Article
Health Care Sciences & Services
Menglan Pang, Robert W. Platt, Tibor Schuster, Michal Abrahamowicz
Summary: The accelerated failure time model is an alternative to the Cox proportional hazards model in survival analysis, but requires meeting specific underlying modeling assumptions for valid conclusions. This model lacks formal investigation of the time ratio and linearity assumptions, while prognostic factors may have time-dependent and nonlinear effects.
STATISTICAL METHODS IN MEDICAL RESEARCH
(2021)
Article
Mathematics, Interdisciplinary Applications
Aliaksandr Hubin, Georg Heinze, Riccardo De Bin
Summary: This paper proposes a framework for fitting multivariable fractional polynomial models as special cases of Bayesian generalized nonlinear models. By applying an adapted version of the genetically modified mode jumping Markov chain Monte Carlo algorithm, the Bayesian version of fractional polynomials can be used in any supervised learning task.
FRACTAL AND FRACTIONAL
(2023)
Article
Mathematical & Computational Biology
Peter C. Austin, Jiming Fang, Douglas S. Lee
Summary: The Cox proportional hazards model, commonly used in clinical and epidemiological research, assumes proportional hazards for variables. When this assumption is violated, there are two methods to allow regression coefficients to vary as a flexible function of time. This flexibility improves the modeling of data and enhances the accuracy of the model.
STATISTICS IN MEDICINE
(2022)
Article
Economics
Xiaoyi Han, Bin Peng, Yanrong Yang, Huanjun Zhu
Summary: This paper studies shrinkage estimation of a general varying-coefficient model using the KLASSO method proposed by Li and Racine (2010), demonstrating its estimation sparsity and oracle efficiency, and providing a BIC-type criterion for parameter selection. Simulation results show that the method performs well in terms of estimation errors and variable selection.
Article
Health Care Sciences & Services
Nadine Marlin, Peter J. Godolphin, Richard L. Hooper, Richard D. Riley, Ewelina Rogozinska
Summary: Linear effect modification (LEM), nonlinear covariate-outcome associations (NL) and nonlinear effect modification (NLEM) are commonly used analysis methods in individual participant data meta-analysis (IPDMA). However, these methods are often susceptible to bias or lack detailed descriptions. Nonlinearity of continuous covariates and power analysis in IPDMA are rarely assessed.
JOURNAL OF CLINICAL EPIDEMIOLOGY
(2023)
Article
Economics
Christoph Breunig, Peter Haan
Summary: In this study, we address the issue of regression with selectively observed covariates in a nonparametric setting using instrumental variables. Identification of the fractional probability weight (FPW) function is achieved through a partial completeness assumption. This method provides a constructive approach for estimation without suffering from the inverse problem.
JOURNAL OF ECONOMETRICS
(2021)
Article
Health Care Sciences & Services
Xiang Li, Yong Ma, Qing Pan
Summary: This paper proposes a novel standardization method for different data types of covariates in sparse penalized regressions, aiming to solve the problem of different selection probabilities caused by traditional methods. The authors illustrate the advantages of the proposed method through simulation studies and empirical analysis.
STATISTICAL METHODS IN MEDICAL RESEARCH
(2023)
Article
Automation & Control Systems
Hamid Reza Marzban
Summary: This research aims to develop a direct transcription approach for solving optimal control problems governed by nonlinear fractional Fredholm integral equations with delays in input and output signals. The new methodology is based on a multi domains decomposition scheme using fractional-order Legendre functions. A new fractional derivative operator associated with the fractional basis is introduced using the Caputo fractional derivative operator. The dynamical system related to the fractional control problem is transformed into a new system of algebraic equations using derivative and delay operators. Various challenging test problems are studied to demonstrate the effectiveness of the designed approach.
