A hybrid blended stacking disease prediction system based on symptoms

作     者：Indupalli, Manjula Rani Pradeepini, G. 

作者机构：Research Scholar Department of Computer Science and Engineering Koneru Lakshmaiah Education Foundation Vaddeswaram Andhra Pradesh522 302 India Professor Department of Computer Science and Engineering Koneru Lakshmaiah Education Foundation Vaddeswaram Andhra Pradesh522 302 India 

出 版 物：《Multimedia Tools and Applications》 (Multimedia Tools Appl)

年 卷 期：2025年第84卷第7期

页      面：3439-3457页

核心收录：

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：The authors thank to KLU for providing characterization supports to complete this research work 

主　　题：Accuracy Blended stacking CATegorical Boost Ensemble Multi classification Naive Bayesian 

摘      要：Traditional disease diagnosis methods often struggle with symptoms-based datasets containing categorical data, leading researchers to favor boosting algorithms like CatBoost for their computational efficiency;despite potential limitations in accuracy. To overcome these challenges, this study proposes a novel approach integrating ensemble boosting with traditional naive Bayesian techniques. Recognizing the risk of overfitting with pure ensemble methods, the model strategically blends non-linear models, including XGBoost, to assign weights to symptoms based on disease impact. The blending process, guided by XGBoost decision-making, significantly improves model accuracy compared to single-boost strategies. Preprocessing involves handling categorical data and missing values, enhancing the model s robustness. During the process of blending, the decision of XGBOOST plays an important role. The proposed model s accuracy of 97.2% is significantly higher than the single-boost strategy. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2025.