To draw real-world evidence about the comparative effectiveness of multiple time-varying treatments on patient survival, we develop a joint marginal structural survival model and a novel weighting strategy to account for time-varying confounding and censoring. Our methods formulate complex longitudinal treatments with multiple start/stop switches as the recurrent events with discontinuous intervals of treatment eligibility. We derive the weights in continuous time to handle a complex longitudinal dataset without the need to discretize or artificially align the measurement times. We further use machine learning models designed for censored survival data with time-varying covariates and the kernel function estimator of the baseline intensity to efficiently estimate the continuous-time weights. Our simulations demonstrate that the proposed methods provide better bias reduction and nominal coverage probability when analyzing observational longitudinal survival data with irregularly spaced time intervals, compared to conventional methods that require aligned measurement time points. We apply the proposed methods to a large-scale COVID-19 dataset to estimate the causal effects of several COVID-19 treatments on the composite of in-hospital mortality and ICU admission.
We propose a method for providing communication network infrastructure in autonomous multi-agent teams. In particular, we consider a set of communication agents that are placed alongside regular agents from the system in order to improve the rate of information transfer between the latter. In order to find the optimal positions to place such agents, we define a flexible performance function that adapts to network requirements for different systems. We provide an algorithm based on shadow prices of a related convex optimization problem in order to drive the configuration of the complete system towards a local maximum. We apply our method to three different performance functions associated with three practical scenarios in which we show both the performance of the algorithm and the flexibility it allows for optimizing different network requirements.
Personalized adaptive interventions offer the opportunity to increase patient benefits, however, there are challenges in their planning and implementation. Once implemented, it is an important question whether personalized adaptive interventions are indeed clinically more effective compared to a fixed gold standard intervention. In this paper, we present an innovative N-of-1 trial study design testing whether implementing a personalized intervention by an online reinforcement learning agent is feasible and effective. Throughout, we use a new study on physical exercise recommendations to reduce pain in endometriosis for illustration. We describe the design of a contextual bandit recommendation agent and evaluate the agent in simulation studies. The results show that adaptive interventions add complexity to the design and implementation process, but have the potential to improve patients' benefits even if only few observations are available. In order to quantify the expected benefit, data from previous interventional studies is required. We expect our approach to be transferable to other interventions and clinical interventions.
This paper investigates the problem of estimating the larger location parameter of two general location families from a decision-theoretic perspective. In this estimation problem, we use the criteria of minimizing the risk function and the Pitman closeness under a general bowl-shaped loss function. Inadmissibility of a general location and equivariant estimators is provided. We prove that a natural estimator (analogue of the BLEE of unordered location parameters) is inadmissible, under certain conditions on underlying densities, and propose a dominating estimator. We also derive a class of improved estimators using the Kubokawa's IERD approach and observe that the boundary estimator of this class is the Brewster-Zidek type estimator. Additionally, under the generalized Pitman criterion, we show that the natural estimator is inadmissible and obtain improved estimators. The results are implemented for different loss functions, and explicit expressions for the dominating estimators are provided. We explore the applications of these results to for exponential and normal distribution under specified loss functions. A simulation is also conducted to compare the risk performance of the proposed estimators. Finally, we present a real-life data analysis to illustrate the practical applications of the paper's findings.
In this work, we consider the problem of goodness-of-fit (GoF) testing for parametric models -- for example, testing whether observed data follows a logistic regression model. This testing problem involves a composite null hypothesis, due to the unknown values of the model parameters. In some special cases, co-sufficient sampling (CSS) can remove the influence of these unknown parameters via conditioning on a sufficient statistic -- often, the maximum likelihood estimator (MLE) of the unknown parameters. However, many common parametric settings (including logistic regression) do not permit this approach, since conditioning on a sufficient statistic leads to a powerless test. The recent approximate co-sufficient sampling (aCSS) framework of Barber and Janson (2022) offers an alternative, replacing sufficiency with an approximately sufficient statistic (namely, a noisy version of the MLE). This approach recovers power in a range of settings where CSS cannot be applied, but can only be applied in settings where the unconstrained MLE is well-defined and well-behaved, which implicitly assumes a low-dimensional regime. In this work, we extend aCSS to the setting of constrained and penalized maximum likelihood estimation, so that more complex estimation problems can now be handled within the aCSS framework, including examples such as mixtures-of-Gaussians (where the unconstrained MLE is not well-defined due to degeneracy) and high-dimensional Gaussian linear models (where the MLE can perform well under regularization, such as an $\ell_1$ penalty or a shape constraint).
The recently published ICH E9 addendum on estimands in clinical trials provides a framework for precisely defining the treatment effect that is to be estimated, but says little about estimation methods. Here we report analyses of a clinical trial in type 2 diabetes, targeting the effects of randomised treatment, handling rescue treatment and discontinuation of randomised treatment using the so-called hypothetical strategy. We show how this can be estimated using mixed models for repeated measures, multiple imputation, inverse probability of treatment weighting, G-formula and G-estimation. We describe their assumptions and practical details of their implementation using packages in R. We report the results of these analyses, broadly finding similar estimates and standard errors across the estimators. We discuss various considerations relevant when choosing an estimation approach, including computational time, how to handle missing data, whether to include post intercurrent event data in the analysis, whether and how to adjust for additional time-varying confounders, and whether and how to model different types of ICE separately.
