Recent Methodological Trends in Epidemiology: No Need for Data-Driven Variable Selection?

Variable selection in regression models is a particularly important issue in epidemiology, where one usually encounters observational studies. In contrast to randomized trials or experiments, confounding is often not controlled by the study design, but has to be accounted for by suitable statistical methods. For instance, when risk factors should be identified with unconfounded effect estimates, multivariable regression techniques can help to adjust for confounders. We investigated the current practice of variable selection in 4 major epidemiologic journals in 2019 and found that the majority of articles used subject-matter knowledge to determine a priori the set of included variables. In comparison with previous reviews from 2008 and 2015, fewer articles applied data-driven variable selection. Furthermore, for most articles the main aim of analysis was hypothesis-driven effect estimation in rather low-dimensional data situations (i.e., large sample size compared with the number of variables). Based on our results, we discuss the role of data-driven variable selection in epidemiology.

Exploring the Relationship between Adherence to Therapy, Treatment Acceptability, and Clinical Outcomes in Adults with Attention-Deficit/Hyperactivity Disorder: Results from the COMPAS Multicenter Randomized Controlled Trial.

INTRODUCTION: Cognitive behavioral therapy and dialectical behavior therapy (DBT) can be effective in treating adults with ADHD, and patients generally consider these interventions useful. While adherence, as measured by attendance at sessions, is mostly sufficient, adherence to therapy skills has not been assessed. Furthermore, the relationship between patient evaluation of therapy effectiveness, treatment adherence, and clinical outcomes is understudied. OBJECTIVE: This study aimed to examine treatment acceptability and adherence in relation to treatment outcomes in a large randomized controlled trial comparing a DBT-based intervention with a nonspecific active comparison, combined with methylphenidate or placebo. METHOD: A total of 433 adult patients with ADHD were randomized. Participants reported how effective they found the therapy, and adherence was measured by attendance at therapy sessions and by self-reports. Descriptive, between-groups, and linear mixed model analyses were conducted. RESULTS: Participants rated psychotherapy as moderately effective, attended 78.40-94.37% of sessions, and used skills regularly. The best-accepted skills were sports and mindfulness. Groups receiving placebo and/or nonspecific clinical management rated their health condition and the medication effectiveness significantly worse than the psychotherapy and methylphenidate groups. Improvements in clinical outcomes were significantly associated with treatment acceptability. Subjective (self-reported) adherence to psychotherapy was significantly associated with improvements in ADHD symptoms, clinical global efficacy and response to treatment. DISCUSSION: These results further support the acceptability of DBT for adult ADHD and suggest the need to address adherence to treatment to maximize clinical improvements. Results may be limited by the retrospective assessment of treatment acceptability and adherence using an ad hoc instrument.

Statistical learning for sparser fine-mapped polygenic models: The prediction of LDL-cholesterol.

Polygenic risk scores quantify the individual genetic predisposition regarding a particular trait. We propose and illustrate the application of existing statistical learning methods to derive sparser models for genome-wide data with a polygenic signal. Our approach is based on three consecutive steps. First, potentially informative loci are identified by a marginal screening approach. Then, fine-mapping is independently applied for blocks of variants in linkage disequilibrium, where informative variants are retrieved by using variable selection methods including boosting with probing and stochastic searches with the Adaptive Subspace method. Finally, joint prediction models with the selected variants are derived using statistical boosting. In contrast to alternative approaches relying on univariate summary statistics from genome-wide association studies, our three-step approach enables to select and fit multivariable regression models on large-scale genotype data. Based on UK Biobank data, we develop prediction models for LDL-cholesterol as a continuous trait. Additionally, we consider a recent scalable algorithm for the Lasso. Results show that statistical learning approaches based on fine-mapping of genetic signals result in a competitive prediction performance compared to classical polygenic risk approaches, while yielding sparser risk models.

Boosting multivariate structured additive distributional regression models.

