INTRODUCTION

Reliable ambulatory seizure detection systems have become a major expectation for persons with epilepsy, their families and their physicians, with the double aim of triggering timely alarms to mitigate the potential harmful consequences of seizures and providing more precise information on seizure frequency to better guide therapy [1]. Indeed, this essential information currently relies on reports from patients or their caregivers, which often prove inaccurate or misleading [2–5]. Accordingly, surveys indicate that over 75% of individuals with epilepsy consider real-time seizure detection as highly important [6]. Yet, recent guidelines from the international league against epilepsy and world federation of clinical neurophysiology have concluded that current wearable devices can only reliably detect generalized tonic-clonic seizures (GTCS) and focal to bilateral tonic-clonic seizures, all of which will be referred as GTCS herein [7▪▪,8▪▪]. Furthermore, there is currently no evidence that using such device decreases the risk of sudden unexpected death in epilepsy (SUDEP) [7▪▪,8▪▪]. 

Box 1:

no caption available

NON-EEG BASED SEIZURE DETECTION Generalized tonic-clonic seizures

A plethora of technologies has been devised to detect seizures without the conventional use of EEG [9–14]. Commercially available devices mostly target the identification of motor seizures [15], and primarily GTCS. Yet, GTCS account for less than 15% of all seizures observed in individuals with uncontrolled epilepsy. This particular seizure type has undergone comprehensive characterization, unveiling distinctive patterns that are readily discernible from nonepileptic physiological activity and psychogenic nonepileptic seizures through the utilization of upper limb 3D-accelerometry (3D-acc) [16–19], electromyography sensors (EMG) [20–23], electrodermal activity [24], and to some extent, electrocardiogram sensors (ECG) [25,26], with each biosignal employed either in isolation or in a multimodal manner.

Currently, only a limited number of devices implementing these biosignal modalities has undergone rigorous testing, clinical validation, and received approval from either the US Food and Drug Administration and/or the European Union, specifically for GTCS detection [15,27,28].

A wrist-worn 3D-acc based seizure detection device (Epi-Care mobile, Danish care technology, Sorø, Denmark) was recently tested in 71 participants (median age 27 years old, range 7–72 years old) in a phase 4 field study. The device achieved a median sensitivity of 90% for the detection of GTCS, with a median false alarm rate (FAR) of 0.1 per day [19], replicating the performance previously observed with the same device in epilepsy monitoring unit (EMU) studies.

Another wrist sensor combines 3D-acc and EDA (Embrace, Empatica Inc., Cambridge, USA) and was tested in a large EMU study involving 67 adults and 85 pediatric participants. The device demonstrated a sensitivity of 91% [95% confidence interval (CI): 84–99%] and a FAR of 0.27 per day (95% CI: 0.18, 0.36) [29▪▪], consistent with the outcomes of prior investigations [30,31].

Another type of 3D-acc sensor is placed on the upper arm and includes photoplethysmography (PPG) [32,33] (NightWatch, LivAssured B.V., Leiden, The Netherlands). Using this device, a recent prospective multicentric phase 4 study, encompassing 2310 nights and totaling 28 174 h of recordings in 53 children, reported a median sensitivity for detecting nocturnal GTCS of 94%, ranging from 71% to 100% across patients, with a FAR of 1.2 per 24 h of recordings [34▪▪]. This device also detected other nocturnal major motor seizure types, though with a lower sensitivity of 86% [34▪▪].

Arm EMG sensors have also proved highly sensitive and specific for detecting GTCS [20–22,35]. In a prospective, multicenter, blinded study employing real-time seizure detection in a cohort of 71 patients (age range: 10–62), one such wearable EMG solution successfully identified 94% of GTCS with an FAR of 0.67 per day [23].

