Title: Deep learning-based medical expert system for diabetes diagnosis on IoT healthcare data
Authors: K. Vijayaprabakaran; K. Sathiyamurthy; S. Sowmya
Addresses: Department of Computer Science, Pondicherry Engineering College, Puducherry, 605 014, India ' Department of Computer Science, Pondicherry Engineering College, Puducherry, 605 014, India ' Department of Computer Science, Pondicherry Engineering College, Puducherry, 605 014, India
Abstract: Machine learning in internet of things (IoT) plays a vital role in diagnosing diseases and predicting the risk level of health by analysing patient health records. Diabetes has been commonly seen in all age groups of peoples. Early diagnosis of diabetes and proper medication will help the patient live normally for long life. In this study, various machine learning techniques were experimented to diagnose diabetes and the results were compared. In order to diagnose diabetes from the health record of the patient, this work proposes a deep learning-based expert system (DL-Expert sys). The DL-Expert system predicts the risk level of the patient and provides the recommendation of the diet to the patient. The experimental results illustrate that the predictive model using deep learning algorithm of RNN with LSTM achieves higher accuracy than logistic regression, Naive Bayes and neural network.
Keywords: machine learning simple linear regression; multiple linear regression; logistic regression; Naïve Bayes; recurrent neural network with GRU; recurrent neural network with LSTM.
International Journal of Social Computing and Cyber-Physical Systems, 2021 Vol.2 No.3, pp.177 - 193
Received: 10 Oct 2019
Accepted: 18 Feb 2020
Published online: 05 Oct 2021 *