Proposed clinical practice guideline for automated seizure detection using wearable devices
A joint EEG Task Force from the ILAE and the International Federation of Clinical Neurophysiology (IFCN) has developed a proposed clinical practice guideline for automated seizure detection using wearable devices. The ILAE guideline process requires obtaining feedback and comments from its members on the proposed guideline. These comments from our international community will be reviewed by the working group before finalizing the guideline.
Please see the draft of the clinical practice guideline on automated seizure detection using wearable devices.
The manuscript is now closed for comments. The revised version of the manuscript is being reviewed by the ILAE and will then be submitted for publication. Responses to public comments.
Thank you for your help in this important effort of the ILAE and IFCN.
10 August 2020
We greatly commend the group on the proposed guidelines to provide recommendations for wearable devices in patients in epilepsy. This is an important first step towards implementing the use of wearable devices in clinical practice and community-based interventions for seizure monitoring and quantification.
Some considerations for the authors:
It does however provide a tenuous link between the community use of wearable devices when most validation has been performed in research settings.
What is clear from the review is that although there is evidence for the effectiveness of wearable devices (high sensitivity) for GTCS and FBTCS detection, mostly from inpatient video-EEG monitoring studies, their clinical validation, acceptability and usability in the real-world is lacking. Thus, well-designed clinical validation studies, particularly in ambulatory settings are required.
Many of the reported studies have a high false alarm rate, which can be problematic for real-world use and clinical interventions.
The group may wish to specify whether multimodal sensors have a higher sensitivity and lower false alarm rate, and if so, which measures (e.g. ECG, accelerometry) perform best.
We also need to better understand which group of patients may benefit from such interventions (e.g. nocturnal GTCS/FBTCSs, high risk of SUDEP).
Patient preferences (e.g. design, cost, efficacy of wearables) and associated stigma should also be taken into consideration.
For Table 1, bottom row “Outcome”, there is no other consideration other than in the broader descriptor of “usability”, but what about the longitudinal nature of signal quality? How consistent is the quality of the primary physiological input?
It may also be good to factor in detection latency in Table 2 to an “Outcome’, rather than as its own column
With best wishes,
Terence O’Brien, Patrick Kwan, Piero Perucca, Shobi Sivathamboo, and Michael Gorman
7 August 2020
Dear Professor Beniczky and the ILAE-IFCN Working Group:
Thank you for concise review regarding automated seizure detection using wearable devices. I read this helpful summary with great interest.
I would suggest that the authors would consider to clarify the definition of seizure “detection”, to distinguish seizure “detection” from seizure “prediction” or seizure “forecast”. Seizure prediction or seizure forecasting can be described as “early seizure detection”. Thus references selected by the word “detection” might include seizure prediction study.
Especially this matters regarding heart rate variability (HRV) study, because change of HRV is not only associated with seizure but often precedes seizure onset. Thus HRV can be utilized for seizure prediction (Fujiwara et al., 2016; https://pubmed.ncbi.nlm.nih.gov/26841385/) as well as seizure detection. The description of detection latency at “Arends J. et al. 2018” in table 2 (3rd row, page 19), “seizures were considered to detected if “within 3 min before” and 5 min after onset.” The difference between “seizure detection” and “seizure prediction” is usually based on the timing of detecting biomarkers associated with epileptic seizures. The definition of seizure detection may be helpful to avoid conceptual confusion of seizure detection and seizure prediction.
Seizure detection and seizure prediction share the same goal of early intervention of seizure which leads SUDEP prevention. On the other hand, seizure prediction has its original prospection of reducing the risk of seizure-related injuries and closed-loop therapy to prevent seizures. Significant progress has been made in the field of seizure prediction in the past decade. Prospective study of HRV based wearable seizure prediction system has been published recently (Yamakawa et al 2020; https://pubmed.ncbi.nlm.nih.gov/32709064/). Large scale clinical trial for seizure prediction using wearable devices are realistic. I wish another guideline for seizure prediction with wearable devices will be also developed in the near feature.
7 August 2020
To Committee members,
The brief ILAE summary accurately reflects the “newness” of mobile seizure detection and the limited number of studies in the area. I suggest several factors also be considered in a guideline to better reflect the current development of mobile seizure detection and clinical needs of patients:
1) FDA cleared devices were tested in epilepsy monitoring units for detection of tonic clonic seizures at rest; false positive rates may be higher for children compared to adults. Device performance should be clearly presented for periods “at rest” and during activity and for adults and children.
