Article
Statistics & Probability
Liugen Xue
Summary: This paper studies the estimation and empirical likelihood (EL) of the parameters in a partially linear single-index varying-coefficient model. A two-stage method is proposed to estimate the regression parameters and the coefficient functions. The asymptotic distributions of the proposed estimators are obtained. A bias-corrected EL ratio for the regression parameters is also proposed. It is shown that the ratio asymptotically follows a standard chi-squared distribution. The proposed method is evaluated through simulation studies and an application example using a real data set.
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY
(2023)
Article
Remote Sensing
Jin Xu, Lindi J. Quackenbush, Timothy A. Volk, Stephen Stehman
Summary: This study evaluated the uncertainty of shrub willow health characterization based on unmanned aerial systems (UAS) data. The results showed that regression models built at different spatial scales could be applied across time, space, and scales. The study also quantified the uncertainty of model parameters and found that the uncertainty increased as pixel size increased. The findings provide guidance for future experimental design to save resources.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)
Article
Multidisciplinary Sciences
Wisunee Puggard, Sa-Aat Niwitpong, Suparat Niwitpong
Summary: This study introduces four methods for constructing confidence intervals for the coefficient of variation (CV) and the difference between CVs of Birnbaum-Saunders (BS) distributions. A Monte Carlo simulation shows that the highest posterior density (HPD) interval performs best overall. The proposed methods were validated using PM 2.5 concentration data for Chiang Mai, Thailand in March and April 2019, with results consistent with the simulation findings.
Article
Computer Science, Artificial Intelligence
Remi Martin, Luc Duong
Summary: Bayesian convolutional neural networks are a cutting-edge computer vision framework that can model two types of uncertainties: epistemic uncertainty and aleatoric uncertainty. Few studies have integrated uncertainty into end-to-end prediction pipelines, but our research proposes a novel way to assess confidence in predictions. We train a model using cross-entropy loss and then use pixel weighting to reduce uncertainty. From the resulting epistemic uncertainty measures, we calculate a histogram and use a final neural network model to determine a confidence interval for predictions. Our approach yields a lower calibration error and higher accuracy compared to existing methods.
MACHINE VISION AND APPLICATIONS
(2023)
Article
Engineering, Multidisciplinary
Xiaofei Wang, Bing Xing Wang, Xin Pan, Yunkai Hu, Yingpei Chen, Junxing Zhou
Summary: This paper discusses interval estimation for the inverse Gaussian distribution, deriving generalized confidence intervals and prediction intervals. Through simulation, it is shown that the proposed intervals have higher coverage probabilities compared to traditional methods like Wald CIs and bootstrap-p CIs. Additionally, the proposed procedures are illustrated using two examples.
QUALITY AND RELIABILITY ENGINEERING INTERNATIONAL
(2021)
Article
Multidisciplinary Sciences
Shu-Fei Wu
Summary: This paper presents interval estimation methods for the scale parameter of two-parameter exponential distribution, and proposes two methods for the joint confidence region. The optimal method of the confidence region is determined by simulation comparison based on confidence region area. The proposed methods are validated through a biometrical example.
Article
History & Philosophy Of Science
Paul Vos, Don Holbert
Summary: Frequentist inference is usually described based on hypothetical repeated sampling, but there are advantages to interpreting it using a single random sample. Recent research has used the perspective of classical probability to interpret probabilities of random phenomena and applied it to statistical inference.
Article
Environmental Sciences
Mamoru Tanaka, Tomoya Kataoka, Yasuo Nihei
Summary: Microplastics, plastic particles smaller than 5 mm in diameter, are becoming widespread pollutants in natural environments, including freshwater ecosystems. However, there is currently no standard method for in situ sampling in freshwater environments. This study proposes a framework based on the Poisson point process to compute confidence intervals for microplastic concentration estimates and validates the method using samples from two urban rivers in Chiba, Japan.
