StatisticAlps is a residential training course on advanced statistical issues of interest in clinical research. It is traditionally held as a residential course in a village in the Alps, with beautiful surroundings. Now in its eleventh edition, StatisticAlps has consistently been fully booked, confirming the strong and continuing demand on courses for high-level education in biostatistics. This is a course which harmonizes lectures on the statistical theory with the “hands on” approach to data analysis. This year StatisticAlps will be in a (late) Winter format.
The course will be delivered by Prof. Cecile Proust-Lima (Bordeaux Inserm Population Health Research Center, Univ. Bordeaux), supported by two tutors.
The presence of an internationally known instructor and the interaction with tutors and colleagues during the four days course are a very important and stimulating feature of this learning experience.
The tutors will be: Matteo Petrosino (Epidemiology Research Unit of the Italian Society for Rheumatology, Milano, Italy) and Francesca Graziano (Biostatistics and Clinical Epidemiology, IRCCS Fondazione San Gerardo dei Tintori, Monza, Italy).
The general aim of this 2027 edition is to address methods for advanced analysis of longitudinal data, rooted in the mixed modeling theory. It is well known that the increasing availability of longitudinal data may contribute to a deeper understanding of the disease course through a more refined prediction of trajectories and associated events over time. Learning how to correctly handle mixed models for longitudinal data analysis is essential in order to account for the hierarchical nature of longitudinal data and accommodate repeated measurements at irregular intervals, also in the presence of missing-at-random information.
Over four days, we will explore three extensions:
- Latent class mixed models to address heterogeneity in developmental trajectories. It identifies latent groups of individuals sharing similar mean patterns and links these clusters to external information.
- Joint modeling to analyze the association between longitudinal outcomes and time-to-event data. It can be used to account for informative truncation, assess the impact of time-varying exposures on events, or derive individual dynamic predictions.
- Latent process modeling to study constructs indirectly measured by one or more outcomes. It extends latent variable models (including Item Response Theory) to analyze trajectories of latent variables over continuous time.
The course will introduce the methods through applications in health and biomedical research, and include practical R sessions using mainly lcmm, IJPM and JMBayes2 packages.
LAURA ANTOLINI
Ponte di Legno (BS)
Executive Course
English
6 days
In presence