We develop a model-based boosting approach for multivariate distributional regression within the framework of generalized additive models for location, scale, and shape. Our approach enables the simultaneous modeling of all distribution parameters of an arbitrary parametric distribution of a multivariate response conditional on explanatory variables, while being applicable to potentially high-dimensional data. Moreover, the boosting algorithm incorporates data-driven variable selection, taking various different types of effects into account. As a special merit of our approach, it allows for modeling the association between multiple continuous or discrete outcomes through the relevant covariates. After a detailed simulation study investigating estimation and prediction performance, we demonstrate the full flexibility of our approach in three diverse biomedical applications. The first is based on high-dimensional genomic cohort data from the UK Biobank, considering a bivariate binary response (chronic ischemic heart disease and high cholesterol). Here, we are able to identify genetic variants that are informative for the association between cholesterol and heart disease. The second application considers the demand for health care in Australia with the number of consultations and the number of prescribed medications as a bivariate count response. The third application analyses two dimensions of childhood undernutrition in Nigeria as a bivariate response and we find that the correlation between the two undernutrition scores is considerably different depending on the child's age and the region the child lives in.

Lack of evidence for effects of lockdowns on stillbirth rates during the SARS-CoV-2 pandemic in Bavaria: analysis of the Bavarian perinatal survey from 2010 to 2020.

BACKGROUND: Internationally, potential effects of national SARS-CoV-2-related lockdowns on stillbirth rates have been reported, but data for Germany, including risk factors for fetal pregnancy outcome, are lacking. The aim of this study is to compare the stillbirth rates during the two first lockdown periods in 2020 with previous years from 2010 to 2019 in a large Bavarian cohort. METHODS: This study is a secondary analysis of the Bavarian perinatal data from 2010 to 2020, including 349,245 births. Univariate and multivariable regression analyses were performed to investigate the effect of two Bavarian lockdowns on the stillbirth rate in 2020 compared to the corresponding periods from 2010 to 2019. RESULTS: During the first lockdown, the stillbirth rate was significantly higher compared to the reference period (4.04 vs. 3.03 stillbirths per 1000 births; P = 0.03). After adjustment for seasonal and long-term trends, this effect can no longer be observed (P = 0.2). During the second lockdown, the stillbirth rate did not differ in univariate (3.46 vs. 2.93 stillbirths per 1000 births; P = 0.22) as well as in multivariable analyses (P = 0.68), compared to the years 2010 to 2019. CONCLUSION: After adjustment for known long-term effects, in this study we did not find evidence that the two Bavarian lockdowns had an effect on the rate of stillbirths.

Effects of treatment with SGLT-2 inhibitors on arginine-related cardiovascular and renal biomarkers.

BACKGROUND: In patients with type 2 diabetes (T2D) sodium-glucose cotransporter 2 (SGLT-2) inhibitors improve glycaemic control as well as cardiovascular and renal outcomes. Their effects on L-arginine (Arg) related risk markers asymmetric and symmetric dimethylarginine (ADMA and SDMA) and the protective biomarker L-homoarginine (hArg) linking T2D to cardiovascular and renal disease have not yet been reported. METHODS: Plasma and 24-h urine samples taken before and after 6 weeks of treatment were available from two prospective, randomized, double-blind, placebo-controlled, cross-over trials with empagliflozin (71 patients analyzed, NCT02471963) and dapagliflozin (59 patients analyzed, NCT02383238). In these samples, concentrations of hArg, Arg, ADMA, SDMA, and creatinine were determined by liquid-chromatography coupled to tandem mass-spectrometry. Additionally, intraindividual changes of the biomarkers in plasma were correlated with intraindividual changes of clinical parameters. RESULTS: Treatment with empagliflozin and dapagliflozin was associated with a reduction of plasma hArg by 17.5% and 13.7% (both p < 0.001), respectively, and increase in plasma SDMA concentration of 6.7% and 3.6%, respectively (p < 0.001 and p < 0.05), while plasma Arg and ADMA concentrations were not significantly altered. 24-h urinary excretion of ADMA was reduced by 15.2% after treatment with empagliflozin (p < 0.001) but not after dapagliflozin treatment, while excretion of the other markers was not significantly altered. Renal clearance of SDMA was reduced by 9.1% and 3.9% for both drugs (both p < 0.05). A reduction in ADMA clearance was observable after empagliflozin treatment only (- 15.5%, p < 0.001), but not after dapagliflozin. Renal clearance of hArg and Arg was not significantly altered. Treatment effects on L-arginine related biomarkers were not constantly correlated with effects on glycated hemoglobin, fasting plasma glucose, body mass index, and systolic blood pressure. CONCLUSIONS: Treatment with SGLT-2 inhibitors has divergent effects on Arg-related biomarkers and could affect risk estimates associated with these markers. The observed effects are unlikely to explain the known cardiovascular and renal benefits of treatment with empagliflozin or dapagliflozin but still may indicate new therapeutic approaches in patients treated with SGLT-2 inhibitors. Trial registration http://www.clinicaltrials.gov : NCT02471963 (registered 15th June 2015, retrospectively registered) and NCT02383238.