In this paper, we develop a novel efficient and robust nonparametric regression estimator under a framework of feedforward neural network. There are several interesting characteristics for the proposed estimator. First, the loss function is built upon an estimated maximum likelihood function, who integrates the information from observed data, as well as the information from data structure. Consequently, the resulting estimator has desirable optimal properties, such as efficiency. Second, different from the traditional maximum likelihood estimation (MLE), the proposed method avoid the specification of the distribution, hence is flexible to any kind of distribution, such as heavy tails, multimodal or heterogeneous distribution. Third, the proposed loss function relies on probabilities rather than direct observations as in least squares, contributing the robustness in the proposed estimator. Finally, the proposed loss function involves nonparametric regression function only. This enables a direct application of existing packages, simplifying the computation and programming. We establish the large sample property of the proposed estimator in terms of its excess risk and minimax near-optimal rate. The theoretical results demonstrate that the proposed estimator is equivalent to the true MLE in which the density function is known. Our simulation studies show that the proposed estimator outperforms the existing methods in terms of prediction accuracy, efficiency and robustness. Particularly, it is comparable to the true MLE, and even gets better as the sample size increases. This implies that the adaptive and data-driven loss function from the estimated density may offer an additional avenue for capturing valuable information. We further apply the proposed method to four real data examples, resulting in significantly reduced out-of-sample prediction errors compared to existing methods.
To improve the robustness of transformer neural networks used for temporal-dynamics prediction of chaotic systems, we propose a novel attention mechanism called easy attention which we demonstrate in time-series reconstruction and prediction. As a consequence of the fact that self attention only makes useof the inner product of queries and keys, it is demonstrated that the keys, queries and softmax are not necessary for obtaining the attention score required to capture long-term dependencies in temporal sequences. Through implementing singular-value decomposition (SVD) on the softmax attention score, we further observe that the self attention compresses contribution from both queries and keys in the spanned space of the attention score. Therefore, our proposed easy-attention method directly treats the attention scores as learnable parameters. This approach produces excellent results when reconstructing and predicting the temporal dynamics of chaotic systems exhibiting more robustness and less complexity than the self attention or the widely-used long short-term memory (LSTM) network. Our results show great potential for applications in more complex high-dimensional dynamical systems. Keywords: Machine Learning, Transformer, Self Attention, Koopman Operator, Chaotic System.
Health-related quality of life (Hr-QoL) scales provide crucial information on neurodegenerative disease progression, help improving patient care, and constitute a meaningful endpoint for therapeutic research. However, Hr-QoL progression is usually poorly documented, as for multiple system atrophy (MSA), a rare and rapidly progressing alpha-synucleinopathy. This work aimed to describe Hr-QoL progression during the natural course of MSA, explore disparities between patients, and identify informative items using a four-step statistical strategy.We leveraged the data of the French MSA cohort comprising annual assessments with the MSA-QoL questionnaire for more than 500 patients over up to 11 years. The four-step strategy (1) determined the subdimensions of Hr-QoL in MSA; (2) modelled the subdimension trajectories over time, accounting for the risk of death; (3) mapped the sequence of item impairments with disease stages; and (4) identified the most informative items specific to each disease stage.Among the 536 patients included, 50% were women and they were aged on average 65.1 years old at entry. Among them, 63.1% died during the follow-up. Four dimensions were identified. In addition to the original motor, nonmotor, and emotional domains, an oropharyngeal component was highlighted. While the motor and oropharyngeal domains deteriorated rapidly, the nonmotor and emotional aspects were already slightly to moderately impaired at cohort entry and deteriorated slowly over the course of the disease. Impairments were associated with sex, diagnosis subtype, and delay since symptom onset. Except for the emotional domain, each dimension was driven by key identified items.Hr-QoL is a multidimensional concept that deteriorates progressively over the course of MSA and brings essential knowledge for improving patient care. As exemplified with MSA, the thorough description of Hr-QoL using the 4-step original analysis can provide new perspectives on neurodegenerative diseases' management to ultimately deliver better support focused on the patient's perspective.
In the present study, we investigate and compare reasoning in large language models (LLM) and humans using a selection of cognitive psychology tools traditionally dedicated to the study of (bounded) rationality. To do so, we presented to human participants and an array of pretrained LLMs new variants of classical cognitive experiments, and cross-compared their performances. Our results showed that most of the included models presented reasoning errors akin to those frequently ascribed to error-prone, heuristic-based human reasoning. Notwithstanding this superficial similarity, an in-depth comparison between humans and LLMs indicated important differences with human-like reasoning, with models limitations disappearing almost entirely in more recent LLMs releases. Moreover, we show that while it is possible to devise strategies to induce better performance, humans and machines are not equally-responsive to the same prompting schemes. We conclude by discussing the epistemological implications and challenges of comparing human and machine behavior for both artificial intelligence and cognitive psychology.
Context: Software start-ups are young companies aiming to build and market software-intensive products fast with little resources. Aiming to accelerate time-to-market, start-ups often opt for ad-hoc engineering practices, make shortcuts in product engineering, and accumulate technical debt. Objective: In this paper we explore to what extent precedents, dimensions and outcomes associated with technical debt are prevalent in start-ups. Method: We apply a case survey method to identify aspects of technical debt and contextual information characterizing the engineering context in start-ups. Results: By analyzing responses from 86 start-up cases we found that start-ups accumulate most technical debt in the testing dimension, despite attempts to automate testing. Furthermore, we found that start-up team size and experience is a leading precedent for accumulating technical debt: larger teams face more challenges in keeping the debt under control. Conclusions: This study highlights the necessity to monitor levels of technical debt and to preemptively introduce practices to keep the debt under control. Adding more people to an already difficult to maintain product could amplify other precedents, such as resource shortages, communication issues and negatively affect decisions pertaining to the use of good engineering practices.