2) Detection performance needs to match the duration of seizures and timing of needed caregiver interventions. We recently measured the average duration of bilateral symmetric tonic clonic seizures as 1 ½ minutes while the duration of post-ictal generalized EEG suppression (PGES) averaged 1 minute; respiratory dysfunction as marked by post-ictal stertor often persisted into post-ictal encephalopathy periods (Carmenate, 2020). Performance testing should note detection timing and alerting that would permit interventions during the typical time-periods of seizures and post-ictal events.
3) In multiple studies, SUDEP risks are greatest immediately following tonic clonic seizures and appear increased during sleep and prone positioning. This suggests current “at rest” tonic clonic seizure detectors may assist caregiver to provide interventions and potentially reduce SUDEP risks. In conjunction with evaluating device performance, there should be more study of effectiveness of caregiver interventions, e.g. the timing and techniques for optimal seizure “first aid”.
1. Carmenate YC, Gutierrez EG, Kang JY, Krauss GL. Postictal stertor: associations with focal and bilateral seizure types. Epilepsy & Behavior, 110, 2020, 107103.
3 August 2020
Thank you very much for the enormous efforts you made for this manuscript which is a very important guideline for this field.
Taking the arguments made before into account and my own thoughts I would suggest to add the following points/changes:
“… especially in unsupervised patients and patients with high frequent GTCS/FBTCS to count the seizures correctly and make acute interventions possible.”
It is important to add that only those devices can be recommended which showed a high sensitivity, everyday suitability and security (e.g. positive CE certification).
“There is need for further research ….”:
4) To increase everyday suitable technical solutions (e.g. wearable everywhere or restricted to home) and comfort in device wearing (leading to higher compliance).
5) To ensure a role based and secure data transfer e.g. from the device to the practitioner. This might also be a usability question.
- The requested sensitivity for phase III studies is here 90%. This seems to be very high and maybe only possible for the detection of GTCS but not for e.g. pure focal seizures. Thinking for future studies, maybe the authors should add here that the requested sensitivity depends on the seizure types which are actually analyzed.
- Some papers are already using “Clusters”, e.g. for identifying seizure detection related “responders” and “non-responders” (e.g. P16: Jeppesen et al. (2019)). There will certainly be more work in the future that cluster the patients, so the criteria for clustering/grouping should be made transparent - so the work remains comparable.
I would suggest to re-arrange the table listing main available devices and the corresponding studies combined the other facts you have in your table. That would make it easier for caregivers to choose the right device for the patient.
31 July 2020
The authors have done a great service for patients and researchers in recommending that wearable devices be used to improve detection of GTCS and FBTCS; we are grateful for their work and their service to the community. Their two concluding recommendations sound
well-supported and good on many levels.
We would like to suggest an area of improvement, however, in the phase categorizations. As currently presented, they risk giving some false impressions about study quality relative to generalizability of the results.
The proposed phase definitions appear to be trying to match what is done in drug trials, not in algorithm studies. In doing so, they omit three very important factors necessary in evaluating the quality of algorithms. For algorithms, these factors are: (1) label quality, (2) double blinding between labeler and developer of the algorithm, and (3) independence of the test data from the training data. For example, FDA required both Brain Sentinel and Empatica (in their pivotal studies) to follow best practices in this area: (1) use gold-standard labels provided by three epileptologists independently labeling the data using Video-EEG, (2) break the blinding between the label and the algorithm output only for performance evaluation and (3) keep the test data entirely separated from the training data (no overlap of people in the two groups of data). However, whether a study fits into Phase 1, 2, 3, or 4 currently ignores whether or not it obeys these criteria.
Because of these omissions, we observe a somewhat cattywampus placement of studies of different qualities. For example, two studies associated with groups that obtained FDA clearance (Halford et al 2017 and Onorati et al 2017) follow these three algorithm evaluation best practices and appear in Phase 2. On the other hand, the recommendation document places into Phase 4 a work that violates separation of the training-testing data by allowing 14 participants from the training-data group to appear in the test-data group, and its “true” seizure labels were neither independent nor gold standard -- they are provided by the trial nurses consulting with the study authors who decide disagreements, and they use only video, not video-EEG (Arends et al 2018).
Thus, we would appreciate if the report would either redefine phases to be more relevant to algorithm evaluation best-practice quality standards, or remove the emphasis on phases as being associated with quality of an evaluation.
