Transfer learning has been highlighted as a promising framework to increase the accuracy of the data-driven model in the case of data sparsity, specifically by leveraging pretrained knowledge to the training of the target model. The objective of this study is to evaluate whether the number of requisite training samples can be reduced with the use of various transfer learning models for predicting, for example, the chemical source terms of the data-driven reduced-order modeling (ROM) that represents the homogeneous ignition of a hydrogen/air mixture. Principal component analysis is applied to reduce the dimensionality of the hydrogen/air mixture in composition space. Artificial neural networks (ANNs) are used to regress the reaction rates of principal components, and subsequently, a system of ordinary differential equations is solved. As the number of training samples decreases in the target task, the ROM fails to predict the ignition evolution of a hydrogen/air mixture. Three transfer learning strategies are then applied to the training of the ANN model with a sparse dataset. The performance of the ROM with a sparse dataset is remarkably enhanced if the training of the ANN model is restricted by a regularization term that controls the degree of knowledge transfer from source to target tasks. To this end, a novel transfer learning method is introduced, Parameter control via Partial Initialization and Regularization (PaPIR), whereby the amount of knowledge transferred is systemically adjusted in terms of the initialization and regularization schemes of the ANN model in the target task.

This study aimed to develop a predictive tool for identifying individuals with high antibody titers crucial for recruiting COVID-19 convalescent plasma (CCP) donors and to assess the quality and storage changes of CCP. A convenience sample of 110 plasma donors was recruited, of which 75 met the study criteria. Using univariate logistic regression and random forest, 6 significant factors were identified, leading to the development of a nomogram. Receiver operating characteristic curves, calibration plots, and decision curve analysis (DCA) evaluated the nomogram’s discrimination, calibration, and clinical utility. The nomogram indicated that females aged 18 to 26, blood type O, receiving 1 to 2 COVID-19 vaccine doses, experiencing 2 symptoms during infection, and donating plasma 41 to 150 days after symptom onset had higher likelihoods of high antibody titres. Nomogram’s AUC was 0.853 with good calibration. DCA showed clinical benefit within 9% ~ 90% thresholds. CCP quality was qualified, with stable antibody titres over 6 months (P > 0.05). These findings highlight developing predictive tools to identify suitable CCP donors and emphasize the stability of CCP quality over time, suggesting its potential for long-term storage.

Despite global efforts to end tuberculosis (TB), the goal of preventing catastrophic health expenditure (CHE) due to TB remains unmet. This cross-sectional study was conducted in Guizhou Province, Southwest China. Data were collected from the Hospital Information System and a survey of TB patients who had completed standardized antituberculosis treatment between January and March 2021. Among the 2 283 participants, the average total expenditure and out-of-pocket expenditure were $1 506.6 (median = $760.5) and $683.6 (median = $437.8), respectively. Health insurance reimbursement reduced CHE by 16.8%, with a contribution rate of 24.9%, and the concentration index changed from -0.070 prereimbursement to -0.099 postreimbursement. However, the contribution of health insurance varied significantly across different economic strata, with contribution rates of 6.4% for the lowest economic group and 53.1% for the highest group. For patients from lower socioeconomic strata, health insurance contributed 10.7% to CHE in the prediagnostic phase and 23.5% during treatment. While social health insurance alleviated the financial burden for TB patients, it did not provide sufficient protection for those in lower economic strata or during the prediagnostic stage. This study underscores the need for more effective and equitable subsidy policies for TB patients .

We show that the twin-width of every $n$-vertex $d$-regular graph is at most $n^{\frac{d-2}{2d-2}+o(1)}$ for any fixed integer $d \geq 2$ and that almost all $d$-regular graphs attain this bound. More generally, we obtain bounds on the twin-width of sparse Erdős–Renyi and regular random graphs, complementing the bounds in the denser regime due to Ahn, Chakraborti, Hendrey, Kim, and Oum.

