At the present, some BEMC Talks are held in a hybrid format (in-person plus live-streatming), and some are held in a virtual only format. All BEMC Talks will be streatmed online as Zoom webinars.
Be sure to register early to secure your spot! Selected BEMC Talks will be recorded and posted on our YouTube channel.
BEMC Talks are generally held on the first Wednesday of each month at 4pm Berlin time (GMT+01:00) with some adjustments for holidays. They consist of an hour-long lecture followed by an interactive group discussion with the speaker.
2022 BEMC Talks
|Date||Speaker & Talk Description|
|February 2, 2022||Regression by Composition|
Anders Huitfeldt, Rhian Daniel & Daniel Farewell, Odense, Denmark & Cardiff, Wales
We introduce regression-by-composition (RBC), a flexible regression modelling framework that includes generalized linear models (GLM) as a special case. The RBC form allows more general maps between conditional outcome means than can be expressed using any GLM link function, and allows predictors (or groups of predictors) to each have their own link function. RBCs have the particular advantage of admitting models for binary outcomes with parameters that can be interpreted as switch risk ratios; this is not possible with GLMs. The switch risk ratio has many attractive properties such as closure, collapsibility and invariance to outcome coding. These properties are not jointly shared by any GLM parameter for a binary outcome. The switch risk ratio is also motivated by a large and growing body of psychological and philosophical work on what causal objects human reasoners expect to be stable between different subgroups.
|March 2, 2022||Utility of Omics at Population Scale|
Claudia Langenberg, Berlin, Germany
Application of different omic technologies is now feasible at population scale. This talk will present examples of how the integration of different omics in large patient and population studies can help to predict disease risk, understand mechanisms, and reveal shared connections between different diseases.
|June 1, 2022||Modeling the Epidemiological Interactions Between Respiratory Viruses and Their Implications for Disease Control|
Matthieu Domenech de Cellès, Berlin, Germany
Interaction—that is, the ability of one pathogen to alter the risk of infection or disease caused by another pathogen—is an intriguing aspect of the biology of many respiratory viruses, including influenza viruses, respiratory syncytial viruses (RSV), and SARS-CoV-2. Despite growing experimental evidence for such interactions, however, their epidemiological consequences remain poorly defined. In this talk, I will present several of my lab’s research projects that aim at elucidating the interactions between respiratory viruses using mathematical models of transmission. I will first describe a general framework to systematically capture the components of pathogen interactions, showing how these models can encapsulate the framework in a mechanistic and parsimonious way. Next, building on the results of a recent study (Domenech de Cellès et al., Proceedings B 2022), I will illustrate how these models can be used to assess the design of epidemiological studies that aim to infer interaction. Specifically, I will show that test-negative designs—or other study designs based on co-detection prevalence data—are expected to frequently fail at estimating the strength, and even the direction, of interaction. Finally, I will present results of an ongoing modeling study of the interactions between influenza viruses and RSV, and explain how such interactions could be leveraged to induce indirect effects of influenza vaccination.
Infectious Disease Epidemiology group at the Max Planck Institute for Infection Biology: Website | Twitter
|August 2022||Summer Break|
|September 7, 2022||Empty Your Research Data File Drawer with fiddle|
René Bernard, Berlin, Germany
Research data production has accelerated over the last decade as evidenced by the rise in the number of publications. However, a large portion of data from well-designed studies is never reported and rots away on inaccessible hard drives or servers. Reasons not to publish are manifold but funders and the public have the right to know study outcomes for which researchers have applied and have received public or private funds. Fiddle is a tool that helps researchers to find an outlet for their data that matches their resource situation as well as the property of the data set and makes them assessable for needed meta-analyses.
> Watch Recording
|September 21, 2022||A Causal Perspective on Age-Education Correction in Cognitive Screening Tests|
Marco Piccininni, Berlin, Germany
Correcting cognitive screening test scores for age and education is a common practice in neuropsychology. Demographic correction quantifies an individual’s test performance relative to the distribution of normative score values obtained from a set of individuals with similar demographic characteristics and without cognitive impairment. Evidence of diminished accuracy of cognitive test scores after applying the correction can be found in the literature, and theoretical criticism of this approach has been raised in the context of cognitive screening. In this talk, I will provide a brief introduction to age-education correction and an overview of the controversy surrounding its use in cognitive screening tests. Then, I will use causal models to shed light on the problem, and I will argue that this commonly-accepted technique should not be used in practice if the aim is to be able to distinguish individuals with cognitive impairment from those without.
|October 5, 2022||Beyond Internal Validity: Field Notes From the Methodological Borderlands|
Daniel Westreich, Chapel Hill, USA
The effect of smoking on risk of death is not the same as the effect of a smoking cessation intervention on risk of death – but we frequently act as if it is when considering policy. In this talk, I will describe the causal impact framework, which considers first internal validity; then external validity; then population intervention impact; and (sometimes) cost-effectiveness analysis. The causal impact framework shows one way forward to making epidemiologic findings more relevant to both public policy and implementation science, and likewise ways in which the epidemiologist can more effectively coordinate with both health behaviorists and health policy scientists.
|November 2, 2022|
Webinar registration opening soon!
|David Q. Rich|
|December 7, 2022||Annika Hoyer|