An end stage kidney disease predictor based on an artificial neural networks ensemble

• Published in: Expert systems with applications 2013-09, Vol.40 (11), p.4438-4445
• Main Authors: Di Noia, Tommaso, Ostuni, Vito Claudio, Pesce, Francesco, Binetti, Giulio, Naso, David, Schena, Francesco Paolo, Di Sciascio, Eugenio
• Format: Article
• Language: English
• Subjects: Applied sciences; Artificial intelligence; Artificial neural networks; Biological and medical sciences; Clinical decision support system (CDSS); Combined surgery. Multiple transplantations; Computer programs; Computer science; control theory; systems; Computer systems and distributed systems. User interface; Computerized, statistical medical data processing and models in biomedicine; Connectionism. Neural networks; End stage kidney disease; Exact sciences and technology; Kidney diseases; Kidneys; Machine learning; Medical; Medical management aid. Diagnosis aid; Medical sciences; Networks; Neural networks ensemble; Patients; Risk; Software; Surgery (general aspects). Transplantations, organ and tissue grafts. Graft diseases
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2013.01.046

► We use an artificial neural networks (ANNs) ensemble to classify patients’ health status potentially leading to end stage kidney disease (ESKD). ► Development based on the largest available cohort worldwide for ESKD with data collected over a period of forty years. ► Evaluation of the classification results based on precision, recall and F-measure. ► Development of a RESTful application to deliver the risk prediction service; both a Web client and an Android application are available for direct evaluation. IgA Nephropathy (IgAN) is a worldwide disease that affects kidneys in human beings and leads to end-stage kidney disease (ESKD) thus requiring renal replacement therapy with dialysis or kidney transplantation. The need for new tools able to help clinicians in predicting ESKD risk for IgAN patients is highly recognized in the medical field. In this paper we present a software tool that exploits the power of artificial neural networks to classify patients’ health status potentially leading to ESKD. The classifier leverages the results returned by an ensemble of 10 networks trained by using data collected in a period of 38years at University of Bari. The developed tool has been made available both as an online Web application and as an Android mobile app. Noteworthy to its clinical usefulness is that its development is based on the largest available cohort worldwide.