Article
Agriculture, Dairy & Animal Science
Pedro Vital Brasil Ramos, Gilberto Romeiro de Oliveira Menezes, Delvan Alves da Silva, Daniela Lourenco, Gustavo Garcia Santiago, Roberto A. A. Torres Junior, Fabyano Fonseca e Silva, Paulo Savio Lopes, Renata Veroneze
Summary: Feed efficiency is crucial for the profitability and sustainability of the beef cattle industry. This study proposed genomic evaluations for feed efficiency traits using random regression models. The models were compared based on goodness-of-fit and genetic parameter behavior. The findings provide evidence for new selection strategies in the beef cattle industry.
JOURNAL OF ANIMAL BREEDING AND GENETICS
(2023)
Article
Multidisciplinary Sciences
Colin Griesbach, Andreas Groll, Elisabeth Bergherr
Summary: The study introduces an improved boosting algorithm for linear mixed models, which appropriately weights random effects, disentangles them from the fixed effects updating scheme, and corrects for correlations with cluster-constant covariates to enhance estimate quality and reduce computational effort. This method outperforms current state-of-the-art approaches in boosting and maximum likelihood inference, as demonstrated through simulations and various data examples.
Article
Mathematical & Computational Biology
Farhad Hatami, Alex Ocampo, Gordon Graham, Thomas E. Nichols, Habib Ganjgahi
Summary: In this article, an optimization technique for fitting continuous time Markov models (CTMM) in the presence of covariates is proposed. This technique combines a stochastic gradient descent algorithm with differentiation of the matrix exponential using a Pade approximation, making it feasible to fit large scale data. Two methods for computing standard errors are presented, one utilizing the Pade expansion and the other using power series expansion of the matrix exponential. Simulation results show improved performance compared to existing CTMM methods, and the method is demonstrated on a large-scale multiple sclerosis NO.MS dataset.
Article
Health Care Sciences & Services
Darsy Darssan, Gita D. Mishra, Darren C. Greenwood, Sven Sandin, Eric J. Brunner, Sybil L. Crawford, Samar R. El Khoudary, Maria Mori Brooks, Ellen B. Gold, Mette Kildevaeld Simonsen, Hsin-Fang Chung, Elisabete Weiderpass, Annette J. Dobson
Summary: Methods for meta-analysis of studies with individual participant data and continuous exposure variables are demonstrated in this study. A two-stage process is used to estimate response curves for each study and average them pointwise over all studies at each value of the exposure. Real data samples and code are provided for result replication.
JOURNAL OF CLINICAL EPIDEMIOLOGY
(2021)
Article
Multidisciplinary Sciences
H. Hassani, J. A. Tenreiro Machado, Z. Avazzadeh, E. Safari, S. Mehrabi
Summary: This article analyzes a fractional order breast cancer competition model under the Caputo fractional derivative, proposing a new set of basis functions and an optimization method for solving the model. Numerical experiments demonstrate the high accuracy, practicality, and computational efficiency of the devised technique.
SCIENTIFIC REPORTS
(2021)
Article
Mathematics, Interdisciplinary Applications
Waleed Mohamed Abd-Elhameed, Youssri Hassan Youssri, Amr Kamel Amin, Ahmed Gamal Atta
Summary: In this study, an innovative approach using a spectral collocation algorithm is proposed to obtain numerical solutions of the nonlinear time-fractional generalized Kawahara equation. The method involves introducing a new set of orthogonal polynomials as fundamental functions and transforming the equation into a set of nonlinear algebraic equations. The effectiveness and reliability of the approach are validated through rigorous analysis and numerical experiments.
FRACTAL AND FRACTIONAL
(2023)
Article
Multidisciplinary Sciences
Fiifi Amoako Johnson
Summary: Childhood stunting in Ghana exhibits geospatial clustering, with varying levels of stunting in different regions. Socio-demographic factors are primarily associated with clustering of districts with high childhood stunting, while socio-ecological factors are mainly associated with clustering of districts with low childhood stunting. Nonlinear associations were found between childhood stunting and socio-ecological factors such as Insecticide Treated Net (ITN) coverage, nightlight composite, travel time to a main settlement, and population density.
