arXiv subjects

Brian Tom

Publications and source records attributed to Brian Tom.

11 recordsLinked to original sources

arXiv2025

Model selection with uncertainty in estimating optimal dynamic treatment regimes

Optimal dynamic treatment regimes (DTRs), as a key part of precision medicine, have progressively gained more attention recently. To inform clinical decision making, interpretable and parsimonious models for contrast functions are preferred, raising concerns about undue misspecification. It is therefore important to properly evaluate the performance of candidate interpretable models and select the one that best approximates the unknown contrast function. Moreover, since a DTR usually involves multiple decision points, an inaccurate approximation at a later decision point affects its estimation at an earlier decision point when a backward induction algorithm is applied. This paper aims to perform model selection for contrast functions in the context of learning optimal DTRs from observed data. Note that the relative performance of candidate models may heavily depend on the sample size when, for example, the comparison is made between parametric and tree-based models. Therefore, instead of investigating the limiting behavior of each candidate model and developing methods to select asymptotically the `correct' one, we focus on the finite sample performance of each model and attempt to perform model selection under a given sample size. To this end, we adopt the counterfactual cross-validation metric and propose a novel method to estimate the variance of the metric. Supplementing the cross-validation metric with its estimated variance allows us to characterize the uncertainty in model selection under a given sample size and facilitates hypothesis testing associated with a preferred model structure.

stat.ME↗

arXiv2025

Competing Risks: Impact on Risk Estimation and Algorithmic Fairness

Accurate time-to-event prediction is integral to decision-making, informing medical guidelines, hiring decisions, and resource allocation. Survival analysis, the quantitative framework used to model time-to-event data, accounts for patients who do not experience the event of interest during the study period, known as censored patients. However, many patients experience events that prevent the observation of the outcome of interest. These competing risks are often treated as censoring, a practice frequently overlooked due to a limited understanding of its consequences. Our work theoretically demonstrates why treating competing risks as censoring introduces substantial bias in survival estimates, leading to systematic overestimation of risk and, critically, amplifying disparities. First, we formalize the problem of misclassifying competing risks as censoring and quantify the resulting error in survival estimates. Specifically, we develop a framework to estimate this error and demonstrate the associated implications for predictive performance and algorithmic fairness. Furthermore, we examine how differing risk profiles across demographic groups lead to group-specific errors, potentially exacerbating existing disparities. Our findings, supported by an empirical analysis of cardiovascular management, demonstrate that ignoring competing risks disproportionately impacts the individuals most at risk of these events, potentially accentuating inequity. By quantifying the error and highlighting the fairness implications of the common practice of considering competing risks as censoring, our work provides a critical insight into the development of survival models: practitioners must account for competing risks to improve accuracy, reduce disparities in risk assessment, and better inform downstream decisions.

cs.LG↗

arXiv2025

Prediction of Survival Outcomes under Clinical Presence Shift: A Joint Neural Network Architecture

Electronic health records arise from the complex interaction between patients and the healthcare system. This observation process of interactions, referred to as clinical presence, often impacts observed outcomes. When using electronic health records to develop clinical prediction models, it is standard practice to overlook clinical presence, impacting performance and limiting the transportability of models when this interaction evolves. We propose a multi-task recurrent neural network that jointly models the inter-observation time and the missingness processes characterising this interaction in parallel to the survival outcome of interest. Our work formalises the concept of clinical presence shift when the prediction model is deployed in new settings (e.g. different hospitals, regions or countries), and we theoretically justify why the proposed joint modelling can improve transportability under changes in clinical presence. We demonstrate, in a real-world mortality prediction task in the MIMIC-III dataset, how the proposed strategy improves performance and transportability compared to state-of-the-art prediction models that do not incorporate the observation process. These results emphasise the importance of leveraging clinical presence to improve performance and create more transportable clinical prediction models.