Wearable ECG patch, designed for monitoring heart rate variability (HRV), also demonstrated high sensitivity in the detection of GTCS. In a cohort of 100 patients, the device successfully detected 17 out of 18 GTCS, resulting in a sensitivity of 100% (95% CI: 79.4–100) with a FAR of 1/day [25]. Similar results were reported in another prospective phase 2 study in 47 patients, with a detection of 9 out 10 GTCS, and a FAR of 0.9/day (90%) [26]. In a more recent phase 2 multicenter trial aimed at detecting all seizure types, particularly nonconvulsive ones, 8 out of 10 GTCS were correctly identified (80%) [36▪▪].

Nonwearable sensors have also been developed to detect seizures, including under-mattress device incorporating a quasi-piezoelectric film integrated with pressure sensors [37]. Clinical studies have shown a sensitivity in detecting GTCS of 85% in children and 89% in adults, with rare false alarms occurring exclusively during daytime [38,39]. A recent retrospective study of 55 adult patients confirmed a lower sensitivity than that provided by wearables, at 78%. FAR was very low, at 0.007 per day [40▪].

Wearable devices detecting GTCS can potentially provide biomarkers of seizure severity, extracted from the different biosignals collected for detection, with the view that such biomarkers could help predicting the risk of sudden unexpected death in epilepsy (SUDEP) [41]. As a first step, two EMUs studies have explored the possibility to use surface EMG and HRV as surrogate indicators of seizure severity. Both biomarkers showed strong association with the presence and duration of a postictal EEG suppression (PGES), with the EMG-based algorithm predicting the presence of PGES longer than 20 s with an accuracy of 85% [42,43]. The presence of a progressive slowing of the clonic phase, which could be captures from 3D-acc, also correlated with the presence and duration of the PGES, as well as with the type of GTCS [44]. The above as well as other wearable sensors might also enable to infer from the duration of GTCS and postictal immobility, as well as the severity of postictal autonomic changes [45].

Focal seizures

In contrast with GTCS, only a limited body of research has explored the utilization of extracerebral biosensors for the detection of focal seizures. As previously discussed, the NightWatch, which combines 3D-acc PPG has been tested for detecting major motor seizures other than GTCS, including 30 seizures with bilateral tonic seizures lasting >30 s, focal onset hyperkinetic seizures (HK), and other not otherwise categorized major motor seizures (OM). Sensitivity for detecting these three seizure types was 53%, 83% and 91% respectively [34▪▪]. Interestingly, the positive (n = 492) and false (n = 1642) alarms were mainly triggered by the 3D-acc (86% for positive alarms and 66% for negative alarms) followed by a rapid increase in heart rate (23% for positive alarms and 36% for negative alarms) and by tachycardia (18% for positive alarms and 6% for negative alarms). A minority of alarms, specifically 27% of true positive alarms and 8% of false positive alarms, were activated by multiple signals simultaneously [34▪▪].

The identification of focal seizures without conspicuous movement or muscle engagement obviously necessitates other biomarkers than those reliant on movement or muscle activity-based devices. Changes in hear rate and HRV have been proposed and recently tested as potential biomarkers for the detection of such seizures [25,26,36▪▪,46–50]. In a recent phase 2 clinical trial of 62 patients, an ECG patch associated with an HRV-based patient-adaptive logistic regression machine learning achieved a sensitivity of 77% for the detection of focal seizures with a FAR of 0.62% [36▪▪]. However, a major and expected difference was observed between patients with an ictal increased in heart rate > 50 bpm (sensitivity of 87%) and those without (sensitivity of 24%). The results confirmed those from a comparable previous study from the same group [26], though with improved FAR believed to reflect the implementation of the patient-adaptive algorithm.

EEG-BASED SEIZURE DETECTION

In recent years, notable advances in ambulatory scalp-EEG recordings have emerged, addressing the challenge of prolonged monitoring spanning weeks, months or even years. Novel systems employing subcutaneous or intra-auricular electrodes have been developed, indicating potential for reliable chronic EEG recordings in the future [9,51–53].

