Development of a Predictive Model for Metabolic Syndrome Using Noninvasive Data and its Cardiovascular Disease Risk...

Area of Science:

• Cardiology
• Metabolic Health
• Machine Learning in Healthcare

Background:

• Metabolic syndrome, characterized by obesity, hypertension, dyslipidemia, and insulin resistance, significantly elevates cardiovascular disease (CVD) risk.
• Rising global prevalence necessitates accessible, scalable screening methods beyond traditional laboratory tests.
• Current screening limitations hinder early identification and intervention, particularly in primary care settings.

Purpose of the Study:

• To develop and validate a noninvasive predictive model for metabolic syndrome utilizing body composition data.
• To assess the model's capability in predicting long-term CVD risk for clinical and public health applications.
• To support early intervention and preventive strategies for cardiometabolic diseases.

Main Methods:

• Developed a machine learning model trained on dual-energy x-ray absorptiometry and validated with bioelectrical impedance analysis data from national cohorts.
• Employed five machine learning algorithms, selecting the best performer based on receiver operating characteristic curve analysis.
• Utilized Cox proportional hazards regression to evaluate the model's association with long-term CVD risk.

Main Results:

• The predictive model demonstrated strong performance, with Area Under the Curve values ranging from 0.8039 to 0.8447 across internal and external validation cohorts.
• Individuals identified by the model as having metabolic syndrome exhibited a 1.51-fold increased risk of developing CVD (HR 1.51, 95% CI 1.32-1.73; P<.001).
• The model's predictions were significantly associated with future cardiovascular risk, indicating its utility for early intervention.

Conclusions:

• A noninvasive predictive model for metabolic syndrome was successfully developed and validated, offering enhanced accessibility for early risk identification.
• The model facilitates large-scale screenings and primary care assessments without requiring laboratory tests.
• Predicting long-term CVD risk supports proactive strategies to mitigate the burden of cardiometabolic diseases, warranting further refinement and broader validation.