ENVIRONMENTAL POLLUTION
(2023)
Article
Engineering, Industrial
Chen Yang, Wanze Lu, Yuanqing Xia
Summary: This study proposes an uncertain optimal attitude-vibration control method for rigid-flexible coupling satellites with reliability constraints based on the interval dimension-wise analysis. The proposed method considers multi-source uncertainties and models them as interval parameters. With the help of the interval dimension-wise analysis method, the uncertain state response of the control system can be accurately predicted. An interval linear quadratic regulator (LQR) is then proposed to optimize the control input and cost function. Additionally, a non-probabilistic time-dependent reliability method is proposed to effectively evaluate the relationship between the uncertain state response and threshold.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2023)
Article
Statistics & Probability
Shu Yang, Chenyin Gao, Donglin Zeng, Xiaofei Wang
Summary: We propose a test-based elastic integrative analysis that combines randomised trial and real-world data to estimate treatment effect heterogeneity. Our approach allows for efficient estimation by using the trial design to decide whether to incorporate real-world data. We provide a data-adaptive procedure for selecting a test threshold that minimises mean square error and offer an elastic confidence interval with good finite-sample coverage.
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY
(2023)
Article
Medicine, Research & Experimental
Arsene Brunelle Sandie, Nicolas Molinari, Anthony Wanjoya, Charles Kouanfack, Christian Laurent, Jules Brice Tchatchueng-Mbougua
Summary: In this study, we proposed two procedures for the non-inferiority test with a flexible margin for continuous endpoints. The procedures were based on a test statistic and confidence interval approaches respectively. Simulations were used to assess the performances and properties of the proposed test procedures. An application on real-world clinical data was conducted to evaluate the efficacy of clinical monitoring alone versus laboratory and clinical monitoring in HIV-infected adult patients.
Article
Statistics & Probability
Chengchun Shi, Rui Song, Wenbin Lu, Runze Li
Summary: This article presents a new estimation and valid inference method for single or low-dimensional regression coefficients in high-dimensional generalized linear models, allowing predictor variables to grow exponentially with respect to sample size. The proposed estimator is computed through solving a score function, and model selection is conducted recursively to reduce dimensionality and construct a score equation. The proposed confidence interval achieves valid coverage without assuming consistency of the model selection procedure, and is shown to be asymptotically narrower than CIs based on other methods in certain cases.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2021)
Article
Computer Science, Software Engineering
Vali Tawosi, Federica Sarro, Alessio Petrozziello, Mark Harman
Summary: The study replicates and extends a previous research on multi-objective software effort estimation, strengthening both internal and external validity. The results confirm the superior performance of CoGEE and its effectiveness across different multi-objective algorithms.
IEEE TRANSACTIONS ON SOFTWARE ENGINEERING
(2022)
Article
Mathematics, Interdisciplinary Applications
Keke Lai
Summary: The study introduces a new method for accurately estimating RMSEA in structural equation model parameters and explains why traditional methods fail in this aspect. Simulation studies indicate that the new method outperforms traditional methods.
STRUCTURAL EQUATION MODELING-A MULTIDISCIPLINARY JOURNAL
(2021)
Review
Medicine, Research & Experimental
Elias Laurin Meyer, Peter Mesenbrink, Tobias Mielke, Tom Parke, Daniel Evans, Franz Koenig
Summary: In recent years, the popularity of platform trials has increased, but there are still many design-related questions that need further exploration. Existing software packages often target single design elements and lack extensive design flexibility. Collaborative efforts are needed to create software that fully utilizes the flexibility of platform trials.
Article
Mathematical & Computational Biology
Rajeshwari Sridhara, Elizabeth Barksdale, Olga Marchenko, Qi Jiang, Yuki Ando, Erick Bloomquist, Michael Coory, Melissa Crouse, Evgeny Degtyarev, Theodor Framke, Boris Freidlin, David E. Gerber, Thomas Gwise, Filip Josephson, Lorenzo Hess, Paul Kluetz, Daniel Li, Sumithra Mandrekar, Martin Posch, Khadija Rantell, Bohdana Ratitch, Andrew Raven, Kit Roes, Kaspar Rufibach, Sinan B. Sarac, Richard Simon, Harpreet Singh, Marc Theoret, Andrew Thomson, Emmanuel Zuber, Yuan Li Shen, Richard Pazdur
Summary: This article summarizes the discussions from the ASA BIOP Section Open Forum, organized by the ASA BIOP Statistical Methods in Oncology Scientific Working Group in collaboration with the US FDA Oncology Center of Excellence and LUNGevity Foundation. The discussions focused on how to incorporate lessons learned during the COVID-19 pandemic into future oncology trial designs, highlighting the importance of flexibility.