In parallel, numerous EEG-based seizure detection algorithms have been developed [54]. These algorithms, which are usually trained and tested on EEG datasets recorded in EMU rather than in ambulatory patients, exhibit sensitivities ranging from 75% to 90%, with false detection rates between 0.1 and 5 per hour [54]. Transitioning these algorithms to ambulatory settings raises critical considerations. On the one hand, long-term ambulatory EEG recordings with noninvasive wearable devices are likely to suffer from much more artefacts than in the EMU, leading to reduced sensitivity and specificity, whereas on the other hand, acceptable rates of false alarms shall be much lower in ambulatory settings than in-hospital where such alarms are primarily managed by the EMU staff.

Yet, some recent data suggest that headband EEG can achieve clinically relevant detection for some seizure type [55,56,57▪,58]. Indeed, a recent multicenter prospective phase 3 clinical trial using a headband embedded with dry electrodes to detect generalized absence seizures in 39 patients achieved a median sensitivity per patient of 93% [interquartile range (IQR) = 66.7–100%] [56]. In contrast, another study focusing on tonic seizures only reached a 41% sensitivity (41%) with a high FAR of 0.75 per hour [58]. Poor findings were also reported for focal impaired awareness (FIA) seizures in both inpatient and outpatients, with 52% and 23% sensitivity, respectively, and FAR up to 7.13 per hour [57▪].

To mitigate these issues, multibiosignal correlations might prove pivotal [59], with a synergistic combination of EEG and non-EEG technologies utilizing wireless-coupled biosensors positioned on the head (e.g., behind or around the ear) and peripheral regions (e.g., the wrist) [60]. Minimally invasive subcutaneous EEG electrodes, which are less susceptible to artifacts than those on the scalp, offer an alternative approach to clinically-relevant EEG-based seizure detection in ambulatory patients [61].

MACHINE LEARNING FOR SEIZURE DETECTION

Machine learning algorithms have become an integral part of seizure detection and prediction systems, with particular emphasis on Deep Learning frameworks [62–68]. They can be developed to operate in real-time when embedded in seizure detection devices or offline for analysis after the recording. Yet, algorithms embedded in seizure detection devices must meet low computing, memory and power requirements that are constrained by the wearable devices [69]. The 2023 ICASSP Seizure Detection Challenge was specifically organized to evaluate algorithms designed for wearable EEG devices. The challenge provided participants with the SeizeIT1 dataset that contains 42 subjects with one week of continuous EEG recorded in the EMU, with four additional electrodes placed behind the ear (Byteflies Sensor Dot) representing potential ambulatory wearable EEG data [70]. While the challenge highlighted some of the state-of-the-art methods for EEG-based wearable seizure detection methods, it also highlighted the continued need for algorithm improvements as the best-performing algorithm obtained a sensitivity of 63% with 15 false-alarms per hour which is not acceptable by patients for real-time seizure alarming devices [71]. At the same time, algorithms designed for use in an EMU setting perform considerably better. A recent assessment of commercially available algorithms for EEG-based EMU seizure detection showed that these algorithms have a sensitivity close to 90% for a false alarm rate of 1.7–5.5 per day [72▪▪].

The development of seizure detection algorithms raises a number of issue, including comparison of performance across studies, transparency and interpretability [73]. Comparison across studies can be promoted by using similar public datasets, as illustrated by the 2023 ICASSP Seizure Detection Challenge. Other publicly available datasets for EEG-based seizure detection are available [74], but remain scarce overall. Recent research has also highlighted the importance of proper cross-validation methodology and the choice of proper performance metrics for the evaluation of algorithms [75,76]. Addressing this issue requires a collective effort from the research community. Collaborative initiatives, akin to those proposed by prominent scholars [73,77▪], are vital in establishing standardized datasets that encompass diverse seizure types, durations, and clinical contexts. These standardized datasets would serve as a common ground for algorithm evaluation, enabling fair and accurate comparisons between different methodologies. Furthermore, a consensus on evaluation metrics, incorporating measures such as sensitivity, specificity, positive predictive value, and false alarm rate [2,3], is imperative. Establishing a set of universally agreed-upon metrics will facilitate the comprehensive assessment of algorithms and enhance the reproducibility of results.

