Machine Learning Risk Prediction Models for Aortic Dissection

Forty studies were systematically reviewed and meta-analyzed to evaluate machine learning models for predicting adverse outcomes in aortic dissection. The analysis synthesizes performance metrics like C-statistics for outcomes including early mortality, long-term mortality, and acute kidney injury. The review was conducted by Yijun Mao and registered with PROSPERO, covering literature from six databases up to September 30, 2025.

Use Cases

• Compare pooled C-statistic performance for early mortality (0.891) versus long-term mortality (0.847) across the 40 included studies
• Analyze the distribution of ML algorithms like random forest, SVM, and XGBoost used for predicting complications such as AKI and neurological complications
• Assess methodological quality by examining PROBAST risk-of-bias results, where 100% of studies were at high or unclear risk
• Evaluate reporting completeness using the average TRIPOD adherence score of 78.7% and specific shortcomings in predictor definitions or sample size justification

Strengths

• Includes 40 studies following PRISMA, CHARMS, and TRIPOD guidelines for systematic review
• Reports specific pooled performance metrics, such as a C-statistic of 0.891 for early mortality prediction
• Covers multiple adverse outcomes including acute kidney injury, neurological complications, and mortality

Limitations

• Significant heterogeneity in meta-analyses with I² values ranging from 61.3% to 78.8%
• All 40 studies were assessed as having high or unclear risk of bias, primarily due to analysis domain deficiencies
• Suboptimal reporting with critical omissions in predictor definitions, sample size justification, and full model specifications

Provenance

Source

Systematic review and meta-analysis registered with PROSPERO (CRD420251154262)

Collection Method

Literature search across six databases (PubMed, Web of Science, Cochrane Library, Embase, CNKI, Wanfang); data extraction followed CHARMS guidelines

Time Range

From database inception to September 30, 2025

Freshness

Literature search covers studies from database inception to September 30, 2025; metadata last updated March 26, 2026.

Geography

null