Article
Statistics & Probability
Sarah Friedrich, Gerd Antes, Sigrid Behr, Harald Binder, Werner Brannath, Florian Dumpert, Katja Ickstadt, Hans A. Kestler, Johannes Lederer, Heinz Leitgob, Markus Pauly, Ansgar Steland, Adalbert Wilhelm, Tim Friede
Summary: Statistics plays a significant role in both the theoretical and practical understanding of artificial intelligence (AI) and in its future development. It contributes to methodological development, planning and design of studies, assessment of data quality and data collection, differentiation of causality and associations, as well as evaluation of uncertainty in results. Integrating statistical aspects into AI teaching is crucial for schools and universities.
ADVANCES IN DATA ANALYSIS AND CLASSIFICATION
(2022)
Review
Biochemical Research Methods
Fabrizio Kuruc, Harald Binder, Moritz Hess
Summary: This study explores the use of different data sets and loss functions to improve parameter learning of deep neural networks. The results show that employing stratified loss functions leads to better predictive power and lower prediction error, with known prognostic genes receiving higher importance.
BRIEFINGS IN BIOINFORMATICS
(2022)
Review
Multidisciplinary Sciences
Christine Wallisch, Paul Bach, Lorena Hafermann, Nadja Klein, Willi Sauerbrei, Ewout W. Steyerberg, Georg Heinze, Geraldine Rauch
Summary: Although regression models are central in medical research, misconceptions still exist and recent statistical advancements are not adequately reflected in publications. Developing statistical guidance for addressing nonlinear effects, model specification, and variable selection is essential to better support medical researchers in performing or interpreting regression analyses.
Article
Health Care Sciences & Services
Willi Sauerbrei, Patrick Royston
Summary: In clinical trials, it is important to investigate whether a treatment effect is consistent in all patients or if certain variables indicate a different response to treatment. Categorizing continuous predictors into groups using cutpoints has several weaknesses. A better approach is to keep the variables continuous and analyze them using methods like Subpopulation Treatment Effect Pattern Plot (STEPP) and multivariable fractional polynomial interaction (MFPI). Meta-analysis techniques can be used to study treatment heterogeneity if individual patient data (IPD) from multiple studies are available. The recently proposed Meta-STEPP method was used to investigate the interaction of estrogen receptors with chemotherapy in eight randomized controlled trials (RCTs) of primary breast cancer patients.
BMC MEDICAL RESEARCH METHODOLOGY
(2022)
Article
Medicine, General & Internal
Willi Sauerbrei, Tim Haeussler, James Balmford, Marianne Huebner
Summary: Most prognostic factor studies are poorly reported and analyzed, which has severe consequences for related systematic reviews and meta-analyses.
Article
Cardiac & Cardiovascular Systems
Natalie Arnold, Iris M. Hermanns, Andreas Schulz, Omar Hahad, Volker H. Schmitt, Marina Panova-Noeva, Juergen H. Prochaska, Harald Binder, Norbert Pfeiffer, Manfred Beutel, Karl J. Lackner, Thomas Muenzel, Philipp S. Wild
Summary: This study investigates the predictive ability of direct plasma renin and aldosterone concentrations as well as their ratio (aldosterone-to-renin, ARR) for incident hypertension in the general population. The study finds that ARR has a stronger predictive value for incident hypertension than renin or aldosterone alone, especially in obese subjects.
CARDIOVASCULAR RESEARCH
(2023)
Article
Multidisciplinary Sciences
Edwin Kipruto, Willi Sauerbrei
Summary: This paper aims to compare variable selection methods and investigate the impact of shrinkage of regression estimates in a simulation study within the framework of a classical linear regression model for low-dimensional data. It emphasizes the importance of conducting neutral comparison studies to address bias and improve the design and reporting of simulation studies in statistical methodology research.
Article
Oncology
Daniel F. Hayes, Willi Sauerbrei, Lisa M. McShane
Summary: In 2005, experts in tumor biomarker research published the REporting recommendations for Tumor MARKer prognostic studies (REMARK) criteria, which, combined with the subsequent Biospecimen Reporting for Improved Study Quality (BRISQ) criteria, provide a framework for transparent reporting of study conduct and analyses.