CONCLUSION

Several commercial wrist or arm wearable sensors, using 3D-acc alone or combined with EDA or PPG, enable a reliable detection of GTCS with acceptable level of false-alarm rates. No such wearable currently offer clinically-relevant solution to detect focal seizures. This conclusion also applies to wearable EEG which utility remains restricted to the detection of generalized absence seizures. Promising developments in the field include the utilization of ECG-based HRV biomarkers and subcutaneous EEG. Improvements are needed in the standardization of datasets and metrics used to trained and tested seizure detection algorithms, as well as in the transparency and interpretability of the underlying machine learning tools.

Acknowledgements

None.

Financial support and sponsorship

This work was supported by the PEDESITE project (Personalized Detection of Epileptic Seizure in the Internet of Things (IoT) Era) funded by the Swiss National Science Foundation Sinergia Grant/Award number SCRSII5-193813/1.

Conflicts of interest

There are no conflicts of interest.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

▪ of special interest

▪▪ of outstanding interest

REFERENCES 1. de Boer HM, Mula M, Sander JW. The global burden and stigma of epilepsy. Epilepsy Behav 2008; 12:540–546. 2. Hoppe C, Poepel A, Elger CE. Epilepsy: accuracy of patient seizure counts. Arch Neurol 2007; 64:1595–1599. 3. Aghaei-Lasboo A, Fisher RS. Methods for measuring seizure frequency and severity. Neurol Clin 2016; 34:383–394. viii. 4. Fisher RS, Blum DE, DiVentura B, et al. Seizure diaries for clinical research and practice: limitations and future prospects. Epilepsy Behav 2012; 24:304–310. 5. Blachut B, Hoppe C, Surges R, et al. Counting seizures: the primary outcome measure in epileptology from the patients’ perspective. Seizure 2015; 29:97–103. 6. Hoppe C, Feldmann M, Blachut B, et al. Novel techniques for automated seizure registration: patients’ wants and needs. Epilepsy Behav 2015; 52:1–7. 7▪▪. Beniczky S, Wiebe S, Jeppesen J, et al. Automated seizure detection using wearable devices: a clinical practice guideline of the International League Against Epilepsy and the International Federation of Clinical Neurophysiology. Epilepsia 2021; 62:632–646. 8▪▪. Beniczky S, Wiebe S, Jeppesen J, et al. Automated seizure detection using wearable devices: a clinical practice guideline of the International League Against Epilepsy and the International Federation of Clinical Neurophysiology. Clin Neurophysiol 2021; 132:1173–1184. 9. Weisdorf S, Gangstad SW, Duun-Henriksen J, et al. High similarity between EEG from subcutaneous and proximate scalp electrodes in patients with temporal lobe epilepsy. J Neurophysiol 2018; 120:1451–1460. 10. Kuhlmann L, Lehnertz K, Richardson MP, et al. Seizure prediction - ready for a new era. Nat Rev Neurol 2018; 14:618–630. 11. Stacey WC. Seizure prediction is possible-now let's make it practical. EBioMedicine 2018; 27:3–4. 12. Baud MO, Rao VR. Gauging seizure risk. Neurology 2018; 91:967–973. 13. Privitera M, Haut SR, Lipton RB, et al. Seizure self-prediction in a randomized controlled trial of stress management. Neurology 2019; 93:e2021–e2031. 14. Baud M, Schindler K. Forecasting seizures: not unthinkable anymore. Epileptologie 2018; 35:156–161. 15. Zelano J, Beniczky S, Ryvlin P, et al. Report of the ILAE SUDEP Task Force on national recommendations and practices around the world regarding the use of wearable seizure detection devices: a global survey. Epilepsia Open 2023; 8:1271–1278. 