Diabetes Mellitus is a group of metabolic disorders associated with abnormally high levels of glucose in the blood. In general, the body controls the concentration of glucose in the blood relatively to the cells with two hormones: insulin and glucagon. In Type 1 Diabetes Mellitus (T1DM) the pancreas produces little to no insulin and in Type 2 Diabetes Mellitus the insulin receptors become resistant to insulin. Both conditions may cause serious complications such as neuropathy, nephropathy, retinopathy and extremity damage when not treated. Patients diagnosed with T1DM need to administer external insulin injections in order to maintain the blood glucose level within normal ranges. Dosage of the insulin injection is a critical decision for the patient to prevent hypoglycemic and hyperglycemic state. Today’s technology uses glucose measuring systems, which allows patients to monitor their blood glucose levels and administer the injections by using an insulin pump.
The purpose of this study is to create an artificial neural network to predict the future blood glucose levels and implement a control algorithm to determine the optimum insulin injection amount to keep the blood glucose levels (BGL) of a T1DM laboratory rat within the control limits of 70 mg/dl to 180mg/dl with the reference value of 125 mg/dl. For the creation of the time series neural network, BGL data were collected from a CGM sensor, insulin injection data from an open source artificial pancreas system and food intake data from a special designed scale. All four streptozotocin-induced diabetic rats were allowed to eat food ad libitum during the 24-hour experiments. Dustless sugar pellets (94.5% carbohydrate) were given to rats as a source of food during the experiments.
Proposed Time Series Neural Network Based Model Predictive Control (NN-MPC) showed promising results. An experiment with the NN-MPC was conducted on the rat. During the 24-hour duration time of the experiment, 89.2% of actually BGL outputs of T1DM rat fell within the desired control limit with a root square mean error of 47.4 mg/dl and mean of 130 mg/dl. Proposed method with further developments and research can be used as the control algorithm for Artificial Pancreas Systems.
|Advisor:||Ko, Hoo S.|
|Commitee:||Lee, Heungsoon F., Kwon, Guim|
|School:||Southern Illinois University at Edwardsville|
|School Location:||United States -- Illinois|
|Source:||MAI 82/3(E), Masters Abstracts International|
|Subjects:||Computer Engineering, Biomedical engineering|
|Keywords:||Blood glucose regulation, Type 1 diabetic rats, Time series neural network|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be