Programme
44th Annual Conference of the International Society for Clinical Biostatistics
Joint Conference with the Italian Region of the International Biometric Society
THE BIOMETRICAL JOURNAL SPECIAL ISSUE
The Biometrical Journal will publish a Special Issue for ISCB2023 with deadline for submission November 30, 2023
Conference Programme
Please see the ISCB44 Program Overview here.
Please see the ISCB44 Program – Oral Presentations here.
Please see the ISCB44 Program – Mini-symposia 1-2 and Early Career Biostatician (ECB) here.
Please see the ISCB44 Program – Poster Presentation here.
President's invited speaker

Vanessa Didelez
Professor of Statistics and Causal Inference at the University of Bremen, Germany, and at the Department of Biometry and Data Management, Leibniz Institute for Prevention Research and Epidemiology
Abstract title:
“On causal inference, estimands and trials in epidemiology and biostatistics”
Keynote speaker

Lisa McShane
Chief of the Biometric Research Program of the National Cancer Institute, USA
Abstract title:
“Statistical adventures in pursuit of precision medicine: secret signatures, sliding subgroups & more”
PRE CONFERENCE COURSES: 27 AUGUST 2023
Separate fees apply for all pre-conference courses. You can book these at the time of booking your delegate place. See the registration page for more details on the cost.
Evaluation of prediction models: from AUC to calibration and decision curve analysis (half-day – afternoon) – SOLD OUT
Teachers:
- Ben van Calster: Department of Development and Regeneration & EPI-Centre, Department of Public Health and Primary Care, KU Leuven; Dept of Biomedical Data Sciences, Leiden University Medical Center
- Ewout Steyerberg: Dept Biomedical Data Sciences, Leiden University Medical Center
Estimands and analyses of longitudinal continuous outcomes in clinical trials (half-day – morning) – SOLD OUT
Teachers:
- Marcel Wolbers, Data & Statistical Sciences Department, Pharma Development, Roche, Basel
- Alessandro Noci, Data & Statistical Sciences Department, Pharma Development, Roche, Basel
Analysis of genomic data: R Bioconductor (half day – afternoon)
Teacher:
- Davide Risso, University of Padua
Beyond classic epidemiological designs (full day)
Teachers:
- Marie Reilly, Karolinska Institute
- Paola Rebora, University of Milano-Bicocca
- Francesca Graziano, University of Milano-Bicocca
Pseudo observations in survival analysis (full day)
Teachers:
- Per Kragh Andersen, Biostatistics, University of Copenhagen
- Henrik Ravn, Biostatistics, Novo Nordisk A/S, Denmark
INVITED SESSION: 28-30 AUGUST 2023
Wednesday 30 August 9:00 – 10:30
Recurrent events and their use in medical studies
There has been increasing interest in more actively using recurrent events in clinical trials to learn more from these compared to a first outcome analysis. Often interest is on the repeated occurrence of an event that will be observed in a setting where subjects can still die and thus have what is termed a terminal event. In addition to terminal events other types of withdrawal such as discontinuation of treatment may also complicate the use of recurrent events data. Recent years have also showed increased methodological research into how to deal with recurrent events and in particular how to handle the terminal events that makes it considerably more complicated to make conclusions. Different uses and summaries of recurrent events data are closely related to specific underlying aims. In addition underlying assumptions and uses of different modelling approaches can be crucial for the interpretation and the use of specific statistical models.
The speakers will address different aspects and perspectives of how to deal with recurrent events.
ORGANIZER
- Thomas Scheike, (University of Copenhagen)
SPEAKERS
- Mouna Akacha (Statistical Methods & Consulting Group, Novartis Pharma AG) – “Estimans for recurrent event endpoints”
- Giuliana Cortese (Department of Statistical Sciences, Padua University) – “Estimating the marginal and conditional means of recurrent events in presence of terminal events“
- Per Kragh Andersen (Department of Biostatistics, University of Copenhagen) – “Dealing with competing risks in the analysis of recurrent events”
Tuesday 29 August 9:00 – 10:30
Evaluation of predictive algorithms and models: uncertainty and impact on medical care
Predictive algorithms and prediction models enjoy increasing popularity with our increasing knowledge on markers, imaging and hopes for Artificial Intelligence / Machine Learning as flexible modeling tools. Careful evaluation of predictions from such models is essential, whether developed by classical statistical methods or novel AI approaches.
The goal of this session is to clarify sources of uncertainty of predictions for individual patients, and approaches to assess impact assessment on medical care.
