase, and Web of Science were systematically searched until September 26, 2023. The risk of bias in the included original studies was appraised utilizing the Prediction Model Risk of Bias Assessment Tool (PROBAST).
Results:
A total of 29 studies were incorporated. Twenty of them used ML to differentiate KD from other febrile illnesses, and involved a total of 103882 study subjects, including 12541 patients with KD. In the validation set, the pooled C-index, sensitivity, and specificity were 0.898 (95% CI:0.874-0.922), 0.91 (95% CI:0.83-0.95), and 0.86 (95% CI:0.80-0.90), respectively. Nine studies used ML for early prediction of the risk of coronary artery lesions in children with KD. They involved a total of 6503 Kawasaki patients, including 986 patients with coronary artery lesions. The pooled C-index in the validation set was 0.787 (95% CI: 0.738-0.835).
Conclusions:
ML is desirably effective in differentiating KD from other febrile illnesses, and it is also able to fairly predict the occurrence of coronary artery lesions. However, these conclusions were drawn based on limited evidence, and the predictive model included in this study was constructed mainly based on random sampling as a validation method. Therefore, more subsequent multicenter studies are desired to validate our conclusions.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.