N6-methyladenosine RNA modification (m6A) is of prognostic value in HPV-dependent vulvar squamous cell carcinoma.

BACKGROUND: Vulvar squamous cell carcinoma (VSCC) is an uncommon gynecologic malignancy but with an increasing incidence in recent years. Etiologically, VSCC is classified into two subtypes: HPV-dependent and HPV-independent. Localized VSCC is treated surgically and/or with radiation therapy, but for advanced, metastatic or recurrent disease, therapeutic options are still limited. N6-methyladenosine (m6A) is the most prevalent post-transcriptional messenger RNA (mRNA) modification and involved in many physiological processes. The group of m6A proteins can be further divided into: 'writers' (METTL3, METTL4, METTL14, WTAP, KIAA1429), 'erasers' (FTO, ALKBH5), and 'readers' (HNRNPA2B1, HNRNPC, YTHDC1, YTHDF1-3). Dysregulated m6A modification is implicated in carcinogenesis, progression, metastatic spread, and drug resistance across various cancer entities. Up to date, however, only little is known regarding the role of m6A in VSCC. METHODS: Here, we comprehensively investigated protein expression levels of a diverse set of m6A writers, readers and erasers by applying immunohistochemical staining in 126 patients with primary VSCC. RESULTS: In the entire study cohort, dominated by HPV-independent tumors, m6A protein expression was not associated with clinical outcome. However, we identified enhanced protein expression levels of the 'writers' METTL3, METTL14 and the 'reader' YTHDC1 as poor prognostic markers in the 23 patients with HPV-dependent VSCC. CONCLUSION: Our study suggests dysregulated m6A modification in HPV-associated VSCC.

Clinical value of pre-operative scoring systems to predict leiomyosarcoma: results of a validation study in 177 patients from the NOGGO-REGSA Registry.

OBJECTIVES: Benign leiomyomas are the most common uterine tumors. In contrast, uterine leiomyosarcomas are malignancies with a poor prognosis due to difficulties in early diagnosis and inappropriate surgical treatment. Most often they are diagnosed incidentally after surgery performed for treatment of leiomyoma. As the appropriate surgical treatment is crucial for survival of the patient, there is a high demand to predict leiomyosarcoma pre-operatively. Available scoring systems to discriminate leiomyoma from leiomyosarcoma are based on retrospective studies with limited numbers of patients and are not implemented in routine clinical practice. METHODS: The aim of our study was to evaluate a recently published score-the pre-operative leiomyosarcoma (pLMS) score-to determine whether it would have been predictive of leiomyosarcoma in 177 patients from the NOGGO-REGSA study, a German register of histologically proven gynecological sarcoma detected during routine clinical investigation. RESULTS: The threshold of the pLMS score for 'leiomyosarcoma not probable' (< -3) failed for 7.5% of the patients and the threshold 'indicator for leiomyosarcoma' (>+1) was true for 39.1% of the patients. 53.4% of the patients were attributed to the group 'additional investigations are recommended' (-3 to +1). The most relevant parameters in our analysis were suspicious sonography and rapid growth, but neither have been quantitatively defined. CONCLUSION: In our validation cohort, the pLMS score seems not to be a reliable tool to predict leiomyosarcoma and therefore we do not recommend its clinical implementation to identify leiomyosarcoma.

Influence of Sonographic Fetal Weight Estimation Inaccuracies in Macrosomia on Perinatal Outcome.

