Modelling Glucose Regulatory System: Adaptive System Dynamic Approach
Permanent address of the item is
Thesis describes a project that starts from evaluation of simulation programs and ends to testing an individually adaptive glucose regulatory system (GRS) model. Thesis presents a modern adaptive approach to model GRS in order to describe each diabetic’s individual causalities. Thesis is divided into three parts; a literature study of diabetes and GRS models, analysis of simulation programs, and building dynamics GRS model and validating it with clinical data. Validation consists literature data for general GRS model and test data from a pilot diabetic who underwent two-week study period. Data collected included glucose values from two continuous glucose monitors (CGM), fingertip blood glucose measurements, meals and exercises. Adaptive parameter identification was applied to the model during 6 days training period and then blood glucose was estimated for the next 24 hours. First part of results show that from four simulation programs analyzed, Simulink was the software best meeting Quattro Folia’s functional requirements and demanded qualities. Therefore, a general GRS model was built with it. Based on literature review, the best model and parts of models were combined for one general model which was validated to function as in previous studies. Second part of results show that with adequate data, blood glucose can be estimated with decent accuracy. Although the material only consist data from one diabetic subject, it gives an indication that blood glucose could be estimated for others also. However, the precision over population is indecisive. To conclude, individual diabetic’s GRS and its functions can be described with adaptive system dynamic model. The model have multiple possible usages from in silico testing to teaching causalities for diabetics or their parents, thus it is useful for research, validation and educational purposes. Its value creators are modularity and wide range of possible usages.