Advertisement

Status epilepticus prevention, ambulatory monitoring, early seizure detection and prediction in at-risk patients

  • Marta Amengual-Gual
    Affiliations
    Division of Epilepsy and Clinical Neurophysiology, Department of Neurology, Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA

    Pediatric Neurology Unit, Department of Pediatrics, Hospital Universitari Son Espases, Universitat de les Illes Balears, Palma, Spain
    Search for articles by this author
  • Adriana Ulate-Campos
    Affiliations
    Department of Neurology, National Children’s Hospital “Dr. Carlos Saenz Herrera”, San José, Costa Rica
    Search for articles by this author
  • Tobias Loddenkemper
    Correspondence
    Corresponding author at: Harvard Medical School, Division of Epilepsy and Clinical Neurophysiology, Fegan 9, Boston Children’s Hospital, 300 Longwood Avenue, Boston, MA 02115, United States.
    Affiliations
    Division of Epilepsy and Clinical Neurophysiology, Department of Neurology, Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA
    Search for articles by this author
Open ArchivePublished:September 18, 2018DOI:https://doi.org/10.1016/j.seizure.2018.09.013

      Highlights

      • Epileptic seizure occurrence follows patient-specific non-random patterns.
      • Ambulatory monitoring systems aim to improve seizure detection.
      • Multimodal devices improve seizure detection compared to unimodal devices.
      • Big data and machine learning may help elucidate detection and predictive algorithms.
      • Closed-loop systems may improve monitoring and treatment of seizures in the future.

      Abstract

      Purpose

      Status epilepticus is an often apparently randomly occurring, life-threatening medical emergency which affects the quality of life in patients with epilepsy and their families. The purpose of this review is to summarize information on ambulatory seizure detection, seizure prediction, and status epilepticus prevention.

      Method

      Narrative review.

      Results

      Seizure detection devices are currently under investigation with regards to utility and feasibility in the detection of isolated seizures, mainly in adult patients with generalized tonic-clonic seizures, in long-term epilepsy monitoring units, and occasionally in the outpatient setting. Detection modalities include accelerometry, electrocardiogram, electrodermal activity, electroencephalogram, mattress sensors, surface electromyography, video detection systems, gyroscope, peripheral temperature, photoplethysmography, and respiratory sensors, among others. Initial detection results are promising, and improve even further, when several modalities are combined. Some portable devices have already been U.S. FDA approved to detect specific seizures. Improved seizure prediction may be attainable in the future given that epileptic seizure occurrence follows complex patient-specific non-random patterns. The combination of multimodal monitoring devices, big data sets, and machine learning may enhance patient-specific detection and predictive algorithms. The integration of these technological advances and novel approaches into closed-loop warning and treatment systems in the ambulatory setting may help detect seizures sooner, and tentatively prevent status epilepticus in the future.

      Conclusions

      Ambulatory monitoring systems are being developed to improve seizure detection and the quality of life in patients with epilepsy and their families.

      Keywords

      1. Introduction

      Status epilepticus (SE) is a time-sensitive and life-threatening medical emergency. The occurrence of seizures and SE appears often random and difficult to predict, reducing the quality of life in patients with epilepsy and their families [
      • Jacoby A.
      Epilepsy and the quality of everyday life. Findings from a study of people with well-controlled epilepsy.
      ]. Here, we summarize the literature on ambulatory seizure detection (focusing on detection modalities and devices for ambulatory monitoring), seizure prediction (focusing on patterns of epileptic seizure and SE occurrence as well as EEG features), and tentatively SE prevention (using closed-loop warning and treatment systems in the future).

      2. Seizure detection

      2.1 Ambulatory monitoring systems

      There are no specific articles focusing on devices for ambulatory monitoring of SE. Most devices are under investigation and, as such, have been mostly tested for detection of isolated seizures, mainly in adult patients, with generalized tonic-clonic seizures (GTCS), in long-term epilepsy monitoring units and small outpatient setting populations. A variety of modalities are used to detect seizures, such as accelerometry (ACM), electrocardiogram (EKG), electrodermal activity (EDA), electroencephalogram (EEG), mattress sensors, surface electromyography (sEMG), video detection systems, gyroscope, peripheral temperature, photoplethysmography, and respiratory sensors, among others.

      2.1.1 Accelerometry

      Accelerometry (ACM) detects changes in velocity and direction, and has been used mainly for detection of seizures with a motor component [
      • Beniczky S.
      • et al.
      Detection of generalized tonic-clonic seizures by a wireless wrist accelerometer: a prospective, multicenter study.
      ,
      • Lockman J.
      • Fisher R.S.
      • Olson D.M.
      Detection of seizure-like movements using a wrist accelerometer.
      ,
      • Nijsen T.M.
      • et al.
      Time-frequency analysis of accelerometry data for detection of myoclonic seizures.
      ,
      • Nijsen T.M.
      • et al.
      The potential value of three-dimensional accelerometry for detection of motor seizures in severe epilepsy.
      ]. A frequent challenge is to differentiate seizures from daily, repetitive, rhythmic movements [
      • Lockman J.
      • Fisher R.S.
      • Olson D.M.
      Detection of seizure-like movements using a wrist accelerometer.
      ,
      • Nijsen T.M.
      • et al.
      The potential value of three-dimensional accelerometry for detection of motor seizures in severe epilepsy.
      ,
      • Dalton A.
      • et al.
      Development of a body sensor network to detect motor patterns of epileptic seizures.
      ]. In addition to GTCS, ACM may also have the potential to detect focal seizures with minimal motor components, unilateral as well as bilateral tonic-clonic, myoclonic, clonic, tonic and hypermotor seizures [
      • Beniczky S.
      • et al.
      Detection of generalized tonic-clonic seizures by a wireless wrist accelerometer: a prospective, multicenter study.
      ,
      • Lockman J.
      • Fisher R.S.
      • Olson D.M.
      Detection of seizure-like movements using a wrist accelerometer.
      ,
      • Nijsen T.M.
      • et al.
      Time-frequency analysis of accelerometry data for detection of myoclonic seizures.
      ,
      • Nijsen T.M.
      • et al.
      The potential value of three-dimensional accelerometry for detection of motor seizures in severe epilepsy.
      ,
      • Van de Vel A.
      • Verhaert K.
      • Ceulemans B.
      Critical evaluation of four different seizure detection systems tested on one patient with focal and generalized tonic and clonic seizures.
      ,
      • Kramer U.
      • et al.
      A novel portable seizure detection alarm system: preliminary results.
      ,
      • Schulc E.
      • et al.
      Measurement and quantification of generalized tonic-clonic seizures in epilepsy patients by means of accelerometry--an explorative study.
      ,
      • Patterson A.L.
      • et al.
      SmartWatch by SmartMonitor: assessment of seizure detection efficacy for various seizure types in children, a large prospective single-center study.
      ]. Sensitivity ranges between 16–100% depending on seizure type and detection algorithm, and one study had a false detection rate (FDR) of 0.2/day [
      • Beniczky S.
      • et al.
      Detection of generalized tonic-clonic seizures by a wireless wrist accelerometer: a prospective, multicenter study.
      ,
      • Lockman J.
      • Fisher R.S.
      • Olson D.M.
      Detection of seizure-like movements using a wrist accelerometer.
      ,
      • Dalton A.
      • et al.
      Development of a body sensor network to detect motor patterns of epileptic seizures.
      ,
      • Kramer U.
      • et al.
      A novel portable seizure detection alarm system: preliminary results.
      ,
      • Schulc E.
      • et al.
      Measurement and quantification of generalized tonic-clonic seizures in epilepsy patients by means of accelerometry--an explorative study.
      ,
      • Patterson A.L.
      • et al.
      SmartWatch by SmartMonitor: assessment of seizure detection efficacy for various seizure types in children, a large prospective single-center study.
      ,
      • Van de Vel A.
      • et al.
      Long-term home monitoring of hypermotor seizures by patient-worn accelerometers.
      ,
      • Jallon P.
      • et al.
      Detection system of motor epileptic seizures through motion analysis with 3D accelerometers.
      ,
      • Velez M.
      • et al.
      Tracking generalized tonic-clonic seizures with a wrist accelerometer linked to an online database.
      ]. Some advantages include relatively low power consumption and good patient tolerance [
      • Patterson A.L.
      • et al.
      SmartWatch by SmartMonitor: assessment of seizure detection efficacy for various seizure types in children, a large prospective single-center study.
      ,
      • Van de Vel A.
      • et al.
      Long-term home monitoring of hypermotor seizures by patient-worn accelerometers.
      ,
      • Jallon P.
      • et al.
      Detection system of motor epileptic seizures through motion analysis with 3D accelerometers.
      ,
      • Velez M.
      • et al.
      Tracking generalized tonic-clonic seizures with a wrist accelerometer linked to an online database.
      ]. One group validated a single wrist-worn accelerometer sensor and found a detection sensitivity for GTCS of 95.2% and a FDR of 0.72/day [
      • Kusmakar S.
      • et al.
      Detection of generalized tonic-clonic seizures using short length accelerometry signal.
      ]. Another recent study using ACM to detect GTCS with spectral analysis had better results than the one that used temporal signal analysis achieving a sensitivity of 100% and a FDR of 2.0/day [
      • Joo H.S.
      • et al.
      Spectral analysis of acceleration data for detection of generalized tonic-clonic seizures.
      ]. Additionally, one study validated the system in a home environment detecting 78.5% of seizures reported by parents, with 0.6 false alarms per night [
      • Jallon P.
      • et al.
      Detection system of motor epileptic seizures through motion analysis with 3D accelerometers.
      ].
      Disadvantages may include restriction to seizures with a motor component, and missed seizures when an obstacle limits free limb movement [
      • Beniczky S.
      • et al.
      Detection of generalized tonic-clonic seizures by a wireless wrist accelerometer: a prospective, multicenter study.
      ]. This modality has good sensitivity, with good night detection rates, and most patients and families found the device user-friendly [
      • Beniczky S.
      • et al.
      Detection of generalized tonic-clonic seizures by a wireless wrist accelerometer: a prospective, multicenter study.
      ,
      • Lockman J.
      • Fisher R.S.
      • Olson D.M.
      Detection of seizure-like movements using a wrist accelerometer.
      ,
      • Schulc E.
      • et al.
      Measurement and quantification of generalized tonic-clonic seizures in epilepsy patients by means of accelerometry--an explorative study.
      ].

