At home EEG monitoring technologies for people with epilepsy and intellectual disabilities: A scoping review

Background: Conducting electroencephalography in people with intellectual disabilities (PwID) can be challenging, but the high proportion of PwID who experience seizures make it an essential part of their care. To reduce hospital-based monitoring, interventions are being developed to enable high-quality EEG data to be collected at home. This scoping review aims to summarise the current state of remote EEG monitoring research, potential benefits and limitations of the interventions, and inclusion of PwID in this research. Methods: The review was structured using the PRISMA extension for Scoping Reviews and the PICOS framework. Studies that evaluated a remote EEG monitoring intervention in adults with epilepsy were retrieved from the PubMed, MEDLINE, Embase, CINAHL, Web of Science, and ClinicalTrials.gov databases. A descriptive analysis provided an overview of the study and intervention characteristics, key results, strengths, and limitations. Results: 34,127 studies were retrieved and 23 were included. Five types of remote EEG monitoring were identified. Common benefits included producing useful results of comparable quality to inpatient monitoring and patient experience. A common limitation was the challenge of capturing all seizures with a small number of localised electrodes. No randomised controlled trials were included, few studies reported sensitivity and specificity, and only three considered PwID. Conclusions: Overall, the studies demonstrated the feasibility of remote EEG interventions for out-of-hospital monitoring and their potential to improve data collection and quality of care for patients. Further research is needed on the effectiveness, benefits, and limitations of remote EEG monitoring compared to in-patient monitoring, especially for PwID.


Background
Electroencephalography (EEG) is a key tool for monitoring epileptiform activity and seizures to diagnose and manage care for epilepsy [1]. Routine, out-patient EEG recordings are often not sufficient for patients who have a low frequency of clinically overt seizure activity [2], so accurate measurement often requires long-term monitoring in hospital. Video-EEG monitoring can provide valuable data to inform patients' care but requires long hospital stays -outside the patients' typical circumstances -and can cost thousands of US dollars equivalent to conduct [3][4][5][6]. Certain population groups -such as people with intellectual disabilities (PwID) -can experience particular difficulties with inpatient video-EEG monitoring.
There is a large population of PwID (approximately 1.5 million [7]) in the UK and there is significant comorbidity between epilepsy and ID. The prevalence of epilepsy in PwID or autism is 22.5%, compared to 0.6% in the general population, and 70% of PwID or autism have treatment resistant epilepsy, compared to 30% in the general population [8]. Despite this, the population of PwID and epilepsy remains underrepresented in research [9] and there is a lack of data on misdiagnosis relating to epilepsy in PwID [10]. PwID experience distinct barriers to seizure-related care, including communication and comprehension difficulties and discomfort or overstimulation in unfamiliar hospital environments [11,12], and are more likely to have brain abnormalities leading to non-seizure linked EEG disturbances and variations which can complicate diagnosis [13]. Remote EEG monitoring -the collection of EEG data over a period of time in an out-of-hospital setting -has the potential to improve the quality of care for PwID and the general population by minimising disruption to their daily lives, reducing the need for hospitalisation, and providing prolonged, high-quality, seizure-activity data.

Rationale
A preliminary review of the literature and previous reviews in this field is detailed in the review protocol [14]. In summary, while some reviews have examined at-home seizure detection [15][16][17][18] the authors did not identify any that included implantable devices. Other reviews examined specific scopes including wearable EEG electrodes [19], ultra long-term wearable or subcutaneous EEG [20], home video-EEG telemetry [21], and the quantification of mobility for remote EEG devices [22]. No reviews were identified -published or registered in the international database of prospectively registered systematic reviews (PROSPERO) -that provided an overview of remote EEG monitoring devices in general or for adult PwID and epilepsy in particular. This gap highlighted the need for an examination of the state of the field regarding remote EEG monitoring interventions and how they are being used to support PwID and epilepsy.

Aim and research questions
The aim of this scoping review was to identify and summarise the current state of research on remote EEG monitoring interventions in general, and for adult PwID and epilepsy in particular, including the types of interventions, their benefits and limitations, and the strengths and weaknesses of the studies investigating them, to inform future directions for research and development. Specifically, the review focused on the research question: What interventions are being evaluated and delivered to enable out-of-hospital EEG monitoring of epileptic seizures in adults, particularly those with intellectual or developmental disorders?

Scope
The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR; Supplemental Material 1) guidance [23]. The Population, Intervention, Comparator, Outcome, and Studies (PICOS) framework was used to develop the search strategy (see Table 1) and structure the review.