31 July 2020
Many thanks for the great efforts to advance the field of automatic seizure detection, and for the opportunity to comment on the recommendations proposed by the ILAE-IFCN Working Group. This is an important step towards the improvement of patients’s autonomy and safety which is more than welcome.
As already mentioned in previous comments, the recommendations focus on detection of GTCS or FBTCS for at least two good reasons: the best performance metrics were achieved only for these seizure types, and nocturnal generalized convulsions are the most important risk factor for sudden unexpected death in epilepsy.
Whilst the current recommendations allow some flexibility and freedom to the practising health care professionels, they also cause significant problems and uncertainties when it comes to their practical application. To facilitate the transition into real life, it may be of interest to discuss and clarify the following issues:
- The patient groups who benefit most from wearable technologies remain to be defined, e.g. those with a lifetime history of GTCS/FBTCS, a history of GTCS/FBTCS during the last 12 months, nocturnal motor seizures, or all „unsupervised“ patients? It might be helpful to discuss this in greater detail and to give advice on this issue.
- The term „unsupervised“ may need more explanation. Does it refer to people living on their own with no partner, or only those who sleep alone? Does the recommendation also include people who are unsupervised during longer time periods during their professional or daily activities? As far as I know, at least one commercially available device can (or should) only be used during nighttime.
- It is stressed in the manuscript that the recommendations aim especially at improving patients‘ safety in the ambulatory setting. It may be emphasized that the detection devices were developed and tested mostly during artificial conditions, i.e. patients lying in bed during video-EEG monitoring with no habitual daily activities (typical inpatient setting). Thus, it remains unclear whether the devices achieve resonable performances during common daily activities in the outpatient setting (and high false-positive rate may weaken the acceptance by people with epilepsy and caregivers).
- A weakness of the cited studies is also the rather low patient numbers and the lack of separate validation groups, which may yield too optimistic performance metrics. A more thorough and consistent larger-scale testing of such devices may be claimed before the use of such devices is recommended (as suggested in Beniczky S, Ryvlin P. Standards for testing and clinical validation of seizure detection devices. Epilepsia. 2018 Jun;59 Suppl 1:9-13).
- The automated detection technology consists of a wearable device that picks up the biosignals, a computing device that determines whether the signals are likely due to a seizure, and a sender that transmits the information on the seizure. As for the detection device, the alerting or messaging system needs to be tested prospectively, ideally in an outpatient setting, as the reliable transmission of a correctly detected seizure is essential for the usefulness of such systems. Were the tranmission technologies also part of the clinical studies included in the recommendation article? Some of my patients have reported failures of transmission e.g. due to weak or no connection or other technical issues.
With best regards,
30 July 2020
The first step in understanding the utility of wearable devices is to define in detail what clinically meaningful outcomes are in different epilepsies. There is broad literature, and regulatory guidelines in the drug development space that define clinically meaningful change in seizure burden. This needs to be translated into the device space. Furthermore, patient preferences as it relates to the quality of life need to be taken into account in the definition of clinically meaningful. The ILAE would be an ideal forum to define not just the specifics of device sensitivity and specificity but more broadly what is needed in the device space to make a clinically significant difference in the lives of people with epilepsy. This should then lead to ILAE guidelines based on the patient, pharmaceutical, and clinician input as to what should be considered as part of clinical validation studies.
29 July 2020
I read with great interest the proposed clinical guideline. The work was done very carefully and I congratulate the authors for this important effort in an area where there has been rapid progress in the last few years. I am surprised however at the main recommendation of the guideline, i.e. that of wearing clinically validated seizure detection devices for GTCS, on the basis that they are effective. The authors state clearly that no clinical benefit has been demonstrated from wearing such devices. How can they recommend usage of devices that have no established clinical utility? This recommendation is not a necessary consequence of the fact that the devices are effective, which is what this study has demonstrated. I would think that the logical recommendation should be that, now we have demonstrably effective devices, studies can and should be performed to determine if the devices bring a clinical benefit.
Recommending the use of seizure detection devices at this stage runs the risk they will be used widely, industry will be happy and nobody will spend the energy necessary to establish clinical utility. Common sense (it makes sense that it is good to detect GTCS) is not always the best guide. Just as an example, the issue of how late in the seizure it is detected has not been discussed much. It may be a critical factor in clinical utility.
20 July 2020
Dear Professor Beniczky and ILAE-IFCN Working Group,
We congratulate you on your considerable efforts in developing the proposed clinical practice guidelines. The recommendations put forward are justified and appropriately measured given the evidence available.
