Diagnostic Prediction Models for Sarcopenia: A Systematic Review and Meta-Analysis

Area of Science:

• Gerontology and Geriatric Medicine
• Biostatistics
• Epidemiology

Background:

• Early detection of sarcopenia is crucial but challenging.
• Existing predictive models lack comprehensive evaluation of their diagnostic performance and quality.
• This study addresses the need for a systematic assessment of sarcopenia prediction models.

Purpose of the Study:

• To systematically evaluate the diagnostic accuracy of prediction models for sarcopenia.
• To compare the performance of different modeling approaches (traditional statistics vs. machine learning).
• To assess model performance across diverse populations and reference standards.

Main Methods:

• Systematic review and meta-analysis of diagnostic test accuracy studies.
• Searched Ovid MEDLINE, Embase, and Cochrane Central databases until June 2024.
• Employed bivariate random-effects meta-analysis and hierarchical summary receiver operating characteristic models.

Main Results:

• Included 13 studies with 122,252 participants.
• Models showed robust performance in development (AUC 0.89) and internal validation (AUC 0.86) sets.
• Traditional models maintained consistent performance, while machine learning models achieved higher specificity in validation sets.

Conclusions:

• Current prediction models demonstrate promising diagnostic accuracy for sarcopenia.
• Different modeling approaches offer complementary strengths.
• Further research on methodologic heterogeneity and external validation is required before clinical implementation.