Accepted for/Published in: Journal of Medical Internet Research
Date Submitted: Sep 18, 2024
Date Accepted: Mar 19, 2025
Predictive modeling of acute respiratory distress syndrome using machine learning: systematic review and meta-analysis
ABSTRACT
Background:
Acute respiratory distress syndrome (ARDS) is a common clinical condition encountered in the intensive care unit (ICU), marked by a high incidence and substantial mortality rate. Early detection and prediction of ARDS could enhance patient outcomes. Machine learning (ML) models are increasingly employed to achieve this goal.
Objective:
This study seeks to conduct a systematic evaluation and meta-analysis of the effectiveness of current ARDS prediction models.
Methods:
We performed a search across six electronic databases for studies developing machine learning predictive models for ARDS, with a cutoff date of March 5, 2024. The risk of bias in these models was evaluated using the Prediction model Risk of Bias Assessment Tool (PROBAST). Meta-analyses and investigations into heterogeneity were carried out using Meta-DiSc software (version 1.4). Furthermore, sensitivity and subgroup analyses were employed to explore the sources of heterogeneity more comprehensively.
Results:
This study incorporated 14 research articles, which included a total of 45 prediction models. The area under the receiver operating characteristic curve (AUROC) for ML models predicting ARDS was 0.7758. The sensitivity of these models was 0.740 (95% CI: 0.733–0.747, P<0.001; I² = 83.2%), while the specificity was 0.662 (95% CI: 0.657–0.667, P<0.001; I² = 98.4%). The diagnostic odds ratio (DOR) was calculated at 7.964 (95% CI: 6.079–10.433, P<0.001; I² = 94.6%). The positive likelihood ratio (PLR) was 3.059 (95% CI: 2.631–3.557, P<0.001; I² = 97.2%), and the negative likelihood ratio (NLR) was 0.435 (95% CI: 0.385–0.490, P<0.001; I² = 89.9%).
Conclusions:
This study evaluates prediction models developed using various machine learning algorithms. The results demonstrate high performance in ARDS prediction; however, attention must be given to quality assessment during model development. Additionally, this study is based on a relatively small sample size, necessitating further research to thoroughly evaluate the performance of the models. Clinical Trial: CRD42024529403
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.