Search strategy
Six databases (PubMed, MEDLINE, Embase, CINAHL, Web of Science, and ClinicalTrials.gov) were searched by one author to identify relevant studies on 23 March 2022. To ensure that no relevant studies were missed or excluded in the first stage of screening, a hand search was conducted by an author with extensive experience in the field and awareness of the relevant literature and included in the full-text screen.
Search terms were developed based on a preliminary review of the data and the search string was structured based on three main themes: population (MeSH OR Keywords) AND epilepsy (MeSH OR Keywords) AND remote EEG monitoring (MeSH OR Keywords). An example of the search string (with the number of results returned) is provided in Supplemental Material B.

Eligibility criteria
The inclusion and exclusion criteria are detailed in Textbox 1. To provide a comprehensive summary of out-of-hospital EEG monitoring of epileptic seizures, we did not restrict the type of EEG monitoring to either implantable or not-implanted EEG electrodes. Despite differences in their use and monitoring time frames, both are used in out-of-hospital contexts to collect EEG data. Although the focus of the review was on PwID and epilepsy, the scope of the review was not limited to that population; to support future evidence generation on EEG monitoring for PwID, it was important to capture and summarise all technologies that are potential options for that population. To limit the scope of the review to current and upcoming remote monitoring technologies, only studies published in the past 10 years were eligible. This was a conservative estimate, as it enabled studies with data collected earlier to still be included, reducing the possibility of eliminating relevant interventions.

Population
The main population of interest was adults (≥18 years old) with intellectual disabilities and epilepsy, but the inclusion criteria included all adults with epilepsy to enable an overview of all potential remote EEG monitoring technologies that could be used in the PwID and epilepsy population Intervention Remote EEG monitoring interventions (including physical devices and software or algorithms used to analyse collected data) Comparator No comparator required Outcome The primary outcome was the evidence regarding seizure detection. As a variety of study types were expected, this outcome was defined broadly (e.g. including studies that evaluated the intervention's ability to detect seizures for the purposes of diagnosis or to accurately count seizures for the purposes of clinical management). Secondary outcomes included study characteristics, strengths, and weaknesses and intervention characteristics, benefits, and limitations.

Study types
All study types that evaluated a relevant intervention were eligible for inclusion. Protocols, reviews, meta-analyses, and conference or poster abstracts where no full text was available were excluded.

Textbox 1
Inclusion and exclusion criteria.

Inclusion criteria
• Any adults (≥18 years old) with epilepsy (including but not limited to adult PwID and epilepsy) • Interventions that provide at-home EEG monitoring of epileptiform activity or seizures; including wearable or implanted devices and studies • Primary or secondary data analysis (e.g. use of previously collected datasets) Exclusion criteria • Studies focusing on paediatric populations • Remote monitoring interventions for epilepsy that do not use EEG (e.g. electronic seizure diaries, motion sensors, video monitoring only) • EEG interventions that are not aimed at providing home-based monitoring • Studies that do not evaluate the intervention (e.g. protocols, reviews, abstracts without available full texts) • Studies not published in English • Studies published before 2011

Screening and article selection
References were exported to the citation management software EndNote X9, which was used to detect and remove duplicates. An initial keyword-based screening of references was conducted using the EndNote X9 search function. The remaining titles and abstracts were screened by four of the authors and a full-text review of the included studies was conducted by three of the authors, who discussed their decisions to determine final eligibility. Hand searches identified relevant papers that were not included in the full text review, so a second screening of all the retrieved references was conducted using keywords from the search terms and eligible papers by one author.

Data extraction
Three reviewers extracted data from the included studies into a predeveloped form working as two independent entities (Table 2). In addition to characteristics about the study and the intervention, we extracted the main results reported by the included studies regarding seizure detection (eg, sensitivity, specificity, false-alarm rate, safety, percentage of seizures captured, success at answering clinical question), any data collected about acceptability or patient perceptions, and the benefits and limitations of the intervention.

Data analysis and synthesis
A descriptive analysis of the data extracted from the studies was conducted by three authors and summarised to provide a scoping overview of the state of the literature, the strengths and weaknesses, and implications for future research.

