Variability in Clinical Parameters as Predictors of Cardiovascular Disease in Type 2 Diabetes: A Machine Learning Approach

Abstract

Background

Cardiovascular disease (CVD) remains a leading cause of morbidity and mortality in patients with type 2 diabetes mellitus (T2DM). Traditional risk prediction models often underperform in diabetic populations. This study aimed to develop and validate a machine learning (ML) model to predict CVD risk in T2DM patients using routinely collected clinical data with focus on parameter variability rather than absolute values.

Methods

We retrospectively analyzed data from the National Health and Nutrition Examination Survey (NHANES), a nationally representative public dataset, selecting 5,426 T2DM patients without prior CVD (2015-2020). Multiple ML algorithms were trained to predict 3-year CVD risk. Input variables included demographic data, comorbidities, medications, and crucially, both median values and ranges (maximum minus minimum) of clinical parameters including HbA1c, creatinine, liver enzymes, and lipid profiles. Models were validated using 5-fold cross-validation.

Results

The random forest model demonstrated superior performance with an area under the receiver operating characteristic curve of 0.81 (95% CI: 0.79–0.83). Parameter variability provided stronger predictive value than median values for key variables. The top five predictors were creatinine variability, HbA1c variability, AST variability, ALP variability, and ALT variability, highlighting the importance of metabolic stability in CVD risk reduction.

Conclusion

This study demonstrates that fluctuations in routine clinical parameters, particularly renal and glycemic markers, outperform traditional static measurements in predicting CVD risk among T2DM patients. Implementation of this ML model could enhance clinical decision-making by identifying high-risk patients who might benefit from more intensive monitoring and earlier therapeutic interventions.