Seizure detection using wearable sensors and machine learning: Setting a benchmark

Abstract : Tracking seizures is crucial for epilepsy monitoring and treatment evaluation. Current epilepsy care relies on caretaker seizure diaries, but clinical seizure monitoring may miss seizures. Wearable devices may be better tolerated and more suitable for long-term ambulatory monitoring. This study evaluates the seizure detection performance of custom-developed machine learning (ML) algorithms across a broad spectrum of epileptic seizures utilizing wrist-and ankle-worn multisignal biosensors. The sensor recorded body temperature, electrodermal activity, accelerometry (ACC), and photoplethysmography, which provides blood volume pulse (BVP). We used electroencephalographic seizure onset and offset as determined by a board-certified epileptologist as a standard comparison.Automatic seizure detection using ML from multimodal wearable sensor data is feasible across a broad spectrum of epileptic seizures.
 ? Epilepsy manifests through seizures which occurs uncontrollably. Several types of seizures exist based on semiology, symptomatic experience and electrophysiological signatures. ? Many challenges exist in designing a reliable system for forecasting seizures from noninvasively recorded data. Training, testing, and validating a forecasting algorithm requires ultra-long duration recordings with an adequate number of seizures. ? Self-reported seizure diaries are the most accessible validation, but the poor reliability of such diaries is widely recognized.
 ? To demonstrate the impact of introducing self-awareness in wearable technologies, we consider the epileptic seizure detection problem, as a case study. ? A machine learning algorithm, which is particularly selected and trained for the target problem, is used to extract the model and detect possible bio-medical abnormalities or pathologies. ? We investigate if this self-aware technique has any negative impact in terms of classification and seizure detection performance.
 • We propose a new generation of self-aware wearable systems to cope with the stringout energy constraints of these bio-medical systems, while at the same time enhancing the detection quality they provide. • In particular, to reduce the energy without sacrificing quality of detection, we propose a two-mode classifier. • We also observe that while this approach improves both sensitivity and specificity, the latter changes more significantly, which indicates that the true positive is increased significantly due to our proposed model refinement.
 ? The main challenge in wearable systems is to increase the battery lifetime, while maintaining the machine-learning performance of the system. ? Self-awareness is perhaps the key to design intelligent systems that can continuously monitor their own performance, adapt to changes, and improve autonomously. ? The high performance of SVM in terms of the accuracy of detecting seizures, SVM is also suitable to be implemented on resource-constrained embedded systems.
Download DOC Download PPT

We have more than 145000 Documents , PPT and Research Papers

Have a question ?

Mail us :