Logistic LASSO Regression for Dietary Intakes and Breast Cancer

Abstract

A multitude of dietary factors from dietary fat to macro and micronutrients intakes have been associated with breast cancer, yet data are still equivocal. Therefore, utilizing data from the large, multi-year, cross-sectional National Health and Nutrition Examination Survey (NHANES), we applied a novel, modern statistical shrinkage technique, logistic least absolute shrinkage and selection operator (LASSO) regression, to examine the association between dietary intakes in women, ≥50 years, with self-reported breast cancer (n = 286) compared with women without self-reported breast cancer (1144) from the 1999–2010 NHANES cycle. Logistic LASSO regression was used to examine the relationship between twenty-nine variables, including dietary variables from food, as well as well-established/known breast cancer risk factors, and to subsequently identify the most relevant variables associated with self-reported breast cancer. We observed that as the penalty factor (λ) increased in the logistic LASSO regression, well-established breast cancer risk factors, including age (β = 0.83) and parity (β = −0.05) remained in the model. For dietary macro and micronutrient intakes, only vitamin B12 (β = 0.07) was positively associated with self-reported breast cancer. Caffeine (β = −0.01) and alcohol (β = 0.03) use also continued to remain in the model. These data suggest that a diet high in vitamin B12, as well as alcohol use may be associated with self-reported breast cancer. Nonetheless, additional prospective studies should apply more recent statistical techniques to dietary data and cancer outcomes to replicate and confirm the present findings.

A Panel of Five-lncRNA Signature as a Potential Biomarker for Predicting Survival in Gastric and Thoracic Cancers

“…Compared with other feature filtering algorithms such as linear regression and ridge regression, the lasso algorithm solves the overfitting problem and can directly reduce some repetitive and unnecessary parameters to zero in the parameter reduction process. Lasso regression performs well in controlling the number of features ( Xu et al, 2018 ; McEligot et al, 2020 ). Further, multivariate Cox regression analysis identified five lncRNAs that could independently predict the OS of patients.…”

“…Univariate Cox regression analysis was performed on the training set to evaluate the correlation between the expression level of each lncRNA and patient overall survival (OS) ( Guo et al, 2019 ). Lasso regression ( Xu et al, 2018 ; McEligot et al, 2020 ), which solves the over-fitting problem and can directly reduce some repetitive unnecessary parameters to zero in the parameter reduction process, was used to further screen the results from univariate Cox regression. The results of lasso algorithm screening were used to construct a multivariate Cox risk regression model, where OS was the dependent variable and the other clinical information was the covariate.…”

Laboratory indicators in COVID-19 and other pneumonias: Analysis for differential diagnosis and comparison of dynamic changes during 400-day follow-up

“…Therefore, whether the response variable is continuous, binary or multivariate discrete can be modeled and predicted by LASSO regression. In clinical applications, if the independent variables have multicollinearity or the number of variables is much larger than the sample size, LASSO regression should be done [26] .…”

Identification of 15 lncRNAs Signature for Predicting Survival Benefit of Advanced Melanoma Patients Treated with Anti-PD-1 Monotherapy

“…Although which module was most relevant to the predictor of melanoma can be identified after WGCNA, we further applied a novel, modern statistical shrinkage technique to examine the association between lncRNAs and the prognosis of melanoma to establish prognostic lncRNAs signature. The logistic LASSO regression model is a shrinkage method that can actively select from a large and potentially multicollinear set of variables in the regression, resulting in a more relevant and interpretable set of predictors [ 29 ]. One interesting property of LASSO is that the estimates of the regression coefficients are sparse, which means that many components are exactly 0 [ 30 ].…”