Sparse Group Penalties for bi-level variable selection.

Many data sets exhibit a natural group structure due to contextual similarities or high correlations of variables, such as lipid markers that are interrelated based on biochemical principles. Knowledge of such groupings can be used through bi-level selection methods to identify relevant feature groups and highlight their predictive members. One of the best known approaches of this kind combines the classical Least Absolute Shrinkage and Selection Operator (LASSO) with the Group LASSO, resulting in the Sparse Group LASSO. We propose the Sparse Group Penalty (SGP) framework, which allows for a flexible combination of different SGL-style shrinkage conditions. Analogous to SGL, we investigated the combination of the Smoothly Clipped Absolute Deviation (SCAD), the Minimax Concave Penalty (MCP) and the Exponential Penalty (EP) with their group versions, resulting in the Sparse Group SCAD, the Sparse Group MCP, and the novel Sparse Group EP (SGE). Those shrinkage operators provide refined control of the effect of group formation on the selection process through a tuning parameter. In simulation studies, SGPs were compared with other bi-level selection methods (Group Bridge, composite MCP, and Group Exponential LASSO) for variable and group selection evaluated with the Matthews correlation coefficient. We demonstrated the advantages of the new SGE in identifying parsimonious models, but also identified scenarios that highlight the limitations of the approach. The performance of the techniques was further investigated in a real-world use case for the selection of regulated lipids in a randomized clinical trial.

Interpretability of bi-level variable selection methods.

Variable selection is usually performed to increase interpretability, as sparser models are easier to understand than full models. However, a focus on sparsity is not always suitable, for example, when features are related due to contextual similarities or high correlations. Here, it may be more appropriate to identify groups and their predictive members, a task that can be accomplished with bi-level selection procedures. To investigate whether such techniques lead to increased interpretability, group exponential LASSO (GEL), sparse group LASSO (SGL), composite minimax concave penalty (cMCP), and least absolute shrinkage, and selection operator (LASSO) as reference methods were used to select predictors in time-to-event, regression, and classification tasks in bootstrap samples from a cohort of 1001 patients. Different groupings based on prior knowledge, correlation structure, and random assignment were compared in terms of selection relevance, group consistency, and collinearity tolerance. The results show that bi-level selection methods are superior to LASSO in all criteria. The cMCP demonstrated superiority in selection relevance, while SGL was convincing in group consistency. An all-round capacity was achieved by GEL: the approach jointly selected correlated and content-related predictors while maintaining high selection relevance. This method seems recommendable when variables are grouped, and interpretation is of primary interest.

Heart rate variability: reference values and role for clinical profile and mortality in individuals with heart failure.

AIMS: To establish reference values and clinically relevant determinants for measures of heart rate variability (HRV) and to assess their relevance for clinical outcome prediction in individuals with heart failure. METHODS: Data from the MyoVasc study (NCT04064450; N = 3289), a prospective cohort on chronic heart failure with a highly standardized, 5 h examination, and Holter ECG recording were investigated. HRV markers were selected using a systematic literature screen and a data-driven approach. Reference values were determined from a healthy subsample. Clinical determinants of HRV were investigated via multivariable linear regression analyses, while their relationship with mortality was investigated by multivariable Cox regression analyses. RESULTS: Holter ECG recordings were available for analysis in 1001 study participants (mean age 64.5 ± 10.5 years; female sex 35.4%). While the most frequently reported HRV markers in literature were from time and frequency domains, the data-driven approach revealed predominantly non-linear HRV measures. Age, sex, dyslipidemia, family history of myocardial infarction or stroke, peripheral artery disease, and heart failure were strongly related to HRV in multivariable models. In a follow-up period of 6.5 years, acceleration capacity [HRperSD 1.53 (95% CI 1.21/1.93), p = 0.0004], deceleration capacity [HRperSD: 0.70 (95% CI 0.55/0.88), p = 0.002], and time lag [HRperSD 1.22 (95% CI 1.03/1.44), p = 0.018] were the strongest predictors of all-cause mortality in individuals with heart failure independently of cardiovascular risk factors, comorbidities, and medication. CONCLUSION: HRV markers are associated with the cardiovascular clinical profile and are strong and independent predictors of survival in heart failure. This underscores clinical relevance and interventional potential for individuals with heart failure. GOV IDENTIFIER: NCT04064450.

A systematic review and evaluation of statistical methods for group variable selection.