BRITISH JOURNAL OF CANCER
(2023)
Article
Computer Science, Information Systems
Urs Alexander Fichtner, Lukas Maximilian Horstmeier, Boris Alexander Bruhmann, Manuel Watter, Harald Binder, Jochen Knaus
Summary: This study aims to investigate the effect of data sharing on researchers' response patterns from the perspective of survey participants. The results show that data sharing does not have an impact on response bias or dropout for the researchers, and provide insights into the experiences and behaviors of the participants regarding data sharing.
JOURNAL OF DOCUMENTATION
(2023)
Article
Mathematical & Computational Biology
Goeran Koeber, Raffael Kalisch, Lara M. C. Puhlmann, Andrea Chmitorz, Anita Schick, Harald Binder
Summary: When modeling longitudinal biomedical data, both dimensionality reduction and dynamic modeling are needed. This study proposes an extension that allows different sets of differential equation parameters for observation subperiods, improving the model's fit with a small dataset. The approach successfully identifies individual-level parameters and promising predictors for predicting resilience and informing future study design.
BIOMETRICAL JOURNAL
(2023)
Article
Medicine, General & Internal
Joerg Rahnenfuehrer, Riccardo De Bin, Axel Benner, Federico Ambrogi, Lara Lusa, Anne-Laure Boulesteix, Eugenia Migliavacca, Harald Binder, Stefan Michiels, Willi Sauerbrei, Lisa McShane
Summary: This review provides an overview of key aspects and methods for analyzing high-dimensional data in biomedical research. It addresses the challenges and opportunities of analyzing high-dimensional data, and offers solutions and references for researchers.
Article
Health Care Sciences & Services
Urs A. Fichtner, Anita Arslanow, Harald Binder, Peter R. Galle, Christian Labenz, Frank Lammert, Julia Ortner, Dominikus Stelzer, Louis Velthuis, Erik Farin-Glattacker
Summary: This study aimed to explore the psychosocial consequences of positive liver screening results and identify influencing factors for perceived strain within a multistage screening program for liver cirrhosis and fibrosis in Germany. The results showed that the screening had emotional and behavioral impacts on the patients. Negative emotional consequences were primarily driven by suboptimal patient-provider communication and could be worsened by a lack of transparent information transfer. Patients sought information and support from their social environment. All patients had positive attitudes towards liver screening. To minimize the potential psychosocial consequences, transparent information and regular health communication should be emphasized during the screening process.
HEALTH EXPECTATIONS
(2023)
Article
Multidisciplinary Sciences
Urs A. Fichtner, Lukas M. Horstmeier, Boris A. Bruehmann, Harald Binder, Jochen Knaus
Summary: The BE-KONFORM study was conducted to investigate employees' needs of the Medical Faculty of the University of Freiburg regarding research data management. The study involved qualitative video interviews with four researchers and a standardized online survey. The final dataset included 236 complete cases, with 90% in German and 10% in English. The data offers the potential for connection with data collected in other contexts.
Article
Health Care Sciences & Services
Tanja Sprave, Michelle Pfaffenlehner, Raluca Stoian, Eleni Christofi, Alexander Ruehle, Daniela Zoeller, Alexander Fabian, Harald Fahrner, Harald Binder, Henning Schaefer, Eleni Gkika, Anca-Ligia Grosu, Felix Heinemann, Nils Henrik Nicolay
Summary: This study aimed to investigate the feasibility of integrating electronic patient-reported outcomes (ePROs) in the treatment surveillance pathway of patients with head and neck cancer (HNC) and assess the influence of app-based ePRO monitoring on patient satisfaction and quality of life. Results showed that ePRO monitoring was feasible and led to an increased reporting of HNC-specific symptom burden, while significantly improving certain domains of patient satisfaction.
JOURNAL OF MEDICAL INTERNET RESEARCH
(2023)