High prevalence of long COVID symptoms has emerged as a significant public health concern. This study investigated the associations between three doses of COVID-19 vaccines and the presence of any and ≥3 types of long COVID symptoms among people with a history of SARS-CoV-2 infection in Hong Kong, China. This is a secondary analysis of a cross-sectional online survey among Hong Kong adult residents conducted between June and August 2022. This analysis was based on a sub-sample of 1,542 participants with confirmed SARS-CoV-2 infection during the fifth wave of COVID-19 outbreak in Hong Kong (December 2021 to April 2022). Among the participants, 40.9% and 16.1% self-reported having any and ≥3 types of long COVID symptoms, respectively. After adjusting for significant variables related to sociodemographic characteristics, health conditions and lifestyles, and SARS-CoV-2 infection, receiving at least three doses of COVID-19 vaccines was associated with lower odds of reporting any long COVID symptoms comparing to receiving two doses (adjusted odds ratio [AOR]: 0.69, 95% CI: 0.54, 0.87, P = .002). Three doses of inactivated and mRNA vaccines had similar protective effects against long COVID symptoms. It is important to strengthen the coverage of COVID-19 vaccination booster doses, even in the post-pandemic era.

Let $T$ be a tree on $t$ vertices. We prove that for every positive integer $k$ and every graph $G$, either $G$ contains $k$ pairwise vertex-disjoint subgraphs each having a $T$ minor, or there exists a set $X$ of at most $t(k-1)$ vertices of $G$ such that $G-X$ has no $T$ minor. The bound on the size of $X$ is best possible and improves on an earlier $f(t)k$ bound proved by Fiorini, Joret, and Wood (2013) with some fast-growing function $f(t)$. Moreover, our proof is short and simple.

Cervical cancer, closely linked to human papillomavirus (HPV) infection, is a major global health concern. Our study aims to fill the gap in understanding HPV vaccine awareness and acceptance in the Middle East, where national immunization programs are often lacking and cultural perceptions hinder acceptance. This systematic review and meta-analysis adhered to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. A comprehensive literature search across several databases was conducted on 5 September 2023. We included quantitative studies on HPV vaccine awareness and acceptance in Middle Eastern countries. Data extraction and quality assessment were conducted independently by multiple reviewers to ensure accuracy. Statistical analyses, including subgroup analyses, were performed using R to calculate pooled estimates, assess heterogeneity, and publication bias. We reviewed 159 articles from 15 Middle Eastern countries, focusing on 93,730 participants, predominantly female and healthcare workers. HPV vaccine awareness was found to be 41.7% (95% CI 37.4%–46.1%), with higher awareness among healthcare workers. The pooled acceptance rate was 45.6% (95% CI 41.3%–50.1%), with similar rates between healthcare and non-healthcare workers. Our study highlights the critical need for increased HPV vaccine awareness and acceptance in the Middle East, emphasizing the importance of integrating the vaccine into national immunization programs and addressing cultural and religious factors to improve public health outcomes.

The expanding application of advanced analytics in insurance has generated numerous opportunities, such as more accurate predictive modeling powered by machine learning and artificial intelligence (AI) methods, the utilization of novel and unstructured datasets, and the automation of key operations. Significant advances in these areas are being made through novel applications and adaptations of predictive modeling techniques for insurance purposes, while, concurrently, rapid advances in machine learning methods are being made outside of the insurance sector. However, these innovations also bring substantial challenges, particularly around the transparency, explanation, and fairness of complex algorithmic models and the economic and societal impacts of their adoption in decision-making. As insurance is a highly regulated industry, models may be required by regulators to be explainable, in order to enable analysis of the basis for decision making. Due to the societal importance of insurance, significant attention is being paid to ensuring that insurance models do not discriminate unfairly. In this special issue, we feature papers that explore key issues in insurance analytics, focusing on prediction, explainability, and fairness.

This article examines the National Health Data Network (RNDS), the platform launched by the Ministry of Health in Brazil as the primary tool for its Digital Health Strategy 2020–2028, including innovation aspects. The analysis is made through two distinct frameworks: Right to health and personal data protection in Brazil. The first approach is rooted in the legal framework shaped by Brazil’s trajectory on health since 1988, marked by the formal acknowledgment of the Right to health and the establishment of the Unified Health System, Brazil’s universal access health system, encompassing public healthcare and public health actions. The second approach stems from the repercussions of the General Data Protection Law, enacted in 2018 and the inclusion of Right to personal data protection in Brazilian’s Constitution. This legislation, akin to the EU’s General Data Protection Regulations, addressed the gap in personal data protection in Brazil and established principles and rules for data processing. The article begins by explanting the two approaches, and then it provides a brief history of health informatics policies in Brazil, leading to the current Digital Health Strategy and the RNDS. Subsequently, it delves into an analysis of the RNDS through the lenses of the two aforementioned approaches. In the final discussion sections, the article attempts to extract lessons from the analyses, particularly in light of ongoing discussions such as the secondary use of data for innovation in the context of different interpretations about innovation policies.