This review condenses the knowledge on variable selection methods implemented in R and appropriate for datasets with grouped features. The focus is on regularized regressions identified through a systematic review of the literature, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. A total of 14 methods are discussed, most of which use penalty terms to perform group variable selection. Depending on how the methods account for the group structure, they can be classified into knowledge and data-driven approaches. The first encompass group-level and bi-level selection methods, while two-step approaches and collinearity-tolerant methods constitute the second category. The identified methods are briefly explained and their performance compared in a simulation study. This comparison demonstrated that group-level selection methods, such as the group minimax concave penalty, are superior to other methods in selecting relevant variable groups but are inferior in identifying important individual variables in scenarios where not all variables in the groups are predictive. This can be better achieved by bi-level selection methods such as group bridge. Two-step and collinearity-tolerant approaches such as elastic net and ordered homogeneity pursuit least absolute shrinkage and selection operator are inferior to knowledge-driven methods but provide results without requiring prior knowledge. Possible applications in proteomics are considered, leading to suggestions on which method to use depending on existing prior knowledge and research question.

FEV1 Predicts Cardiac Status and Outcome in Chronic Heart Failure.

BACKGROUND: COPD is an established predictor of clinical outcome in patients with chronic heart failure (HF). However, little evidence is available about the predictive value of FEV1 in chronic HF. RESEARCH QUESTION: Is pulmonary function related to the progression of chronic HF? STUDY DESIGN AND METHODS: The MyoVasc study (ClinicalTrials.gov Identifier: NCT04064450) is a prospective cohort study of HF. Information on pulmonary and cardiac functional and structural status was obtained by body plethysmography and echocardiography. The primary study end point was worsening of HF. RESULTS: Overall 2,998 participants (age range, 35-84 years) with available FEV1 data were eligible for analysis. Linear multivariate regression analysis revealed an independent relationship of FEV1 (per -1 SD) with deteriorated systolic and diastolic left ventricle (LV) function as well as LV hypertrophy under adjustment of age, sex, height, cardiovascular risk factors (CVRFs), and clinical profile (LV ejection fraction: ß-estimate, -1.63% [95% CI, -2.00% to -1.26%]; E/E' ratio: ß-estimate, 0.82 [95% CI, 0.64-0.99]; and LV mass/height2.7: ß-estimate, 1.58 [95% CI, 1.07-2.10]; P < .001 for all). During a median time to follow-up of 2.6 years (interquartile range, 1.1-4.1 years), worsening of HF occurred in 235 individuals. In Cox regression model adjusted for age, sex, height, CVRF, and clinical profile, pulmonary function (FEV1 per -1 SD) was an independent predictor of worsening of HF (hazard ratio [HR], 1.44 [95% CI, 1.27-1.63]; P < .001). Additional adjustment for obstructive airway pattern and C-reactive protein mitigated, but did not substantially alter, the results underlining the robustness of the observed effect (HRFEV1, 1.39 [95% CI, 1.20-1.61]; P < .001). The predictive value of FEV1 was consistent across subgroups, including individuals without obstruction (HR, 1.55 [95% CI, 1.34-1.77]; P < .001) and nonsmokers (HR, 1.72 [95% CI, 1.39-1.96]; P < .001). INTERPRETATION: FEV1 represents a strong candidate to improve future risk stratification and prevention strategies in individuals with chronic, stable HF. TRIAL REGISTRY: ClinicalTrials.gov; No.: NCT04064450; URL: www.clinicaltrials.gov.

Chronic venous insufficiency, cardiovascular disease, and mortality: a population study.

AIMS: Evidence regarding the health burden of chronic venous insufficiency (CVI), its clinical determinants, and impact on outcome is scarce. METHODS AND RESULTS: Systematic phenotyping of CVI according to established CEAP (Clinical-Etiologic-Anatomic-Pathophysiologic) classification was performed in 12 423 participants (age range: 40-80 years) of the Gutenberg Health Study from April 2012 to April 2017. Prevalence was calculated age- and sex-specifically. Multivariable Poisson regression models were calculated to evaluate the relation of CVI with cardiovascular comorbidities. Survival analyses were carried out to assess the CVI-associated risk of death. Replication of findings was done in an independent cohort study (MyoVasc, NCT04064450). The prevalence of telangiectasia/reticular, varicose veins, and CVI was 36.5% [95% confidence interval (CI), 35.6-37.4%], 13.3% [12.6-13.9%], and 40.8% [39.9-41.7%], respectively. Age, female sex, arterial hypertension, obesity, smoking, and clinically overt cardiovascular disease were identified as clinical determinants of CVI. Higher CEAP classes were associated with a higher predicted 10-year risk for incident cardiovascular disease in individuals free of cardiovascular disease (n = 9923). During a mean follow-up of 6.4 ± 1.6 years, CVI was a strong predictor of all-cause death independent of the concomitant clinical profile and medication [hazard ratio (HR) 1.46 (95% CI 1.19-1.79), P = 0. 0003]. The association of CVI with an increased risk of all-cause death was externally validated in the MyoVasc cohort [HR 1.51 (95% CI 1.11-2.05), P = 0.009]. CONCLUSION: Chronic venous insufficiency is highly prevalent in the population and is associated with the presence of cardiovascular risk factors and disease. Individuals with CVI experience an elevated risk of death, which is independent of age and sex, and present cardiovascular risk factors and comorbidities.