An important aspect of automated seizure detection technologies that is not discussed in this draft guideline is the counting of seizure events to assess the efficacy of therapies. As highlighted by multiple long-term EEG device studies, seizure counts recorded by diaries typically have poor correlation with electrographic events (Cook et al., 2013, Weisdorf et al., 2019). Therefore, seizure detection devices are worthwhile in their potential to advance the evaluation of seizure therapies.
We believe that it is within the scope of the proposed guidelines to make recommendations on the use of wearable devices to count seizure events. Multiple devices available for purchase internationally provide users with event counts, and as such it is timely for evidence-based recommendations to be provided on the clinical use of this information. In support of this, we note that the current standard for counting seizure events in clinical practice and randomized clinical trials (RCTs) is the use of seizure diaries. Hence, seizure counts from wearable devices should be adopted in non real-time settings (clinical practice, RCTs) if they provide a significant improvement over self-reporting.
We would also note that while existing evidence supports the effectiveness of these devices for detecting tonic-clonic seizures, their application should not be limited to patients known to have generalized tonic-clonic seizures given the clinical importance of these events and the potential for patients to be amnestic to them.
We would bring to your attention a recent pilot study (Page, 2019) suggesting electronic wearable devices and smartphone apps may contribute to reducing hospitalization rates and durations.
As with the currently proposed recommendations, further research can establish if seizure counting with wearable devices confers a significant improvement compared to diaries in applications requiring retrospective therapy assessment.
The My Seizure Gauge Team
17 July 2020
Thank you for summarizing the evidence for the use of wearables in automated seizure detection and deriving recommendations. I would like to raise three points:
- As already pointed out by Mark Richardson, the article focuses on "seizure detection" and reports rates on correctly and falsely detected seizure events. Two main intentions to use wearables, however, are not seizure detection per se but rather seizure monitoring on the one hand and ictal/postictal interventions following an alarm on the other hand. Seizure monitoring may have different targets than correct counting, including informing about trajectories of seizure frequencies and assessment of therapeutic efficacy, and benefit in these areas may occur even if absolute numbers are not correctly counted. Seizure alarming for intervention, as would be needed if SUDEP prevention was a key target, on the other hand requires both, rapid online data analysis at very low false detection rates if alarms are prompted to a medical or other observational team. I wonder if discussing possible clinical intentions first and to then specifying which literature provides relevant data for the respective application could be helpful to the reader.
- A number of studies suggest that there is a relevant inter-patient variability in device performance. I.e. there are patients in whom even bilateral tonic-clonic seizures are hard to detect, and there are patients in whom other seizure types (e.g. tonic seizures or autonomic seizures) may detected reliably. A statement that wearables are generally not recommended to monitor seizure types other than bilateral tonic-clonic seizures may exclude patients from access to wearables in whom devices may be clinically helpful.
- Wearables will be of benefit only to the degree that patients accept their long-term and everyday use. There have been a number of studies addressing patients´ preferences and requirements which may be considered worthwhile to be integrated into the recommendation.
With best regards,
16 July 2020
Thanks for the professors and the ILAE-IFCN working group for this nice and helpful review. It's timely to have the guidelines for Automated seizure detection using wearable devices. Thanks for your effort .
Ayat Allah Farouk Ahmad Hussein
16 July 2020
Nice review. Probably should consider citing and including in the table Lockman J, Fisher RS, Olson DM. Detection of seizure-like movements using a wrist accelerometer. Epilepsy & Behavior 2011;20:638-641. It was one of the pioneering studies.
Robert S. Fisher
16 July 2020
I commend Sandor Beniczky and colleagues for a very helpful review and recommendation regarding this rapidly-emerging field.
To my knowledge, the focus of all studies in this area to date has been to accurately detect the occurrence of individual seizures in real time. Although this is important, it is only one of several possible use-cases for wearable detection devices, and would require a very high sensitivity and very low false alarm rate to be acceptable in real life for most people with epilepsy. It is extremely challenging for devices that do not continuously monitor EEG to achieve a sufficiently high sensitivity and low false alarm rate.