Predictions are estimates of outcomes and inherently uncertain. In this session, Ben Van Calster will start with a perspective on different sources of uncertainty in risk modelling. He will focus on epistemic uncertainty beyond aleatoric uncertainty, including approximation uncertainty due to limited sample size and model uncertainty due to subjective modeling choices. Paula Dhiman will continue with on overview of lessons learnt on reporting and methodological quality of machine learning prediction model studies in the field of oncology, including assessments of risk of bias and spin. Finally, Laure Wynants will discuss the potential impact of prediction models in clinical practice.
A key measure for such impact is the Net Benefit of clinical decisions based on predictions from an algorithm or model. Uncertainty in NB includes sampling variability and heterogeneity between populations. Novel concepts will be discussed, such as the probability of usefulness and the Net Benefit Value of Information (NB VOI).
The session will include various medical applications, specifically in oncology. We will have adequate time for discussion on future methodological directions and context of application in prediction research, risk modeling, and
machine learning.
ORGANIZER
- Ewout Steyerberg (Leiden University Medical Center)
SPEAKERS
- Ben van Calster (Catholic University Leuven) – “Sources of uncertainty in clinical prediction models”
- Paula Dhiman:(Oxford University, UK) –“Reporting and Methodological quality of machine learning prediction model studies: an overview of results”
- Laure Wynants, (Maastricht University) – “Measuring clinical utility: uncertainty in Net Benefit“
Monday 28 August 11:00 – 12:30
Advances on causal inference in longitudinal studies
Causal inference is a fundamental and dynamic area of health research that notably aims at assessing treatment effects in the presence of selection and confounding or at decomposing causal mechanisms into pathways. Due to the inherent conceptual complexity of causal questions, approaches were originally developed for cross-sectional settings. However, the majority of phenomena studied in biostatistics and epidemiology (treatments, exposures, mediators, confounders, and outcomes) evolve over time, making causal questions and underlying assumptions even more challenging. In this session, the speakers will take three different directions to tackle causal questions in longitudinal studies.
Mireille Schnitzer will show how to select longitudinal confounders using regularization techniques under sparse setting for time dependent treatment assessment. Eric Tchetgen will discuss proximal causal inference techniques to derive separable effects in the presence of unmeasured confounders. Finally, Ruth Keogh will present how to make individual risk predictions under hypothetical interventions to inform treatment decisions.
ORGANIZER
- Cécile Proust-Lima, (Inserm & Univ. Bordeaux, France)
SPEAKERS
- Mireille Schnitzer, (University of Montreal) – “Longitudinal outcome-adaptive and marginal fused LASSO for confounder selection and model pooling with time-varying treatments“
- Eric Tchetgen Tchetgen (Department of Statistics and Data Science, Wharton School, University of Pennsylvania) – “Proximal Causal Inference for Separable Effects With Applications to Aging Research“
- Ruth Keogh (London School of Hygiene and Tropical Medicine) – “Risk prediction under hypothetical interventions“
Tuesday 29 August 11:00 – 12:30
High-dimensional inference in biostatistics
Since the big data revolution due to novel sequencing technologies, statistical methods for genomics and genetics data have relied heavily on Bayesian methods for inference. In this session we give an overview of recent techniques that tackle fundamental aspects of statistical inference for high-dimensional genetics and genomics data. A particular focus is put on recent techniques for the integration of multiple data types, drug
combination screening, and subpopulation identification, motivated by both patient and cell lines data.
Specifically, in this session will be discussed approaches for the analysis of spatially resolved transcriptome that enable the understanding of the spatial interactions between the cellular environment and transcriptional regulation, modelling approaches that use recently developed Bayesian model for synergy estimation from drug screening studies on cancer cell lines (with full uncertainty quantification), and a Bayesian profile
regression as an outcome-guided model-based clustering approach that makes use of a response in order to guide the clustering toward relevant stratifications, e.g., identification of breast cancer subtypes based on integrative clustering of multiple omics datasets.
ORGANIZER
Francesco C. Stingo, (University of Florence)
SPEAKERS
- Veera Baladandayuthapani, (University of Michigan) – “Spacex: gene co-expression network estimation for spatial transcriptomics“
- Manuela Zucknick, (University of Oslo) – “Bayesian hierarchical models for large-scale pharmacogenomic screens of drug combinations“
- Paul Kirk, (Cambridge University) – “Outcome-guided multi-view bayesian clustering for integrative omic data analysis“
Monday 28 August 13:30 – 15:00
Innovative designs for dose optimization studies
The current paradigm of dose selection in oncology is based on the study of cytotoxic agents, with a focus in phase I on identifying a single maximum tolerated dose. However, the maximum tolerated dose does not always exist in the context of biomarker-targeted therapies, for which a lower dose may be more efficacious than a higher dose, with equal or lesser toxicity. Due to a mismatch in clinical context and statistical design, therapies are moving to phase III study, or going to market, with poorly characterized doses and/or schedules. This can lead to failed phase III trials, or the need for post-market trials to further refine the dosing. There is a need for innovative clinical trial designs to identify the optimal dose and schedule of molecularly-targeted agents in oncology, with an emphasis on efficacy as well as long- term safety and tolerability. This need was recently recognized by the U.S. Food and Drug Administration with the launch of Project Optimus (https://www.fda.gov/about-fda/oncology-center-excellence/project-optimus). This session will include talks on developments in the field of early phase clinical trial designs for dose optimization.
