Early detection of the risk of depressive episodes using a proprietary diagnostic test by new epistemic similarity measures

Abstract

The first primary motivation of this paper is to propose new methods for measuring similarity between interval-valued fuzzy sets based on an order relation of the intervals, which are a possible and necessary similarity measure. To capture the basic idea of the suggested approach, one has to realize that an interval-valued fuzzy set (IVFS) can be considered as a family of closed intervals describing imprecise information, or uncertainty, on the possible belongingness of each element of the universe of discourse and the dependence between them. Consequently, any relation between two interval-valued fuzzy sets could be analyzed using the links between the corresponding families of intervals. The core of the proposed methodology is based on the description of possible and necessary inclusion and the width of the corresponding intervals in epistemic issues, i.e. optimistic (possible) and pessimistic (necessary) measures. Our second motivation is to use our concept of measures to early detect the risk of depressive episodes among students tested for this research. We propose an effective algorithm to estimate the risk of depressive episodes. By using machine learning methods concerning uncertainty, we obtained more sensitivity to diagnosis compared to the currently used tests, e.g., CESD-R. We aim to identify the risk of depressive states before they are indicated in other tests such as CESD-R, which can prevent the appearance of symptoms after early medical consultation. In the initial study, we were able to achieve a 4-week faster diagnosis of at-risk depression in 20% of subjects who had CESD-R at risk, so we are talking about a more sensitive test, in this sense, as a result of this work. The proposed methodology, in the form of a system, will be used in medical centers by general practitioners or entrepreneurs for their subordinates to eliminate depression-related breaks at an early prevention stage.