Diabetes is one of the most widespread diseases in the world. It occurs because of a malfunction of the pancreas, which leads to an increased concentration of glucose in the blood. Artificial intelligence technology is gaining ground in the field of diagnosis and treatment of various diseases. The purpose of this Master's thesis was a systematic review of research that examines and evaluates the use of artificial intelligence in the field of diagnosis, monitoring and support in the treatment of diabetes and associated complications. When creating and designing the systematic review, we considered the checklist and the phase diagram of the PRISMA criteria. Using the selected search profile, we searched for all research on the topic of diabetes and artificial intelligence in the PubMed bibliographic database. We analyzed the research according to the year of publication, the country of implementation, the type of research, the number and age of participants, gender structure of the sample, the type of diabetes, the field of artificial intelligence and the selected algorithm, and according to thematic groups. After considering all the inclusion and exclusion criteria, we included 54 studies in the systematic review. Randomized controlled trials predominated (N = 25, 46.3 %), followed by observational cohort studies (N = 18, 33.3 %) and observational cross-sectional studies (N = 11, 20.4 %). Most studies were conducted in Asia (N = 24, 44.4 %) and North America (N = 23, 42.6 %), including 20 in the USA (37 %) and 9 in China (16.7 %). 17 surveys (31.5 %) were conducted in Europe. The growing trend in the number of publications over the years indicates an increased use of artificial intelligence methods in various areas of diabetes. Among the various methods of artificial intelligence, machine learning plays the most prominent role, which was present in about 85% of the research. Most of the studies involved people with type 2 diabetes (N = 41, 75.9 %). Most studies (N = 21, 38.9 %) studied the role of artificial intelligence as a support in self-care and treatment of people with diabetes, and in 14 studies (25, 9 %) they tested predictive models for the occurrence of diabetes or its complications. 12 studies (22.2 %) dealt with the field of diabetic retinopathy, 5 studies (9.3%) with diabetic nephropathy and 2 studies (3.7 %) with diabetic neuropathy. The results of most of the studies, except for 3, show the positive impact of the use of artificial intelligence and the potential for its further use.
|