      2.1.2 Electrocardiogram

      Electrocardiogram (EKG) can be recorded from a single channel and has a higher signal-to-noise ratio than electroencephalogram (EEG) [
      • Behbahani S.
      • et al.
      Pre-ictal heart rate variability assessment of epileptic seizures by means of linear and non-linear analyses.
      ,
      • Osorio I.
      • Manly B.F.
      Probability of detection of clinical seizures using heart rate changes.
      ]. The pattern of heart rate changes during seizures is highly patient-specific, warranting the development of patient-tailored detection algorithms [
      • Leutmezer F.
      • et al.
      Electrocardiographic changes at the onset of epileptic seizures.
      ,
      • van Elmpt W.J.
      • et al.
      A model of heart rate changes to detect seizures in severe epilepsy.
      ,
      • Kerem D.H.
      • Geva A.B.
      Forecasting epilepsy from the heart rate signal.
      ,
      • Jeppesen J.
      • et al.
      Detection of epileptic seizures with a modified heart rate variability algorithm based on Lorenz plot.
      ,
      • De Cooman T.
      • et al.
      Online detection of tonic-clonic seizures in pediatric patients using ECG and low-complexity incremental novelty detection.
      ]. EKG has been used to detect focal seizures, focal seizures evolving into bilateral tonic-clonic seizures, and GTCS [
      • Jansen K.
      • et al.
      Peri-ictal ECG changes in childhood epilepsy: implications for detection systems.
      ,
      • Varon C.
      • et al.
      Can ECG monitoring identify seizures?.
      ]. Sensitivities in these seizure types ranged from 90% to 100% [
      • Varon C.
      • et al.
      Can ECG monitoring identify seizures?.
      ]. A comparison among hospital EKG, a wearable EKG device, and photoplethysmography in patients with temporal seizures in hospital setting showed sensitivities of 57%, 70% and 32%, respectively [
      • Vandecasteele K.
      • et al.
      Automated epileptic seizure detection based on wearable ECG and PPG in a hospital environment.
      ]. Regarding ambulatory application of EKG monitoring, a group developed a wireless EKG device with a sensitivity of 99.8% and positive predictive value (PPV) of 99.8% [
      • Romero I.
      • Grundlehner B.
      • Penders J.
      Robust beat detector for ambulatory cardiac monitoring.
      ]. Cardiac-based activation vagus nerve stimulation (VNS) is now part of a commercially available closed-loop system, which reports a sensitivity of 80% [
      • Boon P.
      • et al.
      A prospective, multicenter study of cardiac-based seizure detection to activate vagus nerve stimulation.
      ]. The effect on seizure frequency was moderate but there was significant improvement in quality of life [
      • Boon P.
      • et al.
      A prospective, multicenter study of cardiac-based seizure detection to activate vagus nerve stimulation.
      ].
      Some EKG disadvantages include low specificity of heart rate changes, limited stability of externally applied electrodes, and discomfort with long-term use, which could be avoided with wireless, or VNS activated devices [
      • van Elmpt W.J.
      • et al.
      A model of heart rate changes to detect seizures in severe epilepsy.
      ,
      • Ramgopal S.
      • et al.
      Seizure detection, seizure prediction, and closed-loop warning systems in epilepsy.
      ]. Automated seizure detection from a single EKG lead is feasible, especially when parameters are tailored to the individual patient, and results improve when combined with other detection modalities.

      2.1.3 Electrodermal activity

      Changes in skin conductance are referred to as electrodermal activity (EDA). EDA is best detected in close proximity to sweat glands (and usually best detected in the volar surfaces of the fingers and palm of the hand) [
      • Taylor N.A.
      • Machado-Moreira C.A.
      Regional variations in transepidermal water loss, eccrine sweat gland density, sweat secretion rates and electrolyte composition in resting and exercising humans.
      ]. EDA reflects the activity of the sympathetic branch of the autonomic nervous system, and it may be transiently increased during epileptic seizures [
      • Poh M.Z.
      • Swenson N.C.
      • Picard R.W.
      A wearable sensor for unobtrusive, long-term assessment of electrodermal activity.
      ,
      • Poh M.Z.
      • et al.
      Continuous monitoring of electrodermal activity during epileptic seizures using a wearable sensor.
      ], specifically during GTCS and focal seizures with impaired awareness [
      • Poh M.Z.
      • et al.
      Continuous monitoring of electrodermal activity during epileptic seizures using a wearable sensor.
      ]. In a study including seven patients with GTCS and focal seizures with impaired awareness, EDA was significantly elevated immediately after the onset of each EEG seizure [
      • Poh M.Z.
      • et al.
      Continuous monitoring of electrodermal activity during epileptic seizures using a wearable sensor.
      ]. Change in EDA remained higher for longer periods of time in GTCS compared to focal seizures with impaired awareness [
      • Poh M.Z.
      • et al.
      Continuous monitoring of electrodermal activity during epileptic seizures using a wearable sensor.
      ]. In a study including 11 patients, 100% of GTCS were associated with a greater than 2 standard deviation increase from baseline in EDA, and up to 86% of focal seizures with impaired awareness had EDA changes [
      • Poh M.Z.
      • et al.
      Autonomic changes with seizures correlate with postictal EEG suppression.
      ]. EDA recordings may also add the understanding of the pathophysiology of sudden unexpected death in epilepsy (SUDEP) [
      • Poh M.Z.
      • et al.
      Autonomic changes with seizures correlate with postictal EEG suppression.
      ,
      • Sarkis R.A.
      • et al.
      Autonomic changes following generalized tonic clonic seizures: an analysis of adult and pediatric patients with epilepsy.
      ]. Novel frameworks are under investigation to improve processing of EDA signals [
      • Jain S.
      • et al.
      A compressed sensing based decomposition of electrodermal activity signals.
      ]. There are also suggestions to utilize EDA signals in closed-loop biofeedback training, which may have the potential to reduce seizure frequency in patients with temporal seizures [
      • Micoulaud-Franchi J.A.
      • et al.
      Skin conductance biofeedback training in adults with drug-resistant temporal lobe epilepsy and stress-triggered seizures: a proof-of-concept study.
      ,
      • Kotwas I.
      • et al.
      A case-control study of skin conductance biofeedback on seizure frequency and emotion regulation in drug-resistant temporal lobe epilepsy.
      ].
      Disadvantages of EDA include susceptibility to pressure and motion artifacts, variability with temperature, and in selected patients, discomfort [
      • Poh M.Z.
      • Swenson N.C.
      • Picard R.W.
      A wearable sensor for unobtrusive, long-term assessment of electrodermal activity.
      ]. Larger studies on continuous ambulatory autonomic monitoring may provide additional insights to optimize this modality.

      2.1.4 Electroencephalogram

      Video-EEG is the gold standard for the diagnosis of epilepsy. Seizures detected with a traditional ambulatory video-EEG include a wide semiological variety, such as focal seizures with retained awareness, focal seizures with impaired awareness, focal seizures with evolution into bilateral tonic-clonic features, absence seizures [
      • Petersen E.B.
      • et al.
      Generic single-channel detection of absence seizures.
      ], among others.
      With the aim to facilitate portability and patients’ comfort, one group developed an ambulatory noninvasive EEG monitoring device with 2 channels connected via Bluetooth to a smartphone [
      • Myers M.H.
      • et al.
      Ambulatory seizure monitoring: from concept to prototype device.
      ]. The device was tried in 3 patients achieving a detection sensitivity of 75–100% [
      • Myers M.H.
      • et al.
      Ambulatory seizure monitoring: from concept to prototype device.
      ]. Another noninvasive 2-electrode EEG monitoring device is able to track seizures for seven days and allows home monitoring [
      • Lehmkuhle M.
      • et al.
      Development of a discrete, wearable, EEG device for counting seizures (Abstract nr 2.158) 2015.
      ]. Finally, as part of the efforts to simplify EEG recordings and make interpretation easier, another group developed a portable EEG data recorder and a noninvasive 10-electrode headband with rapid setup (approximately 5 min). This device was compared to a commercially available EEG widely used in clinical setting showing similar signal quality, while requiring less setup time and allowing greater portability [
      • Munaretto J.
      • et al.
      Technical analysis of the Ceribell EEG device.
      ,
      • Parvizi J.
      • et al.
      Detecting silent seizures by their sound.
      ].
      Intracranial EEG (iEEG) offers advantages of better localized signals and decreased biological noise [
      • Ray A.
      • et al.
      Localizing value of scalp EEG spikes: a simultaneous scalp and intracranial study.
      ]. Some implantable intracranial devices based on iEEG are detailed in the subsection 4.1 [
      • Morrell M.J.
      • RNS System in Epilepsy Study Group
      Responsive cortical stimulation for the treatment of medically intractable partial epilepsy.
      ,
      • Cook M.J.
      • et al.
      Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study.
      ]. Recently, a multicenter collaboration that worked on high quality intracranial data achieved robust and accurate seizure detection [
      • Baldassano S.N.
      • et al.
      Crowdsourcing seizure detection: algorithm development and validation on human implanted device recordings.
      ]. The development of new prediction methods, mainly based on iEEG data, has yielded prediction accuracies of 75.8–97.5%, and false prediction rates of 0.05–1/h [
      • Bandarabadi M.
      • et al.
      On the proper selection of preictal period for seizure prediction.
      ,
      • Ozdemir N.
      • Yildirim E.
      Patient specific seizure prediction system using Hilbert spectrum and Bayesian networks classifiers.
      ,
      • Parvez M.Z.
      • Paul M.
      Epileptic seizure prediction by exploiting spatiotemporal relationship of EEG signals using phase correlation.
      ,
      • Wang N.
      • Lyu M.R.
      Extracting and selecting distinctive EEG features for efficient epileptic seizure prediction.
      ,
      • Alexandre Teixeira C.
      • et al.
      Epileptic seizure predictors based on computational intelligence techniques: a comparative study with 278 patients.
      ,
      • Zandi A.S.
      • et al.
      Epileptic seizure prediction using variational mixture of Gaussians.
      ,
      • Bandarabadi M.
      • et al.
      Epileptic seizure prediction using relative spectral power features.
      ,
      • Aarabi A.
      • He B.
      Seizure prediction in intracranial EEG: a patient-specific rule-based approach.
      ,
      • Gadhoumi K.
      • Gotman J.
      • Lina J.M.
      Scale invariance properties of intracerebral EEG improve seizure prediction in mesial temporal lobe epilepsy.
      ,
      • Rabbi A.F.
      • Azinfar L.
      • Fazel-Rezai R.
      Seizure prediction using adaptive neuro-fuzzy inference system.
      ,
      • Park Y.
      • et al.
      Seizure prediction with spectral power of EEG using cost-sensitive support vector machines.
      ].
      Potential disadvantages of noninvasive EEG may include that patients often have to wear scalp electrodes and remain attached to EEG equipment, which is uncomfortable, may increase artifacts and may lead to stigmatization, as electrodes are often difficult to hide [
      • Schulze-Bonhage A.
      • et al.
      Views of patients with epilepsy on seizure prediction devices.
      ]. However, in the future, these challenges may be overcome by lighter and smaller monitoring systems, which may also be implanted subcutaneously or in the subgaleal space in selected cases. Limitations of iEEG include risks associated with a surgical procedure as well as those of wearing an implanted device.