Inaccuracy of self-reported seizure diaries is a major problem in epilepsy management, because therapy recommendations are typically based on self-reported (change in) seizure frequency. Wearable devices could make an important contribution to therapy management, if such devices could be shown to provide a better indication of (change in) the number of seizures occurring over a period of time than self reported diary. For such a use case, a wearable detection device would need to be able to reliably detect a change in seizure frequency, but would not necessarily need to detect every seizure, and certainly would not need to make detections in real time. It is crucial to note that reliable detection of a change in seizure frequency would probably require a much lower sensitivity and would probably be compatible with a much higher false alarm rate, than for the use-case of detection of individual seizure occurrences in real time. In my opinion, it is surprising that device manufacturers, and regulators, are not focussed on reliable detection of change in seizure frequency, because this could be transformative for epilepsy management, and the technical challenges for devices and algorithms might be easier than for real-time detection of individual seizures.
16 July 2020
It is good that the guidelines specify that it is for healthcare professionals use. Will the healthcare professionals also have training on each of these devices before advising the patients/carers?
On page 4 in the Key Point Box we have the recommendation that the wearable devices for GTCS and focal to bilateral tonic clonic seizures are recommended for the automatic detection YET on page 3 under the evidence these are weak/conditional recommendations.
In my experience as an occupational therapist there is also a difference between those children and families with ambulatory epilepsy and those children and families that need to support a non-ambulant child with additional learning needs and live in residential settings.
I agree that further research is needed for these devices and the safe uses to help ease the burden of care for the people with epilepsy.
Mrs. Deanna Middleton
16 July 2020
Dear Professor Beniczky and The ILAE-IFCN Working Group:
This is a very well written and concise review of a clinically highly relevant topic. Thanks for the effort.
The authors might consider adding the notion that persons with epilepsy themselves are well aware of the issue of not noticing many of their seizures (Blachut et al. 2015, 2017: https://pubmed.ncbi.nlm.nih.gov/26076850/ and https://pubmed.ncbi.nlm.nih.gov/28139449). In fact, estimations based on subjective evaluations arrive at quite the same figures than obtained by objective assessment during video-EEG monitoring.
16 July 2020
Dear Professor Beniczky and the team,
Thank you for analyzing the vast data in regards to the use of wearable devices for detection of seizures. With the improvement in technology and availability of fitness bands, it is quite conceivable that such devices will be routinely used in near future for epilepsy patients.
The point which is made very clearly is that seizures with generalized tonic clonic component is most reliably detected. A whole lot of devices using different modalities for detection were used. However, was the group able to comment on the modality which is likely to be most successful among heart rate, EMG, EEG etc.
Since the guideline is intended to have implication for practicing clinicians, which device would the group recommend as a first choice?
16 July 2020
Thank you for your marvelous effort about guidlines for wearable device for automated seizures to decrease morbidity and mortality.
Really I did not see devices like these before in my country Egypt, but from the written guidlines it is very helpful for serious and prolonged seizures.
14 July 2020
Dear Professor Beniczky and The ILAE-IFCN Working Group
Thanks for your efforts collecting evidence for automated seizure detection using wearable devices. After having read your guidelines, I am left with the impression that you have focused mostly on alarming devices, although you write you set out to “provide recommendations on the use of wearable devices for automated seizure detection in outpatients with epilepsy in ambulatory setting, to reduce the morbidity and mortality associated with seizures and to improve the objective documentation of seizure frequency”.
While the former part of this CPG certainly should focus on an online alarm, I would say the latter part can very well be covered by an offline system. In Table 2 you provide a lot of prospective as well as retrospective / offline studies, however, their applicability on treatment optimization and potentially patient empowerment is left unsaid.
In the table-overview I also miss the study by Weisdorf et al., 2019, “Ultra long-term subcutaneous home monitoring of epilepsy - 490 days of EEG from nine patients”. This phase 2 study shows how subcutaneous EEG can be used to detect FBTCS but more importantly; focal impaired awareness seizures – an area you stress yourself is greatly missing by the devices and studies you have included in the guideline. While the paper by Weisdorf et al. doesn’t report on sensitivity, false alarm rate and detection latency, I believe it is still worth mentioning as it might provide a modality for treatment optimization that is filling a gap in the clinical management today.
6 July 2020
Thank you for an interesting guideline. The guideline focuses almost exclusively on devices measuring peripheral effects secondary to abnormal brain activity (like muscle activity), while a number of EEG devices that measure brain activity directly is on the market or on the way to market.
