Comparison of Machine Learning Approach in Smart Wearables

Abstract

Smart devices and smart wearables using machine learning techniques and Internet of Things (IoT) are automating the epilepsy seizure predictions and overcoming the limitations of manual epilepsy monitoring and detection using EEG signals. Epilepsy being the most common neurological disorder, physicians usually prescribe antibiotic drugs and surgical treatments to cure epilepsy. The antibiotic drugs and surgical treatments have side effects on the health of the patients. The patients resistant to epileptic medications, to overcome these limitations several smart devices & smart wearable’s making use of machine learning approaches & IoT are being developed. In this paper, we have tried to review and compare such smart wearable devices being used to make epileptic seizure predictions and the machine learning techniques being used by these devices. The devices being reviewed in this paper are (i) Smart Head Bands (ii) Smart Phone Application (iii) Smartphone-Based Detection System (iii) Smart Watch (iv) Mobile multimedia framework (v) Microcontroller and (vi) Wrist-Worn Sensor. The accuracy results of the smart devices and smart wearables have been reviewed and reported. It has been observed that Support Vector Machine is the most used machine learning technique in almost all the smart devices.