      2.1.5 Mattress sensors

      Mattress sensors are usually placed under the patient’s mattress or bedding, and connected to a monitor. The sensor alerts the caregivers when a stimulus above the selected threshold is detected [
      • Carlson C.
      • et al.
      Detecting nocturnal convulsions: efficacy of the MP5 monitor.
      ,
      • Fulton S.
      • et al.
      Prospective study of 2 bed alarms for detection of nocturnal seizures.
      ,
      • Poppel K.V.
      • et al.
      Prospective study of the Emfit movement monitor.
      ]. These sensors use detection techniques based on pressure changes triggered by patient’s movement. Nevertheless, novel sensors may be able to detect air-pressure fluctuations caused by the tiny tremors of heartbeats –heart rate and breathing can be tracked and apneas may already be detected-. The seizures detected with classic mattress sensors are GTCS and focal seizures with impaired awareness and motor onset. The reported sensitivity of three devices ranged from 16.7% to 85%, being better for nocturnal GTCS [
      • Carlson C.
      • et al.
      Detecting nocturnal convulsions: efficacy of the MP5 monitor.
      ,
      • Fulton S.
      • et al.
      Prospective study of 2 bed alarms for detection of nocturnal seizures.
      ,
      • Poppel K.V.
      • et al.
      Prospective study of the Emfit movement monitor.
      ]. Monitoring while the patient is in bed may be helpful, as most cases of SUDEP occur in patients with GTCS, while unsupervised, in bed at night [
      • Lamberts R.J.
      • et al.
      Sudden unexpected death in epilepsy: people with nocturnal seizures may be at highest risk.
      ,
      • Langan Y.
      • Nashef L.
      • Sander J.W.
      Case-control study of SUDEP.
      ]. Furthermore, mattress sensors do not require device attachment to the patient’s body.
      These devices have the disadvantages of currently having weight restrictions, detecting only seizures with rhythmic movements, and often have relatively low sensitivity [
      • Fulton S.
      • et al.
      Prospective study of 2 bed alarms for detection of nocturnal seizures.
      ,
      • Poppel K.V.
      • et al.
      Prospective study of the Emfit movement monitor.
      ]. Individual calibration and testing over a couple of nights in a home setting may be helpful, and might improve sensitivity [
      • Carlson C.
      • et al.
      Detecting nocturnal convulsions: efficacy of the MP5 monitor.
      ].

      2.1.6 Surface electromyography

      Analysis of muscle activity with sEMG is a viable option for seizure detection, mainly for seizures with a motor component [
      • Larsen S.N.
      • et al.
      Detection of tonic epileptic seizures based on surface electromyography.
      ]. sEMG detects muscle activity with one channel [
      • Conradsen I.
      • et al.
      Automated algorithm for generalized tonic-clonic epileptic seizure onset detection based on sEMG zero-crossing rate.
      ,
      • Conradsen I.
      • et al.
      Patterns of muscle activation during generalized tonic and tonic-clonic epileptic seizures.
      ,
      • Conradsen I.
      • et al.
      Evaluation of novel algorithm embedded in a wearable sEMG device for seizure detection.
      ]. sEMG during tonic seizures, recorded over the deltoid muscle, had a sensitivity of 53–63% and a FDR of 1.49–4.03/h, achieving better results when certain parameters were tailored for each patient [
      • Larsen S.N.
      • et al.
      Detection of tonic epileptic seizures based on surface electromyography.
      ]. sEMG electrodes placed on the biceps and triceps detected 95% of GTCS, with only one false positive, during a recording period of 1399 h, but no other seizure types (myoclonic, tonic, absence, and focal seizures with or without loss of awareness) were detected [
      • Szabo C.A.
      • et al.
      Electromyography-based seizure detector: preliminary results comparing a generalized tonic-clonic seizure detection algorithm to video-EEG recordings.
      ]. An EMG-based system well-placed over the belly of the biceps demonstrated 100% sensitivity to detect GTCS (n = 29 episodes) when compared with video-EEG, with a detection delay average of 7.7 s from the onset of bilateral appendicular tonic motor manifestations, and a mean false alarm rate of 1.4/24 h [
      • Halford J.J.
      • et al.
      Detection of generalized tonic-clonic seizures using surface electromyographic monitoring.
      ]. This system has a web-portal that allows to view detection times, and adjustment of detection settings [
      • Halford J.J.
      • et al.
      Detection of generalized tonic-clonic seizures using surface electromyographic monitoring.
      ]. In another similar blinded study including 71 patients, this device had a sensitivity of 93.8% (30 out of 32 GTCS), a median detection latency from the tonic phase onset of 9 s, and a false alarm rate of 0.67/24 h [
      • Beniczky S.
      • et al.
      Automated real-time detection of tonic-clonic seizures using a wearable EMG device.
      ].
      Disadvantages of sEMG sensors may include discomfort and irritation when attached to the skin. As with other sensors, there may also be potential for detachment [
      • Conradsen I.
      • et al.
      Evaluation of novel algorithm embedded in a wearable sEMG device for seizure detection.
      ,
      • Szabo C.A.
      • et al.
      Electromyography-based seizure detector: preliminary results comparing a generalized tonic-clonic seizure detection algorithm to video-EEG recordings.
      ,
      • Halford J.J.
      • et al.
      Detection of generalized tonic-clonic seizures using surface electromyographic monitoring.
      ]. Better results may be achieved when certain parameters are tailored to the individual patient, in particular for tonic seizures [
      • Larsen S.N.
      • et al.
      Detection of tonic epileptic seizures based on surface electromyography.
      ]. Therefore, sEMG reliably detects GTCS and tonic seizures, and could potentially detect other seizure types with a motor component [
      • Beniczky S.
      • et al.
      Automated real-time detection of tonic-clonic seizures using a wearable EMG device.
      ].

      2.1.7 Video detection systems

      Automatic video detection systems use video-based fiducials as well as area, duration, velocity, rotation, oscillation, angular speed, and/or motion trajectory to detect seizures [
      • Pediaditis M.
      • Tsiknakis M.
      • Leitgeb N.
      Vision-based motion detection, analysis and recognition of epileptic seizures--a systematic review.
      ,
      • Lu H.
      • et al.
      Quantifying limb movements in epileptic seizures through color-based video analysis.
      ,
      • Cuppens K.
      • et al.
      Using Spatio-Temporal Interest Points (STIP) for myoclonic jerk detection in nocturnal video.
      ,
      • Kalitzin S.
      • et al.
      Automatic segmentation of episodes containing epileptic clonic seizures in video sequences.
      ,
      • Mandal B.
      • et al.
      Non-intrusive head movement analysis of videotaped seizures of epileptic origin.
      ]. The systems are classified as marker-based or marker-free, depending on whether the cameras track detectable markers placed in determined sites [
      • Pediaditis M.
      • Tsiknakis M.
      • Leitgeb N.
      Vision-based motion detection, analysis and recognition of epileptic seizures--a systematic review.
      ]. Seizure types that can be captured by video include focal, hypermotor, myoclonic, and clonic [
      • Pediaditis M.
      • Tsiknakis M.
      • Leitgeb N.
      Vision-based motion detection, analysis and recognition of epileptic seizures--a systematic review.
      ,
      • Lu H.
      • et al.
      Quantifying limb movements in epileptic seizures through color-based video analysis.
      ,
      • Cuppens K.
      • et al.
      Using Spatio-Temporal Interest Points (STIP) for myoclonic jerk detection in nocturnal video.
      ,
      • Rémi J.
      • et al.
      Quantitative movement analysis differentiates focal seizures characterized by automatisms.
      ], and tentatively in the future other seizure types with more complex motor or automotor features, prominent behavioral arrest, and depending on camera features, seizures with skin color or skin temperature changes, as well as respiratory and heart rate changes, among others. Current sensitivity varies from 75 to 100%, with a PPV over 85%, and a specificity of 53–93% [
      • Pediaditis M.
      • Tsiknakis M.
      • Leitgeb N.
      Vision-based motion detection, analysis and recognition of epileptic seizures--a systematic review.
      ,
      • Lu H.
      • et al.
      Quantifying limb movements in epileptic seizures through color-based video analysis.
      ,
      • Cuppens K.
      • et al.
      Using Spatio-Temporal Interest Points (STIP) for myoclonic jerk detection in nocturnal video.
      ,
      • Kalitzin S.
      • et al.
      Automatic segmentation of episodes containing epileptic clonic seizures in video sequences.
      ]. A video monitoring system in a residential care unit facilitated nocturnal surveillance; in fact, 33% of all observed seizures were only seen on video, including mainly tonic seizures [
      • van der Lende M.
      • et al.
      Value of video monitoring for nocturnal seizure detection in a residential setting.
      ]. However, video monitoring was costly, and without further data processing was not recommended for broad implementation [
      • van der Lende M.
      • et al.
      Value of video monitoring for nocturnal seizure detection in a residential setting.
      ].
      The main disadvantage in marker-based video detection devices is that sensors can be uncomfortable or dislocate with prolonged use [
      • Lu H.
      • et al.
      Quantifying limb movements in epileptic seizures through color-based video analysis.
      ,
      • Mandal B.
      • et al.
      Non-intrusive head movement analysis of videotaped seizures of epileptic origin.
      ]. Marker-free systems to date are limited to detection of seizures with a motor component, and they are also more limited to the area covered by video: the patient must be visible and natural fiducials may need to be on camera [
      • Pediaditis M.
      • Tsiknakis M.
      • Leitgeb N.
      Vision-based motion detection, analysis and recognition of epileptic seizures--a systematic review.
      ]. Seizure detection based on video is feasible, but currently, this may recognize seizures with major movements best.