OBJECTIVE: To evaluate the influence of inaccurate sonographic fetal weight estimation in macrosomia on the mode of delivery and neonatal outcome (NO). METHODS: In 14 633 pregnancies between 2002 and 2016, this retrospective study evaluated the association between sonographic fetal weight estimation, true birth weight (BW), mode of delivery (primary cesarean section [pCS], secondary cesarean section, vaginal delivery, and operative vaginal delivery rates) and NO parameters (5-min Apgar < 7, pH < 7.1, neonatal intensive care unit [NICU] admission, shoulder dystocia). Singleton pregnancies > 37 + 0 weeks with ultrasound-estimated fetal weight (EFW) within 7 days before delivery were included. The study population was divided into four groups: Group 1 (false-negative): EFW < 4000 g/BW ≥ 4000 g; Group 2 (true-positive): EFW ≥ 4000 g/BW ≥ 4000 g; Group 3 (false-positive): EFW ≥ 4000 g/BW < 4000 g; and Group 4 (true-negative): EFW < 4000 g/BW < 4000 g. RESULTS: As expected, the highest secondary cesarean section (sCS) rate was found in Group 2 (true-positive) (30.62 %), compared with only 17.68 % in Group 4 (true-negative). The sCS rate in the false-positive Group 3 was significantly higher (28.48 %) in comparison with the false-negative Group 1 (21.22 %; OR 1.48; 95 % CI, 1.16 to 1.89; P = 0.002). In comparison with the true-negative Group 4, univariate analyses showed significantly higher rates for sCS in all other groups: odds ratio (OR) 2.06 for Group 2 (95 % CI, 1.74 to 2.42; P < 0.001), 1.85 for Group 3 (95 % CI, 1.54 to 2.22, P < 0.001), and 1.25 for Group 1 (95 % CI, 1.05 to 1.49; P < 0.01). No significant differences were found for NO between Groups 1 and 3 for the parameters 5-min Apgar < 7 (P = 0.75), pH < 7.1 (P = 0.28), or NICU admission (P = 0.54). However, there was a significantly higher chance for shoulder dystocia in Group 1 compared with Group 3 (OR 4.58; 95 % CI, 1.34 to 24.30; P = 0.008). CONCLUSION: Sonographic EFW inaccuracies in fetal macrosomia appear to have a greater impact on the mode of delivery than birth weight itself. Underestimation of fetal weight may be associated with a higher probability of shoulder dystocia.

Randomized boosting with multivariable base-learners for high-dimensional variable selection and prediction.

BACKGROUND: Statistical boosting is a computational approach to select and estimate interpretable prediction models for high-dimensional biomedical data, leading to implicit regularization and variable selection when combined with early stopping. Traditionally, the set of base-learners is fixed for all iterations and consists of simple regression learners including only one predictor variable at a time. Furthermore, the number of iterations is typically tuned by optimizing the predictive performance, leading to models which often include unnecessarily large numbers of noise variables. RESULTS: We propose three consecutive extensions of classical component-wise gradient boosting. In the first extension, called Subspace Boosting (SubBoost), base-learners can consist of several variables, allowing for multivariable updates in a single iteration. To compensate for the larger flexibility, the ultimate selection of base-learners is based on information criteria leading to an automatic stopping of the algorithm. As the second extension, Random Subspace Boosting (RSubBoost) additionally includes a random preselection of base-learners in each iteration, enabling the scalability to high-dimensional data. In a third extension, called Adaptive Subspace Boosting (AdaSubBoost), an adaptive random preselection of base-learners is considered, focusing on base-learners which have proven to be predictive in previous iterations. Simulation results show that the multivariable updates in the three subspace algorithms are particularly beneficial in cases of high correlations among signal covariates. In several biomedical applications the proposed algorithms tend to yield sparser models than classical statistical boosting, while showing a very competitive predictive performance also compared to penalized regression approaches like the (relaxed) lasso and the elastic net. CONCLUSIONS: The proposed randomized boosting approaches with multivariable base-learners are promising extensions of statistical boosting, particularly suited for highly-correlated and sparse high-dimensional settings. The incorporated selection of base-learners via information criteria induces automatic stopping of the algorithms, promoting sparser and more interpretable prediction models.

Estimating effective infection fatality rates during the course of the COVID-19 pandemic in Germany.