16. Lockman J, Fisher RS, Olson DM. Detection of seizure-like movements using a wrist accelerometer. Epilepsy Behav 2011; 20:638–641. 17. Beniczky S, Polster T, Kjaer TW, Hjalgrim H. Detection of generalized tonic-clonic seizures by a wireless wrist accelerometer: a prospective, multicenter study. Epilepsia 2013; 54:e58–e61. 18. Patterson AL, Mudigoudar B, Fulton S, et al. SmartWatch by SmartMonitor: assessment of seizure detection efficacy for various seizure types in children, a large prospective single-center study. Pediatr Neurol 2015; 53:309–311. 19. Meritam P, Ryvlin P, Beniczky S. User-based evaluation of applicability and usability of a wearable accelerometer device for detecting bilateral tonic-clonic seizures: a field study. Epilepsia 2018; 59: (Suppl 1): 48–52. 20. Beniczky S, Conradsen I, Pressler R, Wolf P. Quantitative analysis of surface electromyography: biomarkers for convulsive seizures. Clin Neurophysiol 2016; 127:2900–2907. 21. Halford JJ, Sperling MR, Nair DR, et al. Detection of generalized tonic-clonic seizures using surface electromyographic monitoring. Epilepsia 2017; 58:1861–1869. 22. Beniczky S, Conradsen I, Wolf P. Detection of convulsive seizures using surface electromyography. Epilepsia 2018; 59: (Suppl 1): 23–29. 23. Beniczky S, Conradsen I, Henning O, et al. Automated real-time detection of tonic-clonic seizures using a wearable EMG device. Neurology 2018; 90:e428–e434. 24. Casanovas Ortega M, Bruno E, Richardson MP. Electrodermal activity response during seizures: a systematic review and meta-analysis. Epilepsy Behav 2022; 134:108864. 25. Jeppesen J, Fuglsang-Frederiksen A, Johansen P, et al. Seizure detection based on heart rate variability using a wearable electrocardiography device. Epilepsia 2019; 60:2105–2113. 26. Jeppesen J, Fuglsang-Frederiksen A, Johansen P, et al. Seizure detection using heart rate variability: a prospective validation study. Epilepsia 2020; 61: (Suppl 1): S41–S46. 27. Beniczky S, Ryvlin P. Standards for testing and clinical validation of seizure detection devices. Epilepsia 2018; 59: (Suppl 1): 9–13. 28. Hubbard I, Beniczky S, Ryvlin P. The challenging path to developing a mobile health device for epilepsy: the current landscape and where we go from here. Front Neurol 2021; 12:740743. 29▪▪. Onorati F, Regalia G, Caborni C, et al. Prospective study of a multimodal convulsive seizure detection wearable system on pediatric and adult patients in the epilepsy monitoring unit. Front Neurol 2021; 12:724904. 30. Onorati F, Regalia G, Caborni C, et al. Multicenter clinical assessment of improved wearable multimodal convulsive seizure detectors. Epilepsia 2017; 58:1870–1879. 31. Regalia G, Onorati F, Lai M, et al. Multimodal wrist-worn devices for seizure detection and advancing research: focus on the Empatica wristbands. Epilepsy Res 2019; 153:79–82. 32. Arends J, Thijs RD, Gutter T, et al. Multimodal nocturnal seizure detection in a residential care setting: a long-term prospective trial. Neurology 2018; 91:e2010–e2019. 33. Lazeron RHC, Thijs RD, Arends J, et al. Multimodal nocturnal seizure detection: do we need to adapt algorithms for children? Epilepsia Open 2022; 7:406–413. 34▪▪. van Westrhenen A, Lazeron RHC, van Dijk JP, et al. Multimodal nocturnal seizure detection in children with epilepsy: a prospective, multicenter, long-term, in-home trial. Epilepsia 2023; 64:2137–2152. 35. Beniczky S, Conradsen I, Moldovan M, et al. Automated differentiation between epileptic and nonepileptic convulsive seizures. Ann Neurol 2015; 77:348–351. 