      2.1.8 Other potential future modalities

      A state in which seizure occurrence is highly likely must be identified to forecast seizures. Several biophysical or metabolic parameters could be measured to help determine this, and in combination with the modalities mentioned, may achieve optimal seizure detection, and even prediction. Some parameters seem to change during or prior seizures, such as ionic changes in vivo -for example extracellular potassium levels-, pH, oxygen, very fast oscillations, and intracellular NADH/FAD+ [
      • Jirsa V.K.
      • et al.
      On the nature of seizure dynamics.
      ]. In addition, cortisol, mood, orexin, temperature, respiration, sex hormones, glucose, inflammatory markers, time of day, blood oxygen, ketones, among others, have also been proposed as parameters that could enhance a seizure forecasting algorithm [
      • Dumanis S.B.
      • et al.
      Seizure forecasting from idea to reality. Outcomes of the My Seizure Gauge Epilepsy Innovation Institute workshop.
      ]. Devices to monitor these potential biomarkers are currently being developed and refined [
      • Curto V.F.
      Wearable chemo/bio-sensors for sweat sensing in sports applications: combining micro-fluidics and novel materials. PhD thesis.
      ,
      • Cunningham D.D.
      Human body to device biofluid transfer.
      ].

      2.2 Combination of detection modalities

      Multimodal seizure detection devices may have advantages over uni-modal systems, since combination of modalities may at times achieve higher sensitivities with lower FDR [
      • Ulate-Campos A.
      • et al.
      Automated seizure detection systems and their effectiveness for each type of seizure.
      ]. Some available multimodal signal combinations have been tested in selected studies, and may be useful in SE detection and prevention.

      2.2.1 Accelerometry and electrodermal activity

      The combination of EDA and ACM offers improvement in the detection of motor seizures and selected seizures with autonomic involvement, with a potential role in SUDEP risk evaluation and prevention. Wrist-worn devices were used to record EDA and ACM from 69 patients at 6 clinical sites, obtaining 5928 h of data, including 55 convulsive seizures from 22 patients [
      • Onorati F.
      • et al.
      Multicenter clinical assessment of improved wearable multimodal convulsive seizure detectors.
      ]. Three automatic classifiers were then used to detect seizures, and all detected seizures prior to video EEG onset, with comparable latencies [
      • Onorati F.
      • et al.
      Multicenter clinical assessment of improved wearable multimodal convulsive seizure detectors.
      ]. One classifier had a sensitivity of 94.5%, and a false alarm rate of 0.2 events per day [
      • Onorati F.
      • et al.
      Multicenter clinical assessment of improved wearable multimodal convulsive seizure detectors.
      ]. The seizures detected were focal tonic-clonic and focal evolving into bilateral tonic-clonic seizures, and all nocturnal seizures were detected by 2 classifiers [
      • Onorati F.
      • et al.
      Multicenter clinical assessment of improved wearable multimodal convulsive seizure detectors.
      ].

      2.2.2 Accelerometry and surface electromyography

      This combination was proposed since ACM seems to be more sensitive in detecting the clonic phase, and sEMG appears to have greater sensitivity to specifically detect phases of GTCS, such as the tonic phase [
      • Milosevic M.
      • et al.
      Automated detection of tonic-clonic seizures using 3-D accelerometry and surface electromyography in pediatric patients.
      ]. One study in children with tonic-clonic seizures achieved a sensitivity of 91%, with a FDR of 0.5/12 h, utilizing a combination of two ACM and two sEMG sensors [
      • Milosevic M.
      • et al.
      Automated detection of tonic-clonic seizures using 3-D accelerometry and surface electromyography in pediatric patients.
      ].

      2.2.3 Video, accelerometry and radar induced activity (VARIA system)

      This system uses video, accelerometry and radar induced activity to detect seizures based on joint movements [
      • Van de Vel A.
      • et al.
      Long-term accelerometry-triggered video monitoring and detection of tonic-clonic and clonic seizures in a home environment: pilot study.
      ]. A camera and a motion sensor (radar) are attached to a tripod placed near the patient’s bed. Four sensors with 3-axis ACM and streaming wireless communication are placed on elastic bracelets, worn around wrists and ankles. A laptop stores all movement data recorded by camera, radar or ACM sensors [
      • Van de Vel A.
      • et al.
      Long-term accelerometry-triggered video monitoring and detection of tonic-clonic and clonic seizures in a home environment: pilot study.
      ]. VARIA was tested in two patients with tonic-clonic seizures in a home environment, with a sensitivity of 66.9%, and a FDR of 1.16 per night [
      • Van de Vel A.
      • et al.
      Long-term accelerometry-triggered video monitoring and detection of tonic-clonic and clonic seizures in a home environment: pilot study.
      ]. Challenges of this system may include numerous, easily removable bracelets [
      • Van de Vel A.
      • Verhaert K.
      • Ceulemans B.
      Critical evaluation of four different seizure detection systems tested on one patient with focal and generalized tonic and clonic seizures.
      ], requiring potentially further refinement prior to routine and everyday monitoring application.
      Facilitated by advances in device development, further work on modality combinations is in progress [
      • Ulate-Campos A.
      • et al.
      Automated seizure detection systems and their effectiveness for each type of seizure.
      ]. Table 1 summarizes the sensitivity and FDR of the detection modalities outlined above (Table 1).
      Table 1Sensitivity and false detection rate (FDR) of detection modalities.
      ModalitySensitivityFDR
      UnimodalACM16–100%0.2–2.0/day, 0.6/night
      EKG57–100%
      EDA86–100%
      Noninvasive 2/10-electrode EEG75–100%
      Mattress sensors16.7–85%
      sEMG53–100%1.49–4.03/h, 1/1399 h, 0.67–1.4/24 h
      Video detection systems75–100%
      MultimodalACM and EDA94.5%0.2/day
      ACM and sEMG91%0.5/12h
      Video, ACM and radar (VARIA system)66.9%1.16/night
      Legend. ACM: accelerometry. EKG: electrocardiogram. EDA: electrodermal activity. EEG: electroencephalogram. sEMG: surface electromyography. FDR: false detection rate.

      2.3 Training of modalities

      Individualized training of detection modalities and selection of specific sensor combinations for each patient may improve individual seizure detection rates. Regarding device selection, patient’s characteristics and clinical seizure type evaluation may help select the most appropriate monitoring modality. Optimal detection device for selected seizure types may differ, including a combination of sensor modalities, as multimodal devices may yield better results in selected patients. For example, sEMG and EDA may serve as a viable combination for tonic seizures, EDA and EKG for focal seizures with impaired awareness and autonomic changes, and ACM and/or sEMG for myoclonic and clonic seizures.
      Individualization and personalization of seizure forecasting algorithms may be necessary, including multiple additional variables. Seizure generating mechanisms and preictal dynamics might be different among patients and among different types of seizures in the same patient. There is a lot of variability in time patterns in epilepsy patients, but the same patient longitudinally often tends to present with more consistent patterns [
      • Freestone D.R.
      • Karoly P.J.
      • Cook M.J.
      A forward-looking review of seizure prediction.
      ]. An initial training phase may be helpful to tailor the algorithm to the patient’s characteristics, ictal data, and interictal data in order to develop a patient and seizure-type specific forecasting algorithm [
      • Freestone D.R.
      • Karoly P.J.
      • Cook M.J.
      A forward-looking review of seizure prediction.
      ]. Taking all of these features into consideration might improve seizure detection and make limited seizure prediction possible.

      2.4 Portable devices approved by the U.S. FDA

      2.4.1 SPEAC system (Brain Sentinel)

      The Brain Sentinel Monitoring and Alerting System, also known as SPEAC system, is a portable seizure detection and alerting system. SPEAC system is composed of a wireless sEMG monitor attached to a non-invasive sEMG electrode patch placed on the biceps (unimodal device). Since this device measures muscle activity, the system focuses on detection of prominent motor seizures. The system also provides audio recordings and warning alarms to caregivers when events are detected, and includes a seizure diary. Minor motor events not detected in real-time may be identified since the sEMG is recorded. This device is FDA approved to detect GTCS.

      2.4.2 Embrace (Empatica)

      Embrace is a wrist-worn seizure-alerting smartwatch which includes a 3-axis accelerometer, an EDA sensor, a gyroscope, and a peripheral temperature sensor. Embrace transmits data to a paired smartphone via Bluetooth, and the smartphone resends these data to Empatica servers. This system alerts caregivers when events are detected (through an Alert App), it provides a seizure diary, and tracks daily rest - physical activity (through Mate App). Convulsive seizures with rhythmic motor movements involving one or both upper and/or lower extremities may be detectable by this device. This device is FDA approved to detect GTCS.

