In silico modeling systems: learning about the prevalence and dynamics of dementia through virtual experimentation

Abstract

Background: Virtual experimentation using computer modeling creates opportunities for researchers who want to better understand disease processes, foresee effects of future demographics, and evaluate combinations of interventions when applied to larger target groups.

Methods: We created a computer model of dementia prevalence consisting of six population groups representing diagnosed and undiagnosed dementia at mild, moderate, and severe levels. Dynamic transitions between these groups corresponded to the gradual progression of disease. The seventh group represented the general population without dementia aged >60 years from which new dementia cases emerged. Through a series of virtual experiments we estimated future changes in the severity-specific prevalence of dementia in Australia.

Results: The projected total prevalence of dementia in Australia for year 2040 changed from 742,000 to 986,000 (+33%) and to 433,000 (-42%) when the incidence rate was altered by ±50%. Increasing the transition time between mild and moderate dementia from 5 to 7 years and between moderate to severe from 7 to 9 years increased the prevalence of mild dementia by 23% and decreased the prevalence of severe dementia by 24%.

Conclusions: As computer modeling becomes more accepted, in silico experiments are being routinely performed to update demographic projections. Despite its simplicity, the framework of this model integrates a large pool of knowledge and consists of components which are dynamically interconnected. The computational logic underpins series of assumptions and binds them together with demographic data.