The embedding problem of Markov chains examines whether a stochastic matrix$\mathbf{P} $ can arise as the transition matrix from time 0 to time 1 of a continuous-time Markov chain. When the chain is homogeneous, it checks if $ \mathbf{P}=\exp{\mathbf{Q}}$ for a rate matrix $ \mathbf{Q}$ with zero row sums and non-negative off-diagonal elements, called a Markov generator. It is known that a Markov generator may not always exist or be unique. This paper addresses finding $ \mathbf{Q}$, assuming that the process has at most one jump per unit time interval, and focuses on the problem of aligning the conditional one-jump transition matrix from time 0 to time 1 with $ \mathbf{P}$. We derive a formula for this matrix in terms of $ \mathbf{Q}$ and establish that for any $ \mathbf{P}$ with non-zero diagonal entries, a unique $ \mathbf{Q}$, called the ${\unicode{x1D7D9}}$-generator, exists. We compare the ${\unicode{x1D7D9}}$-generator with the one-jump rate matrix from Jarrow, Lando, and Turnbull (1997), showing which is a better approximate Markov generator of $ \mathbf{P}$ in some practical cases.

Turbulent flows are chaotic and multi-scale dynamical systems, which have large numbers of degrees of freedom. Turbulent flows, however, can be modeled with a smaller number of degrees of freedom when using an appropriate coordinate system, which is the goal of dimensionality reduction via nonlinear autoencoders. Autoencoders are expressive tools, but they are difficult to interpret. This article aims to propose a method to aid the interpretability of autoencoders. First, we introduce the decoder decomposition, a post-processing method to connect the latent variables to the coherent structures of flows. Second, we apply the decoder decomposition to analyze the latent space of synthetic data of a two-dimensional unsteady wake past a cylinder. We find that the dimension of latent space has a significant impact on the interpretability of autoencoders. We identify the physical and spurious latent variables. Third, we apply the decoder decomposition to the latent space of wind-tunnel experimental data of a three-dimensional turbulent wake past a bluff body. We show that the reconstruction error is a function of both the latent space dimension and the decoder size, which are correlated. Finally, we apply the decoder decomposition to rank and select latent variables based on the coherent structures that they represent. This is useful to filter unwanted or spurious latent variables or to pinpoint specific coherent structures of interest. The ability to rank and select latent variables will help users design and interpret nonlinear autoencoders.

Usage data on research outputs such as books and journals is well established in the scholarly community. Yet, as research impact is derived from a broader set of scholarly outputs, such as data, code, and multimedia, more holistic usage and impact metrics could inform national innovation and research policy. While usage data reporting standards, such as Project COUNTER, provide the basis for shared statistics reporting practice, mandated access to publicly funded research has increased the demand for impact metrics and analytics. In this context, stakeholders are exploring how to scaffold and strengthen shared infrastructure to better support the trusted, multistakeholder exchange of usage data across a variety of outputs. In April 2023, a workshop on Exploring National Infrastructure for Public Access and Impact Reporting supported by the United States (US) National Science Foundation (NSF) explored these issues. This paper contextualizes the resources shared and recommendations generated in the workshop.

In Chung–Lu random graphs, a classic model for real-world networks, each vertex is equipped with a weight drawn from a power-law distribution, and two vertices form an edge independently with probability proportional to the product of their weights. Chung–Lu graphs have average distance $O(\log\log n)$ and thus reproduce the small-world phenomenon, a key property of real-world networks. Modern, more realistic variants of this model also equip each vertex with a random position in a specific underlying geometry. The edge probability of two vertices then depends, say, inversely polynomially on their distance.

In this paper we study a generic augmented version of Chung–Lu random graphs. We analyze a model where the edge probability of two vertices can depend arbitrarily on their positions, as long as the marginal probability of forming an edge (for two vertices with fixed weights, one fixed position, and one random position) is as in a Chung–Lu random graph. The resulting class contains Chung–Lu random graphs, hyperbolic random graphs, and geometric inhomogeneous random graphs as special cases.

Our main result is that every random graph model in this general class has the same average distance as a Chung–Lu random graph, up to a factor of $1+o(1)$. This shows in particular that specific choices, such as taking the underlying geometry to be Euclidean, do not significantly influence the average distance. The proof also shows that every random graph model in our class has a giant component and polylogarithmic diameter with high probability.

