Diagnostic performance of machine learning models for malignant and non-malignant pleural effusion: Systematic review and meta-analysis.
Luhan Guo, Jiansheng Li, Jiaheng Yao et al.
✦ AI-curated · Sources linked
In 30 seconds
This systematic review and meta-analysis evaluated the diagnostic performance of machine learning models in distinguishing malignant pleural effusion (MPE) from non-malignant pleural effusion. The analysis included 42 studies, revealing a pooled AUC of 0.90, with a sensitivity of 0.80 and specificity of 0.87. The study highlights concerns regarding methodological quality and external validation.
Key findings
- The pooled AUC for machine learning models was 0.90 (95% CI 0.85-0.94).
- Sensitivity was 0.80 (95% CI 0.77-0.83) and specificity was 0.87 (95% CI 0.79-0.92).
- Externally validated studies showed a lower pooled AUC of 0.83 compared to 0.92 for non-validated studies.
- All 42 model evaluations were judged at high risk of bias.
Why it matters
Accurate differentiation between malignant and non-malignant pleural effusions is crucial for patient management. Understanding the limitations of machine learning models can guide clinicians in integrating these tools into diagnostic processes.
What to keep in mind
All 42 development assessments had high overall quality concerns, and all model evaluations were judged at high risk of bias.