Machine learning application development to predict blood glucose level based on real time patient data

摘要

The given paper details the development of a decision support system application to help for people with type 1 diabetes mellitus. The developed solution is based on supervised machine learning and it focuses to predict the future blood glucose level to support decision making of patients with conservative therapy. We applied the Tensorflow and Keras framework during our work to make our solutions embedded system compatible. We applied the AIDA diabetes simulator to generate data for the proof-of-concept. We found that our result are promising and the performance of the developed solutions are able to satisfy the requirements related the proof-of-concept.