Currently submitted to: Journal of Medical Internet Research
Date Submitted: Jun 11, 2026
Open Peer Review Period: Jun 12, 2026 - Aug 7, 2026
(currently open for review)
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Artificial Intelligence for Early Prediction and Mortality Risk Stratification in Acute Pancreatitis-Associated Acute Kidney Injury: A Systematic Review and Network Meta-Analysis
ABSTRACT
Background:
Acute pancreatitis-associated acute kidney injury (AP-AKI) is linked to substantial morbidity and mortality, but the comparative performance of artificial intelligence (AI) models for early AP-AKI prediction and mortality risk stratification remains uncertain.
Objective:
This systematic review and network meta-analysis evaluated the diagnostic performance and comparative ranking of AI models for early prediction of AP-AKI and AP-AKI-related mortality.
Objective:
This systematic review and network meta-analysis evaluated the diagnostic performance and comparative ranking of AI models for early prediction of AP-AKI and AP-AKI-related mortality.
Methods:
PubMed, Embase, Web of Science, Cochrane Library, and IEEE Xplore were searched from inception to February 23, 2026. Eligible studies developed or validated AI, machine learning, or deep learning models for AP-AKI or mortality among patients with AP-AKI and reported reconstructable diagnostic accuracy data. Two reviewers independently extracted study and model characteristics, validation methods, and 2 x 2 diagnostic data. Risk of bias was assessed with PROBAST+AI, certainty of evidence with GRADE, pooled accuracy with bivariate random-effects models, and algorithm rankings with Bayesian diagnostic network meta-analysis.
Results:
Fourteen studies were included. For AP-AKI prediction, 11 studies with 35 validation datasets yielded pooled sensitivity of 0.76, specificity of 0.85, and area under the receiver operating characteristic curve of 0.87; XGBoost ranked highest for sensitivity and diagnostic odds ratio. For mortality prediction, three studies with 66 validation datasets yielded pooled sensitivity of 0.73, specificity of 0.77, and area under the receiver operating characteristic curve of 0.81; support vector machine ranked highest for sensitivity and diagnostic odds ratio. Certainty of evidence was mostly low or very low, mainly because of heterogeneity, retrospective designs, limited external validation, and incomplete calibration or clinical utility reporting.
Conclusions:
AI models show promising but heterogeneous performance for AP-AKI prediction and mortality risk stratification. Prospective multicenter validation, calibration assessment, workflow evaluation, and clinical utility studies are needed before routine implementation. Clinical Trial: PROSPERO CRD420261360696.
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.