Study on insulin infusion estimation based on continuous blood glucose data

Insulin pump is the key equipment for intensive insulin treatment of diabetes, but it often has failures such as tube blockage and leaking tubes. Such failures can lead to insufficient insulin infusions, which make it difficult for patients to lower blood glucose and endanger the health. Estimating the insulin infusion is helpful for fault diagnosis of insulin pumps. This paper proposes a method for estimating insulin infusion based on dynamic blood glucose data. Firstly, a glucose-insulin model is built based on physiological mechanisms. Secondly, a particle filter estimator is established. Finally, based on the monitored glucose data, insulin infusion is estimated dynamically. It is verified by simulation that the method can estimate the real insulin infusion rate effectively. The method has the potential to provide a path for insulin pump fault diagnosis.