As part of the 4th Midwest Healthcare Conference, we are hosting a Mini Data Challenge, which offers a valuable opportunity for participants to apply causal inference methodologies to real-world data. By focusing on the causal effects of glucocorticoids on COVID-19 survival, participants contribute to a critical area of research with significant clinical implications.
Structural causal models (SCMs) provide a framework for understanding and analyzing cause-and-effect relationships within complex systems. Using directed acyclic graphs, the causal structure between variables can be represented to distinguish correlation from causation, identify causal pathways, predict outcomes under interventions, and estimate effects.
The challenge aims to optimize the use of SCMs to estimate the causal effects of glucocorticoids and hydroxychloroquine on in-hospital survival rates among COVID-19 patients using real-world data.
The primary goal is to assess how participants can develop SCMs to estimate the causal effects of glucocorticoids and hydroxychloroquine on 28-day survival rates among COVID-19 patients using a large de-identified COVID-19 dataset, stratified by disease severity.
Estimate the causal effects of glucocorticoids and hydroxychloroquine on the 28-day survival of COVID-19 patients stratified by COVID-19 disease severity (low, moderate, severe) using a structural causal model and the provided real-world dataset.
Participants submit causal diagrams (formally, SCMs) that describe the causal factors of all-cause 28-day survival in the context of COVID-19 infection. The SCMs define variables that may introduce confounding bias, such as age, race, and gender, that must be handled appropriately. Participants may use inverse probability of treatment weights or doubly robust models.
An observational dataset collected at a large U.S. health system during the early waves of the global pandemic is provided to registered participants. Only data provided by organizers may be used.
Submitted causal diagrams are used to assess the impact of glucocorticoids and hydroxychloroquine on 28-day all-cause survival rates. Estimates are expressed as relative risk ratios with bootstrapped confidence intervals and stratified by disease severity. Participants also submit documentation describing their modeling strategy for reproducibility.
The competition is held on the cStructure platform. Teams register, obtain platform access, analyze the data, build and refine their model, and submit a final entry by the deadline.
Data are de-identified to ensure patient privacy. The challenge emphasizes model transparency and ethical compliance in data handling.
The primary metric is alignment between causal effect estimates produced by team SCMs and findings from high-quality randomized controlled trials. Secondary criteria include rigor, plausible causal relationships, creativity, precision, and clear documentation.