Abstract for: When accuracy matters: Prevalence, incidence and mortality rate in an Agent Based model of dementia management

An agent-based (AB) model of dementia management required that an accurate number of virtual patients be initialised at the beginning of the simulation, that new patients are added as time progresses and that some groups of patients are gradually removed from the simulation. This is the equivalent of prevalence, incidence and mortality in demographic analysis of the disease. We built a model in which these required parameters were taken from static look-up tables containing average data published in the literature. Our test results revealed that due to the probabilistic nature of the AB model it is difficult to accurately calibrate the model using average estimates of these three parameters over time. The proposed solution is based on implementation of a System Dynamics (SD) model that can drive the agent-based model. The possible benefits of such approach are further discussed and include the implementation of a feedback loop between the AB and SD parts of the model.