A novel hybridized prediction model for detection of chronic renal diseases using Multilayer Perceptron-Stochastic Gradient Descent based boosted classifiers

DOI: https://doie.org/10.1127/Jbse.2024145743

Alamma B.H., Manjula Sanjay Koti, C.H. Vanipriya

Keywords:

CKD, Prediction, Detection, Robustness, Anova, Correlation.

Abstract:

Chronic Kidney Disease (CKD) is a critical health condition that affects millions worldwide, necessitating effective early diagnosis to mitigate its progression and associated mortality. The primary challenge in CKD prediction lies in handling complex, high-dimensional medical data and addressing issues of data imbalance, feature selection, and integration of various predictive models. Traditional single-model approaches often fall short in capturing the intricate patterns necessary for accurate diagnosis. The proposed research aims to develop and build a novel hybridized CKD prediction model leveraging an ensemble of advanced machine learning techniques to enhance diagnostic accuracy and reliability. This hybrid model integrates Multilayer Perceptron (MLP), Stochastic Gradient Descent (SGD), Adaptive Boosting (Adaboost), Logistic Regression, and Random Forest, fortified by a series of robust methodologies and Clinical Prediction Models (CPMs). Therefore, a hybridized model, combining the strengths of various algorithms, is proposed to achieve a more comprehensive and robust CKD prediction system.
To improve model performance, feature selection methods such as ANOVA, Pearson correlations, and Cramer’s V tests are applied. Incorporating deep stacked autoencoder networks allows for effective learning from multimedia data, enhancing the model’s ability to process and interpret complex medical images and signals. Integrating CPMs provides a clinical context to the predictions, making the model’s output more relevant and actionable in real-world medical settings. This comprehensive approach not only enhances diagnostic accuracy but also provides a framework that can be adapted to other complex medical prediction tasks.
 

PDF