BACKGROUND: The infection fatality rate (IFR) of the Coronavirus Disease 2019 (COVID-19) is one of the most discussed figures in the context of this pandemic. In contrast to the case fatality rate (CFR), the IFR depends on the total number of infected individuals - not just on the number of confirmed cases. In order to estimate the IFR, several seroprevalence studies have been or are currently conducted. METHODS: Using German COVID-19 surveillance data and age-group specific IFR estimates from multiple international studies, this work investigates time-dependent variations in effective IFR over the course of the pandemic. Three different methods for estimating (effective) IFRs are presented: (a) population-averaged IFRs based on the assumption that the infection risk is independent of age and time, (b) effective IFRs based on the assumption that the age distribution of confirmed cases approximately reflects the age distribution of infected individuals, and (c) effective IFRs accounting for age- and time-dependent dark figures of infections. RESULTS: Effective IFRs in Germany are estimated to vary over time, as the age distributions of confirmed cases and estimated infections are changing during the course of the pandemic. In particular during the first and second waves of infections in spring and autumn/winter 2020, there has been a pronounced shift in the age distribution of confirmed cases towards older age groups, resulting in larger effective IFR estimates. The temporary increase in effective IFR during the first wave is estimated to be smaller but still remains when adjusting for age- and time-dependent dark figures. A comparison of effective IFRs with observed CFRs indicates that a substantial fraction of the time-dependent variability in observed mortality can be explained by changes in the age distribution of infections. Furthermore, a vanishing gap between effective IFRs and observed CFRs is apparent after the first infection wave, while an increasing gap can be observed during the second wave. CONCLUSIONS: The development of estimated effective IFR and observed CFR reflects the changing age distribution of infections over the course of the COVID-19 pandemic in Germany. Further research is warranted to obtain timely age-stratified IFR estimates, particularly in light of new variants of the virus.

Generalizability of polygenic prediction models: how is the R2 defined on test data?

BACKGROUND: Polygenic risk scores (PRS) quantify an individual's genetic predisposition for different traits and are expected to play an increasingly important role in personalized medicine. A crucial challenge in clinical practice is the generalizability and transferability of PRS models to populations with different ancestries. When assessing the generalizability of PRS models for continuous traits, the R 2 is a commonly used measure to evaluate prediction accuracy. While the R 2 is a well-defined goodness-of-fit measure for statistical linear models, there exist different definitions for its application on test data, which complicates interpretation and comparison of results. METHODS: Based on large-scale genotype data from the UK Biobank, we compare three definitions of the R 2 on test data for evaluating the generalizability of PRS models to different populations. Polygenic models for several phenotypes, including height, BMI and lipoprotein A, are derived based on training data with European ancestry using state-of-the-art regression methods and are evaluated on various test populations with different ancestries. RESULTS: Our analysis shows that the choice of the R 2  definition can lead to considerably different results on test data, making the comparison of R 2  values from the literature problematic. While the definition as the squared correlation between predicted and observed phenotypes solely addresses the discriminative performance and always yields values between 0 and 1, definitions of the R 2 based on the mean squared prediction error (MSPE) with reference to intercept-only models assess both discrimination and calibration. These MSPE-based definitions can yield negative values indicating miscalibrated predictions for out-of-target populations. We argue that the choice of the most appropriate definition depends on the aim of PRS analysis - whether it primarily serves for risk stratification or also for individual phenotype prediction. Moreover, both correlation-based and MSPE-based definitions of R 2 can provide valuable complementary information. CONCLUSIONS: Awareness of the different definitions of the R 2 on test data is necessary to facilitate the reporting and interpretation of results on PRS generalizability. It is recommended to explicitly state which definition was used when reporting R 2 values on test data. Further research is warranted to develop and evaluate well-calibrated polygenic models for diverse populations.

Immune Profiling of Vulvar Squamous Cell Cancer Discovers a Macrophage-rich Subtype Associated with Poor Prognosis.

