SGA-DT: An adaptive fusion framework for missing data imputation and interpretable healthcare classification

Read full text Summarise with VIRAT
Authors: Monalisa Jena, Satchidananda Dehuri, Sung‐Bae Cho
Cited by: 0
References: 36

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Despite advances in machine learning and medical data processing, handling missing values remains a critical and complex challenge in healthcare analytics. Missing data, especially in non-class attributes can severely compromise model accuracy, clinical reliability, and interpretability. In sensitive domains such as healthcare, improper imputation may lead to biased outcomes or delayed interventions. To address this challenge, we propose SGA-DT, an adaptive and interpretable learning framework that combines the best features of genetically optimized support vector regression (SVR) with a decision tree (DT) classifier for robust healthcare prediction. The framework adaptively selects an imputation strategy based on the level of missingness. It uses standard SVR for low, iterative SVR for moderate, and k-Nearest Neighbor (KNN) followed by SVR refinement for high missingness. Genetic algorithm (GA) is used to select the best SVR kernel and tune its hyperparameters, enhancing imputation accuracy across different data patterns. The complete dataset is then classified using DT, providing both robustness and transparency in prediction. The SGA-DT framework is evaluated on three healthcare datasets, Breast Cancer, Mammographic, and Hepatitis, along with other real-world and synthetic datasets. For interpretability analysis, decision trees are generated under varying missingness levels to support clinical transparency. Comparative results show that SGA-DT consistently outperforms multiple integrated frameworks across accuracy, precision, recall, and F-measure, demonstrating its robustness, interpretability, and generalizability in healthcare prediction tasks.

Fields of Study & Subjects

Artificial Intelligence in Healthcare Machine Learning and Data Classification Data Mining and Machine Learning Applications

Publication details

Journal / source
PLoS ONE
Volume
21
Issue
3
Pages
e0343619–e0343619
Publication date
26 March 2026
Language
EN