Dr. Jinbo Chen, Professor of Biostatistics, recently led a four-session workshop on statistical methods for risk modeling and model evaluation, bringing together biostatistics and clinical colleagues to address challenges in developing and evaluating health algorithms.
“As more and more statistical, machine learning and AI algorithms get to operate in health systems, we want to make sure that they are accurate, fair, and consistent in performance over time,” said Chen.
Organized through the Statistical Center for Translational Research in Medicine (SC-TRM), which Chen directs, the workshop explored challenges in risk modeling using Penn EHR and biobank data, as well as the fairness and sustainability of algorithms used in the Penn and VA health systems.
We designed this workshop with two main purposes. First, to raise awareness of the complexities involved in risk modeling and evaluation, for example, modeling with Penn EHR and biobank data, and ensuring fairness and sustainability of algorithms deployed in the Penn and VA health systems. Second, to showcase the in-house statistical methodologies we have developed to address these challenges.
Professor of Biostatistics
The workshop was well attended, with participants identifying applications for the methods in their own research. Chen hopes to expand opportunities for trainees, postdoctoral fellows, and students in the future.
“My vision is for the Statistical Center for Translational Research in Medicine to serve as a central hub for training in statistical methodology for risk modeling, model evaluation, and algorithm fairness and reliability.”