The incidence rates of vulvar squamous cell cancer (VSCC) have increased over the past decades, requiring personalized oncologic approaches. Currently, lymph node involvement is a key factor in determining prognosis and treatment options. However, there is a need for additional immune-related biomarkers to provide more precise treatment and prognostic information. Here, we used IHC and expression data to characterize immune cells and their spatial distribution in VSCC. Hierarchical clustering analysis identified distinct immune subtypes, of which the macrophage-rich subtype was associated with adverse outcome. This is consistent with our findings of increased lymphogenesis, lymphatic invasion, and lymph node involvement associated with high macrophage infiltration. Further in vitro studies showed that VSCC-associated macrophages expressed VEGF-A and subsequently induced VEGF-A in the VSCC cell line A-431, providing experimental support for a pro-lymphangiogenic role of macrophages in VSCC. Taken together, immune profiling in VSCC revealed tumor processes, identified a subset of patients with adverse outcome, and provided a valuable biomarker for risk stratification and therapeutic decision making for anti-VEGF treatment, ultimately contributing to the advancement of precision medicine in VSCC. SIGNIFICANCE: Immunoprofiling in VSCC reveals subtypes with distinct clinical and biological behavior. Of these, the macrophage-rich VSCC subtype is characterized by poor clinical outcome and increased VEGF-A expression, providing a biomarker for risk stratification and therapeutic sensitivity.

Robust statistical boosting with quantile-based adaptive loss functions.

We combine robust loss functions with statistical boosting algorithms in an adaptive way to perform variable selection and predictive modelling for potentially high-dimensional biomedical data. To achieve robustness against outliers in the outcome variable (vertical outliers), we consider different composite robust loss functions together with base-learners for linear regression. For composite loss functions, such as the Huber loss and the Bisquare loss, a threshold parameter has to be specified that controls the robustness. In the context of boosting algorithms, we propose an approach that adapts the threshold parameter of composite robust losses in each iteration to the current sizes of residuals, based on a fixed quantile level. We compared the performance of our approach to classical M-regression, boosting with standard loss functions or the lasso regarding prediction accuracy and variable selection in different simulated settings: the adaptive Huber and Bisquare losses led to a better performance when the outcome contained outliers or was affected by specific types of corruption. For non-corrupted data, our approach yielded a similar performance to boosting with the efficient L 2 loss or the lasso. Also in the analysis of skewed KRT19 protein expression data based on gene expression measurements from human cancer cell lines (NCI-60 cell line panel), boosting with the new adaptive loss functions performed favourably compared to standard loss functions or competing robust approaches regarding prediction accuracy and resulted in very sparse models.

Comprehensive immunohistochemical analysis of N6-methyladenosine (m6A) writers, erasers, and readers in endometrial cancer.

PURPOSE: N6-methyladenosine (m6A) is the most frequent type of messenger RNA (mRNA) modification and is implicated in diverse physiological processes. The procedure of m6A RNA modification is regulated by a dynamic interaction of writers (METTL3, METTL4, METTL14, WTAP, KIAA1429), erasers (FTO, ALKBH5), and readers (HNRNPA2B1, HNRNPC, YTHDC1, YTHDC1, YTHDF1-3). In the oncological context, alterations in m6A were identified to be critically involved in tumorigenesis, proliferation, angiogenesis, and drug resistance across diverse cancer entities including endometrial cancer (EC). METHODS: In this study, we comprehensively examined the protein expression of m6A writers, readers and erasers by immunohistochemical staining in a cohort of N = 65 EC patients. Protein expression data were analyzed with regard to clinical outcomes. RESULTS: We identified enhanced protein expression levels of METTL3, METTL14, FTO, HNRNPA2B1, and HNRNPC, respectively to be of prognostic value and linked to a shortened overall survival in EC. CONCLUSION: Overall, our study points toward dysregulated m6A modification in EC and its possibility to serve as a promising prognostic biomarker.

Lung Ultrasound in Predicting Outcomes in Patients with COVID-19 Treated with Extracorporeal Membrane Oxygenation.

