Julie Higle, Ph.D.
Department of Industrial and Systems Engineering, University of Southern California
Abstract: Modeling for Medical Decision Making: Data, Models, Decisions and Unintended Consequences
Increasingly often, model-based analyses of screening and treatment strategies for medical conditions are used to inform health policy and its implementation. They permit an exploration of a broader range of strategies than might be tested in clinical studies, including those that are configured hypothetically.
A central component of a model-based analysis is the natural history model, which represents the evolution of disease in the absence of medical intervention. The development of a natural history model can be a delicate process, requiring significant navigation around issues involving “data” and “model parameters”. The construction of the model involves various data sources, and modeling techniques are necessary to estimate data that are not available through clinical studies.
This seminar will discuss experiences with natural history models for cervical and ovarian cancers, and will segue into an exploration of the way that results and recommendations are impacted by the way that we explore our models.
A healthy discussion of the pros and cons of various modeling decisions will undoubtedly emerge.