Screening/Diagnosing Sarcopenia with Machine Learning-Powered Risk Assessment: The SARCO X Study

Area of Science:

• Gerontology
• Biomedical Engineering
• Artificial Intelligence in Healthcare

Background:

• Sarcopenia presents a significant health burden, necessitating efficient screening and diagnostic tools.
• Early identification of sarcopenia is crucial for managing associated morbidity and healthcare costs.

Purpose of the Study:

• To develop and validate a machine learning (ML)-based algorithm for sarcopenia screening and diagnosis.
• To enhance the accuracy and efficiency of sarcopenia diagnosis compared to traditional methods.

Main Methods:

• A multicenter, cross-sectional case-control study involving participants aged 45 years and older.
• Collected demographic and clinical data, diagnosing sarcopenia using basic and ML-based algorithms incorporating muscle mass, chair stand test (CST), and hand grip strength (HGS).
• Employed Gradient Boosting Classifier (GBC) for the ML model, evaluating its performance on holdout test data.

Main Results:

• The ML-based model identified age, weight, height, education, exercise status, hypertension, and diabetes as significant factors associated with sarcopenia.
• The Gradient Boosting Classifier (GBC) achieved high performance, with recall of 0.979, precision of 0.926, and accuracy of 0.980.
• The ML-augmented algorithm reduced the need for HGS and ultrasound by 38.1% and 49.5%, respectively, optimizing diagnostic pathways.

Conclusions:

• The developed ML algorithm significantly decreases the requirement for extensive testing and imaging in sarcopenia diagnosis.
• This tool facilitates earlier sarcopenia identification in primary and secondary care, reducing unnecessary referrals for further evaluation.