An Economical AI for Diabetes Management

Lead Research Organisation: University College London
Department Name: Institute of Health Informatics

Abstract

Aims:
1. To develop an AI system to predict glucose levels from a few samples
2. Use this prediction system for glucose control and management
3. Use the insight gained in solving this problem to make advancements in general machine
learning capabilities that impact both this and other problems

Project Motivations

Diabetes is a chronic disease that affects an estimated 422 million people in the world, estimated to be 1 in 11 of the world's adult population [1]. 90% of people living with diabetes have Type 2 diabetes (T2D), which is a serious condition where insulin is either ineffective, or the pancreas can't make enough insulin. To manage glucose levels, people with diabetes usually use self- monitored blood glucose (SMBG) to assess glucose and inform decision-making by testing the blood several time a day. With the development of new continuous glucose monitoring (CGM), more and more people use CGM to monitor their real-time glucose level, due to the inconvenience and pain associated with fingersticks. It also changes the clinical diagnosis of diabetes from pure glycated hemoglobin (A1C) test to time in target range (TIR) of glucose level.
CGM is good. However, only a few patients can afford it. It costs c. 1000 pounds for a standalone device and 60 pounds for sensors (they last for 2 weeks). The NHS provides CGM only to people having T1D with severe conditions (<20% in US). Some clinics even lend patients CGMs for a short time to help patients look for patterns in their blood glucose levels. The high cost of CGM and shortage of funding mean many people with T2D do not receive proper close glucose monitoring or prognosis, which can lead to many complications. Thus, there is an opportunity to use AI to develop an economic glucose monitoring system that can fill the gap between expensive/real-time CGM, and cheap/several-samples-a-day SMBG.

Publications

10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/S021612/1 01/04/2019 30/09/2027
2418752 Studentship EP/S021612/1 28/09/2020 30/09/2024 Edward Rees