      2.4.3 Ceribell EEG system (Ceribell)

      Ceribell is a noninvasive EEG system composed of a 10-electrode headband and a ‘Brain Stethoscope function’ which converts EEG to sound in real-time. Ceribell allows rapid access to EEG −5 min to set up-, as well as rapid and easy interpretation by listening, without requiring any epileptologist or EEG technician [
      • Parvizi J.
      • et al.
      Detecting silent seizures by their sound.
      ]. Its quality is equivalent to conventional EEG and may present even higher sensitivity [
      • Parvizi J.
      • et al.
      Detecting silent seizures by their sound.
      ]. Ceribell is able to detect any type of seizure but, since the system selects individual channels, focal seizures in other channels are not detected, and therefore it is mainly intended to detect generalized or hemispheric patterns. Its use in ED and ICU may allow earlier and easier seizure detection and to select patients who should undergo a formal EEG study for better seizure classification [
      • Parvizi J.
      • et al.
      Detecting silent seizures by their sound.
      ]. Ceribell is FDA approved as an EEG device.

      3. Seizure prediction

      3.1 Patterns of epileptic seizure and SE occurrence: big data and machine learning

      Seizure occurrence appears random because seizure patterns may often be too complex to be described by any simple intuitive model, but seizures may follow complex non-random patterns [
      • Amengual-Gual M.
      • Sánchez Fernández I.
      • Loddenkemper T.
      Patterns of epileptic seizure occurrence.
      ]. There are examples of patterns of seizure occurrence [
      • Amengual-Gual M.
      • Sánchez Fernández I.
      • Loddenkemper T.
      Patterns of epileptic seizure occurrence.
      ] based on seizure onset [
      • Kaleyias J.
      • et al.
      Sleep-wake patterns of seizures in children with lesional epilepsy.
      ,
      • Loddenkemper T.
      • et al.
      Circadian patterns of pediatric seizures.
      ,
      • Ramgopal S.
      • et al.
      Predicting diurnal and sleep/wake seizure patterns in paediatric patients of different ages.
      ], seizure semiology [
      • Loddenkemper T.
      • et al.
      Circadian patterns of pediatric seizures.
      ,
      • Zarowski M.
      • et al.
      Circadian distribution and sleep/wake patterns of generalized seizures in children.
      ,
      • Ramgopal S.
      • et al.
      Diurnal and sleep/wake patterns of epileptic spasms in different age groups.
      ], seizure evolution [
      • Sánchez Fernández I.
      • et al.
      Clinical evolution of seizures: distribution across time of day and sleep/wakefulness cycle.
      ,
      • Ramgopal S.
      • et al.
      Circadian patterns of generalized tonic-clonic evolutions in pediatric epilepsy patients.
      ], triggers such as hormonal factors [
      • Herzog A.G.
      Catamenial epilepsy: update on prevalence, pathophysiology and treatment from the findings of the NIH progesterone treatment trial.
      ] and weather variation, among others [
      • Rakers F.
      • et al.
      Weather as a risk factor for epileptic seizures: a case-crossover study.
      ,
      • Haut S.R.
      • et al.
      Seizure occurrence: precipitants and prediction.
      ,
      • Rüegg S.
      • et al.
      Association of environmental factors with the onset of status epilepticus.
      ,
      • Gunn B.G.
      • Baram T.Z.
      Stress and seizures: space, time and hippocampal circuits.
      ]. The pattern of seizure occurrence is most often described over a 24 h period, and may be influenced by time of day and sleep-wake stage [
      • Loddenkemper T.
      • et al.
      Circadian patterns of pediatric seizures.
      ,
      • Quigg M.
      Circadian rhythms: interactions with seizures and epilepsy.
      ]. However, the pattern of seizure occurrence seems to be more irregular and complex than a simple circadian pattern, as well as highly individual-specific [
      • Cook M.J.
      • et al.
      Human focal seizures are characterized by populations of fixed duration and interval.
      ,
      • Karoly P.J.
      • et al.
      Interictal spikes and epileptic seizures: their relationship and underlying rhythmicity.
      ]. Some studies suggest that patient-specific ultradian and infradian rhythms may also contribute to the distribution of seizure occurrence [
      • Karoly P.J.
      • et al.
      Interictal spikes and epileptic seizures: their relationship and underlying rhythmicity.
      ]. Regarding the pattern of SE occurrence, it may simply reflect the pattern of seizure occurrence or may have its individual and more complex distribution.
      Studying complex patterns with occasional events, such as SE, requires numerous data points. This allows for the introduction of big data and machine learning in seizure detection and prediction. The increasing use of large databases –from large multicenter networks [
      • Cock H.R.
      • ESETT Group
      Established Status Epilepticus Treatment Trial (ESETT).
      ,
      • Sánchez Fernández I.
      • et al.
      Gaps and opportunities in refractory status epilepticus research in children: a multi-center approach by the Pediatric Status Epilepticus Research Group (pSERG).
      ] and large patient self-reported databases [
      • Goldenholz D.M.
      • et al.
      Confusing placebo effect with natural history in epilepsy: a big data approach.
      ] – and machine learning techniques may help to better define patterns of SE and seizure occurrence in individual patients, and therefore, may permit prediction [
      • Sánchez Fernández I.
      • et al.
      Prediction of time of occurrence and length of seizures based on basic demographic and clinical data using machine learning algorithms.
      ]. Simple learning algorithms may serve as basic building blocks that lead to more advanced approaches, like deep learning and neural network approaches. The integration of learning algorithms in wearable detection device algorithms may lead to closed-loop systems for seizure detection and prediction [
      • Ulate-Campos A.
      • et al.
      Automated seizure detection systems and their effectiveness for each type of seizure.
      ]. However, if the input data is not of sufficient quality, the result may also be less than perfect (‘garbage in, garbage out’) [
      • Beam A.L.
      • Kohane I.S.
      Big data and machine learning in health care.
      ], and machine learning may not be able to easily resolve or make up for poor data quality.
      Seizure susceptibility assessment, based on simultaneously collected diary and longitudinally evaluated seizure susceptibility intervals, may also provide opportunities for chrono-pharmacological approaches [
      • Ramgopal S.
      • et al.
      Seizure detection, seizure prediction, and closed-loop warning systems in epilepsy.
      ].

      3.2 EEG analysis

      Cerebral electrical neuronal activity may exemplify a chaotic system, allowing for epileptic seizure description and seizure occurrence prediction through nonlinear differential equations, similarly to other dynamic systems [
      • Amengual-Gual M.
      • Sánchez Fernández I.
      • Loddenkemper T.
      Patterns of epileptic seizure occurrence.
      ,
      • Litt B.
      • Echauz J.
      Prediction of epileptic seizures.
      ,
      • Mormann F.
      • et al.
      Seizure prediction: the long and winding road.
      ,
      • Gadhoumi K.
      • et al.
      Seizure prediction for therapeutic devices: a review.
      ,
      • Iasemidis L.D.
      • Sackellares J.C.
      Chaos theory and epilepsy.
      ]. For this reason, long-term EEG recordings may benefit from nonlinear analysis, and ultimately may provide practical tools in epilepsy care and management [
      • Iasemidis L.D.
      • Sackellares J.C.
      Chaos theory and epilepsy.
      ,
      • Lehnertz K.
      Non-linear time series analysis of intracranial EEG recordings in patients with epilepsy--an overview.
      ,
      • Elger C.E.
      • et al.
      Nonlinear EEG analysis and its potential role in epileptology.
      ,
      • Venkataraman V.
      • et al.
      Brain dynamics based automated epileptic seizure detection.
      ]. In fact, following nonlinear EEG analysis based on chaos theory, several chaotic levels depending on the epileptic state have been discovered: the ictal state corresponds to a certain degree of order, the postictal state corresponds to chaos, and the pre-ictal state corresponds to an intermediate level between chaos and order [
      • Iasemidis L.D.
      • Sackellares J.C.
      Chaos theory and epilepsy.
      ]. In addition, different chaotic levels are registered from epileptogenic and non-epileptogenic areas during interictal states [
      • Iasemidis L.D.
      • Sackellares J.C.
      Chaos theory and epilepsy.
      ,
      • Drury I.
      • et al.
      Seizure prediction using scalp electroencephalogram.
      ].
      Furthermore, the relationship between some EEG features and seizure onset also supports the concept of non-randomness in brain neuronal activity. Examples of EEG features prior to seizure onset include long-term energy bursts [
      • Litt B.
      • et al.
      Epileptic seizures may begin hours in advance of clinical onset: a report of five patients.
      ], missing ordinal patterns in deterministic dynamics [
      • Schindler K.
      • et al.
      Forbidden ordinal patterns of periictal intracranial EEG indicate deterministic dynamics in human epileptic seizures.
      ], reduction of sleep spindles [
      • Tezer F.I.
      • et al.
      A reduction of sleep spindles heralds seizures in focal epilepsy.
      ], and subject-specific changes in spike rate [
      • Karoly P.J.
      • et al.
      Interictal spikes and epileptic seizures: their relationship and underlying rhythmicity.
      ]. Additionally, interictal epileptiform activity may also follow rhythmic patterns [
      • Anderson C.T.
      • et al.
      Day-night patterns of epileptiform activity in 65 patients with long-term ambulatory electrocorticography.
      ]. Moreover, focal seizures may be characterized by seizure groups of fixed duration and interval also supporting non-randomness [
      • Cook M.J.
      • et al.
      Human focal seizures are characterized by populations of fixed duration and interval.
      ]. Regarding end of seizures, self-terminating seizures seem to end through a common dynamical mechanism via a critical electrophysiological transition, in contrast to SE that does not cross the critical transition despite repeated approaches [
      • Kramer M.A.
      • et al.
      Human seizures self-terminate across spatial scales via a critical transition.
      ]. Therefore, both onset and end of seizures may be potentially predicted.