Our article, "A new era in EEG monitoring? Sub-scalp devices for ultra long-term recordings" was just accepted in Epilepsia. You can see more at https://brain-professor.com/EEG.html
The authors are:
Duun-Henriksen, Jonas; UNEEG medical, Epilepsy Science
Baud, Maxime; Hopitaux Universitaires de Geneve Departement des Neurosciences Cliniques, Neurology
Richardson, Mark; King’s College London, Department of Clinical Neuroscience
Cook, Mark; University of Melbourne, Medicine;
Kouvas, George; Wyss Center for Bio and Neuroengineering, CTO
Heasman, John; Cochlear Ltd, Epi-Minder
Friedman, Daniel; NYU Langone Health, Department of Neurology
Peltola, Jukka; Tampere University, Faculty of Medicine and Health Technology; Tampere University Hospital, Neurology
Zibrandtsen, Ivan; Zealand University Hospital Roskilde, Center of Neurophysiology
Kjaer, Troels W; Zealand University Hospital Roskilde, Department of Neurology
Troels Wesenberg Kjaer
30 June 2020
This is a well written guideline addressing two clinical questions “Can automated devices accurately detect GTCS/ FBTCS? Can automated devices detect impaired awareness seizures without GTCS?
28 studies were included, but only 3 phase III trials contribute to the recommendations for GTCS/ FBTCS, while for seizures without TCS 8 phase II studies were used.
Analysis of false alarm rate and device deficiency times could have been added to the scope of the review and all studies, including Phase II could have been utilized for the above outcomes even if they were considered at risk of bias for efficacy or effectiveness.
You may consider adding the following study "Automated Video-Based Detection of Nocturnal Motor Seizures in Children" by. Anouk van Westrhenen et al. Epilepsia. 2020 May.
In TABLE 4: The questions asked are pertinent, but the answers for the two questions viz. “Values and preferences” and “Wise use of resources” are not based on the evidence generated in this review.
Non wrist worn Wearable technologies (which can incorporate EEGs) including vests and caps are likely to be available in near future, with increased detection rate of seizures with fewer false negatives.
29 June 2020
I think the article is timely and should be published ASAP. MY one comment for the authors to consider:
Our metric for therapy success, whether medical or otherwise, seizure diaries, has not evolved for several thousand years, and all of our current methods to approve medicines are based on these diaries. IS there a role of automatic devices in monitoring if ASDs actually work in patients with epilepsy and if the devices should be used in approving new medications and therapies. This comment also applies to surgery and devices used to augment epilepsy like VNS DBS and Neuropace.
26 June 2020
I'm a family caregiver and, in our non-profit organization, we are looking for seizure detection devices since 2008. We are buying and testing the devices in real-life conditions, first by family and then by the persons suffering of epilepsy. Our conclusions are quite different from those published, for daily-life but quite similar for nights. Reason : all validations are made in EEG (or video-EEG) conditions, which is important to be sure that the event detected (or not detected) is a seizure, but ... it is not real life. In real life, on non-epileptic people, the false alarms are high ! I think that it must be the 1st test, as we are sure that all alarms on non-epileptic people are false alarms. Example of false alarms : brush teeth, bicycle, masturbation, clean a table, electrical rasor, etc.
The second point that must not be forgoten is that a high percentage of persons having frequent seizures, specially GTCS, have intellectual disability, maybe also autistic troubles or other troubles. So, an evaluation of the robustness and acceptability of the device is mandatory. 2 examples : The EMFIT matress is fully OK for these people : no need to wear it, no impact on their way to go to bed. . The device named smartwach is a nightmare : vibration on the arm at each detection (unacceptable of most of these persons, no way to disable this), need to have a smartphone very near of the smartwatch (on the pocket or something like), guess how long will it take before it will be broken (due to a seizure falling or to the person's troubles). The device named nighwatch is OK for disabled people, if they accept it (robust, easy to wear, no vibration, etc).
I would like to read studies with 1/false alarm on no-epileptic people in real life conditions and 2/robustness and acceptability of the device (and mandatory tools around) by a disabled person. But I've never found something on that 2 points. These two points must be noted in this paper. From our point of view we are far from devices usable during ordinsry life, but some devices are OK for GTCS detection during sleeping-time, if they are "long-enough" (eg. 10s min for EMFIT)
22 June 2020
Good evening Professor Sándor,
I have gone through the draft of Automated Seizure detection using wearable devices.
Very well written.
Man Mohan Mehndiratta
Subscribe to the ILAE Newsletter
To subscribe, please click on the button below.
Please send me information about ILAE activities and other
information of interest to the epilepsy community