A Two-Layer Ensemble Method for Detecting Epileptic Seizures Using a Self-Annotation Bracelet With Motor Sensors

Abstract

published version features the final layout of the paper including the volume, issue and page numbers. Link to publication General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the "Taverne" license above, please follow below link for the End User

Machine Learning Techniques for Developing Remotely Monitored Central Nervous System Biomarkers Using Wearable Sensors: A Narrative Literature Review

“…As filter methods are independent of any model that is to be used in later steps, they are typically faster to implement and reduce the need for repeating feature selection for different ML models. In our selected studies, we found five studies that used Analysis of Variance (ANOVA), Pearson's Correlation, or Spearman's Correlation to identify features that were statistically significant predictors of the outcomes [24,[93][94][95][96]. p-value based feature selection, while commonly used in clinical studies, is not always suitable for training a ML model.…”

“…When similar hyperparameter tuning processes can be used for different ML algorithms for different datasets, researchers can then identify the optimal ML model. Among the selected studies, 25 discussed which hyperparameters were considered for their models [23,24,34,43,44,46,53,69,73,86,87,94,95,[107][108][109][110]114,138,158,159,[181][182][183][184], of which one stated they used the default hyperparameters of the models [69]. Only nine studies discussed how they selected or optimized their hyperparameters.…”

Seizure detection based on wearable devices: A review of device, mechanism, and algorithm

“…Based on the time domain and frequency domain features of ACM, many machine learning algorithms such as KNN and RF were used for the detection of seizures. The sensitivity and FDR of the models were between 76.84%–100% and 0.01–0.05/h, respectively 56–58 …”

“…The sensitivity and FDR of the models were between 76.84%-100% and 0.01-0.05/h, respectively. [56][57][58] Surface electromyography is generally used for the detection of motor seizures, as well. However, compared with ACM, the acquisition instrument of sEMG needs to be in direct contact with the skin.…”

“…Based on some time domain and frequency domain features, they used RF to detect these seizures, with a sensitivity of 88.01% and FDR of 0.01/h 57 . In 2022, Dong et al 58 continued to collect ACM data of 7 people and obtained 547 motor seizures. After using the two‐layer ensemble model, the sensitivity and FDR achieved 76.84% and 0.04/h, respectively.…”