Optimization Of Insulin Pump Therapy Using Swarm Intelligence Algorithm

Diabetes is a common endocrine disease, which affects the daily life and harms the health of patients seriously, and will cause a series of complications in long-term hyperglycemia. Currently, the common-used intensive insulin therapies can be categorized as multiple daily injection and continuous subcutaneous insulin injection (insulin pump). Continuous subcutaneous insulin injection is more close to people’s physiological model. Insulin pump controls patients’blood glucose by adjusting basal insulin and controls postprandial hyperglycemia by adjusting bolus insulin. Although the application of insulin pump is more widely, there is less infusion mode researches on basal insulin and bolus insulin of insulin pump. Clinically, the basal insulin and bolus insulin often rely on the experience of physician and formulas of handbook of insulin pump; however, there are great differences between various patients, e.g., weight, insulin sensitivity, and so on, so it is difficult or even impossible to design a fix and reasonable basal and bolus for all patients. Hence, the optimization on basal insulin and bolus insulin is very important.This paper proposes a novel particle swarm optimization (PSO) algorithm with an intelligent weighting mechanism, termed as WPSO. The bright point is a framework composed by any search methods, and it gets the further optimized value by linked with a time-varying weight. It is important to select the weight. In this study, the weight selection is proportional to the optimized performance of the independent search method. This paper adopts the non-uniform mutation operator, differential mutation operator and random local search algorithm, non-uniform mutation has the advantages of local searching, differential mutation operator has the advantage of keeping diversity and stochastic local search algorithm has the ability of keeping balance of the local search and global search. Therefore, the improved particle swarm algorithm has local searching and global searching ability. The proposed algorithm was compared with standard PSO and other four famous optimization algorithms on 15 well-known benchmark optimization functions. The optimization results show that the proposed algorithm can enhance the searching efficiency and improve the searching quality effectively. Moreover, this paper proposes a method to regulate basal insulin infusion rate and bolus insulin automatically based on the proposed WPSO, without human intervention. This paper gives a simulation on 10 virtual subjects, and the simulation results show that the proposed WPSO can rapidly control the blood glucose in a safe range and has superior performance in adjusting of basal and bolus insulin for Type 1 diabetes mellitus.