36▪▪. Jeppesen J, Christensen J, Johansen P, Beniczky S. Personalized seizure detection using logistic regression machine learning based on wearable ECG-monitoring device. Seizure 2023; 107:155–161. 37. Ulate-Campos A, Coughlin F, Gainza-Lein M, et al. Automated seizure detection systems and their effectiveness for each type of seizure. Seizure 2016; 40:88–101. 38. Poppel KV, Fulton SP, McGregor A, et al. Prospective study of the emfit movement monitor. J Child Neurol 2013; 28:1434–1436. 39. Narechania AP, Garic II, Sen-Gupta I, et al. Assessment of a quasi-piezoelectric mattress monitor as a detection system for generalized convulsions. Epilepsy Behav 2013; 28:172–176. 40▪. Nouboue C, Selfi S, Diab E, et al. Assessment of an under-mattress sensor as a seizure detection tool in an adult epilepsy monitoring unit. Seizure 2023; 105:17–21. 41. Ryvlin P, Ciumas C, Wisniewski I, Beniczky S. Wearable devices for sudden unexpected death in epilepsy prevention. Epilepsia 2018; 59: (Suppl 1): 61–66. 42. Arbune AA, Conradsen I, Cardenas DP, et al. Ictal quantitative surface electromyography correlates with postictal EEG suppression. Neurology 2020; 94:e2567–e2576. 43. Arbune AA, Jeppesen J, Conradsen I, et al. Peri-ictal heart rate variability parameters as surrogate markers of seizure severity. Epilepsia 2020; 61: (Suppl 1): S55–S60. 44. Vlachou M, Ryvlin P, Arbune AA, et al. Progressive slowing of clonic phase predicts postictal generalized EEG suppression. Epilepsia 2022; 63:3204–3211. 45. Beniczky S, Arbune AA, Jeppesen J, Ryvlin P. Biomarkers of seizure severity derived from wearable devices. Epilepsia 2020; 61: (Suppl 1): S61–S66. 46. Cogan D, Heydarzadeh M, Nourani M. Personalization of NonEEG-based seizure detection systems. Annu Int Conf IEEE Eng Med Biol Soc 2016; 2016:6349–6352. 47. Vandecasteele K, De Cooman T, Gu Y, et al. Automated epileptic seizure detection based on wearable ECG and PPG in a hospital environment. Sensors (Basel, Switzerland) 2017; 17:2338. 48. De Cooman T, Kjær TW, Van Huffel S, Sorensen HB. Adaptive heart rate-based epileptic seizure detection using real-time user feedback. Physiol Meas 2018; 39:014005. 49. Pavei J, Heinzen RG, Novakova B, et al. Early seizure detection based on cardiac autonomic regulation dynamics. Front Physiol 2017; 8:765. 50. van Elmpt WJ, Nijsen TM, Griep PA, Arends JB. A model of heart rate changes to detect seizures in severe epilepsy. Seizure 2006; 15:366–375. 51. Bleichner MG, Debener S. Concealed, unobtrusive ear-centered EEG acquisition: cEEGrids for transparent EEG. Front Hum Neurosci 2017; 11:163. 52. Remvig LS, Duun-Henriksen J, Furbass F, et al. Detecting temporal lobe seizures in ultra long-term subcutaneous EEG using algorithm-based data reduction. Clin Neurophysiol 2022; 142:86–93. 53. Pal Attia T, Viana PF, Nasseri M, et al. Seizure forecasting using minimally invasive, ultra-long-term subcutaneous EEG: generalizable cross-patient models. Epilepsia 2022; 64:S114–S123. 54. Baumgartner C, Koren JP. Seizure detection using scalp-EEG. Epilepsia 2018; 59: (Suppl 1): 14–22. 55. Shum J, Friedman D. Commercially available seizure detection devices: a systematic review. J Neurol Sci 2021; 428:117611. 56. Japaridze G, Loeckx D, Buckinx T, et al. Automated detection of absence seizures using a wearable electroencephalographic device: a phase 3 validation study and feasibility of automated behavioral testing. Epilepsia 2022; 64:S40–S46. 