cs.LG↗

arXiv2024

Identifying treatment response subgroups in observational time-to-event data

Identifying patient subgroups with different treatment responses is an important task to inform medical recommendations, guidelines, and the design of future clinical trials. Existing approaches for treatment effect estimation primarily rely on Randomised Controlled Trials (RCTs), which tend to feature more homogeneous patient groups, making them less relevant for uncovering subgroups in the population encountered in real-world clinical practice. Subgroup analyses established for RCTs suffer from significant statistical biases when applied to observational studies, which benefit from larger and more representative populations. Our work introduces a novel, outcome-guided, subgroup analysis strategy for identifying subgroups of treatment response in both RCTs and observational studies alike. It hence positions itself in-between individualised and average treatment effect estimation to uncover patient subgroups with distinct treatment responses, critical for actionable insights that may influence treatment guidelines. In experiments, our approach significantly outperforms the current state-of-the-art method for subgroup analysis in both randomised and observational treatment regimes.

stat.ME↗

arXiv2023

Exploring the big data paradox for various estimands using vaccination data from the global COVID-19 Trends and Impact Survey (CTIS)

Selection bias poses a challenge to statistical inference validity in non-probability surveys. This study compared estimates of the first-dose COVID-19 vaccination rates among Indian adults in 2021 from a large non-probability survey, COVID-19 Trends and Impact Survey (CTIS), and a small probability survey, the Center for Voting Options and Trends in Election Research (CVoter), against benchmark data from the COVID Vaccine Intelligence Network (CoWIN). Notably, CTIS exhibits a larger estimation error (0.39) compared to CVoter (0.16). Additionally, we investigated the estimation accuracy of the CTIS when using a relative scale and found a significant increase in the effective sample size by altering the estimand from the overall vaccination rate. These results suggest that the big data paradox can manifest in countries beyond the US and it may not apply to every estimand of interest.

stat.AP↗

arXiv2023

Neural Fine-Gray: Monotonic neural networks for competing risks

Time-to-event modelling, known as survival analysis, differs from standard regression as it addresses censoring in patients who do not experience the event of interest. Despite competitive performances in tackling this problem, machine learning methods often ignore other competing risks that preclude the event of interest. This practice biases the survival estimation. Extensions to address this challenge often rely on parametric assumptions or numerical estimations leading to sub-optimal survival approximations. This paper leverages constrained monotonic neural networks to model each competing survival distribution. This modelling choice ensures the exact likelihood maximisation at a reduced computational cost by using automatic differentiation. The effectiveness of the solution is demonstrated on one synthetic and three medical datasets. Finally, we discuss the implications of considering competing risks when developing risk scores for medical practice.

cs.LG↗

arXiv2022

Imputation Strategies Under Clinical Presence: Impact on Algorithmic Fairness

Machine learning risks reinforcing biases present in data and, as we argue in this work, in what is absent from data. In healthcare, societal and decision biases shape patterns in missing data, yet the algorithmic fairness implications of group-specific missingness are poorly understood. The way we address missingness in healthcare can have detrimental impacts on downstream algorithmic fairness. Our work questions current recommendations and practices aimed at handling missing data with a focus on their effect on algorithmic fairness, and offers a path forward. Specifically, we consider the theoretical underpinnings of existing recommendations as well as their empirical predictive performance and corresponding algorithmic fairness measured through subgroup performances. Our results show that current practices for handling missingness lack principled foundations, are disconnected from the realities of missingness mechanisms in healthcare, and can be counterproductive. For example, we show that favouring group-specific imputation strategy can be misguided and exacerbate prediction disparities. We then build on our findings to propose a framework for empirically guiding imputation choices, and an accompanying reporting framework. Our work constitutes an important contribution to recent efforts by regulators and practitioners to grapple with the realities of real-world data, and to foster the responsible and transparent deployment of machine learning systems. We demonstrate the practical utility of the proposed framework through experimentation on widely used datasets, where we show how the proposed framework can guide the selection of imputation strategies, allowing us to choose among strategies that yield equal overall predictive performance but present different algorithmic fairness properties.

