Reasoning with graded information: the case of diagnostic rating scales in healthcare

TL;DR

This paper investigates the use of fuzzy logic and prototype approaches to formalize reasoning with graded information in medical assessment questionnaires, aiming to mathematically justify scoring methods in healthcare diagnostics.

Contribution

It introduces two formal frameworks—fuzzy logic and prototype-based models—for reasoning with graded questionnaire data in medical diagnosis, providing logical and conceptual justifications.

Findings

Developed a fuzzy logic-based deduction system for score calculation.

Proposed a prototype approach framework with a well-justified model.

Compared the formal properties and applicability of both approaches.

Abstract

In medicine one frequently deals with vague information. As a tool for reasoning in this area, fuzzy logic suggests itself. In this paper we explore the applicability of the basic ideas of fuzzy set theory in the context of medical assessment questionnaires, which are commonly used, for instance, to support the diagnosis of psychological disorders. The items of a questionnaire are answered in a graded form; patients are asked to choose an element on a linear scale. The derived diagnostic hypotheses are graded as well. This leads to the question whether there is a logical formalism that is suitable to capture the score calculation of medical assessment questionnaires and thereby provides a mathematical justification of the way in which the calculation is typically done. We elaborate two alternative approaches to this problem. First, we follow the lines of mathematical fuzzy logic. For the proposed logic, which can deal with the formation of mean values, we present a Hilbert-style deduction system. In addition, we consider a variant of the prototype approach to vagueness. In this case we are led to a framework for which to obtain a logical calculus turns out to be difficult, yet our gain is a model that is conceptually comparably well-justifiable.