Machine learning algorithms for the rupture risk assessment in intracranial aneurysms: a diagnostic meta-analysis

Several machine learning algorithms have been increasingly applied to predict the rupture risk of intracranial aneurysms (IAs). We performed this diagnostic meta-analysis to comprehensively evaluate the diagnostic value of machine learning algorithms for assessing the rupture risk of IAs. We systematically searched three electronic databases including Medline (via PubMed), the Cochrane Register of Controlled Trials (CENTRAL) (via OVID) and EMBASE (via Elseiver) to retrieve eligible studies from their inception through March 2021. The latest update was made in June 2021. StataMP version 14 was used to estimate all pooled diagnostic values. A total of 4 studies involving 6 reports were considered to meet the inclusion criteria finally. Our diagnostic meta-analysis generated the following pooled diagnostic values: sensitivity of 0.84 (95% confidence interval [CI], 0.75 to 0.90), specificity of 0.78 (95% CI, 0.68 to 0.85), positive likelihood ratio (PLR) of 3.8 (95% CI, 2.4 to 5.9), negative likelihood ratio (NLR) of 0.21 (95% CI, 0.12 to 0.35), diagnostic odd ratio (DOR) of 18 (95% CI, 7 to 46), and the area under the summary receiver operating characteristic curve (AUC) of 0.88 (95% CI, 0.85 to 0.90). The result demonstrates that the diagnostic performance of machine learning algorithms for the rupture risk assessment of AIs was excellent. Considering the negative impact resulted from the limited number of eligible studies, we suggest developing more well-designed studies with large sample sizes to establish our finding.