Study characteristics
The database and register search retrieved 34,127 references (see Supplemental Material 2). The EndNote X9 software was used to remove duplicates and conduct initial keyword screening (see Supplemental Material 3). The titles and abstracts of 301 studies were manually screened by one reviewer in Rayyan. Of these articles, 48 were selected for the full-text review and 11 were determined to be eligible for inclusion. Hand-searches by one author identified 8 more studies that were potentially eligible; after independent full-text review by two authors, 6 were eligible for inclusion. To refine the screening process and ensure these and other relevant papers were included, a second screening was conducted. In the second screening, 731 references were identified for title and abstract review after EndNote screening (see Supplemental Material 3). These references were manually screened using the Rayyan software by one reviewer; 177 duplicates were removed and 483 were excluded. The full texts of 58 papers that were not already included in the study were reviewed and 6 were included, for an overall total of 23 included studies. The reasons for exclusion in the full-text review stage are detailed in Fig. 1.
A variety of study designs were used to evaluate the remote EEG monitoring interventions, but cohort designs were the most common (Table 3). Participant sample sizes ranged from 2 [24] to 324 [25] (excepting two studies that used large retrospective databases [6,26]) with three quarters of the studies (17/23) having sample sizes smaller than 30. Most of the studies did not specify a target population beyond having epilepsy or needing prolonged EEG monitoring, but 4 studies specified drug-resistant (refractory) epilepsy [24,[27][28][29], 2 specified focal epilepsy [30,31], and 1 specified mesial temporal lobe epilepsy [32]. Only 1 out of 17 studies specifically looked at a population with ID [33], but two other studies did include some patients with learning disabilities [34,35].
Most of the studies did not specify the degree of mobility associated with the type of remote EEG monitoring intervention they were evaluating. Several can be inferred to have unrestricted mobility -the subscalp EEG [29,32,41,43,44] and intracranial EEG [24,27,28,31] implants and the single-electrode wearables [36,38,40]. One of the ambulatory EEG studies claimed that patients could continue their normal daily activities [30], although the ambulatory EEGs and EEG-NIRS use a cap of EEG electrodes, often wired to a computer or control module that, based on included photos, would appear to hamper mobility to at least some degree [24,39].

Seizure detection
The primary outcome of our review was the evidence reported by the studies about the remote EEG monitoring intervention's ability to detect seizures. This was operationalised differently amongst the studies; a wide range of outcome measures were used, including sensitivity, specificity, accuracy, false prediction rates, usefulness of the data for clinical management, and recording quality. Five of the studies did not evaluate seizure detection [6,24,27,37,39]; two were validation studies, one was examining signal variability post-implant, one was an economic analysis [6], and one used the remote EEG intervention as the gold standard to evaluate patients' self-assessment [37].
Typically, to evaluate the validity of a diagnostic measure, sensitivity and specificity are used to assess how well a tool can correctly identify positives and negatives. Sensitivity, specificity, and other associated measures were what we expected to find for outcomes evaluating the ability of a tool to accurately detect seizures. However, only five of the studies conducted comparisons of the performance of an intervention that enabled reporting of sensitivity, specificity, or false prediction rates -three studies of wearables, and one each of an intracranial EEG and an ambulatory EEG (Table 5) [28,33,36,38,42].
Two studies compared automated seizure detection to experts. One found that the algorithm performed with comparable accuracy to experts (0.84 and 0.80, respectively) [31] while in the other, the algorithm's mean sensitivity performed better than individual or consensus  review, even when clinical decision support was provided [38]. However, it did have a higher false detection rate. Two studies reported accuracy and area under the curve / areaunder-precision-recall curve (AUC / AUPRC) scores, which reflect how well the model distinguishes between conditions. One seizure forecaster had high accuracy at distinguishing seizure from non-seizure hours (83%, AUC = 0.88), although it was only tested on 1 patient [41]. Another reported high model accuracy detecting ictal patterns in existing recordings (AUPRC = 0.84) and prospectively on new patients (AUPRC = 0.80) [31].
Six studies evaluated usefulness for clinical management; one study found that the clinical usefulness of seizure prediction using intracranial EEG was inconclusive [28] but the other five found positive impacts of the interventions. Two studies compared home and inpatient video-EEG and found that their clinical usefulness was comparable -in one study, both produced conclusive findings for 80% of patients (4/5 each) [35], in the other, both enabled accurate interpretation in over 90% of cases (97% or 40/41 for home patients and 91% or 58/64 for inpatients) [34]. The third and fourth studies found that A-EEG was useful for clinical management in 72% (73/101) [5] and 51% (112/219) [25] of cases, respectively. The remaining study qualitatively assessed seizure detection performance and observed that the seizure count collected provided information that would not have been captured from patient seizure diaries [44].
The two studies comparing home and inpatient video-EEG also assessed EEG recording quality and found no significant difference between them [34,35], with home video-EEG as good, if not better, on all the measures except nighttime video quality in one study [34]. Two other studies also assessed EEG signal quality. A study of Epitel's epilogue wearable sensor classified EEG data by bandwidth (which indicates signal quality by estimating the highest frequency where noise and signal are significantly different) and found that 21.4% were good (>75 Hz) while 45.3% were 'marginal' (<35 Hz) [40] and two studies of UNEEG's SubQ subcutaneous EEG found similar spectral characteristics to scalp EEG and stable signal quality over months [32,43].
One study examined the impact of age on the seizure detection ability of A-EEG and found an age difference in EEG detection of interictal epileptiform abnormalities when patients were awake, but not when they were asleep, with a lower sensitivity in older adults [30]. Table 6 Benefits and limitations of remote EEG monitoring by intervention type.

