Giampiero Marra

Low birth weight and preterm birth are key indicators of neonatal health, influencing both immediate and long‐term infant outcomes. While low birth weight may reflect fetal growth restrictions, preterm birth captures disruptions in gestational development. Ignoring the potential interdependence between these variables may lead to an incomplete understanding of their shared determinants and underlying dynamics. To address this, a copula distributional regression framework is adopted to jointly model both indicators as flexible functions of maternal characteristics and geographic effects. Applied to female birth data from North Carolina, the methodology identifies shared factors of low birth weight and preterm birth, and reveals how maternal health, socioeconomic conditions and geographic disparities shape neonatal risk. The joint modeling approach provides a more nuanced understanding of these birth metrics, offering insights that can inform targeted interventions, prenatal care strategies and public health planning.

Doctor and non‐doctor visit frequencies are key indicators of healthcare access, utilization and individual health‐seeking behavior. While doctor visits reflect engagement with formal medical services, non‐doctor visits, such as to nurses, physiotherapists or alternative providers, offer insights into patient preferences and system adaptability. Modeling these outcomes separately can hide relevant interdependencies and hence lead to incomplete conclusions. To address this, we employ a copula additive distributional regression framework to jointly model doctor and non‐doctor visits as flexible functions of demographic, socioeconomic and health‐related covariates. The estimation approach allows all the distributional parameters, including location, scale and the dependence structure, to vary with covariates via additive predictors. Application of the model to data from the 2012 Medical Expenditure Panel Survey reveals key determinants of physician and non‐physician visits, such as age, income and health status. Importantly, the method allows for the modeling of shared unobserved heterogeneity and effectively captures how changes in one type of utilization influence the other, thereby yielding a deeper understanding of healthcare behavior.

This paper introduces a copula additive distributional regression framework for mixed count–continuous outcomes, with a focus on simultaneous modelling of the number and cost of physician visits. Traditionally analysed separately, these outcomes are inherently interdependent, and modelling them jointly uncovers relationships that would otherwise be overlooked. The approach employs a zero-truncated count distribution for visit frequency, a continuous skewed distribution with positive support for costs, allows for flexible covariate effects through additive predictors and captures the dependence between the responses using copulae. Alongside model development, model-based statistics, such as conditional expectations, are derived to improve interpretability. Applied to Medical Expenditure Panel Survey data, the methodology provides valuable insights into the determinants of visit frequency and healthcare costs, as well as their association, highlighting its potential as a useful tool for decision-makers. To enhance reproducibility and transparency, the modelling framework has been incorporated into the R package GJRM.

Advancements in single‐cell RNA‐sequencing (scRNA‐seq) technologies generate a wealth of gene expression data that provide exciting opportunities for studying gene‐gene interactions systematically at individual cell resolution. Genetic interactions within a cell are tightly regulated and often highly dynamic in response to internal cellular signals and external stimuli. Evidence of these dynamic interactions can often be observed in scRNA‐seq data by examining conditional co‐expression changes. Existing approaches for studying these dynamic interaction changes in scRNA‐seq data do not address the multi‐subject hierarchical design commonly considered in single‐cell experiments. In this paper, we propose a Mixed‐effects framework for differential Coexpression and transcriptional interaction modeling in Single‐Cell RNA‐seq (scCOSMiX) to account for the cell‐cell correlation from the same individual. The proposed copula‐based approach allows the zero‐inflation, marginal, and association parameters to be modeled as functions of covariates with subject‐level random effects, to enable analyses to be tailored to the data under consideration. A series of simulation analyses were conducted to evaluate and compare the performance of scCOSMiX to other existing approaches. We applied the proposed method to both droplet and plate‐based scRNA‐seq data sets GSE266919 and GSE108989 to illustrate its applicability across distinct scRNA‐seq experimental protocols.

We introduce a general framework for latent variable modeling, named Generalized Latent Variable Models for Location, Scale, and Shape parameters (GLVM-LSS). This framework extends the generalized linear latent variable model beyond the exponential family distributional assumption and enables the modeling of distributional parameters other than the mean (location parameter), such as scale and shape parameters, as functions of latent variables. Model parameters are estimated via maximum likelihood. We present two real-world applications on public opinion research and educational testing, and evaluate the model’s performance in terms of parameter recovery through extensive simulation studies. Our results suggest that the GLVM-LSS is a valuable tool in applications where modeling higher-order moments of the observed variables through latent variables is of substantive interest. The proposed model is implemented in the R package glvmlss, available online.

The rapid advancements of single-cell RNA sequencing (scRNAseq) technology provide high-resolution views of transcriptomic activity within a single cell. Most routine analyses of scRNAseq data focus on individual genes; however, the one-gene-at-a-time analysis is likely to miss meaningful genetic interactions. Gene co-expression analysis addresses this issue by identifying coordinated gene expression changes in response to cellular conditions, such as developmental or temporal trajectory. Identifying differential co-expression gene combinations along the cell temporal trajectory using scRNAseq data can provide deeper insight into the biological processes. Existing approaches for gene co-expression analysis assume a restrictive linear change of gene co-expression. In this paper, we propose a copula-based approach with proper data-driven smoothing functions to model non-linear gene co-expression changes along cellular temporal trajectories. Our proposed approach provides flexibility to incorporate characteristics such as over-dispersion and zero-inflation rate observed in scRNAseq data into the modeling framework. We conducted a series of simulation analyses to evaluate the performance of the proposed algorithm. We demonstrate the implementation of the proposed algorithm using a scRNAseq dataset and identify differential co-expression gene pairs along cell temporal trajectory in pituitary embryonic development comparing Nxn ^{− / −} mutated versus wild-type mice.

Mains libres est la revue de formation continue de référence pour les médecins praticiens.