Model-assisted estimators for time-to-event data from complex surveys

We develop model-assisted estimators for complex survey data for the proportion of a population that experienced some event by a specified time t. Theory for the new estimators uses time-to-event models as the underlying framework but have both good model-based and design-based properties. The estimators are compared in a simulation to traditional survey estimation methods and are also applied to a study of nurses' health. The new estimators take advantage of covariates predictive of the event and reduce standard errors compared to conventional alternatives.     

Subthalamic nucleus deep brain stimulation improves sleep in Parkinson's disease patients: a retrospective study and a meta-analysis

BackgroundMost Parkinson's patients suffered from sleep problems. There is increasing evidence that Subthalamic Nucleus Deep Brain Stimulation (STN-DBS) has a positive effect on several sleep parameters, improving overall sleep quality in patients with PD. However, the results are controversial.MethodsWe performed a retrospective study and meta-analysis to assess the Parkinson's disease sleep scale (PDSS) in Parkinson's patients.ResultsWe reviewed our data of patients who underwent STN-DBS, and then extracted five other trials to perform a meta-analysis. The pooled results showed an advantage on post-operative PDSS in both our medical center and pooled results (MD = 20.41, 95% CI = [13.03, 27.79], I² = 61%, P < 0.001). There was a significant difference in Unified Parkinson's Disease Rating Scale (UPDRS)-Ⅲ score between pre and post-operation (MD = −12.59, 95% CI = [−14.70, −10.49], I² = 90%, P < 0.001). What's more, Parkinsonian medication was significantly lower in the post-operative groups after DBS (MD = −314.71, 95% CI = [−468.13, −161.28], I² = 53%, P < 0.001).ConclusionIn the retrospective study and meta-analysis of 6 trials, we found that DBS can significantly increase sleep quality. Furthermore, motor function improved and Parkinsonian medication was significantly decreased postoperatively. The sample size was enough and no further investigations would change the conclusion.     

Learning to Discretize: Solving 1D Scalar Conservation Laws via Deep Reinforcement Learning

Conservation laws are considered to be fundamental laws of nature. It has broad applications in many fields, including physics, chemistry, biology, geology, and engineering. Solving the differential equations associated with conservation laws is a major branch in computational mathematics. The recent success of machine learning, especially deep learning in areas such as computer vision and natural language processing, has attracted a lot of attention from the community of computational mathematics and inspired many intriguing works in combining machine learning with traditional methods. In this paper, we are the first to view numerical PDE solvers as an MDP and to use (deep) RL to learn new solvers. As proof of concept, we focus on 1-dimensional scalar conservation laws. We deploy the machinery of deep reinforcement learning to train a policy network that can decide on how the numerical solutions should be approximated in a sequential and spatial-temporal adaptive manner. We will show that the problem of solving conservation laws can be naturally viewed as a sequential decision-making process, and the numerical schemes learned in such a way can easily enforce long-term accuracy. Furthermore, the learned policy network is carefully designed to determine a good local discrete approximation based on the current state of the solution, which essentially makes the proposed method a meta-learning approach. In other words, the proposed method is capable of learning how to discretize for a given situation mimicking human experts. Finally, we will provide details on how the policy network is trained, how well it performs compared with some state-of-the-art numerical solvers such as WENO schemes, and supervised learning based approach L3D and PINN, and how well it generalizes.     

Detection of lactate in the striatum without contamination of macromolecules by J‐difference editing MRS at 7 T