57▪. Macea J, Bhagubai M, Broux V, et al. In-hospital and home-based long-term monitoring of focal epilepsy with a wearable electroencephalographic device: diagnostic yield and user experience. Epilepsia 2023; 64:937–950. 58. Proost R, Macea J, Lagae L, et al. Wearable detection of tonic seizures in childhood epilepsy: an exploratory cohort study. Epilepsia 2023; 64:3013–3024. 59. Beniczky S, Karoly P, Nurse E, et al. Machine learning and wearable devices of the future. Epilepsia 2021; 62: (Suppl 2): S116–s124. 60. Nielsen JM, Kristinsdottir AE, Zibrandtsen IC, et al. Out-of-hospital multimodal seizure detection: a pilot study. BMJ Neurol Open 2023; 5:e000442. 61. Ryvlin P, Beniczky S. Seizure detection and mobile health devices in epilepsy: update and future developments. Epilepsia 2018; 59: (Suppl 1): 7–8. 62. Kramer U, Kipervasser S, Shlitner A, Kuzniecky R. A novel portable seizure detection alarm system: preliminary results. J Clin Neurophysiol 2011; 28:36–38. 63. Teijeiro T, García CA, Castro D, Félix P. Abductive reasoning as a basis to reproduce expert criteria in ECG atrial fibrillation identification. Physiol Meas 2018; 39:084006. 64. Fei-Fei L, Fergus R, Perona P. Learning generative visual models from few training examples: an incremental Bayesian approach tested on 101 object categories. Comput Vis Image Understand 2007; 106:59–70. 65. Ravi S, Larochelle H. Optimization as a model for few-shot learning. In Proceedings of the International Conference on Learning Representations 2016. 66. Hariharan B, Girshick RB. Low-shot visual recognition by shrinking and hallucinating features. 2017 IEEE International Conference on Computer Vision (ICCV) 2016; 3037–3046. 67. Burrello A, Schindler KA, Benini L, Rahimi A. One-shot learning for iEEG seizure detection using end-to-end binary operations: local binary patterns with hyperdimensional computing. 2018 IEEE Biomedical Circuits and Systems Conference (BioCAS) 2018; 1–4. 68. Kozerawski J, Turk MA. CLEAR: Cumulative LEARning for One-Shot One-Class Image Recognition. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition 2018; 3446–3455. 69. Al-Hajjar ALN, Al-Qurabat AKM. An overview of machine learning methods in enabling IoMT-based epileptic seizure detection. J Supercomput 2023; 1–48. 70. Chatzichristos C, Claro Bhagubai M. SeizeIT1. Draft version. KU Leuven RDR; 2023. 71. Al-Hussaini I, Mitchell CS. SeizFt: interpretable machine learning for seizure detection using wearables. Bioengineering (Basel) 2023; 10:918. 72▪▪. Reus EEM, Visser GH, van Dijk JG, Cox FME. Automated seizure detection in an EMU setting: are software packages ready for implementation? Seizure 2022; 96:13–17. 73. Reus EEM, Visser GH, Sommers-Spijkerman MPJ, et al. Automated spike and seizure detection: are we ready for implementation? Seizure 2023; 108:66–71. 74. Wong S, Simmons A, Rivera-Villicana J, et al. EEG datasets for seizure detection and prediction – a review. Epilepsia Open 2023; 8:252–267. 75. Shafiezadeh S, Duma GM, Mento G, et al. Methodological issues in evaluating machine learning models for EEG seizure prediction: good cross-validation accuracy does not guarantee generalization to new patients. Appl Sci 2023; 13:4262. 76. Pale U, Teijeiro T, Atienza D. Importance of methodological choices in data manipulation for validating epileptic seizure detection models. ArXiv 2023; abs/2302.10672. 77▪. Koren J, Hafner S, Feigl M, Baumgartner C. Systematic analysis and comparison of commercial seizure-detection software. Epilepsia 2021; 62:426–438.