Insulin Recommender Systems for T1DM: A Review

In type 1 diabetes mellitus (T1DM) pancreas beta-cells do not segregate insulin. This hormone is necessary to convert glucose into energy. Thus, people with diabetes are required to maintain blood glucose (BG) levels within a safe range using external control solutions. Insulin recommender systems (IRS's) provide the precise amount of insulin to the patient when needed, reducing the effects of the disease. The goal of this paper is to review and summarize all current proposals of IRS's and, with this purpose, 70 papers have been analysed. The analysis of the works was performed taking the following aspects into account: (i) technology of the recommendation process, (ii) control procedures, (iii) complementary processes, (iv) hardware, testing and assessment, (v) pricing and (vi) results. Those are our main conclusions after the review: There is a lack of published research works providing real experimentation together with simulation processes. Information about the IRS's features is also lacking in a remarkable percentage of the publications. Due to the variability in how experiments are performed and results are presented, research work comparisons become difficult. In summary, this topic requires standards to be able to perform comparison analysis of published papers and therefore, progress adequately.