arXiv subjects

Steven G. Gilmour

Publications and source records attributed to Steven G. Gilmour.

12 recordsLinked to original sources

arXiv2026

Factorial clinical trials in the presence and absence of plausible statistical interactions between treatment factors. A historical review of the methodological literature

Factorial trials can be conducted when statistical interactions between two or more treatment factors are not anticipated, but also when they are. A k-in-1 factorial trial answers k single-factor questions with the same number of units as one parallel-group trial if there are no interactions. Literature on k-in-1 factorial trials has dominated trialists understanding of factorial trials in the UK. However, factorial experiments originated from the Design of Experiments field to enable interactions to be robustly estimated. Seemingly conflicting guidance from these literatures poses a source of confusion and misunderstanding for trialists. We bring these literatures together to provide clarity on the arguments that have been used to recommend use of factorial trials in the presence and absence of interactions. We outline motivating examples. We summarise the rationales for using factorial trials, the treatment contrasts of interest, and the properties of their estimators, for the two schools of thought. We describe the debate, going back to 1935, and use an empirical example to illustrate the impact of different analysis approaches. We conclude that it is vital that trialists carefully and clearly specify their objectives. Estimands of interest and treatment contrasts follow, with properties of estimators dictated by this choice.

stat.ME↗

arXiv2025

Pseudo-Bayesian Optimal Designs for Fitting Fractional Polynomial Response Surface Models

Fractional polynomial models are potentially useful for response surfaces investigations. With the availability of routines for fitting nonlinear models in statistical packages they are increasingly being used. However, as in all experiments the design should be chosen such that the model parameters are estimated as efficiently as possible. The design choice for such models involves the known nonlinear models' design difficulties but \cite{gilmour_trinca_2012b} proposed a methodology capable of producing exact designs that makes use of the computing facilities available today. In this paper, we use this methodology to find Bayesian optimal exact designs for several fractional polynomial models. The optimum designs are compared to various standard designs in response surface problems.

stat.ME↗

arXiv2025

The Generalized Word Count in Two-Level Fractional Factorial Designs

Unreplicated two-level factorial designs are often used in screening experiments to determine which factors out of a large plausible set are active. A theorem regarding the generalized word count pattern is stated and proved for unreplicated designs. It is shown that a phenomenon regarding optimal designs seen in the recent literature can be explained by the theorem obtained.

stat.ME↗

arXiv2025

$Q_B$-Optimal Two-Level Designs

Two-level designs are widely used for screening experiments where the goal is to identify a few active factors which have major effects. Orthogonal two-level designs in which all factors are level-balance and each of the four level combinations of any pair of factors appears equally often are commonly used. In this paper, we apply the model-robust $Q_B$ criterion introduced by Tsai, Gilmour and Mead (2007) to the selection of optimal two-level screening designs without the requirements of level-balance and pairwise orthogonality. The criterion incorporates experimenter's prior belief on how likely a factor is to be active and recommends different designs under different priors, and without the requirement of level-balance and pairwise orthogonality, a wider range of designs is possible. A coordinate exchange algorithm is developed for the construction of $Q_B$-optimal designs for given priors.

stat.ME↗

arXiv2024

MOODE: An R Package for Multi-Objective Optimal Design of Experiments

We describe the R package MOODE and demonstrate its use to find multi-objective optimal experimental designs. Multi-Objective Optimal Design of Experiments (MOODE) targets the experimental objectives directly, ensuring that the full set of research questions is answered as economically as possible. In particular, individual criteria aimed at optimizing inference are combined with lack-of-fit and MSE-based components in compound optimality criteria to target multiple and competing objectives reflecting the priorities and aims of the experimentation. The package implements either a point exchange or coordinate exchange algorithm as appropriate to find nearly optimal designs. We demonstrate the functionality of MOODE through the application of the methodology to two case studies of varying complexity.

stat.CO↗

arXiv2024

Response Surface Designs for Crossed and Nested Multi-Stratum Structures

Response surface designs are usually described as being run under complete randomization of the treatment combinations to the experimental units. In practice, however, it is often necessary or beneficial to run them under some kind of restriction to the randomization, leading to multi-stratum designs. In particular, some factors are often hard to set, so they cannot have their levels reset for each experimental unit. This paper presents a general solution to designing response surface experiments in any multi-stratum structure made up of crossing and/or nesting of unit factors. A stratum-by-stratum approach to constructing designs using compound optimal design criteria is used and illustrated. It is shown that good designs can be found even for large experiments in complex structures.

