Wed Oct 23, 2013 7:00pm EDT —
Wed Oct 23, 2013 8:00pm EDT
Accurate estimates of prognosis are crucial for clinical decision-making as well as risk adjustment. Current methods for prognostication, based on comorbidities captured in administrative datasets, can identify significant predictors of mortality at the aggregate level but fail to produce meaningful estimates for individual patients. We begin with an example of what is at stake, both for patients and the health system, by showing preliminary data on the costs of caring for patients with end-stage cancer in the US. We then illustrate the challenges of applying current predictive models to individuals and propose directions for future research.