Lactate levels are measurable by MRS and are related to neural activity. Therefore, it is of interest to accurately measure lactate levels in the basal ganglia networks. If sufficiently stable, lactate measurements may be used to investigate alterations in dopaminergic signalling in the striatum, facilitating the detection and diagnosis of metabolic deficits. The aim of this study is to provide a J‐difference editing MRS technique for the selective editing of lactate only, thus allowing the detection of lactate without contamination of overlapping macromolecules. As a validation procedure, macromolecule nulling was combined with J‐difference editing, and this was compared with J‐difference editing with a new highly selective editing pulse. The use of a high‐field (7T) MR scanner enables the application of editing pulses with very narrow bandwidth, which are selective for lactate. We show that, despite the sensitivity to B₀ offsets, the use of a highly selective editing pulse is more efficient for the detection of lactate than the combination of a broad‐band editing pulse with macromolecule nulling. Although the signal‐to‐noise ratio of uncontaminated lactate detection in healthy subjects is relatively low, this article describes the test–retest performance of lactate detection in the striatum when using highly selective J‐difference editing MRS at 7 T. The coefficient of variation, σ_w and intraclass correlation coefficients for within‐ and between‐subject differences of lactate were determined. Lactate levels in the left and right striatum were determined twice in 10 healthy volunteers. Despite the fact that the test–retest performance of lactate detection is moderate with a coefficient of variation of about 20% for lactate, these values can be used for the design of new studies comparing, for example, patient populations with healthy controls. Copyright © 2015 John Wiley & Sons, Ltd.     

Glioblastoma in England: 2007–2011

AimsGlioblastoma (GBM) is the most common and aggressive primary malignant brain tumour in adults, with a poor prognosis. Changing treatment paradigms suggest improved outcome, but whole nation data for England is scarce. The aim of this report is to examine the incidence of patients with glioblastoma in England, and to assess the influence of gender, age, geographical region and treatment on outcome.MethodsA search strategy encompassing all patients coded with GBM and treated from January 2007 to December 2011 was obtained from data linkage between the National Cancer Registration Service and Hospital Episode Statistics for England.ResultsThere were 10,743 patients coded with GBM in this 5-year period (6451 male, 4292 female), giving an overall national age standardised incidence of 4.64/100,000/year. Incidence increases with age. Median survival overall was 6.1 months. One, 2 and 5-year survivals, were 28.4%, 11.5% and 3.4% respectively. Age stratified median survivals decreased significantly (p < 0.0001) with increasing age from 16.2 months for the 20–44 year age group, to 7.9 months for the 45–69 years, and 3.2 months for 70+ years. In the maximal treatment subgroup, patients aged up to 69 years had a median survival of 14.9 months. Patients over 60 years were less likely to receive maximal combination treatment but median survival was better with maximal treatment at all ages.ConclusionsThe overall outcome for patients with GBM remains poor. However, aggressive treatment at every age group is associated with extended survival similar to that described in clinical trials.     

Opinion about seasonal influenza vaccination among the general population 3 years after the A(H1N1)pdm2009 influenza pandemic

ObjectiveTo assess the opinions of the French general population about seasonal influenza vaccination three years after the A(H1N1)pdm 09 pandemic and identify factors associated with a neutral or negative opinion about this vaccination.Study designThe study was conducted using data collected from 5374 participants during the 2012/2013 season of the GrippeNet.fr study. The opinion about seasonal influenza vaccination was studied on three levels (“positive”, “negative” or “neutral”). The link between the participant's characteristics and their opinion regarding the seasonal influenza vaccination were studied using a multinomial logistic regression with categorical variables. The “positive” opinion was used as the reference for identifying individuals being at risk of having a “neutral” or a “negative” opinion.ResultsAmong the participants, 39% reported having a positive opinion about seasonal influenza vaccine, 39% a neutral opinion, and 22% a negative opinion. Factors associated with a neutral or negative opinion were young age, low educational level, lack of contact with sick or elderly individuals, lack of treatment for a chronic disease and taking a homeopathic preventive treatment.ConclusionsThese results show that an important part of the French population does not have a positive opinion about influenza vaccination in France. Furthermore, it allows outlining the profiles of particularly reluctant individuals who could be targeted by informative campaigns.     

The response to nutritional labels: Evidence from a quasi-experiment

This paper evaluates a UK policy that aimed to improve dietary information provision by introducing nutrition labelling on retailers’ store-brand products. Exploiting the differential timing of the introduction of Front-of-Pack nutrition labels as a quasi-experiment, our findings suggest that labelling led to a reduction in the quantity purchased of labelled store-brand foods, and an improvement in their nutritional composition. More specifically, we find that households reduced the total monthly calories from labelled store-brand foods by 588 kcal, saturated fats by 14 g, sugars by 7 g, and sodium by 0.8 mg.