Article
Mathematical & Computational Biology
Shiyang Ma, Michael P. McDermott
Summary: Adaptive designs in learning-phase clinical trials can be efficient and highly informative. This article extends the MCP-Mod procedure with GMCTs for two-stage adaptive designs for proof-of-concept. The results of an interim analysis guide adaptations to candidate dose-response models and dosages studied in the second stage, with GMCTs used in both stages to obtain and combine stage-wise p-values for an overall p-value. Simulation studies show advantages of adaptive designs over nonadaptive designs when candidate dose-response models are not well-informed by preclinical and early-phase evidence.
BIOMETRICAL JOURNAL
(2022)
Article
Computer Science, Hardware & Architecture
Xuejing Zhou, Wanli Peng, Boya Yang, Juan Wen, Yiming Xue, Ping Zhong
Summary: This article proposes a novel linguistic steganographic model based on adaptive probability distribution and generative adversarial network, which can hide secret messages in the generated text and improve the security performance of steganography.
IEEE TRANSACTIONS ON DEPENDABLE AND SECURE COMPUTING
(2022)
Article
Health Care Sciences & Services
Ekkehard Glimm, David S. Robertson
Summary: Response-adaptive randomization adjusts treatment allocation probabilities based on previously observed response data. This article proposes an improved method that guarantees non-negative weights for each block of data and provides a substantial power advantage in practice, especially when patients are allocated in blocks using response-adaptive randomization.
STATISTICAL METHODS IN MEDICAL RESEARCH
(2023)
Article
Economics
Heng Lei, Minggao Xue, Huiling Liu
Summary: This study proposes a hybrid model for forecasting the conditional probability distribution of carbon allowance prices. It utilizes singular spectrum analysis to process non-stationary signals and a non-crossing composite quantile regression neural network algorithm for accurate quantile forecasts. The model outperforms benchmark models in terms of forecasting accuracy and has applications in risk management.
Article
Computer Science, Information Systems
Muhammad Aslam Mohd Safari, Nurulkamal Masseran, Muhammad Hilmi Abdul Majid
Summary: A robust and efficient estimator of the exponential distribution parameter was proposed in this study, based on the probability integral transform statistic. The new estimator offers reasonable protection against outliers and is simple to compute.
CMC-COMPUTERS MATERIALS & CONTINUA
(2021)
Article
Engineering, Marine
Ali Tian, Xufeng Shu, Jiaming Guo, Haoyun Li, Renchuan Ye, Peng Ren
Summary: This paper presents a statistical modeling approach to explore the dependence of extreme values in multi-site tidal water levels using hourly data from six tidal stations in the Kansai region, Japan. The proposed method utilizes a multi-site conditional extreme value model and a peak over threshold model with yearly order grouping to handle extreme value dependence and marginal modeling of tidal CEVM, respectively. The results demonstrate the effectiveness of the proposed approach in analyzing extreme value dependence among multiple sites and provide a more accurate basis for engineering construction and disaster prevention.
Article
Computer Science, Artificial Intelligence
Elham Hatefi, Hossein Karshenas, Peyman Adibi
Summary: This paper proposes a novel method to reduce inter-domain differences in domain adaptation, leading to improved accuracy in unsupervised domain adaptation. Experimental studies demonstrate significant improvement compared to other state-of-the-art methods in benchmark image classification tasks.
EXPERT SYSTEMS WITH APPLICATIONS
(2024)
Article
Computer Science, Interdisciplinary Applications
Yifan Li, Jinhua Qin, Chunjie Wu
Summary: The adaptive EWMA control chart has good overall performance in monitoring Gaussian processes but often fails under non-Gaussian processes with limited information about the underlying distribution. This paper develops a new robust adaptive EWMA control chart that discount outliers and achieve excellent detection performance, even with little information. It proposes a nearly optimal and distribution-free parameter design strategy and demonstrates superior performance in process monitoring through simulation and real examples.
COMPUTERS & INDUSTRIAL ENGINEERING
(2023)
Article
Energy & Fuels
Lingzhi Wang, Lai Wei, Jun Liu, Fucai Qian
Summary: The maximum entropy distribution model is used to fit wind power fluctuations in this study. Comparing with other distribution models, the maximum entropy distribution model shows the best fitting effect and performance, making it more suitable for describing the fluctuation characteristics of wind power.
Article
Mathematical & Computational Biology
David S. Robertson, Babak Choodari-Oskooei, Munya Dimairo, Laura Flight, Philip Pallmann, Thomas Jaki
Summary: In adaptive clinical trials, the conventional end-of-trial point estimate of a treatment effect is prone to bias. This article is the second part of a series that examines the issue of bias in point estimation for adaptive trials. It discusses the negative impact of bias on standard estimators, presents guidelines for researchers, and emphasizes the importance of considering bias throughout the entire lifecycle of an adaptive design.
STATISTICS IN MEDICINE
(2023)
Article
Automation & Control Systems
Bernard Vau, Ioan Dore Landau
Summary: The stability of adaptive disturbance rejection schemes using Youla-Kucera parametrization and the internal model principle in the presence of plant model uncertainties is investigated in this study. Utilizing over parametrization of the Youla-Kucera filter is crucial for solving the IMP and stability problems, as discussed with known disturbance cases. The approach is extended for unknown disturbances using parameter adaptation algorithm with projection, showing promising results for handling plant-model mismatch.
Article
Automation & Control Systems
Likang Feng, Weihai Zhang
Summary: This paper proposes a method for practical tracking control and disturbance rejection for a class of discrete-time stochastic linear systems, by designing a dynamic controller to achieve practical tracking in mean square sense and disturbance rejection while maintaining system stability. The feasibility of the method is demonstrated by applying it to a mass-spring-damper system, and the efficiency is illustrated through simulation results.
INTERNATIONAL JOURNAL OF CONTROL
(2021)
Article
Health Care Sciences & Services
Dario Zocholl, Cornelia U. Kunz, Geraldine Rauch
Summary: In oncology, phase II clinical trials often adopt single-arm two-stage designs with binary endpoints, such as progression-free survival after 12 months, and allow for stopping for futility after the first stage. However, these trials may experience undesirably long interruptions due to the required follow-up time. To address this problem, a short-term endpoint, such as progression-free survival after 3 months, can be used for the interim decision. We propose two approaches based on conditional power and Bayesian posterior predictive probability of success to guide the interim decision using all available short-term and long-term assessments, which demonstrate improved statistical power compared to the existing approach in slow patient recruitment settings. The software code for implementing these methods is publicly available.
STATISTICAL METHODS IN MEDICAL RESEARCH
(2023)
Article
Operations Research & Management Science
Bogdan Grechuk, Michael Zabarankin, Alexander Mafusalov, Stan Uryasev
Summary: This study introduces the concept of superdistribution and its application to random variables, proposing the buffered CDF and reduced CDF for multivariate probability distribution problems. The efficiency of the algorithm is demonstrated through a case study on optimizing a collateralized debt obligation.
OPTIMIZATION LETTERS
(2023)
Article
Biology
Hoseung Song, Hao Chen
Summary: This article extends a graph-based framework for changepoint detection to address challenges posed by repeated observations, deriving analytical formulas to control Type I error for new methods, making them applicable to large datasets.