Abstract
In the modern world, technology plays a major role in making processes easy, efficient, and mostly automated no matter where they take place. Technology such as artificial intelligence, the Internet of Things, blockchain, and deep learning has revolutionized the growth of many fields. One of the best examples is the medical field, in which timely and accurate diagnoses must be made routinely; traditional systems are of little help due to their lack of accuracy and the time delays they introduce. Instead, advancements in deep learning (DL) and the Internet of Things (IoT) are useful in building effective models for timely and accurate diagnosis and developing a smart health care system. In this paper, we propose a disease diagnosis model using DL in combination with IoT. The stages involved in the model are as follows: (a) Data are collected from various IoT wearable devices, in which sensors play a vital role in collecting data and relaying these data to DL systems for accurate diagnosis. (b) These medical data are preprocessed, as they contain noise. (c) The preprocessed data are passed to an isolation forest (iForest) for outlier detection with linear time complexity and high precision. (d) The data undergo a classification process, in which we use an integration of the particle swarm optimization algorithm and DenseNet169 (PSO-DenseNet169) to diagnose diseases; the parameters are tuned to improve accuracy. When we compared our proposed model to existing models such as SVM, KNN, NB-A, and J-48 based on performance parameters such as sensitivity, accuracy, and specificity, we found that our model outperformed the state of the art by 96.16% and 97.26% in diagnosing the heart and thyroid, respectively.