May BEMC Talk- Student summary

Dear BEMCers,

We are pleased to be able to share a summary prepared by a student. A warm thank you to the anonymous student for letting us share this recap.

On May 8th, Christoph Lippert, a Professor in the area of digital health and machine learning at the Hasso Plattner Institute and at the University Potsdam held a lecture on “Machine Learning for Population-Based Health Studies”. The starting point of this lecture was to show a graphical representation of a typical patient flow through a health care system. This includes a patient’s contact with the health care system due to some symptom. Later on, some tests will be carried out by a physician and, after a lengthy period, the patient will receive an invasive treatment. It can also be the case that it will eventually be too late for any treatment. The current procedure was compared with what it is expected to be a patient flow in the future, where an individual undergoes continuous monitoring for diseases (e.g. through genetic risk assessment). Thus, it is expected that an individual will constantly receive early warnings, which lead him/her to involve a doctor at an early stage. Christoph highlighted how machine learning has the potential to help moving from the current patient flow to a better one, where people know early on their risks for developing certain conditions and can act accordingly. He highlighted how genetic differences across individuals can shed light on an individual’s risk factors for diseases, thus allowing for better and more individualised treatment options. Christoph proceeded by explaining basic concepts about genome-wide association studies and their goal of finding causal variants and/or markers that explain variance. He provided several examples, such as how phenotypes and population structures are correlated. The discussion was heated and insightful. Several attendees pointed out the need to consider the ethical implications of using genetics to predict health conditions. It was also mentioned how genome testing is cheap but, on a population level, this data does not provide meaningful information (i.e.. on a public health perspective). Furthermore, a question left hanging in the air concerned the degree of impact of lifestyle factors versus genes on an individual’s health status.

Announcing May BEMC Talk

Dear Berlin-area Epidemiological Methods Enthusiasts,

You are invited to our next BEMC Talk on Wednesday, May 8th.

BEMC Talk: Wednesday, May 8th, 2019 @ 4pm

“Machine learning for population-based health studies” – Christoph Lippert, Potsdam

  • Please register online: https://docs.google.com/forms/d/e/1FAIpQLSe2EfMPM2lXX1etpLblhLzDcfM0In9wLoTc3wUSWHD3Q5iVGw/viewform?usp=sf_link
    • If the registration link does not work for you please send us an email at bemcolloquium(at)charite.de with your full name and email address and short note that you’d like to be added to the registration list. Thanks!
  • Description: “In my talk, I’ll introduce our research at the interface of machine learning and statistics towards new methods for large-scale epidemiologic studies. We are working on models and algorithms that allow us to analyze high-dimensional genotypes and phenotypes from imaging and sequencing at scale. I’ll be talking about our work on mixed models for confounder correction, quantitative phenotyping using deep learning, as well as our works on novel statistical tests on deep learning embeddings of images.”
  • Location: Seminar room of the Neurology Clinic; Bonhoefferweg 3 entrance, 3rdfloor, Charité – Campus Mitte

Upcoming Berlin Epi Events:

  • May 15th – BEMC JClub – Paper posted online
  • June 5th – BEMC Talk – “Cool applications in R for epidemiologists” – Jochen Kruppa, Berlin
  • June 19th – BEMC JClub – Paper will be posted online
  • July 3rd – BEMC Talk Suzanne Cannigieter, Leiden, Netherlands
  • July 19th – BEMC JClub – Paper will be posted online in late May
  • August 2019 – no BEMC Talk or JClub
  • September 19th – BEMC JClub – Paper will be posted online in late July
  • October 2nd – BEMC JClub – Paper will be posted online in late August
  • Wednesday, October 23rd – IPH Lecture Partner event– John Gill from Vancouver

Interested in other Institute of Public Health events? Visit our calendar to check out upcoming conferences & short courses!

Follow BEMC on Twitter and leave questions for our speakers: @BEMColloquium

Pamela Rist, “One Size Does Not Fit All: Teaching Introductory Epidemiology” — Student summary

Dear BEMCers,

We are pleased to be able to share a summary prepared by three students who are currently earning credit for participating in the BEMC. A warm thank you to the three anonymous students for letting us share a compilation of their summaries!