STATISTICS IN BIOPHARMACEUTICAL RESEARCH
(2023)
Editorial Material
Mathematical & Computational Biology
Martin Posch, Robin Ristl, Franz Koenig
STATISTICS IN BIOPHARMACEUTICAL RESEARCH
(2023)
Editorial Material
Biochemistry & Molecular Biology
Ulrich Mansmann, Clara Locher, Fabian Prasser, Tracey Weissgerber, Ulrich Sax, Martin Posch, Evelyne Decullier, Ioana A. A. Cristea, Thomas P. A. Debray, Leonhard Held, David Moher, John P. A. Ioannidis, Joseph S. S. Ross, Christian Ohmann, Florian Naudet
Summary: Data sharing improves the value of medical research and promotes trust in clinical trials, but more biomedical researchers need training in approaches such as meta-research, data science, and ethical, legal, and social issues.
Article
Multidisciplinary Sciences
Paul Zajic, Michael Eichinger, Michael Eichlseder, Barbara Hallmann, Gabriel Honnef, Tobias Fellinger, Barbara Metnitz, Martin Posch, Martin Rief, Philipp G. H. Metnitz
Summary: This study aims to investigate whether immediate or delayed extubation of tracheally intubated patients admitted to ICU postoperatively is associated with different outcomes. A retrospective analysis of an Austrian ICU registry revealed that immediate extubation is associated with favorable outcomes for tracheally intubated patients, suggesting that it may be preferable.
Article
Statistics & Probability
David Azriel, Yosef Rinott, Martin Posch
Summary: This study aims to design multi-armed parallel group clinical trials to estimate personalized treatment rules. Results show that, in the case of two treatments, the optimal allocation rule does not depend on the value of the covariates but only on the variances of the responses. However, for three or more treatments, the optimal treatment allocation depends on the values of the covariates as well as the true regression coefficients.
SCANDINAVIAN JOURNAL OF STATISTICS
(2023)
Letter
Mathematical & Computational Biology
Olga Marchenko, Rajeshwari Sridhara, Qi Jiang, Elizabeth Barksdale, Yuki Ando, Dinesh De Alwis, Katie Brown, Laura Fernandes, Mark T. J. van Bussel, Qiuyi Choo, Michael Coory, Elizabeth Garrett-Mayer, Thomas Gwise, Lorenzo Hess, Rong Liu, Sumithra Mandrekar, Daniele Ouellet, Jose Pinheiro, Martin Posch, Nam Atiqur Rahman, Khadija Rerhou Rantell, Andrew Raven, Sarem Sarem, Suman Sen, Mirat Shah, Yuan Li Shen, Richard Simon, Marc Theoret, Ying Yuan, Richard Pazdur
Summary: This article summarizes the discussions of the ASA BIOP Section Open Forums held on March 18th, June 10th, and July 8th, 2022. The forums were organized by the ASA BIOP Statistical Methods in Oncology Scientific Working Group, the FDA Oncology Center of Excellence, and the LUNGevity Foundation, and involved various stakeholders in the field of cancer drug development. The discussions focused on strategies and designs of dose-optimization studies, highlighting the need for a paradigm shift in oncology drug development that improves dose optimization without delaying patient access to potentially effective treatments.
STATISTICS IN BIOPHARMACEUTICAL RESEARCH
(2023)
Review
Gastroenterology & Hepatology
Juan M. A. Pericas, Nicholas A. M. Di Prospero, Quentin M. Anstee, Peter S. Mesenbrinck, Mette S. Kjaer, Jesus Rivera-Esteban, Franz Koenig, Elena Sena, Raluca Pais, Ramiro Manzano, Joan Genesca, Frank Tacke, Vlad Ratziu
Summary: Non-alcoholic steatohepatitis (NASH) is an important unmet medical need with a growing field of clinical research and drug development. Platform trials (PT) can accelerate drug development and create a more patient-centric environment. This review provides a comprehensive assessment of the NASH clinical development landscape.
ALIMENTARY PHARMACOLOGY & THERAPEUTICS
(2023)
Article
Mathematical & Computational Biology
David S. Robertson, James M. S. Wason, Franz Koenig, Martin Posch, Thomas Jaki
Summary: Platform trials evaluate multiple experimental treatments under a single master protocol, with the potential risk of inflating the overall type I error rate due to multiple treatment comparisons and non-pre-specified, time-dependent hypotheses. Online error rate control provides a solution to the problem of multiplicity for platform trials, testing each hypothesis one-by-one over time without knowledge of future tests.
