Commissioning procedure for predictive control based on ARX models of Type 1 Diabetes Mellitus patients

Abstract

This work presents a systematic way to assist on the commissioning of the Predictive Functional Control (PFC) algorithm to be implemented in the context of the Artificial Pancreas (AP) for patients with Type 1 Diabetes Mellitus (T1DM). PFC has demonstrated its high potentiality in process industry since several decades ago. Its use for AP has been specifically adapted and tested in the well validated model of the endocrine system, known as the Uva/Padova model. Firstly, the tests were done in silico using the parameters of 30 patients: 10 children, 10 adolescents and 10 adults. The performance of the controller was evaluated by means of the Control Variability Grid Analysis (CVGA) giving excellent results. We propose here the use of data driven models, as a second step, to perform a PFC final tuning to be practically ready to use in AP implementations. The closed loop responses are tested through dynamic simulations with the obtained models when different meals intake and insulin dosages are given to the patient Nº 5004 from the Center of Diabetes Technology (UVa/USA). In this paper an ARX model is build based on the patient’s own historical data recorded during 27 days. The ARX model was used giving good results at 30 min. prediction such as 89.1% of FIT, RMSE=10.312 mg/dl, and the Clarke Grid 97.58% at Zone A and 2.42% at B, while for 60 min. prediction, the FIT was 65.5%, RMSE 25.945 mg/dl and the Clarke Grid 88.93% at Zone A and 11.07% at B. The simplified models used in the PFC structure were obtained from the ARX model to isolate the insulin impact from the carbohydrates effects on the blood glucose variations. Finally, the full closed loop simulations, taking into account the constraints related to the insulin pump, were performed to conclude the PFC design. Closing the work, comparisons between the blood glucose behavior without control and with PFC and PID implementations are given to drive the final conclusions and discuss future works.