Type of intervention Benefits Limitations
Ambulatory EEG (including EEG-NIRS) • Can produce good results in patients with a variety of indications [5] • Can help diagnose epilepsy and inform care management changes [5,25] • Enables monitoring during activity [24] • Improved mobility and potential to undergo monitoring at home, compared to in-hospital monitoring [24] • Outpatient A-EEG can be more cost-effective than inpatient A-EEG [25] • Scalp EEG not always effective at detecting certain seizure types [5] • Lack of video recording, need to rely on patient/family history [5,25] • No opportunity for trained clinician to observe ictal, pre-, and postictal condition [5] • Artefact recognition is a potential issue [5,25] • System involves cap, wires, and control module -not ideal for normal activity [24] Home video-EEG telemetry • Recording conditions are close to real-life context [37] • Reduced travel time and associated costs or lost income [21,26] • Mitigates barriers if patient has caring responsibilities [21] • New patient groups, who could not have done inpatient video-EEG, can be identified and assessed [21,34] • Better suited for patients with special needs (home and usual carers better equipped to support patient) [21] • Opportunity to consult community professionals and provide multidisciplinary care (particularly in cases of people with learning disabilities) [21] • More cost effective than inpatient video-EEG (app. 2 / 3 the cost) [6,21,26,34] • Quieter and more familiar home environment beneficial for sleep monitoring [34] • Wait times and duration of monitoring can be shorter than for inpatient video EEG monitoring [6,26] • Only a short 1-3 day "snapshot" view possible [37] • Home video quality slightly inferior than at hospital [21] • Can be more difficult for technicians to set up in patients' homes (no sedation if needed, must conduct risk assessment) [21] • Travel time (lost working time) for technician [21,34] • Patients can experience difficulties using video-EEG equipment [26,34] • Not suitable for patients who require sleep deprivation or antiepileptic drug withdrawal pre-EEG for safety reasons [34] Wearables • Non-invasive remote monitoring option with low profile device [36] • More comfortable than multiple electrodes [36,38] • Potential for long-term monitoring, which could enable establishment of chronicity and seizure prediction (providing warning to patients with possible benefits for quality of life) [36,38] • Potential to be widely available [38] • Does not require much maintenance by patients or technicians (up to a week) [38] • Water resistant [38] • Potential to be used as a first step, to avoid unnecessary hospitalisation for monitoring [38,42] • Expected to be more cost-effective (but unproven) [38,42] • The placement of the electrode that best predicts seizures varies between individuals [36] • Single (or few) electrode(s) means that data can be affected by artefacts or by a seizure originating in a different area [36,38,42] • Might not be suitable for seizure onset localisation [38] • Limited memory [36] Sub-scalp EEG • Minimally-invasive, with reduced implantation effect affecting early data [41] • Enables ultra long-term monitoring, which could benefit diagnosis and management and enable establishment of chronicity and seizure prediction [32,41,44] • Less noisy than scalp EEG [41] • High-quality signal comparable to scalp EEG [41] • Sensitive to small neurological events (e.g. sleep transients) [41,44] • No impact on mobility or need for long-term hospitalisation [41,43,44] • Does not require regular maintenance from technicians [43] • Susceptible to noise and artifacts from muscle activity [41] • Implantation not acceptable for all patients [41] • Data review requires time-intensive analysis a trained neurophysiologist to review, so extensive use will require effective seizure detection algorithms [29,41] • Only cover a small area of the cortical surface [32,44] Intracranial EEG • Can be used not just for monitoring but for seizure prediction, therapeutic electrical stimulation, responsive neurostimulation, and adaptive deep brain stimulation [27,31] • Allow for long-term monitoring [27,31] • No impact on mobility or need for long-term hospitalisation [27] • Implantation effect (trauma, inflammation, or induced epileptiform activity or seizures) can affect early data recording [27] • Variability in signal properties [27] • Transmission quality from implant needs to be high [28]