STATISTICS IN MEDICINE
(2023)
Review
Medicine, Research & Experimental
Marta Bofill Roig, Cora Burgwinkel, Ursula Garczarek, Franz Koenig, Martin Posch, Quynh Nguyen, Katharina Hees
Summary: Platform trials offer flexibility by allowing new arms to enter after the trial starts. The use of shared control groups improves trial efficiency, but using non-concurrent controls may introduce bias. Statistical methods for incorporating non-concurrent controls are available, but regulatory guidance is limited.
Article
Multidisciplinary Sciences
Paul Zajic, Michael Hiesmayr, Peter Bauer, David M. Baron, Anastasiia Gruber, Michael Joannidis, Martin Posch, Philipp G. H. Metnitz
Summary: This retrospective study analysed the trends and outcomes of COVID-19 patient admissions in Austrian hospitals. The study found that mortality was associated with age, with older individuals having a higher risk. ICU or IMCU admission was more likely for individuals between 55 and 74 years old. The study also found regional differences in outcomes and higher mortality rates in the first half of 2020 and the second half of 2021.
SCIENTIFIC REPORTS
(2023)
Article
Mathematical & Computational Biology
Benjamin Hofner, Elina Asikanius, Wolfgang Jacquet, Theodor Framke, Katrien Oude Rengerink, Lukas Aguirre Davila, Maria Gruenewald, Florian Klinglmueller, Martin Posch, Finbarr P. Leacy, Thomas Lang, Armin Koch, Jorg Zinserling, Kit Roes
Summary: The COVID-19 pandemic has driven a significant research effort to develop vaccines and therapies. The European Medicines Agency (EMA) implemented rolling reviews to accelerate the evaluation process. This article discusses the clinical trial designs, statistical approaches, and reporting of results for the first four vaccines approved by the EMA. The rapid accumulation of COVID-19 cases in vaccine efficacy trials necessitated multiple data updates, posing challenges for evaluation and interpretation.
STATISTICS IN BIOPHARMACEUTICAL RESEARCH
(2023)
Article
Physiology
Pierre Raeven, Gerhard Hagn, Laura Niederstaetter, Jonas Brugger, Sophia Bayer-Blauensteiner, Christoph Domenig, Konrad Hoetzenecker, Martin Posch, Gerda Leitner, Christopher Gerner, David M. M. Baron
Summary: This study analyzed the changes in the levels of eicosanoids in the plasma of postoperative ICU patients before and after transfusion. The results showed that the levels of these substances increased in the stored red blood cell component and were widely present in the plasma before transfusion. This study demonstrates the importance of further research on the role of eicosanoids in transfusion-related immunomodulation.
FRONTIERS IN PHYSIOLOGY
(2023)
Article
Psychiatry
H. Katschnig, C. Strassmayr, F. Endel, M. Posch, I. Steiner
Summary: In this retrospective cohort study, the potential preventability of early psychiatric re-hospitalisations was investigated using nationwide electronic claims data. The results showed a significant association between post-discharge ambulatory physician contacts and decreased psychiatric re-hospitalisation rate. However, caution should be exercised in interpreting this association causally due to potential confounding factors.
INTERNATIONAL JOURNAL OF METHODS IN PSYCHIATRIC RESEARCH
(2023)
Article
Computer Science, Software Engineering
Pavla Krotka, Katharina Hees, Peter Jacko, Dominic Magirr, Martin Posch, Marta Bofill Roig
Summary: Platform trials evaluate the efficacy of multiple treatments through sharing controls, and utilizing non-concurrent controls can improve the power of treatment-control comparisons. The R-package NCC is presented for designing and analyzing platform trials using non-concurrent controls. It allows for simulating platform trials and evaluating analysis methods in various settings. The package provides main NCC functions and demonstrates how to simulate and analyze platform trials with specific examples.
Article
Surgery
Robin Ristl, Johannes Klopf, Andreas Scheuba, Anna Sotir, Florian Wolf, Christoph M. Domenig, Anders Wanhainen, Christoph Neumayer, Martin Posch, Christine Brostjan, Wolf Eilenberg
Summary: The study aimed to evaluate the use of supplementing volume measurements for monitoring abdominal aortic aneurysms (AAAs) and compare the growth rates of maximum diameter and volume at the patient level. The results showed that while there is a substantial association between volume and maximum diameter at the population level, AAAs in individual patients grow at different paces in different dimensions.
INTERNATIONAL JOURNAL OF SURGERY
(2023)