ORGANIZER
- Emily Zabor, (Cleveland Clinic)
SPEAKERS
- Alex Kaizer (Department of Biostatistics and Informatics, University of Colorado Anschutz Medical Campus) – “The Use of Master Protocol Designs with Dose-Optimization Studies“
- Ruitao Lin (Department of Biostatistics, MD Anderson Cancer Center) – “DEMO: Bayesian Adaptive Dose Exploration–Monitoring–Optimization Design based on Short, Intermediate, and Long-term Outcomes“
- Hakim-Moulay Dehbi (Comprehensive Clinical Trials Unit, University College London) – “Controlled amplification in oncology dose-finding trials“
Wednesday 30 August 15:30 – 17:00
Marginal versus conditional effects in clinical trials
Clinical trials have traditionally been analysed using either a simple unadjusted analysis comparing outcomes between treatment groups or a covariate adjusted analysis using a suitable regression model. The former targets a so called marginal effect, whereas the latter targets a conditional effect. In recent years, methods have been developed from within the causal inference sphere which target marginal effects but which exploit baseline covariates for improved efficiency. This has led to some vigorous debate about which type of effect we ought to be estimating. Some argue that outcome regression approaches which report a single conditional effect are bound to be incorrect, since they assume that the treatment effect is the same across all values of the covariates which are adjusted for. Others argue that the marginal effect is not what is relevant to most stakeholders, since we usually make treatment decisions at the individual patient level rather than at the population level, and that the marginal effect will typically vary depending on the covariate distribution of the patients in the trial, which will almost always differ to that in a target population of interest. The topic’s importance has been further increased recently by the FDA’s publication of guidance on methods for covariate adjustment and the ICH E9 estimand addendum, which specifies that estimands should be summary measures, potentially implying trials should be targeting marginal, rather than conditional, effects.
ORGANIZER
- Jonathan Bartlett, (London School of Hygiene & Tropical Medicine)
SPEAKERS
- David Benkeser, (Emory University) – “A value system for evaluating estimands in randomized trials”
- Michael Rosenblum, (Johns Hopkins University) – “Conditional vs. marginal effects in randomized trials: tradeoffs”
- Stephen Senn, (UK) – “Why do we worry about marginal inference?”
- Ewout Steyerberg, (Leiden University) – “Covariate adjustment and exploiting ordinality: simulations of power and a rieview of neurological trials“
Wednesday 30 August 13:30 – 15:00
Quantification of safety signals in clinical trials: Estimand, estimation, and how would good look like in ten years?
Analyses of adverse events (AEs) are an important aspect of the evaluation of experimental therapies. Introduction of the estimand framework in the ICH E9(R1) estimands addendum recommends use of estimand framework also for safety analyses. Applying the framework starts with clarity on the question to be answered. Standard safety analyses in support of drug approval is about detecting risks associated with the use of a new treatment and its risk quantification.
Not only triggered by introduction of the estimand framework, it became clear that currently used methods for estimation of probability of an AE, such as as crude incidence proportions, have many problems. For example, they are only unbiased under strong and implausible assumptions, analyses of safety remain mainly driven through established data collection schemes that are not necessarily aligned with the estimand and/or estimator, or that potential intercurrent events such as treatment discontinuation are either not considered at all or the strategies that are used for them are again not aligned with data collection. The aim of this session is to bring together statisticians with various backgrounds to discuss the current status of these aspects of safety analyses and to sketch “how would good look like” in this field in, say, ten years.
ORGANIZER
- Kaspar Rufibach, (Roche, Basel)
SPEAKERS
- Kaspar Rufibach (Roche) – “Principled approach to time-to-event endpoints with competing risks, with a focus on analysis of aes“
- Anja Loos (Merck) – “Estimands for safety – one size fits all?“
- Laura Antolini (Università di Milano-Bicocca) – “Adverse events with survival outcomes: from clinical questions to methods for statistical analysis”
- Kit Roes (Radboud MC and EMA Methodology WG) – “Regulatory perspective on the analysis of safety in clinical trials and beyond“
Monday 28 August 15:30 – 17:00
Vaccination programmes: post-implementation assessment of protection, benefits and risks
The Covid-19 pandemic brought into sharp focus the complexity of population-wide vaccination and of the associated evaluation of the risks and benefits.
