Diabetes remains a major global health challenge and requires diagnostic systems capable of handling uncertainty and sometimes conflicting clinical evidence. In this study, a Disc T-Spherical Fuzzy (DT-SF) TOPSIS framework is proposed for diabetes risk assessment, where the radius parameter is used to encode the confidence associated with each diagnostic attribute. The methodology also integrates the Analytic Hierarchy Process (AHP) to determine the relative importance of several key risk factors, including blood glucose, body mass index, family history, lifestyle factors, and clinical symptoms. One important feature of the proposed approach is the ternary classification scheme, which categorizes patients as Non-diabetic (N), Prediabetic (P), or Diabetic (D). In particular, this scheme allows the explicit identification of patients located in a grey zone (Class P), where early monitoring and preventive intervention may be beneficial. The proposed framework is evaluated using the Pima Indians Diabetes Dataset (PIDD). The obtained results show that the ternary DT-SF TOPSIS model achieves 89.14% accuracy, while the conventional binary thresholding method reaches 75.91%. Further analysis of the Closeness Coefficient (CC) distributions, together with threshold sensitivity examination, supports the robustness and interpretability of the proposed framework. Overall, the findings indicate that the DT-SF TOPSIS model provides a practical, confidence-weighted, and uncertainty-aware tool for multi-criteria diabetes risk assessment, with possible applications to other chronic diseases.
Alagal et al. (Thu,) studied this question.