A5012023996- Statistical Methods and Inference
- Statistical Methods and Bayesian Inference
- Advanced Statistical Methods and Models
- Bayesian Methods and Mixture Models
- Advanced Statistical Modeling Techniques
- Data Analysis with R
- Anomaly Detection Techniques and Applications
- Animal Behavior and Welfare Studies
- Neural Networks and Applications
- Time Series Analysis and Forecasting
- Food Allergy and Anaphylaxis Research
- Allergic Rhinitis and Sensitization
- Stochastic Gradient Optimization Techniques
- Advanced Causal Inference Techniques
- Soil Geostatistics and Mapping
- Economic and Environmental Valuation
- Control Systems and Identification
- Gaussian Processes and Bayesian Inference
- Topological and Geometric Data Analysis
- Genetic and phenotypic traits in livestock
- Statistical Methods in Clinical Trials
- Advanced Control Systems Optimization
- Advanced Clustering Algorithms Research
- Domain Adaptation and Few-Shot Learning
- Bayesian Modeling and Causal Inference
Ludwig-Maximilians-Universität München
2014-2023
Heidelberg (Poland)
2018-2022
University of Hohenheim
2017-2022
LMU Klinikum
2009-2021
University of Milano-Bicocca
2018-2019
Albany Research Institute
2018
Clausthal University of Technology
2016
Klinikum rechts der Isar
2016
Institut für Angewandte Statistik
2015
Institute of Mathematical Statistics
2013
Ischemic stroke of undetermined cause is a major health issue because its high frequency and clinical relevance. Histopathologic analysis human thrombi, retrieved from patients with large-vessel occlusion during mechanical thrombectomy, may provide information about underlying pathologies. This study examines the relationship between causes histological clot composition to identify specific patterns that might help distinguish cryptogenic stroke.Thrombi 145 consecutive were collected...
We propose an extensive framework for additive regression models correlated functional responses, allowing multiple partially nested or crossed random effects with flexible correlation structures for, e.g., spatial, temporal, longitudinal data. Additionally, our includes linear and nonlinear of scalar covariates that may vary smoothly over the index response. It accommodates densely sparsely observed responses predictors which be additional error both spline-based principal component-based...
Distributed lag non-linear models (DLNMs) are a modelling tool for describing potentially and delayed dependencies. Here, we illustrate an extension of the DLNM framework through use penalized splines within generalized additive (GAM). This offers built-in model selection procedures possibility accommodating assumptions on shape structure specific penalties. In addition, this includes, as special cases, simpler previously proposed linear relationships (DLMs). Alternative versions DLNMs...
We introduce the functional generalized additive model (FGAM), a novel regression for association studies between scalar response and predictor. link-transformed mean as integral with respect to t of F{X(t), t} where F(·,·) is an unknown function X(t) covariate. Rather than having in finite number principal components Müller Yao (2008), our incorporates predictor directly thus can be viewed natural extension models. estimate using tensor-product B-splines roughness penalties. A pointwise...
Researchers are increasingly interested in regression models for functional data. This article discusses a comprehensive framework additive (mixed) responses and/or covariates based on the guiding principle of reframing terms corresponding scalar data, allowing adaptation large body existing methods these novel tasks. The encompasses many as well new models. It includes ‘generalized’ mean regression, quantile generalized location, shape and scale (GAMLSS) admits flexible linear, smooth or...
Abstract: This tutorial article demonstrates how time-to-event data can be modelled in a very flexible way by taking advantage of advanced inference methods that have recently been developed for generalized additive mixed models. In particular, we describe the necessary pre-processing steps transforming such into suitable format and show variety effects, including smooth nonlinear baseline hazard, potentially nonlinearly time-varying estimated interpreted. We also present useful graphical...
The habitat-amount hypothesis challenges traditional concepts that explain species richness within habitats, such as the habitat-patch hypothesis, where number is a function of patch size and isolation. It posits effects isolation are driven by sample area, thus at site basically total habitat amount surrounding this site. We tested for saproxylic beetles their dead wood using an experiment comprising 190 plots with manipulated sizes situated in forested region high variation (i.e., density...
Fabian Scheipla, Ludwig Fahrmeira & Thomas Kneibb a Department of Statistics , Ludwig-Maximilians-Universität München Munich Germany b Economics Georg-August-Universität Göttingen
We propose a comprehensive framework for additive regression models non-Gaussian functional responses, allowing multiple (partially) nested or crossed random effects with flexible correlation structures for, e.g., spatial, temporal, longitudinal data as well linear and nonlinear of scalar covariates that may vary smoothly over the index response. Our implementation handles responses from any exponential family distribution many others like Beta- scaled shifted $t$-distributions. Development...
Abstract Background Proteins are an essential part of medical nutrition therapy in critically ill patients. Guidelines almost universally recommend a high protein intake without robust evidence supporting its use. Methods Using large international database, we modelled associations between the hazard rate in-hospital death and live hospital discharge (competing risks) three categories (low: < 0.8 g/kg per day, standard: 0.8–1.2 high: > 1.2 day) during first 11 days after ICU admission...
Treatment failure during venom immunotherapy (VIT) may be associated with a variety of risk factors.Our aim was to evaluate the association baseline serum tryptase concentration (BTC) and other parameters frequency VIT maintenance phase.In this observational prospective multicenter study, we followed 357 patients established honey bee or vespid allergy after dose had been reached. In all patients, effectiveness either verified by sting challenge (n = 154) patient self-reporting outcome field...
The functional linear array model (FLAM) is a unified class for regression models including function-on-scalar, scalar-on-function and function-on-function regression. Mean, median, quantile as well generalized additive or scalar responses are contained special cases in this general framework. Our implementation features broad variety of covariate effects, such as, linear, smooth interaction effects grouping variables, covariates. Computational efficiency achieved by representing the model....
The R package <b>spikeSlabGAM</b> implements Bayesian variable selection, model choice, and regularized estimation in (geo-)additive mixed models for Gaussian, binomial, Poisson responses. Its purpose is to (1) choose an appropriate subset of potential covariates their interactions, (2) determine whether linear or more flexible functional forms are required the effects respective covariates, (3) estimate shapes. Selection regularization terms based on a novel spike-and-slab-type prior...
Regression models with functional responses and covariates constitute a powerful increasingly important model class. However, regression data poses well known challenging problems of non-identifiability. This non-identifiability can manifest itself in arbitrarily large errors for coefficient surface estimates despite accurate predictions the responses, thus invalidating substantial interpretations fitted models. We offer an accessible rephrasing these identifiability issues realistic...