Epilepsy is known as one of the neurological disorders that will cause seizures, it is known as a sudden abnormal electrical activity that happened in the brain.Seizures usually cause abnormal muscle activity, sensations, and even the loss of consciousness. These symptoms usually last for a few minutes. Hence, an epileptic seizure might cause the loss of life when the occurrence of the seizure causes deadly accidents such as road accidents or fall into death.According to the World Health Organization (WHO), there are around 50 million people around the world suffered from epilepsy. A journal published in Frontiers in Epidemiology: Mortality, and life expectancy in Epilepsy and Status Epilepticuscurrent trends and future aspects (2023) stated that about 125,000 people die each year due to epilepsy. The risk of premature mortality for people with epilepsy is estimated at now three times that of the general population. The high premature mortality has far-reaching consequences, profoundly affecting not only individual families but also society and nations as a whole. First, epileptic seizures might cause the loss of life. Besides that, it will also reduce the quality of life, emotional and psychological burdens on the families and caregivers. This even has significant economic consequences loss of productivity and reduced economic outputs along with the financial loss associated with healthcare, treatment, and managing epilepsy-related complications.However, there is a fact showing that an estimated 70% of epileptic patients could live seizure-free if they are diagnosed in the early stage and received proper medical treatment. The common method in diagnosing epilepsy includes blood tests and neurological exams. The entire process of diagnosis is very time-consuming. Besides, the diagnosis also requires medical experts and doctors who have great experience in the diagnosis process. However, we are facing a situation in the medical field which is the shortage of neurologists.In the face of a shortage of medical officers, the implementation of our proposed AI-powered seizure detection system MindAlert can serve as a promising alternative. This technology can offer an effective solution for the timely and accurate detection of seizures, aiding in the diagnosis and treatment of patients. By leveraging artificial intelligence algorithms, the system can analyze EEG data and quickly identify seizure activity, potentially reducing the burden on medical officers and ensuring prompt intervention when needed.In this MindAlert system, the datasets used to train the AI model are open-source datasets contributed by the Rochester Institute of Technology that are available on the University of California Irvine (UCI) machine learning repository. Why do we choose to use this database? This is because it is a well-established database- widely recognized and extensively used in the field of epilepsy research and seizure detection. Hence, it became the research community benchmark in this research area.In this MindAlert system, we developed our own AI model namely Metric Learning Based Convolutional Neural Network (MLBCNN). Unlike the training process of the conventional CNN using the gradient descent method, the training process of the MLBCNN in this research is modified based on the metric learning concept. The reason for modifying the training process using metric learning is because the EEG waveform of both normal and epileptic seizure in resting state are highly similar to each other. Thus, this cause high data sparsity where the feature embedding vectors of both classes are distributed sparsely and not clustering among the respective class in the metric space. This property causes increases the difficulty in classifying the EEG signal accurately. Hence, as inspired by the metric learning that is usually applied in unsupervised learning, we apply this technique as the training algorithm of the deep neural network CNN as it has the ability to cluster features that are in the same class.In this research, we benchmark the MLBCNN deep learning models with the machine learning models as well as the common deep learning models. As a result, the developed MLBCNN outperformed the other machine learning models and the common deep learning models with the best classification performance with an accuracy of 97.4%, AUC of 99.6%, recall rate of 96.9%, precision rate of 91.3%, and specificity of 97.3%. The method has been filed with 1 patent (PI2020006519) and 2 copyrights too.In conclusion, the MLBCNN has the highest classification performance which is up to 97.4% accuracy in detecting the seizure from EEG signals. It is the current best deep learning model in seizure EEG signal classification. Hence, with such a high accuracy and recall rate in classifying the seizure EEG signal, it is quite convincing that this MindAlert system can be implemented as part of the epileptic seizure diagnostic tool to assist doctors in diagnosing accurately with a shorter diagnosis time. This will definitely benefit not only the doctors but the patients too, as earlier diagnosis may increase the chance to stay in seizure-free life. This brings good news to the nation and globally too as the significant economic impact caused by the epileptic seizure disease can be greatly reduced.
