S3M1P4R

Objective:

"The S3M1P4R project focuses on the development of new statistical methods within the framework of semi-parametric regression in order to address several challenging statistical problems, with a particular interest in applications in different fields such as health and medicine, sports science and the impact of climate change, as well as the development of open source software for the scientific community. The research team consists of mathematicians and applied statisticians with a strong background in theory, computation and applied research in the development of statistical methods that will be used to provide advanced solutions in other scientific fields. The team is composed of foreign experts in the development of methods for mixed non-linear models and researchers in the field of biomedical engineering with experience in the use of machine learning techniques. Semiparametric regression methods are a wide range of statistical techniques that allow the incorporation of different types of structures, such as correlations, repeated measurements or more complex structures, into the data modelling. The recent popularity of these methods is due to the fact that they extend parametric regression (generalised linear models) by allowing the inclusion of smooth non-linear effects in a very flexible way (penalized regression) through the inclusion of random effects and thanks also to existing Open Source software. However, there is still ample scope to address interesting research topics and to develop new statistical methodologies. The S3M1P4R project will cover a wide range of topics. From time-to-event data analysis (also known as survival analysis) with recurrent events applied to sports medicine and injury prevention, statistical modelling of growth curves, measurements of patient-reported outcomes used to assess health-related quality of life of patients with chronic diseases, to the modelling of survey data under complex sampling designs, prediction models for clinical practice, and the development of a statistical methodology for the construction of metamodels to combine predictions from different models to be used in applications related to quantify the impact of climate change on agriculture and the Earth Science. In the methodological part, new techniques will be developed for the selection of variables in highdimensional problems, new efficient estimation methods in non-linear mixed effect models, Bayesian inference methods for spatial compositional data and explore the connection between semiparametric methods and machine learning techniques in biomedical applications."