Machine Learning for Risk Factor Identification and Cardiovascular Mortality Prediction Among Patients with...

Area of Science:

• Cardiology
• Gerontology
• Biomedical Engineering

Background:

• Clinical decision-making increasingly relies on risk prediction tools.
• Existing models often lack specificity for targeted patient groups, such as those with osteoporosis.
• Osteoporosis patients face an elevated risk of cardiovascular mortality.

Purpose of the Study:

• To develop and validate a cardiovascular mortality risk prediction model specifically for individuals with osteoporosis.
• To compare the performance of machine learning (ML) models against established expert-based models.
• To identify key data-driven risk factors for cardiovascular death in this population.

Main Methods:

• Development and internal validation of ML-based cardiovascular mortality risk prediction models.
• Comparison of ML models with existing expert-based tools.
• Evaluation of models based on risk factor identification, discrimination, and calibration.

Main Results:

• Machine learning models demonstrated superior performance compared to existing cardiovascular mortality risk prediction tools for osteoporosis patients.
• The developed models identified important, data-driven risk factors for cardiovascular death.
• ML models showed better discrimination and calibration for the target population.

Conclusions:

• Tailored machine learning models offer improved cardiovascular mortality risk prediction for individuals with osteoporosis.
• These models provide valuable insights into specific risk factors within this patient group.
• External validation of the proposed models is recommended for broader clinical application.