We consider the performance of Glauber dynamics for the random cluster model with real parameter $q\gt 1$ and temperature $\beta \gt 0$. Recent work by Helmuth, Jenssen, and Perkins detailed the ordered/disordered transition of the model on random $\Delta$-regular graphs for all sufficiently large $q$ and obtained an efficient sampling algorithm for all temperatures $\beta$ using cluster expansion methods. Despite this major progress, the performance of natural Markov chains, including Glauber dynamics, is not yet well understood on the random regular graph, partly because of the non-local nature of the model (especially at low temperatures) and partly because of severe bottleneck phenomena that emerge in a window around the ordered/disordered transition. Nevertheless, it is widely conjectured that the bottleneck phenomena that impede mixing from worst-case starting configurations can be avoided by initialising the chain more judiciously. Our main result establishes this conjecture for all sufficiently large $q$ (with respect to $\Delta$). Specifically, we consider the mixing time of Glauber dynamics initialised from the two extreme configurations, the all-in and all-out, and obtain a pair of fast mixing bounds which cover all temperatures $\beta$, including in particular the bottleneck window. Our result is inspired by the recent approach of Gheissari and Sinclair for the Ising model who obtained a similar flavoured mixing-time bound on the random regular graph for sufficiently low temperatures. To cover all temperatures in the RC model, we refine appropriately the structural results of Helmuth, Jenssen and Perkins about the ordered/disordered transition and show spatial mixing properties ‘within the phase’, which are then related to the evolution of the chain.

For coherent systems with components and active redundancies having heterogeneous and dependent lifetimes, we prove that the lifetime of system with redundancy at component level is stochastically larger than that with redundancy at system level. In particular, in the setting of homogeneous components and redundancy lifetimes linked by an Archimedean survival copula, we develop sufficient conditions for the reversed hazard rate order, the hazard rate order and the likelihood ratio order between two system lifetimes, respectively. The present results substantially generalize some related results in the literature. Several numerical examples are presented to illustrate the findings as well.

Can trust norms within the African moral system support data gathering for Generative AI (GenAI) development in African society? Recent developments in the field of large language models, such as GenAI, including models like ChatGPT and Midjourney, have identified a common issue with these GenAI models known as “AI hallucination,” which involves the presentation of misinformation as facts along with its potential downside of facilitating public distrust in AI performance. In the African context, this paper frames unsupportive data-gathering norms as a contributory factor to issues such as AI hallucination and investigates the following claims. First, this paper explores the claim that knowledge in the African context exists in both esoteric and exoteric forms, incorporating such diverse knowledge as data could imply that a GenAI tailored for Africa may have unlimited accessibility across all contexts. Second, this paper acknowledges the formidable challenge of amassing a substantial volume of data, which encompasses esoteric information, requisite for the development of a GenAI model, positing that the establishment of a foundational framework for data collection, rooted in trust norms that is culturally resonant, has the potential to engender trust dynamics between data providers and collectors. Lastly, this paper recommends that trust norms in the African context require recalibration to align with contemporary social progress, while preserving their core values, to accommodate innovative data-gathering methodologies for a GenAI tailored to the African setting. This paper contributes to how trust culture within the African context, particularly in the domain of GenAI for African society, propels the development of Afro-AI technologies.

Nontyphoidal Salmonella enterica infections are a leading cause of enteric disease in Canada, most commonly associated with foodborne exposures. Raw frozen breaded chicken products (FBCP) have been implicated in 16 Salmonella outbreaks between 2017 and 2019. This study quantified the impact of the 1 April 2019 requirement by the Canadian Food Inspection Agency (CFIA) for manufacturers to reduce Salmonella in raw FBCP. An intervention study approach utilizing the pre–post intervention data with a comparison group methodology was used to: (1) estimate the reduction in FBCP Salmonella prevalence using retail meat FoodNet Canada data; (2) estimate the reduction in the human salmonellosis incidence rate using data from the Canadian National Enteric Surveillance Program; and (3) estimate the proportion of reported cases attributed to FBCP if the human exposure to Salmonella through FBCP was completely eliminated. The FBCP Salmonella prevalence decreased from 28% observed before 1 April 2019 to 2.9% after the requirement implementation. The CFIA requirement was estimated to reduce the human salmonellosis incidence rate by 23%. An estimated 26% of cases during the pre-intervention period can be attributed to FBCP. The CFIA requirement was successful at significantly reducing Salmonella prevalence in retail FBCP, and at reducing salmonellosis burden.