As a reminder, we will be having BEMC JClub next Wednesday. A link to the article is available under JCLUB.

On Wednesday, April 3rd, Pamela Rist, Assistant Professor of Medicine at the Harvard Medical School and Assistant Professor of Epidemiology at the Harvard T.H. Chan School of Public Health, held a lecture about different ways to teach introductory courses in epidemiology. She teaches intro epi classes for medical staff and public health professionals, at undergraduate, postgraduate and doctoral levels.

Rist started her lecture by presenting how she first introduces the concept of confounding in her classes. Touching only the most basic aspects, the concept is simple to understand. However, it becomes inaccurate when moving on to more complex causal relations: It does not consider (open) backdoor paths, for instance. This, according to Rist, illustrates the difficulty of teaching: One has to introduce topics in a way that is understandable to the audience but does not cause problems later on for those who progress to substantial classes. Ultimately, there can be a trade-off between an intuitive and a correct definition.

She further explained that what should be thought of in an epidemiology introductory course depends on the type of audience, the skill set they need to learn in their career tracks, the goals of the course and structural issues. This appears to be particularly challenging since these courses are usually characterized by a considerable audience heterogeneity: students have different backgrounds, different coursework and different degrees of interest.

A successful technique to capture an audience’s attention, and to let students understand the importance of the epidemiology in their own field, might be to use different examples and papers based on the audience’s field of interest. In addition, Rist stated that even the way epidemiological concepts are explained should be flexible: using not only medical but also general knowledge examples to explain complex concepts (she provided a striking example explaining the Selection Bias using NBA players as subject).

Lastly, Rist introduced different teaching forms she uses: seminars, online lectures, inverted classrooms and discussed possible advantages as well as drawbacks. Seminars are a good learning environment, however small groups are required for discussion. Therefore, several rooms need to be available as well as teaching assistants to lead the seminars. If the setting does not allow this, Rist advised including live polls in large lectures and polling twice: The first time the students answer the question alone, the second time after discussion with neighbors. In this way, an active element can be incorporated into the lectures and students can learn from one-another. Rist also gave voice to her concern that setting up online lectures for the first time requires intensive preparation.

Discussion:
Q: Where do you find the papers for the discussion with students?
A: Science and Health sections of Times, JAMA, NEJM, Lancet. Talking to family and friends about current issues.
Q: Do you discuss good or bad papers?
A: I try not to focus only on poorly written papers. Sometimes I use also older papers to discuss the constraints of older methodology.
Q: What role does the student evaluation play for your next course?
A: Our course evaluations are done externally and anonymously, but I do read them. I find them to be useful and definitely try to adjust based on feedback where possible.
Q: How do you measure the success of your teaching? Do you have a special format for the exams (multiple choice etc.)?
A: Formally, the school measures the success of the lecture on 1-5 scale (teachers must score at least 3.5 to be allowed to teach the following semester). Informally, by student reactions. In terms of exams: we use tests with multiple choice, true/false, and fill in the blank questions for Master’s students (the class size is quite large and there isn’t much time between the final exam and when grades are due) and essays for PhD students.

Announcing April BEMC Talk

Dear Berlin-area Epidemiological Methods Enthusiasts,

You are invited to our next BEMC Talk on Wednesday, April 3rd.

BEMC Talk: Wednesday, April 3rd, 2019 @ 4pm

“One Size Does Not Fit All: Teaching Introductory Epidemiology” – Pamela Rist, Boston, USA

  • Please register online: https://goo.gl/forms/pbE7UfFAEX2V1oxG2
  • Description: “As epidemiological methods have advanced, there has been debate over whether introductory epidemiology courses should be updated to include more “modern” concepts or if the material covered in introductory courses should remain constant.  Additionally, introductory courses greatly differ in length and scope, and since they are offered to students with diverse backgrounds and career goals, it is unlikely that one static course will fit all students’ needs. Sometimes, ‘teaching’ epidemiology happens more informally outside the classroom, such as in small consultation sessions or even when coauthoring a manuscript. In this lecture, we will discuss important considerations when developing introductory epidemiology course material, examples of ways to incorporate modern epidemiology concepts, and strategies to tailor the instruction to fit your audience’s needs.”
  • Location: Neurology Seminar Room, Charite Campus Mitte, Bonhoefferweg 3, 1. Etage (look for our BEMC signs)