Vaccination programmes face numerous challenges at every stage, from initial implementation, through montoring and evaluation, to the assessement of adverse events and the magnitude and duration of protection. Statistical methods have played a key role in the assesment of safety and effectiveness of many of the standard vaccines that are in routine use today for children, for seasonal illnesses such as influenza and other infectious diseases. In particular, adverse events following vaccination can be studied using a case-only approach based on the vaccination records of just those who experience the event. While such safety assessments usually focus on a short time-period following vaccination, when adverse events are most likely to occur, the ultimate protection offered by a vaccine requires longer follow-up time.
Thus the decisions that need to taken by health authorities and regulatory bodies may need to rely on earlier evidence of protection through the study of surrogate endpoints, such as immunological markers.
This invited session presents some of the real-world challenges of the montoring and evaluation of Covid-19 vaccine programs and statistical methods that enable a vaccine to be validly assessed for its safety (in the shoter term) and effectiveness (in the longer term). The knowledge and application of these methods and further developments to meet the challenges of available data, is of increasing importance for evidence-based decision making in a time of global health challenges. The session will be aimed at biostatisticians with an interest in vaccine studies and a knowledge of regression models for count and time-to-event data.
ORGANIZER
- Marie Reilly, (Karolinska Institutet, Stockholm)
SPEAKERS
- Heather Whitaker (Health Security Agency, UK) – “Statistical methods for the epidemiological evaluation of vaccine safety“
- Susan Hahne (National Institute for Public Health and the Environment, Netherlands) – “Monitoring and evaluating Covid-19 vaccination programmes: real world challeges“
- Andrea Callegaro (GSK Vaccines, Belgium) – “Statistical methods to assess immunological surrogate endpoints for vaccines“
MINI-SYMPOSIA: 31 AUGUST 2023
There are 2 mini-symposia taking place on the 31th August:
Mini-symposium 1: Ten years STRATOS initiative – brief summary of progress and plans for the future (full day)
Coordinators:
Willi Sauerbrei, Institute of Medical Biometry and Statistics, Medical Center, University of Freiburg, Freiburg im Breisgau, Germany
Federico Ambrogi, Department of Clinical Sciences and Community Health, University of Milan, Milan, Italy.
Description:
The STRATOS (STRengthening Analytical Thinking for Observational Studies) initiative, is a large collaboration of experts in many different areas of biostatistical research. It was launched at a half-day Mini-Symposium at ISCB 2013. In Milan it will commemorate the 10th anniversary with a full-day mini-symposium.
The objective of STRATOS is to provide accessible and accurate guidance in the design and analysis of observational studies (www.stratos-initiative.org).
We will reflect on our experiences and progress and discuss about the future of the STRATOS initiative and its ability to improve research in data science, specifically in the health sciences.
Program:
Session 1 (9.00-10.40)
9.00-9.50
Experience and progress with developing guidance for the analysis of key topics in observational research
Sauerbrei W, Abrahamowicz M, Le Cessie S, Huebner M, Keogh R, Carpenter J for the STRATOS initiative
9:50-10.15
Level 1 guidance on conducting and reporting sensitivity analyses for missing data
Lee K, Carpenter J for TG1
10.15-10.40
Aims of the new Open Science panel
Luijken K, Hoffmann S, Boulesteix A-L for the Open Science panel
Session 2 (11.00-12.40)
11.00-11.25
Ongoing research towards state-of-the-art in variable and functional form selection for statistical models
Heinze G, Perperoglou A, Sauerbrei W for TG2
11.25-11.50
How to include time-varying exposures prone to measurement error in survival analyses
Proust-Lima C, Philipps, Deffner V, Boshuizen H, Freedman L, Thiébaut A for TG4
11.50-12.15
Evaluating the impact of covariate measurement error on functional form estimation in regression modelling
Perperoglou A, Abrahamowicz M, Gustafson P, Kipnis V, Thiébaut A, Ferreira Guerra S,
Freedman L for TG2 and TG4
12.15-12-40
Statistical analysis of high-dimensional biomedical data: A gentle introduction to analytical goals, common approaches and challenges
Ambrogi F, Rahnenfuehrer J, De Bin R, McShane L for TG9
Session 3 (13.30-15.10)
13.30-13.55
The slowly changing landscape of predictive modeling in biomedicine
Lusa L, Kappenberg F, Schmid M, Sauerbrei W, Rahnenführer J
13.55-14.20
Counterfactual prediction for personalized healthcare using observational data
Van Geloven N, Steyerberg EW, Wang J, Didelez V, Keogh RH for TG6 and TG7
14.20-14.45
Recommendations to handle patient reported outcome data in oncology cancer trials
Le Cessie S, Goetghebeur E, Thomassen D on behalf of work package 3 of the SISAQOL-IMI consortium
14.45-15.10
Comparing quality of life – while alive – between treatment and (external) controls: methods for real world analysis in clinical trials
Goetghebeur E, Reynders D, Thomassen D, le Cessie S on behalf of work package 3 of the SISAQOL-IMI consortium
Session 4 (15.30-16.30)
Panel discussion about the future of STRATOS
Chair: Carpenter J (London, UK)
Mini-symposium 2: Novel approaches to complex data and predictive modeling in healthcare research (half-day)
Coordinators:
Emanuele DiAngelantonio, Health Data Science Center, Human Technopole, Milan, Italy
emanuele.diangelantonio@fht.org
Francesca Ieva, Department of Mathematics, Politecnico di Milano, Milan, Italy.