cs.AI↗

arXiv2022

DeepJoint: Robust Survival Modelling Under Clinical Presence Shift

Observational data in medicine arise as a result of the complex interaction between patients and the healthcare system. The sampling process is often highly irregular and itself constitutes an informative process. When using such data to develop prediction models, this phenomenon is often ignored, leading to sub-optimal performance and generalisability of models when practices evolve. We propose a multi-task recurrent neural network which models three clinical presence dimensions -- namely the longitudinal, the inter-observation and the missingness processes -- in parallel to the survival outcome. On a prediction task using MIMIC III laboratory tests, explicit modelling of these three processes showed improved performance in comparison to state-of-the-art predictive models (C-index at 1 day horizon: 0.878). More importantly, the proposed approach was more robust to change in the clinical presence setting, demonstrated by performance comparison between patients admitted on weekdays and weekends. This analysis demonstrates the importance of studying and leveraging clinical presence to improve performance and create more transportable clinical models.

cs.LG↗

arXiv2022

A comparison of two frameworks for multi-state modelling, applied to outcomes after hospital admissions with COVID-19

We compare two multi-state modelling frameworks that can be used to represent dates of events following hospital admission for people infected during an epidemic. The methods are applied to data from people admitted to hospital with COVID-19, to estimate the probability of admission to ICU, the probability of death in hospital for patients before and after ICU admission, the lengths of stay in hospital, and how all these vary with age and gender. One modelling framework is based on defining transition-specific hazard functions for competing risks. A less commonly used framework defines partially-latent subpopulations who will experience each subsequent event, and uses a mixture model to estimate the probability that an individual will experience each event, and the distribution of the time to the event given that it occurs. We compare the advantages and disadvantages of these two frameworks, in the context of the COVID-19 example. The issues include the interpretation of the model parameters, the computational efficiency of estimating the quantities of interest, implementation in software and assessing goodness of fit. In the example, we find that some groups appear to be at very low risk of some events, in particular ICU admission, and these are best represented by using "cure-rate" models to define transition-specific hazards. We provide general-purpose software to implement all the models we describe in the "flexsurv" R package, which allows arbitrarily-flexible distributions to be used to represent the cause-specific hazards or times to events.

stat.ME↗

arXiv2022

Latent class mixed modelling for phenotypic stratification of primary biliary cholangitis patients on first line treatment

In patients with primary biliary cholangitis (PBC), the serum liver biochemistry measured during treatment with ursodeoxycholic acid (the UDCA response) accurately predicts long-term outcome. In this study we sought to use liver biochemistry, and in particular alkaline phosphatase (ALP), as a surrogate marker of disease activity, for phenotypic stratification in PBC using a computational modelling approach. Our aim here was to identify distinct disease subgroups of patients with distinct disease trajectories. Methods: We used longitudinal ALP results from 1,601 PBC patients on first line treatment with UDCA, and applied latent class mixed modelling (LCMM), to identify distinct phenotypic subgroups, each with distinct disease trajectories, and risks of end stage liver disease (ESLD). Results: We identified four well discriminated phenotypic subgroups within our PBC cohort, each with distinct disease trajectories.

stat.AP↗

arXiv2017

Two-part models with stochastic processes for modelling longitudinal semicontinuous data: computationally efficient inference and modelling the overall marginal mean

Several researchers have described two-part models with patient-specific stochastic processes for analysing longitudinal semicontinuous data. In theory, such models can offer greater flexibility than the standard two-part model with patient-specific random effects. However, in practice the high dimensional integrations involved in the marginal likelihood (i.e. integrated over the stochastic processes) significantly complicates model fitting. Thus non-standard computationally intensive procedures based on simulating the marginal likelihood have so far only been proposed. In this paper, we describe an efficient method of implementation by demonstrating how the high dimensional integrations involved in the marginal likelihood can be computed efficiently. Specifically, by using a property of the multivariate normal distribution and the standard marginal cumulative distribution function identity, we transform the marginal likelihood so that the high dimensional integrations are contained in the cumulative distribution function of a multivariate normal distribution, which can then be efficiently evaluated. Hence maximum likelihood estimation can be used to obtain parameter estimates and asymptotic standard errors (from the observed information matrix) of model parameters. We describe our proposed efficient implementation procedure for the standard two-part model parameterisation and when it is of interest to directly model the overall marginal mean. The methodology is applied on a psoriatic arthritis data set concerning functional disability.

stat.AP↗