Acceptability
Nine of the 23 studies reported some type of patient experience or acceptability data. Only one study used a specific questionnaire (8-item Client Satisfaction Questionnaire), but this was for patients' overall satisfaction with the comprehensive program and was not specific to the A-EEG evaluated [5]. Three studies examined patients preferences and found that most, but not all, participants preferred home to inpatient video-EEG [34,35] and that wrist-worn devices (which did not measure EEG) were preferred to the single-electrode EEG wearable [40].
Other studies collected patient feedback anecdotally and found that, for the seizure advisory device, higher satisfaction was associated with lower proportions of time in the high likelihood advisory [28], that patients "wore [the EEG-NIRS] cap for several hours without annoyance" [24], and that sub-scalp EEG implants were generally well-tolerated and acceptable [41,44].
Three studies captured adverse events associated with implanted EEGs. One study of the SubQ sub-scalp EEG identified 13 adverse events, none of which were serious [44] and another (case study) reported that the device was well accepted "after a short adaptation period" with no serious adverse events [29]. One study of an intracranial EEG implant identified 11 adverse events in the first four months (2 of which were serious) and an additional 2 serious events within the first year [28].

Benefits and limitations of the interventions
Most of the studies detailed some of the potential benefits and limitations of the type of remote EEG monitoring intervention investigated. These have been compiled in Table 6, divided by the type of intervention.

Remote EEG monitoring in PWID
Although only three studies referred to PwID in their analyses [33][34][35] and only one was specifically focused on seizure detection in that population [33], they all reported positive perspectives on the use of remote EEG monitoring for PwID and epilepsy. Two of the studies highlighted potential benefits of the adoption of remote monitoring methods for PwID relating to the identification of new patient groups and provision of care to those who otherwise could not access it, patients' experience (less traumatic at home in familiar environment with normal care support than in hospital), and the opportunity for multidisciplinary consultations [34,35]. The third study used a retrospective dataset of scalp EEG recordings from PwID to develop an automated seizure detection algorithm. They found that although there were large differences in detection performance across patients, the seizure detector performed promisingly on a seizure pattern common in PwID (discharge with EMG activity) and had good sensitivity for seizures lasting over a minute [33].

Summary of findings
The review identified 23 studies that evaluated remote EEG monitoring interventions for adults with epilepsy, although only 3 studies considered the impact of remote EEG monitoring on PwID and epilepsy. There were multiple types of remote EEG monitoring technologies identified, including ambulatory EEG, home video-EEG, wearable electrodes, sub-scalp EEG, and intracranial EEG. At-home wearable EEG monitoring can range from being as bulky as standard outpatient EEG caps and wires, but with attached control modules that make it possible to remain at home, to single electrode patches that can be placed discreetly on the scalp. Implants range from minimally invasive subscalp electrodes to full intracranial electrode arrays. This highlights the range of remote EEG monitoring options available, varying in invasiveness, signal quality, and enabled mobility. The studies examined a variety of different outcome measures, with only five examining the interventions' sensitivity at detecting seizures. About half of the studies reported something about patient acceptability, but none measured it robustly. There were a variety of benefits and limitations for each of the intervention types; in general, remote EEG monitoring produced useful results, often largely comparable to inpatient recordings, and patient mobility and experience were frequently mentioned as benefits. Two studies highlighted potential cost benefits of home, compared to inpatient, video-EEG monitoring. A common limitation was that a small number of localised electrodes might not be able to capture all seizures, depending on where they originate.

