A PREDICTIVE ALGORITHM FOR EARLY DETECTION OF NEUROPATHIC DIABETES USING MACHINE LEARNING

Subject Area: HEALTH

This paper presents a predictive algorithm for the early detection of neuropathic diabetes using machine learning. The study aimed at developing a machine learning model which can classify features of neuropathic diabetes from clinical data and then diagnose the patient. The methodology for the system development is the structural system analysis and structural design model. The machine learning used is the neural network which was trained which data collected from Niger foundation hospital, Enugu, Nigeria, to generate the predictive model which was used to model a neuropathic diabetes prediction system. The result when tested and validated using the tenfold cross validation technique showed an average mean square error performance value of 4.5992e-11 and a regression value of 9.881. The result implied good training performance for the neurons and also good detection of neuropathic diabetes signs. The algorithm when compared with other state-of-the-art predictive algorithms achieved better regression performance with a percentage improvement of 0.08%.