      4. Status epilepticus prevention

      4.1 Closed-loop detection-treatment systems

      Combinations of seizure detection sensors based on EEG and extra-cerebral signals may be integrated into portable devices to facilitate seizure detection in ambulatory settings. The device may be able to access patients’ healthcare data, and other pertinent data points. When a seizure is detected, the caregiver may be informed and this may lead to an intervention or corrective response, which could imply abortive pharmacotherapy, neurostimulation or even micro-pump systems that deliver medication, as well as acute seizure care and transport to an emergency room if needed [
      • Ulate-Campos A.
      • et al.
      Automated seizure detection systems and their effectiveness for each type of seizure.
      ]. In the future, development of closed-loop prediction, detection, and treatment systems in clinical routine may help identify and treat seizures earlier, and therefore, prevent SE reducing related morbidity and mortality [
      • Van de Vel A.
      • et al.
      Non-EEG seizure-detection systems and potential SUDEP prevention: state of the art.
      ,
      • Gaínza-Lein M.
      • et al.
      Association of time to treatment with short-term outcomes for pediatric patients with refractory convulsive status epilepticus.
      ,
      • Sánchez Fernández I.
      • et al.
      Factors associated with treatment delays in pediatric refractory convulsive status epilepticus.
      ,
      • Sánchez Fernández I.
      • et al.
      Time from convulsive status epilepticus onset to anticonvulsant administration in children.
      ].
      Some implantable intracranial devices based on iEEG provided opportunities for closed-loop monitoring in patients with refractory focal epilepsy [
      • Morrell M.J.
      • RNS System in Epilepsy Study Group
      Responsive cortical stimulation for the treatment of medically intractable partial epilepsy.
      ,
      • Cook M.J.
      • et al.
      Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study.
      ]. However, the risk of neurosurgery and of wearing an implanted device currently limits widespread application. We outline selected devices subsequently:

      4.1.1 NeuroVista

      This implanted advisory system, comprised of two silicon implantable leads each with eight platinum iridium contacts, was placed over the epileptogenic area in patients with refractory focal epilepsy [
      • Cook M.J.
      • et al.
      Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study.
      ]. A patient-specific algorithm was developed during data collection and the recorded EEG was analyzed by a portable device [
      • Cook M.J.
      • et al.
      Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study.
      ]. After four months in the advisory phase, the median sensitivity of seizure prediction in the high likelihood performance was 60% (10 adults) -based on correlated clinical seizures- with a median time in high advisory of 27.5%. The clinical usefulness of seizure prediction was inconclusive, mainly because of the variability in warning times and difficulties adapting to the system [
      • Cook M.J.
      • et al.
      Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study.
      ]. Nevertheless, the system was useful to detect and predict seizures in selected patients, and provided new detailed information on epileptogenic focus and previously undetected seizures [
      • Cook M.J.
      • et al.
      Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study.
      ]. Within one year after implantation, four patients among 15 had serious adverse effects, including device migration and infection [
      • Cook M.J.
      • et al.
      Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study.
      ]. While feasibility of such a system was demonstrated, reportedly funding to develop and market the system was not available, and therefore, this system is currently not available.

      4.1.2 NeuroPace

      This implanted device consists of a programmable neurostimulator that detects abnormal electrocorticographic activity and provides responsive cortical stimulation [
      • Morrell M.J.
      • RNS System in Epilepsy Study Group
      Responsive cortical stimulation for the treatment of medically intractable partial epilepsy.
      ]. This system is connected to 1 or 2 recording and stimulating depth or subdural strip leads placed on the epileptogenic area, and its detection and stimulation parameters can be tailored individually [
      • Morrell M.J.
      • RNS System in Epilepsy Study Group
      Responsive cortical stimulation for the treatment of medically intractable partial epilepsy.
      ]. In a multicenter, double-blind, randomized controlled trial (n = 191 adults), patients with refractory focal epilepsy randomized to receive stimulation in response to abnormal electrocorticographic activity presented greater reduction in seizure frequency than patients randomized to no stimulation. Furthermore, there were no differences in adverse events, and quality of life improved [
      • Morrell M.J.
      • RNS System in Epilepsy Study Group
      Responsive cortical stimulation for the treatment of medically intractable partial epilepsy.
      ].
      Efforts to improve epileptic seizure detection and prediction may positively affect SE prevention, since acute treatment may be administered earlier and chronic medication may be tailored to the individual patient (for example, using chrono-pharmacological approaches, such as differential antiepileptic dosing). As portable devices use wireless technology and batteries, signal dropout may need to be anticipated, and may represent a current shortcoming without implementation of adequate backup systems.

      5. Conclusions

      Seizures tend to follow complex and patient-specific distributions despite their apparently random occurrence. The combination of multimodal monitoring devices, big data sets, machine learning and other potential novel approaches –such as biomarkers at seizure prone states- may enhance patient-specific seizure forecasting algorithms and allow for an improved detection and sooner intervention. The implementation of closed-loop systems in ambulatory settings may help prevent SE in the future, and therefore, this may potentially reduce morbidity and mortality in epilepsy and ultimately improve quality of life in patients and caregivers.

      Contributors

      Marta Amengual-Gual, Adriana Ulate-Campos, and Tobias Loddenkemper participated in literature review, article design and outline, manuscript development, manuscript drafting, manuscript writing, and writing supervision.

      Funding

      This study was supported by the Epilepsy Research Fund. Marta Amengual-Gual is funded by a grant for the study of status epilepticus from “Fundación Alfonso Martín Escudero”.

      Ethics

      This study complied with biomedical research ethical standards.

      Declaration of interest

      Marta Amengual-Gual reports no conflict of interest.
      Adriana Ulate-Campos reports no disclosures.
      Tobias Loddenkemper serves on the Laboratory Accreditation Board for Long Term (Epilepsy and Intensive Care Unit) Monitoring, on the Council (and as President) of the American Clinical Neurophysiology Society, on the American Board of Clinical Neurophysiology, as an Associate Editor for Seizure, and as an Associate Editor for Wyllie’s Treatment of Epilepsy 6th and 7th edition. He is part of pending patent applications to detect and predict seizures and to diagnose epilepsy. He receives research support from the Epilepsy Research Fund, the American Epilepsy Society, the Epilepsy Foundation of America, the Epilepsy Therapy Project, PCORI, the Pediatric Epilepsy Research Foundation, CURE, HHV-6 Foundation, and received research grants from Lundbeck, Eisai, Upsher-Smith, Mallinckrodt and Pfizer. He serves as a consultant for Zogenix, Upsher Smith, Sunovion, Engage, and Advance Medical. He performs video electroencephalogram long-term and ICU monitoring, electroencephalograms, and other electrophysiological studies at Boston Children's Hospital and affiliated hospitals and bills for these procedures and he evaluates pediatric neurology patients and bills for clinical care. He has received speaker honorariums from national societies including the AAN, AES and ACNS, and for grand rounds at various academic centers. His wife, Dr. Karen Stannard, is a pediatric neurologist and she performs video electroencephalogram long-term and ICU monitoring, electroencephalograms, and other electrophysiological studies and bills for these procedures and she evaluates pediatric neurology patients and bills for clinical care.
      The authors report no potential conflicts of interest.