stat.ME↗

arXiv2024

$Q_B$ Optimal Two-Level Designs for the Baseline Parameterization

We have established the association matrix that expresses the estimator of effects under baseline parameterization, which has been considered in some recent literature, in an equivalent form as a linear combination of estimators of effects under the traditional centered parameterization. This allows the generalization of the $Q_B$ criterion which evaluates designs under model uncertainty in the traditional centered parameterization to be applicable to the baseline parameterization. Some optimal designs under the baseline parameterization seen in the previous literature are evaluated and it has been shown that at a given prior probability of a main effect being in the best model, the design converges to $Q_B$ optimal as the probability of an interaction being in the best model converges to 0 from above. The $Q_B$ optimal designs for two setups of factors and run sizes at various priors are found by an extended coordinate exchange algorithm and the evaluation of their performances are discussed. Comparisons have been made to those optimal designs restricted to level balance and orthogonality conditions.

stat.ME↗

arXiv2023

Optimal response surface designs in the presence of model contamination

Complete reliance on the fitted model in response surface experiments is risky and relaxing this assumption, whether out of necessity or intentionally, requires an experimenter to account for multiple conflicting objectives. This work provides a methodological framework of a compound optimality criterion comprising elementary criteria responsible for: (i) the quality of the confidence region-based inference to be done using the fitted model (DP-/LP-optimality); (ii) improving the ability to test for the lack-of-fit from specified potential model contamination in the form of extra polynomial terms; and (iii) simultaneous minimisation of the variance and bias of the fitted model parameters arising from this misspecification. The latter two components have been newly developed in accordance with the model-independent 'pure error' approach to the error estimation. The compound criteria and design construction were adapted to restricted randomisation frameworks: blocked and multistratum experiments, where the stratum-by-stratum approach was adopted. A point-exchange algorithm was employed for searching for nearly optimal designs. The theoretical work is accompanied by one real and two illustrative examples to explore the relationship patterns among the individual components and characteristics of the optimal designs, demonstrating the attainable compromises across the competing objectives and driving some general practical recommendations.

stat.ME↗

arXiv2023

Statistical Design and Analysis for Robust Machine Learning: A Case Study from COVID-19

Since early in the coronavirus disease 2019 (COVID-19) pandemic, there has been interest in using artificial intelligence methods to predict COVID-19 infection status based on vocal audio signals, for example cough recordings. However, existing studies have limitations in terms of data collection and of the assessment of the performances of the proposed predictive models. This paper rigorously assesses state-of-the-art machine learning techniques used to predict COVID-19 infection status based on vocal audio signals, using a dataset collected by the UK Health Security Agency. This dataset includes acoustic recordings and extensive study participant meta-data. We provide guidelines on testing the performance of methods to classify COVID-19 infection status based on acoustic features and we discuss how these can be extended more generally to the development and assessment of predictive methods based on public health datasets.

cs.SD↗

arXiv2021

Designs with complex blocking structures and network effects for agricultural field experiments

We propose a novel model-based approach for constructing optimal designs with complex blocking structures and network effects, for application in agricultural field experiments. The potential interference among treatments applied to different plots is described via a network structure, defined via the adjacency matrix. We consider a field trial run at Rothamsted Research and provide a comparison of optimal designs under various different models, including the commonly used designs in such situations. It is shown that when there is interference between treatments on neighbouring plots, due to the spatial arrangement of the plots, designs incorporating network effects are at least as, and often more efficient than, randomised row-column designs. The advantage of network designs is that we can construct the neighbour structure even for an irregular layout by means of a graph to address the particular characteristics of the experiment. The need for such designs arises when it is required to account for treatment-induced patterns of heterogeneity. Ignoring the network structure can lead to imprecise estimates of the treatment parameters and invalid conclusions.

stat.ME↗

arXiv2019

Optimal block designs for experiments on networks

We propose a method for constructing optimal block designs for experiments on networks. The response model for a given network interference structure extends the linear network effects model to incorporate blocks. The optimality criteria are chosen to reflect the experimental objectives and an exchange algorithm is used to search across the design space for obtaining an efficient design when an exhaustive search is not possible. Our interest lies in estimating the direct comparisons among treatments, in the presence of nuisance network effects that stem from the underlying network interference structure governing the experimental units, or in the network effects themselves. Comparisons of optimal designs under different models, including the standard treatment models, are examined by comparing the variance and bias of treatment effect estimators. We also suggest a way of defining blocks, while taking into account the interrelations of groups of experimental units within a network, using spectral clustering techniques to achieve optimal modularity. We expect connected units within closed-form communities to behave similarly to an external stimulus. We provide evidence that our approach can lead to efficiency gains over conventional designs such as randomized designs that ignore the network structure and we illustrate its usefulness for experiments on networks.

stat.ME↗

arXiv2019

Prediction properties of optimum response surface designs

Prediction capability is considered an important issue in response surface methodology. Following the line of argument that a design should have several desirable properties we have extended an existing compound design criterion to include prediction properties. Prediction of responses and of differences in response are considered. Point and interval predictions are allowed for. Extensions of existing graphical tools for inspecting prediction performances of the designs in the whole region of experimentation are also introduced. The methods are illustrated with two examples.

stat.ME↗