An effective application of insulin treatment in diabetic patients requires ability of the patient to select proper insulin doses, which compensate ingested meals. This is difficult, time consuming and in some cases outreach patient’s abilities.
The main objective of the developed VoiceDiab system is to help patients in calculation of the proper insulin dose based on voice description of meal. The system consists of several software applications (on smartphone and MS Windows server) working with developed linguistic database and nutrient database. The system uses commercial speech recognition software which was tuned for the meal description and its efficacy in speech translation is up to 97%. The algorithm for transition from the text description of meal into amount of carbohydrates, protein and fat can work in advanced mode when patient has to describe all meal segments and all units of measure, and all numbers related to each product, or it can work in simple mode when only one product, one unit of measure and one number should appear in the meal description. The module, for calculation of insulin dosing for users of insulin pen or insulin pump, is working based on the amount of carbohydrates, protein and fat earlier identified in the meal.
Implementation of the system should lead to better metabolic control and as a consequence to prevention of late diabetic complications, giving medical, social and economical effects.