Strengths and weaknesses of the studies
Few of the studies reported their strengths, which is a limitation, as it impedes interpretation of study quality by readers. Certain study design strengths were identified and extracted by reviewers, including the use of a comparison -whether between patients and healthy controls [40], home or clinically-based EEG monitoring [21], or epileptologist or algorithmic review [38], relatively large sample sizes or datasets [5,31,34], and the inclusion of a variety of epilepsy types to improve generalisability [41].
The most prevalent methodological weakness identified was a small sample size. Only two studies had sample sizes greater than 100 participants [5,34] and most had fewer than 30 participants. Even amongst the studies with larger sample sizes, subgroup analyses had fewer participants, limiting the validity of results. This is a significant problem because it affects how well we can interpret the results. Other limitations included a lack of clarity in reporting -for example, the order of intervention types, whether drug changes during the study were tracked, and whether inpatient or remote EEG monitoring data was used in the analysis. Other limitations were particular to the study design; as there were a variety of study types included in the review, there were a variety of limitations reported by the individual studies. Examples of reported limitations included: early conclusion of the study [28], a limited number of channels which limited interpretation of the data [5], a lack of statistical analysis and generalisability due to sample size, population, or epilepsy type included [6,29,32,34,38,42,44], a lack of reviewer experience using and interpreting data from the monitoring system [38,39], limited validation or analysis due to a lack of time, reviewers, or resources [38,41,44], a lack of randomisation (resulting in potential sampling bias) [33,34], data about the remote monitoring system collected in a clinical setting rather than at home [24,38,39], and different implant locations between healthy subjects and controls [43].
Another potential limitation is that many of the studies only evaluated the remote EEG monitoring intervention over a limited period of time (the duration of use for half of the interventions was less than 10 days). While this duration was often limited by logistical factors (such as battery life), these interventions provide a more limited snapshot of epileptic brain activity than longer-term monitoring interventions. However, long-term monitoring may not be necessary for all patients, so certain remote monitoring types may be more or less appropriate depending on the particular clinical case.

Strengths and limitations of the review
This review is the first to provide a comprehensive overview of remote EEG monitoring interventions designed for use in non-clinical settings. A strength is that the review was conducted by multiple authors working independently and the collaboration between researchers with expertise in digital health, epilepsy, and PwID, to ensure that the technical aspects of the studies were interpreted correctly. However, the limited number of included studies considering PwID and the heterogeneity of study designs and remote EEG monitoring intervention types made it difficult to conduct a rigorous comparison of the effectiveness, benefits, and limitations of certain types of intervention compared to others. As a review of published literature, it also may have excluded newly-developed commercial systems that may not have been clinically investigated yet.
The search was designed to focus on out-of-hospital EEG monitoring; however, a few of the studies reviewed at the full-text screening stage were examining interventions intended for home use in clinical settings. After discussion, the authors came to a consensus to retain these studies because we decided that, at this stage, this information was relevant and useful to include to demonstrate the state of the field. As the main focus of the search and screening was to identify evaluations of out-of-hospital EEG monitoring interventions, this review does not claim to have captured all studies that examined EEG monitoring devices (which could potentially be used out-of-hospital) in clinical settings.

Conclusion and future research
There is a growing number of studies examining remote EEG monitoring tools, particularly in recent years. This review demonstrates the variety of different types of remote EEG monitoring tools available, which fall into two main categories -wearable or implantable. Several studies examined the potential use of this remote monitoring opportunity to provide seizure prediction that would not be possible with only short-term, episodic EEG monitoring. This is likely to become an even bigger element of the research, as the emphasis on personalised and preventive medicine grows.
Likewise, some studies were beginning to examine the potential of remote EEG monitoring to support underserved population groups, such as PwID and epilepsy. This is a particularly important area of investigation, as a large portion of PwID also have epilepsy; PwID also have a higher risk of Sudden Unexpected Death in Epilepsy (SUDEP) than the general population [45][46][47]. PwID have a high prevalence of key risk factors, such as nocturnal seizures and a lack of nighttime surveillance [46]. The use of home-based remote monitoring systems could potentially be used to support PwID in the long-term and improve our understanding of SUDEP, risk factors in the PwID population, and how they could potentially be mitigated. It will be important to investigate which remote EEG monitoring interventions are best suited for PwID, both in terms of accurately detecting epileptic activity and seizures, which can be more complex in PwID, and in terms of acceptability. Acceptability will be a key area for future research of the use of such remote monitoring interventions, especially for PwID, and research is needed to address the particular challenges of implementing such interventions in that population (for example, how the intervention can be explained most clearly to get their consent or assent, and how the intervention can be implemented so that it causes minimal interference to their lives). Future research should also continue to explore the benefits and challenges of such remote EEG interventions for different types of people, so that inpatient resources can be prioritised to those who will most benefit from them and people who experience difficulties and disruption due to inpatient stays can receive high-quality care in a more familiar setting.

Declaration of Competing Interest
The funder, UNEEG Medical UK Ltd, manufactures the 24/7 EEG™ SubQ device; a long-term subcutaneous implant for remote EEG monitoring of epilepsy. JDH and LB are employees of UNEEG medical.