Lasso Analysis of Smoking Cessation Predictors

This project implements advanced regularization techniques to analyze smoking cessation outcomes among individuals with major depressive disorder. Using data from a 2x2 factorial clinical trial (n=300), I employed Lasso regression with bootstrap confidence intervals to identify key predictors of successful smoking cessation and potential treatment effect moderators. The analysis revealed that nicotine dependence scores and nicotine metabolism rates significantly predict cessation success, while validating varenicline’s effectiveness as a cessation aid. Implemented in R using glmnet, mice, and custom functions for model diagnostics, this project showcases expertise in handling missing data and statistical modeling with clinical trial data.

Keywords: Academic project, Regression analysis, Clinical trial

View project on GitHub​​​​​​​​​​​​​​​​