A decision-making system based on case-based reasoning for predicting stroke rehabilitation demands in heterogeneous information environment

Abstract

Recently, the increasing number of stroke survivors has created a greater demand for rehabilitation services. How to predict rehabilitation demands is important for policy makers in various countries. The demand prediction is essentially a multi-attribute decision-making problem. The diverse forms of information or data are often neglected or simplified in previous studies of decision-making problems. In this paper, heterogeneous multi-attributes including crisp numbers, interval numbers, and linguistic variables are defined to represent decision-relevant information concerning the patient's condition in cases when applied to real-life scenarios. Moreover, a deviation minimization optimization model combining CRITIC method and H2TLWMSM operator is proposed to assign the attribute weights with dual information. The WHMACBR system can be used to predict stroke rehabilitation demands for making the diagnosis and treatment decisions. The proposed decision-making system based on the CBR approach simulates the operations of human memory and reasoning, thus assisting physicians in addressing the ever-increasing demand for rehabilitation services of stroke survivors. Finally, the effectiveness and applicability of the case retrieval process of the presented CBR model are demonstrated through a stroke rehabilitation case study (top 3 cases ranked by similarity to the target case are C6, C2, and C3) and comparison analysis with relevant existing CBR models. This indicates that the proposed decision-making system, by considering heterogeneous multi-attributes and their weights, is capable of providing more comprehensive and rational decision-related information in the field of rehabilitation medicine.