Pharmacological therapy selection of type 2 diabetes based on the SWARA and modified MULTIMOORA methods under a fuzzy environment

Publication date: Available online 30 March 2018
Source:Artificial Intelligence in Medicine
Author(s): M. Eghbali-Zarch, R. Tavakkoli-Moghaddam, F. Esfahanian, M.M. Sepehri, A. Azaron
Medication selection for Type 2 Diabetes (T2D) is a challenging medical decision-making problem involving multiple medications that can be prescribed to control the patient's blood glucose. The wide range of hyperglycemia lowering agents with varying effects and various side effects makes the decision quite difficult. This paper presents computer-aided medical decision support using a fuzzy Multi-Criteria Decision-Making (MCDM) model that hybridizes a Step-wise Weight Assessment Ratio Analysis (SWARA) method with a modification of Fuzzy Multi-Objective Optimization on the basis of a Ratio Analysis plus the full multiplicative form (FMULTIMOORA) method for pharmacological therapy selection of T2D. It makes the use of SWARA for obtaining the relative significance of every selected criterion by soliciting experts' opinions and FMULTIMOORA method for evaluation of each alternative according to all criteria based on a published clinical guideline. In this paper, an extended reference point approach is considered in the proposed hybrid MCDM model that resolves the classic reference point limitations and improves the FMULTIMOORA ranking procedure. Computational results indicate that Metformin is confirmed as the first-line medication and Sulfonylurea as the second-line add-on therapy. The Glucagon-like peptide-1 receptor agonist, Dipeptidyl peptidase-4 inhibitor, and Insulin are placed 3rd, 4th, and 5th, respectively. A sensitivity analysis is conducted to validate the model performance by comparing its result with studies in the literature, other fuzzy MCDM techniques and an interval MULTIMOORA method based on an observational dataset. The close correspondence between the final rankings of anti-diabetic agents resulted from the proposed hybrid model and other methodologies provide significant implications for endocrinologists to refer.



https://ift.tt/2uujt4g