By integrating cutting-edge statistical research with diverse applications, this book serves as both a reference and an inspiration for those interested in advancing Bayesian methodologies. This volume brings together a collection of research contributions that highlight the versatility and power of Bayesian methods in tackling complex problems across a variety of fields. The chapters reflect the latest advances in Bayesian theory, methodology, and computation, offering novel approaches to analyze data characterized by high dimensionality, structural dependencies, and dynamic behavior. From segmenting mass spectrometry imaging data to modeling dynamic networks and assessing macroeconomic tail risks, this book showcases how advanced Bayesian methods can provide transformative insights while maintaining interpretability and computational feasibility. Whether its addressing challenges in biomedicine, where data often come with hierarchical structures and non-standard distributions, or in economics, where time-varying risks demand adaptive models, the contributions in this book demonstrate the unparalleled capacity of Bayesian methods to model, predict, and interpret complex phenomena. Importantly, they also address the need for theoretical guarantees and computational efficiency, making these methods accessible for real-world applications. This volume highlights the versatility of Bayesian methods in tackling diverse, complex problems across disciplines. The chapters reflect the latest advances in statistical theory, computational techniques, and real-world applications. Readers will find innovative solutions for high-dimensional data analysis, clinical trial design, dynamic network modeling, macroeconomic risk assessment, and more. By integrating theory and practice, this book serves as a valuable resource for statisticians, researchers, and practitioners seeking to explore the frontiers of Bayesian inference.
The volume gathers contributions presented at the Bayesian Young Statisticians Meeting (BAYSM) 2023, the official conference of j-ISBA, the junior section of the International Society for Bayesian Analysis, together with some more invited papers from additional contributors. This prestigious event provides a platform for early-career researchers to showcase innovative work and engage in discussions that shape the future of Bayesian statistics. The inclusion of some additional contributions highlights the vibrancy and creativity of the next generation of Bayesian statisticians, offering a glimpse into cutting-edge methodologies and their diverse applications. The discussions and feedback from BAYSM 2023 have undoubtedly enriched these works, underscoring the collaborative and dynamic nature of the Bayesian research community.
Introduction.- F. Denti, C. Balocchi, G. Capitoli, Segmenting Brain
MALDI-MSI Data under Separate Exchangeability.- M. Giordano, A Bayesian
Approach with Gaussian Priors to the Inverse Problem of Source Identification
in Elliptic PDEs.-
M. Chapman-Rounds, M. Pereira, Phase I Dose Escalation Trials in Cancer
Immunotherapy: Modifying the Bayesian Logistic Regression Model for Cytokine
Release Syndrome.- A. Avalos-Pacheco, A. Lazzerini, M. Lupparelli, F. Claudio
Stingo, A Bayesian Multiple Ising Model.- R. H. Mena, M. Ruggiero, A. Singh,
Bayesian Nonparametric Estimation of Time-Varying Macroeconomic Tail Risk.-
M. Dalla Pria, M. Ruggiero, D. Spanņ, A MetropolisHastings Algorithm for
Sampling Coagulated Partitions.- F. Gaffi, Conditionally Partially
Exchangeable Partitions for Dynamic Networks.
Alejandra Avalos Pacheco is a tenure-track Universitätsassistentin at the Institute of Applied Statistics at JKU Linz, Austria, and an affiliated member of the Harvard-MIT Center for Regulatory Science at Harvard University. She earned her PhD in Statistics through the joint CDT program between the University of Warwick and the University of Oxford. Her thesis received the prestigious Savage Award in Applied Methodology. She has held postdoctoral positions at Harvard University and worked at the Dana-Farber Cancer Institute. Additionally, she served as a research fellow at the University of Florence and a non-tenure-track Universitätsassistentin at TU Wien. Her research focuses on creating interpretable, computationally efficient models for large, complex data, particularly in medical applications such as cancer. She specializes in Bayesian and probabilistic machine learning, with expertise in high-dimensional inference, dimensionality reduction, graphical models, data integration and clinical trials. Fan Bu is a tenure-track Assistant Professor in Biostatistics at the University of Michigan. She completed her Ph.D. in Statistics at Duke University and was previously a postdoctoral research fellow at UCLA, where she developed Bayesian methods for large-scale observational health data. Her research spans Bayesian modeling for temporal and spatio-temporal processes, networks, and federated data, with applications in health data science and observational studies for comparative effectiveness and safety and has appeared in leading journals such as the Journal of the American Statistical Association and Statistics in Medicine. An active member of the Observational Health Data Sciences and Informatics (OHDSI) collaborative, Bu contributes to statistical methods development and leads large-scale network studies to improve health decisions and patient care. Beatrice Franzolini is a Researcher at the Bocconi Institute for Data Science and Analytics, Bocconi University, Milan, Italy. She is a statistician specializing in Bayesian statistical theory, methods and application, with a particular focus on Bayesian nonparametrics. Her research encompasses random probability measures, species sampling models, dependent random partitions, and dynamic models. She has published in leading journals such as Biometrika and The Annals of Applied Statistics. Franzolini holds a Ph.D. from Bocconi University and has held research positions at the Agency for Science, Technology, and Research in Singapore, as well as the Division of Biomedical Data Science at the National University of Singapore's medical school. Beniamino Hadj-Amar is a Postdoctoral Fellow in the Department of Statistics at Rice University, Houston, TX. His research focuses on Bayesian methods for analyzing complex dynamical time series, with expertise in latent structure identification, non-stationary and non-linear processes, and sparse data structures. He holds a Ph.D. from the Oxford-Warwick Statistics Programme (OxWaSP). Hadj-Amars methodological toolkit includes switching models, change-point detection, Bayesian nonparametrics, graphical models, and statistical spectral analysis. His work is applied to neuromodulation, respiratory research, and circadian studies, leveraging diverse datasets such as electrophysiological signals, wearable device data, and fMRI. His contributions have appeared in prestigious journals such as the Journal of the American Statistical Association and The Annals of Applied Statistics.
