Predictive modeling using a nationally representative database to identify patients at risk of developing...

Area of Science:

• Nephrology
• Cardiology
• Biostatistics

Background:

• Microalbuminuria (MA) is an early indicator of cardiovascular disease.
• Predictive models aid in stratifying patient risk for targeted interventions.
• Accurate identification of MA is crucial for cardiovascular risk management.

Purpose of the Study:

• To develop a patient data-driven predictive model for microalbuminuria (MA).
• To create a risk-score assessment tool to enhance MA identification.
• To improve early detection of cardiovascular disease risk through MA assessment.

Main Methods:

• Utilized the 2007-2008 National Health and Nutrition Examination Survey (NHANES) for model development.
• Employed multivariate logistic regression for model creation.
• Validated the model using internal (2007-2008 NHANES) and external (2012-2013 NHANES) datasets.
• Assessed model performance using ROC curves, pseudo-R(2) values, and Hosmer-Lemeshow goodness-of-fit tests.

Main Results:

• Developed a predictive model incorporating systolic blood pressure, fasting glucose, C-reactive protein, blood urea nitrogen, and alcohol consumption.
• The model demonstrated strong performance and validated well across datasets.
• A risk score chart was developed, correlating scores with MA probability.

Conclusions:

• The developed predictive model identifies novel variables associated with MA.
• Clinicians can utilize the model to identify at-risk patients and personalize treatment strategies.
• The risk score facilitates quantitative assessment of individual MA risk.