Setback in Ranking Fuzzy Numbers: A Study in Fuzzy Risk Analysis in Diabetes Prediction Publisher



Sotoudehanvari M¹ ; Sotoudehanvari A² Show Affiliations
Source: Artificial Intelligence Review Published:2023

Abstract

The prevalence of diabetes is terribly increasing worldwide and every 5 s a person dies from this disease. (The IDF Diabetes Atlas, Tenth Edition (https://diabetesatlas.org/) Last visit: 7.14.2022)). Hence, timely detection of diabetes is very vital to prevent or delay the complications. Many researchers have believed that to deal with the vagueness in knowledge and complexity of diabetes, fuzzy set needs to be incorporated into diabetes prediction models to provide more realistic results. In a fuzzy environment, ranking of fuzzy numbers is a crucial prerequisite to make a decision. In this paper, a new phenomenon in ranking fuzzy numbers that we call “setback” is introduced. In setback which indicates the most confusing state in fuzzy ranking, the ranking order of fuzzy numbers is completely reversed when a new ranking method is applied to the same problem. Also a fuzzy risk analysis problem in diabetes prediction is studied in which person with the highest risk of diabetes is replaced with the lowest one and vice versa. With this result in hand, we reveal that defective ranking results in a medical problem have the potential to lead to disastrous effects on human health. We expose some potential causes of “setback” and to alleviate this problem two evaluation criteria and a benchmark are suggested. These criteria help researchers avoid severe pitfalls when introducing their own ranking methods. To achieve this goal, we first address serious drawbacks of some fuzzy ranking methods which have been recently published by reputed journals. Although the aim of this paper is not to challenge the theoretical advances in ranking methods, this paper shows that this field requires deeper studies to be proper to apply in actual cases, particularly medical decision-making. Hence, if the users do not pay attention to various aspects of ranking methods, they may fall into the trap of paradoxical results. This paper is expected to be useful for the medical researchers working in this field and for the scholars thinking about realization of fuzzy sets and ranking fuzzy numbers for real world applications. © 2022, The Author(s), under exclusive licence to Springer Nature B.V.