      References

        • Jacoby A.
        Epilepsy and the quality of everyday life. Findings from a study of people with well-controlled epilepsy.
        Soc Sci Med. 1992; 34: 657-666
        • Beniczky S.
        • et al.
        Detection of generalized tonic-clonic seizures by a wireless wrist accelerometer: a prospective, multicenter study.
        Epilepsia. 2013; 54: e58-e61
        • Lockman J.
        • Fisher R.S.
        • Olson D.M.
        Detection of seizure-like movements using a wrist accelerometer.
        Epilepsy Behav. 2011; 20: 638-641
        • Nijsen T.M.
        • et al.
        Time-frequency analysis of accelerometry data for detection of myoclonic seizures.
        IEEE Trans Inf Technol Biomed. 2010; 14: 1197-1203
        • Nijsen T.M.
        • et al.
        The potential value of three-dimensional accelerometry for detection of motor seizures in severe epilepsy.
        Epilepsy Behav. 2005; 7: 74-84
        • Dalton A.
        • et al.
        Development of a body sensor network to detect motor patterns of epileptic seizures.
        IEEE Trans Biomed Eng. 2012; 59: 3204-3211
        • Van de Vel A.
        • Verhaert K.
        • Ceulemans B.
        Critical evaluation of four different seizure detection systems tested on one patient with focal and generalized tonic and clonic seizures.
        Epilepsy Behav. 2014; 37: 91-94
        • Kramer U.
        • et al.
        A novel portable seizure detection alarm system: preliminary results.
        J Clin Neurophysiol. 2011; 28: 36-38
        • Schulc E.
        • et al.
        Measurement and quantification of generalized tonic-clonic seizures in epilepsy patients by means of accelerometry--an explorative study.
        Epilepsy Res. 2011; 95: 173-183
        • Patterson A.L.
        • 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
        • Van de Vel A.
        • et al.
        Long-term home monitoring of hypermotor seizures by patient-worn accelerometers.
        Epilepsy Behav. 2013; 26: 118-125
        • Jallon P.
        • et al.
        Detection system of motor epileptic seizures through motion analysis with 3D accelerometers.
        Conf Proc IEEE Eng Med Biol Soc, 2009. 2009: 2466-2469
        • Velez M.
        • et al.
        Tracking generalized tonic-clonic seizures with a wrist accelerometer linked to an online database.
        Seizure. 2016; 39: 13-18
        • Kusmakar S.
        • et al.
        Detection of generalized tonic-clonic seizures using short length accelerometry signal.
        Conf Proc IEEE Eng Med Biol Soc. 2017; 2017: 4566-4569
        • Joo H.S.
        • et al.
        Spectral analysis of acceleration data for detection of generalized tonic-clonic seizures.
        Sensors (Basel). 2017; 17
        • Behbahani S.
        • et al.
        Pre-ictal heart rate variability assessment of epileptic seizures by means of linear and non-linear analyses.
        Anadolu Kardiyol Derg. 2013; 13: 797-803
        • Osorio I.
        • Manly B.F.
        Probability of detection of clinical seizures using heart rate changes.
        Seizure. 2015; 30: 120-123
        • Leutmezer F.
        • et al.
        Electrocardiographic changes at the onset of epileptic seizures.
        Epilepsia. 2003; 44: 348-354
        • van Elmpt W.J.
        • et al.
        A model of heart rate changes to detect seizures in severe epilepsy.
        Seizure. 2006; 15: 366-375
        • Kerem D.H.
        • Geva A.B.
        Forecasting epilepsy from the heart rate signal.
        Med Biol Eng Comput. 2005; 43: 230-239
        • Jeppesen J.
        • et al.
        Detection of epileptic seizures with a modified heart rate variability algorithm based on Lorenz plot.
        Seizure. 2015; 24: 1-7
        • De Cooman T.
        • et al.
        Online detection of tonic-clonic seizures in pediatric patients using ECG and low-complexity incremental novelty detection.
        Conf Proc IEEE Eng Med Biol Soc. 2015; 2015: 5597-5600
        • Jansen K.
        • et al.
        Peri-ictal ECG changes in childhood epilepsy: implications for detection systems.
        Epilepsy Behav. 2013; 29: 72-76
        • Varon C.
        • et al.
        Can ECG monitoring identify seizures?.
        J Electrocardiol. 2015; 48: 1069-1074
        • Vandecasteele K.
        • et al.
        Automated epileptic seizure detection based on wearable ECG and PPG in a hospital environment.
        Sensors (Basel). 2017; 17
        • Romero I.
        • Grundlehner B.
        • Penders J.
        Robust beat detector for ambulatory cardiac monitoring.
        Conf Proc IEEE Eng Med Biol Soc. 2009; 2009: 950-953
        • Boon P.
        • et al.
        A prospective, multicenter study of cardiac-based seizure detection to activate vagus nerve stimulation.
        Seizure. 2015; 32: 52-61
        • Ramgopal S.
        • et al.
        Seizure detection, seizure prediction, and closed-loop warning systems in epilepsy.
        Epilepsy Behav. 2014; 37: 291-307
        • Taylor N.A.
        • Machado-Moreira C.A.
        Regional variations in transepidermal water loss, eccrine sweat gland density, sweat secretion rates and electrolyte composition in resting and exercising humans.
        Extrem Physiol Med. 2013; 2: 4
        • Poh M.Z.
        • Swenson N.C.
        • Picard R.W.
        A wearable sensor for unobtrusive, long-term assessment of electrodermal activity.
        IEEE Trans Biomed Eng. 2010; 57: 1243-1252
        • Poh M.Z.
        • et al.
        Continuous monitoring of electrodermal activity during epileptic seizures using a wearable sensor.
        Conf Proc IEEE Eng Med Biol Soc. 2010; 2010: 4415-4418
        • Poh M.Z.
        • et al.
        Autonomic changes with seizures correlate with postictal EEG suppression.
        Neurology. 2012; 78: 1868-1876
        • Sarkis R.A.
        • et al.
        Autonomic changes following generalized tonic clonic seizures: an analysis of adult and pediatric patients with epilepsy.
        Epilepsy Res. 2015; 115: 113-138
        • Jain S.
        • et al.
        A compressed sensing based decomposition of electrodermal activity signals.
        IEEE Trans Biomed Eng. 2017; 64: 2142-2151
        • Micoulaud-Franchi J.A.
        • et al.
        Skin conductance biofeedback training in adults with drug-resistant temporal lobe epilepsy and stress-triggered seizures: a proof-of-concept study.
        Epilepsy Behav. 2014; 41: 244-250
        • Kotwas I.
        • et al.
        A case-control study of skin conductance biofeedback on seizure frequency and emotion regulation in drug-resistant temporal lobe epilepsy.
        Int J Psychophysiol. 2018; 123: 103-110
        • Petersen E.B.
        • et al.
        Generic single-channel detection of absence seizures.
        Conf Proc IEEE Eng Med Biol Soc. 2011; 2011: 4820-4823
        • Myers M.H.
        • et al.
        Ambulatory seizure monitoring: from concept to prototype device.
        Ann Neurosci. 2016; 23: 100-111
        • Lehmkuhle M.
        • et al.
        Development of a discrete, wearable, EEG device for counting seizures (Abstract nr 2.158) 2015.
        in: 69th Annual Meeting of the American Epilepsy Society, 2015 Dec 4–8; Philadelphia (PA), AES2015
        • Munaretto J.
        • et al.
        Technical analysis of the Ceribell EEG device.
        American Epilepsy Society Meeting 2017. 2017
        • Parvizi J.
        • et al.
        Detecting silent seizures by their sound.
        Epilepsia. 2018; 59: 877-884
        • Ray A.
        • et al.
        Localizing value of scalp EEG spikes: a simultaneous scalp and intracranial study.
        Clin Neurophysiol. 2007; 118: 69-79
        • Morrell M.J.
        • RNS System in Epilepsy Study Group
        Responsive cortical stimulation for the treatment of medically intractable partial epilepsy.
        Neurology. 2011; 77: 1295-1304
        • Cook M.J.
        • et al.
        Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study.
        Lancet Neurol. 2013; 12: 563-571
        • Baldassano S.N.
        • et al.
        Crowdsourcing seizure detection: algorithm development and validation on human implanted device recordings.
        Brain. 2017; 140: 1680-1691
        • Bandarabadi M.
        • et al.
        On the proper selection of preictal period for seizure prediction.
        Epilepsy Behav. 2015; 46: 158-166
        • Ozdemir N.
        • Yildirim E.
        Patient specific seizure prediction system using Hilbert spectrum and Bayesian networks classifiers.
        Comput Math Methods Med. 2014; 2014: 572082
        • Parvez M.Z.
        • Paul M.
        Epileptic seizure prediction by exploiting spatiotemporal relationship of EEG signals using phase correlation.
        IEEE Trans Neural Syst Rehabil Eng. 2016; 24: 158-168
        • Wang N.
        • Lyu M.R.
        Extracting and selecting distinctive EEG features for efficient epileptic seizure prediction.
        IEEE J Biomed Health Inform. 2015; 19: 1648-1659
        • Alexandre Teixeira C.
        • et al.
        Epileptic seizure predictors based on computational intelligence techniques: a comparative study with 278 patients.
        Comput Methods Programs Biomed. 2014; 114: 324-336
        • Zandi A.S.
        • et al.
        Epileptic seizure prediction using variational mixture of Gaussians.
        Conf Proc IEEE Eng Med Biol Soc. 2011; 2011: 7549-7552
        • Bandarabadi M.
        • et al.
        Epileptic seizure prediction using relative spectral power features.
        Clin Neurophysiol. 2015; 126: 237-248
        • Aarabi A.
        • He B.
        Seizure prediction in intracranial EEG: a patient-specific rule-based approach.
        Conf Proc IEEE Eng Med Biol Soc. 2011; 2011: 2566-2569
        • Gadhoumi K.
        • Gotman J.
        • Lina J.M.
        Scale invariance properties of intracerebral EEG improve seizure prediction in mesial temporal lobe epilepsy.
        PLoS One. 2015; 10: e0121182
        • Rabbi A.F.
        • Azinfar L.
        • Fazel-Rezai R.
        Seizure prediction using adaptive neuro-fuzzy inference system.
        Conf Proc IEEE Eng Med Biol Soc. 2013; 2013: 2100-2103
        • Park Y.
        • et al.
        Seizure prediction with spectral power of EEG using cost-sensitive support vector machines.
        Epilepsia. 2011; 52: 1761-1770
        • Schulze-Bonhage A.
        • et al.
        Views of patients with epilepsy on seizure prediction devices.
        Epilepsy Behav. 2010; 18: 388-396
        • Carlson C.
        • et al.
        Detecting nocturnal convulsions: efficacy of the MP5 monitor.
        Seizure. 2009; 18: 225-257
        • Fulton S.
        • et al.
        Prospective study of 2 bed alarms for detection of nocturnal seizures.
        J Child Neurol. 2013; 28: 1430-1433
        • Poppel K.V.
        • et al.
        Prospective study of the Emfit movement monitor.
        J Child Neurol. 2013; 28: 1434-1436
        • Lamberts R.J.
        • et al.
        Sudden unexpected death in epilepsy: people with nocturnal seizures may be at highest risk.