Pulmonary involvement due to SARS-CoV-2 infection can lead to acute respiratory distress syndrome in patients with COVID-19. Consequently, pulmonary imaging is crucial for management of COVID-19. This study aimed to evaluate the prognostic value of lung ultrasound (LUS) with a handheld ultrasound device (HHUD) in patients with COVID-19 treated with extracorporeal membrane oxygenation (ECMO). Therefore, patients underwent LUS with a HHUD every two days until they were either discharged from the intensive care unit or died. The study was conducted at the University Hospital of Bonn's anesthesiological intensive care ward from December 2020 to August 2021. A total of 33 patients (median [IQR]: 56.0 [53-60.5] years) were included. A high LUS score was associated with a decreased P/F ratio (repeated measures correlation [rmcorr]: -0.26; 95% CI: -0.34, -0.15; p < 0.001), increased extravascular lung water, defined as fluid accumulation in the pulmonary interstitium and alveoli (rmcorr: 0.11; 95% CI: 0.01, 0.20; p = 0.030), deteriorated electrolyte status (base excess: rmcorr: 0.14; 95% CI: 0.05, 0.24; p = 0.004; pH: rmcorr: 0.12; 95% CI: 0.03, 0.21; p = 0.001), and decreased pulmonary compliance (rmcorr: -0.10; 95% CI: -0.20, -0.01; p = 0.034). The maximum LUS score was lower in survivors (median difference [md]: -0.35; 95% CI: -0.55, -0.06; p = 0.006). A cutoff value for non-survival was calculated at a LUS score of 2.63. At the time of maximum LUS score, P/F ratio (md: 1.97; 95% CI: 1.12, 2.76; p < 0.001) and pulmonary compliance (md: 18.67; 95% CI: 3.33, 37.15; p = 0.018) were higher in surviving patients. In conclusion, LUS with a HHUD enables continuous evaluation of cardiopulmonary function in COVID-19 patients receiving ECMO support therapy and provides prognostic value in determining the patients' likelihood of survival.

Estimating the course of the COVID-19 pandemic in Germany via spline-based hierarchical modelling of death counts.

We consider a retrospective modelling approach for estimating effective reproduction numbers based on death counts during the first year of the COVID-19 pandemic in Germany. The proposed Bayesian hierarchical model incorporates splines to estimate reproduction numbers flexibly over time while adjusting for varying effective infection fatality rates. The approach also provides estimates of dark figures regarding undetected infections. Results for Germany illustrate that our estimates based on death counts are often similar to classical estimates based on confirmed cases; however, considering death counts allows to disentangle effects of adapted testing policies from transmission dynamics. In particular, during the second wave of infections, classical estimates suggest a flattening infection curve following the "lockdown light" in November 2020, while our results indicate that infections continued to rise until the "second lockdown" in December 2020. This observation is associated with more stringent testing criteria introduced concurrently with the "lockdown light", which is reflected in subsequently increasing dark figures of infections estimated by our model. In light of progressive vaccinations, shifting the focus from modelling confirmed cases to reported deaths with the possibility to incorporate effective infection fatality rates might be of increasing relevance for the future surveillance of the pandemic.

A statistical boosting framework for polygenic risk scores based on large-scale genotype data.

Polygenic risk scores (PRS) evaluate the individual genetic liability to a certain trait and are expected to play an increasingly important role in clinical risk stratification. Most often, PRS are estimated based on summary statistics of univariate effects derived from genome-wide association studies. To improve the predictive performance of PRS, it is desirable to fit multivariable models directly on the genetic data. Due to the large and high-dimensional data, a direct application of existing methods is often not feasible and new efficient algorithms are required to overcome the computational burden regarding efficiency and memory demands. We develop an adapted component-wise L 2-boosting algorithm to fit genotype data from large cohort studies to continuous outcomes using linear base-learners for the genetic variants. Similar to the snpnet approach implementing lasso regression, the proposed snpboost approach iteratively works on smaller batches of variants. By restricting the set of possible base-learners in each boosting step to variants most correlated with the residuals from previous iterations, the computational efficiency can be substantially increased without losing prediction accuracy. Furthermore, for large-scale data based on various traits from the UK Biobank we show that our method yields competitive prediction accuracy and computational efficiency compared to the snpnet approach and further commonly used methods. Due to the modular structure of boosting, our framework can be further extended to construct PRS for different outcome data and effect types-we illustrate this for the prediction of binary traits.

Deselection of base-learners for statistical boosting-with an application to distributional regression.

We present a new procedure for enhanced variable selection for component-wise gradient boosting. Statistical boosting is a computational approach that emerged from machine learning, which allows to fit regression models in the presence of high-dimensional data. Furthermore, the algorithm can lead to data-driven variable selection. In practice, however, the final models typically tend to include too many variables in some situations. This occurs particularly for low-dimensional data (p