Program:
- H 9.00: Introduction (Chairs: E. Di Angelantonio, MG Valsecchi)
- H 9.15-9.45
Catalina Vallejos – “PREDICTING EMERGENCY ADMISSIONS IN SCOTLAND”
- H 9.45-10.15
Mihaela van der Schaar – “TIME: THE NEXT FRONTIER IN MACHINE LEARNING FOR HEALTHCARE”
- H 10.15-10.45 coffee break
- H 10.45-11.15
Marteen van Smeden – “RAGE AGAINST THE MACHINE LEARNING”
- H 11.15-11.45
Davide Bernasconi – “REGRESSION AND ML APPROACHES FOR EVALUATION OF BIOMARKERS WITH APPLICATION TO PRIMARY BILIARY CHOLANGITIS”
- H 11.45 – 12.15 Panel discussion (Chair: F. Ieva)
ECB DAY: 31 AUGUST 2023
We welcome all students and researchers starting their journey in Biostatistics to join the Early Career Biostatisticians’ (ECB) Day that will take place on Thursday, 31 August.
The aim of the day is to share personal experiences relating to biostatistical research and discuss how to deal with the potential pitfalls of the research process to become better researchers in the process.
Whether you are just about to graduate or have already experience working as a researcher or biostatistical consultant, you will benefit from meeting your peers, exchanging your thoughts and ideas, and getting to know more about how to shape a career in biostatistics.
Thus, this day will complement the main conference which mainly focuses on research results.
TIME (CEST)
9:00 – 9:45
PRESENTER
- Valeria Edefonti (invited speaker)
9:45 – 10:00
- Rushani Wijesuriya
TITLE OF TALK
A NETWORK OF MENTORS: LEVERAGING THE POWER OF NETWORKING AND MENTORING TO ACCELERATE YOUR CAREER
10:00 – 10:15
- Judith ter Schure
10:15 – 10:30
- Bethany Hillier
ARE STATISTICAL AND SCIENTIFIC ASSESSMENTS OF RAPID SELF-TEST DIAGNOSTICS RELIABLE?
10:30 – 11:00 – BREAK
11:00 – 11:15
PRESENTER
- Elena Albu
11:15 – 11:30
PRESENTER
- Alexandra Hunt
11:30 – 11:45
PRESENTER
- Autumn O’Donnell
11:45 – 12:30
PRESENTER
- Katherine Lee (invited speaker)
12:30 – 12:45
PANEL DISCUSSION /Q&A
Registration
You can register for the ECB Day through the main conference registration system (www.iscb2023.promoest.com).
Participation will be free of charge for delegates also attending the main conference, and will be 35 EUR for ECBs attending the day only.
Invited speakers
- Valeria Edefonti (University of Milan) – “SUSTAINING A CULTURE OF REPRODUCIBILITY IN RESEARCH: A PERSONAL CREDO FOR EARLY CAREER BIOSTATISTICIANS”
- Katherine Lee (Murdoch Children’s Reseach Institute and University of Melbourne) – “NAVIGATING THE WORLD OF BIOSTATISTICS“
We look forward to meeting you in Milan!
Please note that Pre Conference courses will not be available to virtual participants either during the conference dates or after the completion of the Conference.
For more information contact
Organizing Secretariat - Promoest s.r.l.
Barbara Colonnello and Ilaria Magnani
Call us: