Pre-conference Courses

Overview

The following full-day and half-day (morning or afternoon) pre-conference courses will be held in-person-only on Sunday, 24th of August 2025 (the day before the start of the main conference). Note that main conference registration is a prerequisite to be able to book pre-conference courses. See the registration page for details on the registration costs for the pre-conference courses.

Time Course Instructors
Full-day Targeted Learning Oliver Dukes,
Stijn Vansteelandt,
Shaun Seaman
Full-day The Design of Simulation Studies Tim P. Morris,
Ian R. White
Full-day Network Meta-Analysis:
From Key Concepts to Advanced Methods
Virginia Chiocchia,
Konstantina Chalkou,
Orestis Efthimiou,
Tasnim Hamza,
Georgia Salanti
Full-day Good Software Engineering Practice for R Packages Audrey Te-ying Yeo,
Alessandro Gasparini,
Daniel Sabanés Bové
Morning Bayesian Methods for Precision Medicine Peter F. Thall
Afternoon Multi-State Models:
Theory, Applications and New Developments
Liesbeth de Wreede,
Hein Putter

Targeted Learning

Instructors:

  • Oliver Dukes, Ghent University, Belgium
  • Stijn Vansteelandt, Ghent University, Belgium
  • Shaun Seaman, University of Cambridge, United Kingdom

Evaluating treatment effects using observational data or trials with complex intercurrent events may require accounting for high-dimensional confounding. This course will describe how, in these challenging situations, machine learning and variable selection procedures can be used to infer causal effects. The first part is a high level overview of how and why this methodology works, touching on recent developments in ‘double machine learning’ and ‘targeted maximum likelihood estimation’. In the second part the participants will exercise the concepts and methods explained during the first part via the analysis of real data sets.

This introductory course is aimed at researchers in the (pharmaceutical) industry and academia working with observational as well as trial data; a basic understanding of causal inference can be helpful but is not necessary. We foresee a mix of lectures and hands-on exercises using R.

The Design of Simulation Studies

Instructors:

  • Tim P. Morris, UCL, United Kingdom
  • Ian R. White, UCL, United Kingdom

Simulation studies are an invaluable tool for biostatistical research, as you will see at the ISCB46 conference this week, where they will play a prominent role in many presentations. You may even be presenting one of your own. Do you have the confidence to thoughtfully critique someone’s simulation study, or to defend yours? Although it is tempting to think of a simulation studies as a mere coding exercise, we have a rather different view. This short course will focus on the design of simulation studies, following the ADEMP framework (Aims, Data-generating mechanisms, Estimands, Methods of Analysis, Performance measures). Practical sessions will be interactive rather than computer-based, involving group discussions and debates on issues around the design of simulation studies.

Network Meta-Analysis: From Key Concepts to Advanced Methods

Instructors:

  • Virginia Chiocchia, Institute of Social and Preventive Medicine, University of Bern, Switzerland
  • Konstantina Chalkou, Institute of Social and Preventive Medicine, University of Bern, Switzerland
  • Orestis Efthimiou, Institute of Primary Health Care (BIHAM), University of Bern, Switzerland
  • Tasnim Hamza, Institute of Social and Preventive Medicine, University of Bern, Switzerland
  • Georgia Salanti, Institute of Social and Preventive Medicine, University of Bern, Switzerland

Network meta-analysis is an extension of pairwise meta-analysis that allows us to compare three or more interventions simultaneously, by combining direct and indirect evidence from a network of studies. Network meta-analysis can be used to estimate the relative treatmenteffects between any pair of interventions in the network, it increases precision compared to using only direct evidence, and it can produce a hierarchy of the interventions.

In the morning session of this full-day course we will demonstrate the assumptions and methods of network meta-analysis and network meta-regression with interactive lectures and practical exercise. In the afternoon we will introduce advanced topics in network meta-analysis, such as the use of individual participant data, component network meta-analysis for composite interventions, and dose-response analysis.

The course is designed for participants who are familiar with meta-analysis and Bayesian statistics. By the end of the course, participants will be able to:

  • Understand and assess the assumptions underlying the validity of indirect comparisonsand network meta-analysis

  • Estimate the relative treatment effects between any pair of interventions within a network of studies and present them in a transparent way

  • Assess and test for inconsistency within a network of interventions

  • Formulate a network meta-regression model and interpret the results and output

  • Obtain an overview of advanced methods and extensions in network meta-analysis

The practical exercises will be performed in the statistical software R.

Good Software Engineering Practice for R Packages

Instructors:

  • Audrey Te-ying Yeo, Independent
  • Alessandro Gasparini, Red Door Analytics AB
  • Daniel Sabanés Bové, RCONIS

The vast majority of statisticians in academia and industry alike write statistical software daily. Nonetheless, software engineering principles are often neglected in biostatistics: most biostatisticians know a programming language (such as R) but lack formal training in writing reusable and reliable code.

This course aims to equip participants with the essential software engineering practices required to develop and maintain robust R packages. With the growing demand for reproducible research and the increasing complexity of statistical methods developed for multidimensional data, writing high-quality R packages has become a critical skill for statisticians to prototype, develop, and disseminate novel methods and push their adoption in practice. The course will focus on the key principles of software engineering, such as workflows, modular design, version control, testing, documentation, and quality indicators. Focussing on these aspects ensures the reliability and sustainability of R packages.

Participants will learn how to structure their R packages following best practices and making use of tools that streamline the development process. The course will also cover version control using Git, allowing participants to manage code changes effectively and collaborate with others. A significant emphasis will be placed on writing and running unit tests, ensuring that packages are error-free and behave as expected across different environments and over time.

Furthermore, the course will cover quality indicators for R packages and explore techniques for writing effective documentation, enabling users to pick, understand, and use statistical software packages effectively.

By the end of the course, participants will have a solid understanding of good software engineering principles tailored to R package development, enabling them to build packages that are not only functional but also reliable, reusable, and easy to maintain.

Bayesian Methods for Precision Medicine

Instructor:

  • Peter F. Thall, M.D. Anderson Cancer Center, United States of America

This half day short course will present statistical concepts and methods related to precision, or ‘personalized’ medicine, which uses individual patient covariates to choose treatment or doses. The topics are drawn from the book, ‘Bayesian Precision Medicine’ published by Chapman and Hall in 2024. To start, the problem of comparing immunotherapy to prayer for treating a severe disease will be discussed. Basic concepts of causal inference will be reviewed, including bias correction methods for analyzing observational data, causal diagrams, with both toy and real-world illustrative examples. Two clinical trial designs will be reviewed that aim to identify optimal subgroup-specific doses or treatments in particular medical settings, each using a utility of a multivariate outcome. The first is a phase 1-2 design that uses the joint utility of five time-to-event outcomes to optimize patient subgroup-specific natural killer cell doses for treating advanced leukemia or lymphoma. The second design does phase 2 treatment screening and selection, illustrated by a randomized three-arm trial to compare targeted agents. Two data analyses that apply Bayesian nonparametric regression models to identify optimal covariate-specific treatments then will be presented. The first analysis uses observational data to identify optimal covariate-specific doses of intravenous busulfan as part of the preparative regimen for allogeneic stem cell transplantation. The second is a re-analysis of a published dataset from a randomized trial, with a joint utility of progression free survival time and total toxicity burden used to choose optimal personalized targeted therapies for advanced breast cancer.

Multi-State Models: Theory, Applications and New Developments

Instructors:

  • Liesbeth de Wreede, Leiden University Medical Center, Netherlands
  • Hein Putter, Leiden University Medical Center, Netherlands

The Multi-state models play an increasingly important role in the analysis of time-to-event data. They provide a comprehensive framework to analyze and understand complex medical phenomena, making them invaluable in research aimed at improving patient care, guiding public health policies, and advancing medical science. Extensions of the Nelson-Aalen estimator of the cumulative hazard and of the Kaplan-Meier estimator of the survival function allow for a detailed assessment of the dynamics of complex disease processes and patient trajectories, and the effect of covariates on these patterns.

In the first half we offer a brief coverage of basic concepts and techniques in multi-state models, focusing on non- and semi-parametric (Cox-model based) Markov models. We start with an introduction to important concepts, in particular transition intensities (rates) and transition probabilities (risks) and the relation between them, viewing multi-state models as an extension of competing risks models. We continue with methods for assessment of the effect of covariates on the transition intensities through proportional hazards models. Throughout the course, all steps needed for a multi-state analysis will be illustrated with examples and syntax based on the mstate package in R.

In the second half we focus on two selected topics related to more recent advancements in multi-state modelling. The first concerns the Markov assumption. We discuss formal tests for the Markov assumption. We explore estimation of transition probabilities that are consistent also when the Markov assumption is violated, in particular the landmark Aalen-Johansen estimator, and extensions like the hybrid landmark Aalen-Johansen estimator. The second topic concerns incorporation of relative survival in multi-state models. This allows to split mortality in excess and background mortality with and without intermediate events. Two different models for assessing the impact of covariates on the excess hazard will be introduced: the Cox model and Aalen’s additive hazards model.