CHAOS BASED BLOOD GLUCOSE PREDICTION AND INSULIN ADJUSTMENT FOR DIABETES MELLITUS

摘要

Many diabetes mellitus (DM) patients are concerned about unstable blood glucose, even with regular monitoring by a family doctor. The insulin preparation shows peak action at several hours after subcutaneous administration. For this reason, unstable blood glucose is often caused by intensive insulin therapy with self-monitored blood glucose (SMBG). The insulin requirement is determined in proportion to fasting blood glucose (FBG) in the sliding scale method. A fixed amount of insulin is administered regardless of FBG level in the constant insulin method. It is indispensable to estimate FBG at peak time, when insulin works the hardest, to obtain the appropriate effect of insulin administration. We employed the local fuzzy reconstruction method based on chaos theory for predicting FBG at peak time. The direction to change the FBG at peak time is predicted with a 70-90% success rate. The amount of insulin administration is adjusted based on predicted FBG level. After predictive grycemic control (PGC) for around one year, the FBG average approached the normal range, standard deviation (SD) reduced by half, and hyperglycemia decreased.