Upcoming Berlin Epi Events:

  • April 17th – BEMC JClub – Paper will be posted online
  • May 8th – BEMC Talk – “Machine Learning for Population-Based Health Studies” – Christoph Lippert, Potsdam
  • May 15th – BEMC JClub – Paper will be posted online
  • June 5th – BEMC Talk – “Cool applications in R for epidemiologists” – Jochen Kruppa, Berlin
  • June 19th – BEMC JClub – Paper will be posted online
  • July 3rd – BEMC Talk – Tentative title: “Pragmatic trials and lessons from venous thrombosis” – Suzanne Cannigieter, Leiden

Interested in other Institute of Public Health events? Visit our calendar to check out upcoming conferences & short courses!

Follow BEMC on Twitter and leave questions for our speakers: @BEMColloquium

Maarten van Smeden, Regression shrinkage: better answers to causal questions

Dear BEMCers,

We are pleased to be able to share a summary prepared by a student earning credit for participating in the BEMC. A warm thank you to Ana Sofia Oliveira Gonḉalves for letting us share her summary!

Maarten’s slides can be found online: https://www.slideshare.net/MaartenvanSmeden/regression-shrinkage-better-answers-to-causal-questions

As a reminder, we will be having BEMC JClub on Wednesday this week. A link to the article is available under JCLUB.

On Wednesday March 6th, Maarten van Smeden, a senior researcher from the Leiden University Medical Center (NL) shared with the audience valuable insights on coefficient shrinkage in regression, both in a prediction and in a causal research context, with a focus on the latter. The starting point of this lecture was to question the appropriateness of traditional ways of computing the odds ratios (e.g. in a 2×2 table or by standard logistic regressions based on maximum likelihood estimation). Maarten explained that the maximum likelihood estimators for regression coefficients in generalized linear models are biased but consistent.

Throughout the lecture, Maarten used simple logistic regression model as an example. Based on such a model, he provided us with graphical representations of his simulations to show the properties of such estimators. With this simulation he intended to stress the difference between the two concepts of lack of bias and consistency. He then presented us with a solution for the reduction of bias for maximum likelihood estimators: Firth’s correction.

Firth’s correction is a penalized estimation procedure that shrinks regression coefficients, thereby removing a large part of the finite sample bias. The Firth’s correction can be readily implemented in a causal research context and packages in statistical programs already exist. Maarten mentioned other shrinkage estimators, such as Ridge or LASSO can conducted for prediction purposes, since biased coefficients are better suited for this purpose. Nevertheless, he warned the audience against its use in a causal inference context, since these approaches are designed to create bias in coefficient estimators, rather than to remove it.

Announcing March BEMC Talk

Dear Berlin-area Epidemiological Methods Enthusiasts,

You are invited to our next BEMC Talk on Wednesday, March 6th. Also join us this week for the BEMC Journal Club! The paper is already posted online.

BEMC Talk: Wednesday, March 6th, 2019 @ 4pm

“Regression shrinkage: better answers to causal questions” – Maarten van Smeden, Leiden

  • Please register online: https://goo.gl/forms/EdISo4yoZQhepGNY2
  • Description: “In this lecture, I will discuss regression shrinkage in the context of causal epidemiologic research. Although regression shrinkage methods have predominantly been used in prediction research (i.e. not to answer causal questions) to avoid statistical overfitting, I will posit the view that some of these methods can also improve estimates of exposure-outcome relationships. In particular, I will illustrate how a simple-to-apply approach known as Firth’s correction removes an often overlooked estimation bias in conventional regression analyses such as logistic regression.
  • Location: Seminar room of the Neurology Clinic; Bonhoefferweg 3 entrance, 3rdfloor, Charité – Campus Mitte

Upcoming Berlin Epi Events:

  • February 20st– BEMC JClub – Paper posted online
  • March 6th– BEMC Talk  “Shrinkage for causal inference” – Maarten van Smeden, Leiden
  • March 20th – BEMC JClub – Paper posted online
  • April 3rd – BEMC Talk –”One Size Does Not Fit All: Teaching Introductory Epidemiology” –Pamela Rist, Boston
  • April 17th – BEMC JClub – Paper will be posted online in late February
  • May 8th– BEMC Talk – “Machine Learning for Population-Based Health Studies” – Christoph Lippert, Potsdam
  • May 15th – BEMC JClub – Paper will be posted online in late March
  • June 5th – BEMC Talk – “Cool applications in R for epidemiologists” – Jochen Kruppa, Berlin

Interested in other Institute of Public Health events? Visit our calendar to check out upcoming conferences & short courses!

Follow BEMC on Twitter and leave questions for our speakers: @BEMColloquium

Lars Andersen, Time-dependent propensity scores: an application

Dear BEMCers,

This month, we are sharing a summary prepared by a student earning credit for the BEMC together with a photo. Thanks to the student, who prefers to remain anonymous!

This past Wednesday, Lars Andersen from Aarhus, Denmark, lectured on the use of time-dependent propensity score matching to address changes of treatment and covariates over time at the BEMC. After giving a general outline, Andersen introduced the immortal time bias as well as the related concept of “resuscitation bias” and explained how the latter can be avoided using risk set matching. As time-variant confounding remains an issue in the analysis phase, Lars Andersen introduced time-dependent propensity score matching as a strategy to deal with such a challenge.

As a clinician, Andersen explained how the nature of his clinical field and research questions inform his methodological choices. To illustrate these methods, Andersen elaborated on one of his papers dealing with tracheal intubation during cardiac arrests. In the paper, the authors could demonstrate that intubation lowered the chance of survival in an analysis using time-dependent propensity score matching while the unadjusted analysis revealed the opposite.

lars_andersenbemc

See you in March!

Announcing February BEMC Talk

Dear Berlin-area Epidemiological Methods Enthusiasts,

Thank you for participating in our 2018 BEMC Talks! We will be taking a short break in January and look forwarding to seeing you again in February!

You are invited to our next BEMC Talk on Wednesday, February 6th.

BEMC Talk: Wednesday, February 6th, 2019 @ 4pm

“The use of time-depending propensity score matching to address changes of treatment and covariates over time” – Lars Andersen, Aarhus, Denmark

  • Please register online: https://goo.gl/forms/jKhS5kVGIR0fTvX03
  • Description: “Time-related biases, such as immortal time bias, are well-known within the epidemiology literature. However, they continue to be a problem in the published medical literature. The lecture will focus on time-related bias and how they can be effectively addressed using a combination of risk set matching and time-dependent propensity scores. Studies with cardiac arrest, where a specific type of time-related bias (“resuscitation time bias”) is a problem, will be used as examples.”
  • Location: Virchow-Saal, Philippstr. 11, Charité – Campus Mitte (this location is only for December 2018 and February 2019)

Upcoming Berlin Epi Events:

  • February 20th – BEMC JClub  – Paper posted online
  • March 6th – BEMC Talk  “Shrinkage for causal inference” Maarten van Smeden, Leiden
  • March 20th – BEMC JClub – Paper will be posted online in late January
  • April 3rd – BEMC Talk Pamela Rist, Boston
  • April 17th – BEMC JClub – Paper will be posted online in late February
  • Tuesday, May 7th – IPH Lecture (no BEMC Talk in May)– Frits Rosendaal from Leiden

Interested in other Institute of Public Health events? Visit our calendar to check out upcoming conferences & short courses!

Follow BEMC on Twitter and leave questions for our speakers: @BEMColloquium

End of 2018: wrap-up

A big thanks to all of our fantastic speakers and of course, our engaged Berlin-area epidemiology community for another great year of BEMC! Here a few impressions from the last several months. We just finalized our 2019 BEMC Talks line-up. Next year, we’re bringing in epi talent from Aarhus, Leiden, Boston, Berlin, Los Angeles, Vancouver, Toulouse, & Tirol. Check out the ‘Talks’ tab. Wishing you a relaxing last few days off and, of course, guten Rutsch!

-Jess, Magdalen, Bob & Tobias