Improving prediction of glycaemia course after different meals—New individualized approach

Abstract

Motivation and objectives: Diabetes is one of the biggest medical problems nowadays, having different forms and different mechanism of development but the ultimate result is the same—hyperglycaemia. Hyperglycaemia leads to development of chronic diabetic complications, which are the most frequent cause of worsening patient’s life quality and often shortening life expectancy. All diabetes mellitus (DM) type 1 patients and some of DM type 2 patients require full insulin substitution. It is not simple to adjust insulin dose to different meals and different daily activities. To help patients with this challenge we started to develop an application for smart phones having new features in comparison with existing applications. We concentrated on individual response to different types of meals (division based on glycaemic index), to physical activity and individually different basal metabolic rate. Material and methods: So far 24 patients, mostly insulin pump users, were enrolled. Patients used during the study at least 4, weeks RT-CGM and during this period they were asked to document all food and drinks containing carbohydrates by smart phone camera. Patients wrote during this time a detailed logbook as well. The detailed nutritional analysis of patient’s food was done as well as evaluation of other condition (level of depression, measurement of basal metabolic rate). Results: The quality of photos was problematic but the biggest problem was to analyze mixed meal from the photography. It was not possible without at least short patient’s description. Patient’s diet was unhealthy (high fat content etc.) and patients despite remedial nutritional reeducation made mistakes in carbohydrates counting which was reflected in their glycaemia profiles. Conclusion: It seems that using photos with brief notes is an acceptable solution and adding a personalized database of favourite meals with correct nutritional data (which we are developing now) may be very helpful. Then the patient only confirms selected meal and does not need to insert all data again. Based on data from the insulin pump and the glucose sensor and inserted information about the planned meal from the patient, the application can recommend the prandial bolus to be injected before meal.