        Epilepsia. 2012; 53: 253-257
        • Langan Y.
        • Nashef L.
        • Sander J.W.
        Case-control study of SUDEP.
        Neurology. 2005; 64: 1131-1133
        • Larsen S.N.
        • et al.
        Detection of tonic epileptic seizures based on surface electromyography.
        Conf Proc IEEE Eng Med Biol Soc. 2014; 2014: 942-945
        • Conradsen I.
        • et al.
        Automated algorithm for generalized tonic-clonic epileptic seizure onset detection based on sEMG zero-crossing rate.
        IEEE Trans Biomed Eng. 2012; 59: 579-585
        • Conradsen I.
        • et al.
        Patterns of muscle activation during generalized tonic and tonic-clonic epileptic seizures.
        Epilepsia. 2011; 52: 2125-2132
        • Conradsen I.
        • et al.
        Evaluation of novel algorithm embedded in a wearable sEMG device for seizure detection.
        Conf Proc IEEE Eng Med Biol Soc. 2012; 2012: 2048-2051
        • Szabo C.A.
        • et al.
        Electromyography-based seizure detector: preliminary results comparing a generalized tonic-clonic seizure detection algorithm to video-EEG recordings.
        Epilepsia. 2015; 56: 1432-1437
        • Halford J.J.
        • et al.
        Detection of generalized tonic-clonic seizures using surface electromyographic monitoring.
        Epilepsia. 2017; 58: 1861-1869
        • Beniczky S.
        • et al.
        Automated real-time detection of tonic-clonic seizures using a wearable EMG device.
        Neurology. 2018; 90: e428-e434
        • Pediaditis M.
        • Tsiknakis M.
        • Leitgeb N.
        Vision-based motion detection, analysis and recognition of epileptic seizures--a systematic review.
        Comput Methods Programs Biomed. 2012; 108: 1133-1148
        • Lu H.
        • et al.
        Quantifying limb movements in epileptic seizures through color-based video analysis.
        IEEE Trans Biomed Eng. 2013; 60: 461-469
        • Cuppens K.
        • et al.
        Using Spatio-Temporal Interest Points (STIP) for myoclonic jerk detection in nocturnal video.
        Conf Proc IEEE Eng Med Biol Soc. 2012; 2012: 4454-4457
        • Kalitzin S.
        • et al.
        Automatic segmentation of episodes containing epileptic clonic seizures in video sequences.
        IEEE Trans Biomed Eng. 2012; 59: 3379-3385
        • Mandal B.
        • et al.
        Non-intrusive head movement analysis of videotaped seizures of epileptic origin.
        Conf Proc IEEE Eng Med Biol Soc. 2012; 2012: 6060-6063
        • Rémi J.
        • et al.
        Quantitative movement analysis differentiates focal seizures characterized by automatisms.
        Epilepsy Behav. 2011; 20: 642-647
        • van der Lende M.
        • et al.
        Value of video monitoring for nocturnal seizure detection in a residential setting.
        Epilepsia. 2016; 57: 1748-1753
        • Jirsa V.K.
        • et al.
        On the nature of seizure dynamics.
        Brain. 2014; 137: 2210-2230
        • Dumanis S.B.
        • et al.
        Seizure forecasting from idea to reality. Outcomes of the My Seizure Gauge Epilepsy Innovation Institute workshop.
        eNeuro. 2017; 4
        • Curto V.F.
        Wearable chemo/bio-sensors for sweat sensing in sports applications: combining micro-fluidics and novel materials. PhD thesis.
        Dublin City University, 2013 (http://doras.dcu.ie/19310/1/PhD_thesis_VFC.pdf)
        • Cunningham D.D.
        Human body to device biofluid transfer.
        in: Li D. Encyclopedia of microfluidics and nanofluidics. Springer, New York, NY2015
        • Ulate-Campos A.
        • et al.
        Automated seizure detection systems and their effectiveness for each type of seizure.
        Seizure. 2016; 40: 88-101
        • Onorati F.
        • et al.
        Multicenter clinical assessment of improved wearable multimodal convulsive seizure detectors.
        Epilepsia. 2017; 58: 1870-1879
        • Milosevic M.
        • et al.
        Automated detection of tonic-clonic seizures using 3-D accelerometry and surface electromyography in pediatric patients.
        IEEE J Biomed Health Inform. 2016; 20: 1333-1341
        • Van de Vel A.
        • et al.
        Long-term accelerometry-triggered video monitoring and detection of tonic-clonic and clonic seizures in a home environment: pilot study.
        Epilepsy Behav Case Rep. 2016; 5: 66-71
        • Freestone D.R.
        • Karoly P.J.
        • Cook M.J.
        A forward-looking review of seizure prediction.
        Curr Opin Neurol. 2017; 30: 167-173
        • Amengual-Gual M.
        • Sánchez Fernández I.
        • Loddenkemper T.
        Patterns of epileptic seizure occurrence.
        Brain Res. 2018; (pii: S0006-8993(18)30102-1. [Epub ahead of print]. PMID: 29481793)https://doi.org/10.1016/j.brainres.2018.02.032
        • Kaleyias J.
        • et al.
        Sleep-wake patterns of seizures in children with lesional epilepsy.
        Pediatr Neurol. 2011; 45: 109-113
        • Loddenkemper T.
        • et al.
        Circadian patterns of pediatric seizures.
        Neurology. 2011; 76: 145-153
        • Ramgopal S.
        • et al.
        Predicting diurnal and sleep/wake seizure patterns in paediatric patients of different ages.
        Epileptic Disord. 2014; 16: 56-66
        • Zarowski M.
        • et al.
        Circadian distribution and sleep/wake patterns of generalized seizures in children.
        Epilepsia. 2011; 52: 1076-1083
        • Ramgopal S.
        • et al.
        Diurnal and sleep/wake patterns of epileptic spasms in different age groups.
        Epilepsia. 2012; 53: 1170-1177
        • Sánchez Fernández I.
        • et al.
        Clinical evolution of seizures: distribution across time of day and sleep/wakefulness cycle.
        J Neurol. 2013; 260: 549-557
        • Ramgopal S.
        • et al.
        Circadian patterns of generalized tonic-clonic evolutions in pediatric epilepsy patients.
        Seizure. 2012; 21: 535-539
        • Herzog A.G.
        Catamenial epilepsy: update on prevalence, pathophysiology and treatment from the findings of the NIH progesterone treatment trial.
        Seizure. 2015; 28: 18-25
        • Rakers F.
        • et al.
        Weather as a risk factor for epileptic seizures: a case-crossover study.
        Epilepsia. 2017; 58: 1287-1295
        • Haut S.R.
        • et al.
        Seizure occurrence: precipitants and prediction.
        Neurology. 2007; 69: 1905-1910
        • Rüegg S.
        • et al.
        Association of environmental factors with the onset of status epilepticus.
        Epilepsy Behav. 2008; 12: 66-73
        • Gunn B.G.
        • Baram T.Z.
        Stress and seizures: space, time and hippocampal circuits.
        Trends Neurosci. 2017; 40 (Epub 2017 Sep 12. PMID: 28916130): 667-679https://doi.org/10.1016/j.tins.2017.08.004
        • Quigg M.
        Circadian rhythms: interactions with seizures and epilepsy.
        Epilepsy Res. 2000; 42: 43-55
        • Cook M.J.
        • et al.
        Human focal seizures are characterized by populations of fixed duration and interval.
        Epilepsia. 2016; 57: 359-368
        • Karoly P.J.
        • et al.
        Interictal spikes and epileptic seizures: their relationship and underlying rhythmicity.
        Brain. 2016; 139: 1066-1078
        • Cock H.R.
        • ESETT Group
        Established Status Epilepticus Treatment Trial (ESETT).
        Epilepsia. 2011; 52: 50-52
        • Sánchez Fernández I.
        • et al.
        Gaps and opportunities in refractory status epilepticus research in children: a multi-center approach by the Pediatric Status Epilepticus Research Group (pSERG).
        Seizure. 2014; 23: 87-97
        • Goldenholz D.M.
        • et al.
        Confusing placebo effect with natural history in epilepsy: a big data approach.
        Ann Neurol. 2015; 78: 329-336
        • Sánchez Fernández I.
        • et al.
        Prediction of time of occurrence and length of seizures based on basic demographic and clinical data using machine learning algorithms.
        American Epilepsy Society Meeting. 2016 (Houston, Tx)
        • Beam A.L.
        • Kohane I.S.
        Big data and machine learning in health care.
        JAMA. 2018; 319: 1317-1318
        • Litt B.
        • Echauz J.
        Prediction of epileptic seizures.
        Lancet Neurol. 2002; 1: 22-30
        • Mormann F.
        • et al.
        Seizure prediction: the long and winding road.
        Brain. 2007; 130: 314-333
        • Gadhoumi K.
        • et al.
        Seizure prediction for therapeutic devices: a review.
        J Neurosci Methods. 2016; 260: 270-282
        • Iasemidis L.D.
        • Sackellares J.C.
        Chaos theory and epilepsy.
        The Neuroscientist. 1996; 2: 118-126
        • Lehnertz K.
        Non-linear time series analysis of intracranial EEG recordings in patients with epilepsy--an overview.
        Int J Psychophysiol. 1999; 34: 45-52
        • Elger C.E.
        • et al.
        Nonlinear EEG analysis and its potential role in epileptology.
        Epilepsia. 2000; 41: S34-S38
        • Venkataraman V.
        • et al.
        Brain dynamics based automated epileptic seizure detection.
        Conf Proc IEEE Eng Med Biol Soc. 2014; 2014: 946-949
        • Drury I.
        • et al.
        Seizure prediction using scalp electroencephalogram.
        Exp Neurol. 2003; 184: S9-18
        • Litt B.
        • et al.
        Epileptic seizures may begin hours in advance of clinical onset: a report of five patients.
        Neuron. 2001; 30: 51-64
        • Schindler K.
        • et al.
        Forbidden ordinal patterns of periictal intracranial EEG indicate deterministic dynamics in human epileptic seizures.
        Epilepsia. 2011; 52: 1771-1780
        • Tezer F.I.
        • et al.
        A reduction of sleep spindles heralds seizures in focal epilepsy.
        Clin Neurophysiol. 2014; 125: 2207-2211
        • Anderson C.T.
        • et al.
        Day-night patterns of epileptiform activity in 65 patients with long-term ambulatory electrocorticography.
        J Clin Neurophysiol. 2015; 32: 406-412
        • Kramer M.A.
        • et al.
        Human seizures self-terminate across spatial scales via a critical transition.
        Proc Natl Acad Sci U S A. 2012; 109: 21116-21121
        • Van de Vel A.
        • et al.
        Non-EEG seizure-detection systems and potential SUDEP prevention: state of the art.
        Seizure. 2013; 22: 345-355
        • Gaínza-Lein M.
        • et al.
        Association of time to treatment with short-term outcomes for pediatric patients with refractory convulsive status epilepticus.
        JAMA Neurol. 2018; 75: 410-418https://doi.org/10.1001/jamaneurol.2017.4382
        • Sánchez Fernández I.
        • et al.
        Factors associated with treatment delays in pediatric refractory convulsive status epilepticus.
        Neurology. 2018;
        • Sánchez Fernández I.
        • et al.
        Time from convulsive status epilepticus onset to anticonvulsant administration in children.
        Neurology. 2015; 84: 2304-2311