The HyperSoft Set (HSS) is a powerful tool for Multi-Criteria Group Decision-Making (MCGDM) problems because it expands on the concept of the soft set by combining many sets of qualities. The function F in this framework is a multi-argument function. The importance of uncertainty in medical practice is becoming more widely recognized, yet research on this topic remains fragmented across various disciplines. Considering several attributes and their sub-divisions, ambiguity, imprecision, and uncertainty make the Data Mining (DM) complex. The Fuzzy HyperSoft Set (FHSS) combined with the Weight-Based Support Vector Machine (WSVM) algorithm is presented in this study to overcome those complex problems. This study mainly emphasizes detecting critical symptoms to diagnose diseases. Initially, the K-Means Clustering (KMC) algorithm was employed to pre-process the dataset. The noise from the data can be effectively eliminated by this KMC method. This process significantly improved the accuracy of medical Data Classification (DC). This uncertainty became a basic feature of people's lives. Each attribute is attributed to a group of possible objects in the discourse world. The FHSS method uses the Fuzzy Membership (FM) to handle uncertain data. This integration will also support expressing those data in detail, and DM was also enhanced. For medical diagnosis, the WSVM algorithm is then employed. Classification outcomes were improved by employing this WSVM method in a dataset. Experimental outcomes indicate that the suggested FHSS-WSVM algorithm executes better than the current Accuracy, precision, recall, and F-measure methods. The model was evaluated using the Cleveland heart disease dataset, comprising 303 patient records with 13 diagnostic attributes. Comparative analysis is conducted against conventional classifiers such as standard SVM, Random Forest, and fuzzy soft set-based methods. Experimental results demonstrate the superior performance of FHSS-WSVM, achieving 92.3 accuracy, 91.6 precision, 90.8 recall, and an F-measure of 91.1, outperforming baseline models by statistically significant margins (p < 0.05). © 2025 Elsevier B.V., All rights reserved.