Combined Discrete-event Simulation and Ant Colony Optimisation Approach for Selecting Optimal Screening Policies for Diabetic Retinopathy

Abstract

In this paper we present the first application to a healthcare problem of discrete-event simulation (DES) embedded in an ant colony optimisation (ACO) model. We are concerned with choosing optimal screening policies for retinopathy, a sight-threatening complication of diabetes. The early signs of retinopathy can be detected by screening before the patient is aware of symptoms, and blindness prevented by laser treatment. In this paper we describe the methodology used to combine the purpose-written DES model with the ACO algorithm. We simulate the effects of different screening strategies on a population of diabetic patients, and compare them in terms of two objective functions: Min C/E, cost-effectiveness (minimum incremental cost per year of sight saved, compared with a no-screening baseline) and Max E, maximum effectiveness (years of sight saved). We describe how ACO is used to optimise these two objectives, and discuss the issues involved in optimising stochastic variables. We present results for a range of different assumptions and scenarios about the format of screening programmes, using realistic data, and make policy recommendations on the basis of our findings.