article · Applied Sciences
Chronic kidney disease involves a gradual loss of kidney function over time, making early detection vital for guiding treatments such as medication, dialysis, or organ transplantation. Computational tools using artificial intelligence offer high accuracy for medical diagnosis, though their effectiveness relies on selecting suitable algorithms and relevant clinical indicators. An ensemble deep learning approach was developed to detect the disease by combining pretrained deep learning models with a support vector machine operating as a metalearner. Multiple feature selection techniques were tested to identify the most informative diagnostic variables, alongside an evaluation of their medical significance. Tested on a dataset of 400 patient records from the UCI machine learning repository, the ensemble system demonstrated efficient predictive performance compared to alternative models, achieving its highest diagnostic results when paired with the mutual information classification feature selection method.
Early identification of chronic kidney disease allows healthcare providers to intervene before severe organ failure occurs. By improving how artificial intelligence selects clinical features and combines diagnostic algorithms, computer-aided screening tools can offer more reliable predictions. This supports clinicians in choosing timely therapies, potentially delaying disease progression and reducing the need for intensive treatments like dialysis or transplantation.
This research points toward clinical decision support software that assists healthcare practitioners in diagnosing chronic kidney disease from patient data. Given that the system has only been validated on a standard benchmark dataset of 400 records from a public repository, the technology remains at an early, laboratory-based stage of research and would require clinical trials and validation on broader real-world patient cohorts before practical healthcare deployment.
AI-generated from the published abstract. Always read the original work before citing.
Chronic kidney disease (CKD) refers to the gradual decline of kidney function over months or years. Early detection of CKD is crucial and significantly affects a patient’s decreasing health progression through several methods, including pharmacological intervention in mild cases or hemodialysis and kidney transportation in severe cases. In the recent past, machine learning (ML) and deep learning (DL) models have become important in the medical diagnosis domain due to their high prediction accuracy. The performance of the developed model mainly depends on choosing the appropriate features and suitable algorithms. Accordingly, the paper aims to introduce a novel ensemble DL approach to detect CKD; multiple methods of feature selection were used to select the optimal selected features. Moreover, we study the effect of the optimal features chosen on CKD from the medical side. The proposed ensemble model integrates pretrained DL models with the support vector machine (SVM) as the metalearner model. Extensive experiments were conducted by using 400 patients from the UCI machine learning repository. The results demonstrate the efficiency of the proposed model in CKD prediction compared to other models. The proposed model with selected features using mutual_info_classi obtained the highest performance.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.3390/app13063937
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
