Spring 2020
Past Circle Meetings
January 17 – Kerollos Wanis and Arin Madenci discussed IP weighting for continuous treatments. Issa Dahabreh presented results on causal meta-analysis.
Fall 2019
October 25 – Aaron Sarvet discussed the advantages of using SWIGs compared with DAGs with nodes for counterfactual variables.
September 27 – Lan Wen described several available g-formula estimators for survival analysis and discussed their relative advantages and disadvantages. Xinkun Nie explained how computer scientists refer to the various g-formula estimators.
September 20 – Anthony Matthews (postdoc at Karolinska Institutet) presented his proposed research for randomized-observational partnerships in registry-based databases. Barbra Dickerman presented her latest findings on case-control studies for target trial emulation.
Spring 2019
May 17 – Barbra Dickerman discussed the challenges to emulate a target trial in published case-control designs of statins and cancer. Guilin Li and Xiaohui Yu presented their research to use different versions of (diabetes) treatment to infer causal effects on cardiovascular disease.
April 26 – Jose Zubizarreta and his junior collaborators in the Department of Health Care Policy at HMS presented a series of lightning talks. They covered causal inference topics on which they conducted research, including covariate balancing, machine learning and propensity scores, etc.
April 19 – No meeting.
April 12 – Kerollos Wanis and Aaron Sarvet discussed settings in which partial exchangeability may be a reasonable condition, and Lan Wen presented the newly available g-formula estimator based on iterated expectation.
March 29 – No meeting.
March 22 – Justin Bohn and Darren Toh from Harvard Pilgrim presented a project using negative controls in EHR data.
March 15 – Barbra Dickerman and Lucia Petito co-lead a discussion of practical issues related to the use of inverse probability weighting.
March 1 – Yu-Han Chiu lead a discussion of how to handle competing risks. She described an application to the effects of fertility therapies on neonatal outcomes using real data. The methods she presented integrate the latest methodological advances by Jessica Young and Mats Stensrud.
February 8 – Discussion held about the use of electronic health records to compare complex (like, really complex) dynamic treatment strategies using inverse probability weighting. Kate Birnie and Jonathan Sterne from the University of Bristol will describe their ongoing research using UK data to compare the effectiveness of anemia management strategies in individuals with renal failure.
January 25 – Barbra Dickerman discussed her work-in-progress to emulate a target trial using electronic medical records (not claims). Time pending, Yu-Han Chiu will discuss her work to handling competing risks in a follow-up study.
Fall 2018
December 14 – Kun-Hsing Yu from the Department of Biomedical Informatics at the Harvard Medical School lead a discussion on “Ten simple rules for causal inference” in the spirit of the Ten Simple Rules section (https://collections.plos.org/ten-simple-rules) of PLOS Computational Biology.
November 30 – Kerollos Wanis lead a discussion on how to emulate a target trial using an administrative database. Ilya Shpitser (Johns Hopkins University – Biostatistics) briefly introduced the field of “learning causal structure from data.”
November 16 – Louisa Smith presented her proposed research on the impact of different fertility treatment strategies on child’s outcomes. Daniel Chicharro (from Harvard Medical School) presented his approach to causal inference in neuroscience by exploiting conditional variance homogeneity.
October 26 – Michalis Katsoulis presented his work studying physical activity, diet, and obesity using electronic medical record data. Paloma Rojas presented her work using The Rotterdam Study.
October 19 – Jessica Young and Mats Stensrud presented an overview of causal inference in the presence of competing risks, and their proposed solution to the problem.
October 11 – Sara Lodi and Miguel Hernan lead a discussion on the role of “cloning” when comparing sustained treatment strategies and strategies with a grace period.
September 21 – This discussion was framed around the use of observational databases to emulate hypothetical target trials. Lucia Petito discussed her work on the emulation of trials of cancer treatments and Kerollos Wanis presented his work on the emulation of trials of organ transplantation.
September 14 – Eleanor Murray and Sara Lodi lead a conversation about their work on per-protocol effects in randomized trials and their analogs in observational studies.
