What is the clinical evidence from this study?

Study design: Review. Population: Dyslipidemia and Cardiovascular Disease (n=2577). Intervention: LASSO regression analysis vs. Traditional regression methods. Primary outcome: Variable selection for stroke prediction.

January 1, 2025Open Access

LASSO Regression Analysis: Applications in Dyslipidemia and Cardiovascular Disease Research

Key Result

LASSO regression effectively selected key predictors for stroke, including age, sex, HDL-C, systolic blood pressure, smoking, diabetes, and blood pressure medication use.

Structured PICO

Population

2,577 participants from the Framingham Heart Study NHLBI Teaching Dataset were analyzed to demonstrate the application of LASSO regression for stroke prediction.

Intervention

LASSO regression analysis for variable selection and risk prediction

Comparator

Traditional linear/logistic regression and other machine learning models (e.g., Random Forest, Elastic Net)

Outcome

Stroke occurrence (in the educational example)

LASSO regression is a valuable statistical tool in cardiovascular research for handling high-dimensional data and multicollinearity by performing simultaneous variable selection and regularization.

Limitations

Decreased performance when the number of variables exceeds the number of samples (p > n)
Difficulty in selecting related variables, as it may select only a single variable from a group of highly correlated predictors
Difficulty in choosing the optimal regularization parameter λ
Instability with high-dimensional data where it may excessively shrink variables

Abstract

Dyslipidemia and atherosclerosis are major contributors to cardiovascular disease (CVD), necessitating the development of effective risk assessment models. Traditional regression methods often encounter limitations in handling high-dimensional data and multicollinearity, highlighting the need for advanced statistical techniques. This study discusses the theoretical background of least absolute shrinkage and selection operator (LASSO) regression and presents an example of its use with data from the Framingham Heart Study to identify the most predictive clinical variables and construct a robust CVD risk prediction model. Data from patients with dyslipidemia were analyzed, including lipid profiles, inflammatory markers, and additional metabolic indicators. Model performance was evaluated using cross-validation and benchmarked against conventional regression approaches. LASSO regression effectively selected key predictors, such as low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, C-reactive protein, and body mass index. The proposed model exhibited superior predictive accuracy and generalizability compared to traditional methods. LASSO regression is a valuable tool in cardiovascular research, offering improved variable selection and enhanced prediction performance. Its application in dyslipidemia-related CVD risk assessment holds promise for optimizing clinical decision-making and advancing personalized treatment strategies.

Mark Helpful

Bookmark

Relay

View Full Paper