Meta-Analysis of Machine Learning Models for Aortic Dissection Risk Prediction

40 studies from a systematic review synthesize performance metrics for machine learning models predicting adverse outcomes in aortic dissection. The analysis covers outcomes like early mortality, long-term mortality, and acute kidney injury, reporting pooled C-statistics. It was conducted by Yijun Mao following PRISMA and CHARMS guidelines, with data extracted up to September 30, 2025.

Use Cases

• Analyze the pooled C-statistic performance across different adverse outcomes like early mortality and acute kidney injury.
• Compare the discriminative performance of specific ML algorithms such as random forest, SVM, and XGBoost as reported in the included studies.
• Assess methodological quality using the PROBAST risk of bias results and TRIPOD adherence rates reported for the 40 studies.
• Investigate heterogeneity metrics (I² values) and publication bias analyses from the meta-analysis results.

Strengths

• Synthesizes data from 40 included studies on a clinically critical topic.
• Reports specific pooled performance metrics, such as a C-statistic of 0.891 for early mortality prediction.
• Follows established guidelines (PRISMA, CHARMS, TRIPOD) and was registered with PROSPERO.
• Provides detailed subgroup and sensitivity analyses based on the extracted study data.

Limitations

• The underlying studies exhibit significant heterogeneity (I² = 61.3–78.8%) and high risk of bias per PROBAST.
• Limited external validation, with only 5 of the 40 included studies performing robust external validation.
• Incomplete reporting in primary studies, with 82.5% lacking sample size justification and 72.5% omitting full model specifications.

Provenance

Source

Systematic review and meta-analysis by author Yijun Mao.

Collection Method

Data extracted from 40 studies identified via searches in six databases (PubMed, Web of Science, Cochrane Library, Embase, CNKI, Wanfang).

Time Range

Literature search from database inception to September 30, 2025.

Freshness

Last updated March 2026, with literature search up to September 30, 2025.

Geography

null