User login
HOUSTON—Mobile or wearable devices that observe autonomic and motor changes may provide seizure detection, according to an overview presented at the 70th Annual Meeting of the American Epilepsy Society. Devices that monitor heart rate, electrodermal changes, and electromyogram (EMG) changes also may provide accurate detection of convulsive seizures. Many such devices are in development, and some already are available. They may be convenient for patients to use and could provide opportunities for interventions to prevent sudden unexpected death in epilepsy (SUDEP).
What Do Patients Want?
In 2015, Hoppe and colleagues published the results of their survey of patients’ preferences for automated seizure-detection devices. The authors found that patients want convenient devices such as wearable wristbands, but do not want devices that are conspicuous. Patients want devices that perform automatic seizure registration and make emergency calls to family members or caregivers. They do not, however, want devices that notify their doctors automatically about seizures, preferring to maintain control of their own medical data. On the other hand, patients do want to get help during medical appointments with controlling and using the devices. Finally, patients want devices that predict seizures, but this task “probably requires intracranial recording,” said Gregory Krauss, MD, Professor of Neurology at Johns Hopkins University School of Medicine in Baltimore. 
Another survey asked patients about the accuracy that they wanted from these devices. One survey suggested that patients would accept a false-positive or false-negative rate of less than 25%. Seizure-free patients would accept a false-positive rate of less than one per week. Patients with recurring seizures would accept a higher false-positive rate. “Most patients said [that] as long as the ratio of false detections to true detections was not more than 1:1, they would find that acceptable,” said Dr. Krauss.
Patients also want rapid detection of seizures so that they can protect themselves or quickly get help. “The majority of patients would like detection to occur with alerting within one minute” of seizure onset, said Dr. Krauss. Some patients want detection to occur within 30 seconds of seizure onset, and about 14% of patients said that detection within three minutes of onset would be sufficient.
Investigators have pursued various methods for seizure detection. The major methods are physiologic (eg, heart rate, pulse oximetry, and electrodermal activity), motion-related (eg, accelerometers, mattress sensors, EMG, and video motion detection), and responsive testing (eg, testing memory and responsiveness). The best technique may be to combine several of these methods to minimize the rate of false positives, said Dr. Krauss.
SmartWatch
One device that currently is available is SmartWatch, which is manufactured by SmartMonitor. The SmartWatch device is a wrist accelerometer that predominantly detects convulsive seizures. Data from epilepsy monitoring units (EMUs) on the device’s efficacy have been mixed. In a Stanford University study, SmartWatch recorded 62 seizures in 27 patients. Nine patients had 13 tonic–clonic seizures, and the device detected 12 of them. The device had a false positive rate of 87%. In a University of Tennessee study involving 41 patients who had 191 seizures, the SmartWatch device detected 31% of tonic–clonic seizures. These studies suggest that the device has “somewhat limited sensitivity without individualizing settings,” and its accuracy during real-world use is unclear, said Dr. Krauss.
Embrace Watch
The device furthest along in development is Embrace Watch, which is being designed by Empatica. Embrace Watch is a smart watch that performs electrodermal and accelerometer measurements and is connected to a smart phone. It uses an app that graphically displays information about the user’s activity, stress level, and sleep. The watch is intended to detect convulsive seizures and send alerts to caregivers.
One reason that Embrace Watch measures electrodermal activity is that the latter correlates with postictal EEG suppression. Research indicates that postictal generalized EEG suppression corresponds with the risk of SUDEP. “That [finding] raises the possibility that the Empatica device may be useful for intervening in SUDEP,” said Dr. Krauss.
An ongoing study is evaluating Embrace Watch’s efficacy in an EMU. Investigators collected video EEG data for 69 patients and monitored electrodermal activity and movement with the Embrace Watch. The investigators reviewed the detection of 55 convulsive seizures in 22 patients. They tested two classifier models, which were different combinations of electrodermal activity and heart rate changes. One classifier model had 100% sensitivity, but a moderately high rate of false alarms (1.26/day). The other classifier model had a sensitivity of 95% and a rate of 0.2 false alarms per day, which may be acceptable for a patient with uncontrolled seizures, said Dr. Krauss. A separate case study suggests that Embrace Watch accurately detects major seizures in patients with Dravet syndrome.
Brain Sentinel
The FDA is reviewing the Brain Sentinel seizure-monitoring system, which uses EMG recordings to detect convulsive seizures. The device takes advantage of a characteristic firing pattern on EMG that is associated with convulsive seizures. Research has indicated that, when used properly, the system detects seizures with near-perfect accuracy. The system’s accuracy was slightly higher in adults than in a group of adults and children. The average time between seizure onset and alarm was 5.34 seconds in adults and 7.70 seconds in adults and children. The rate of false positives per eight-hour period was 0.51 in adults and 0.48 in adults and children. Most false alarms occurred in a minority of participants. “This device is probably going to be effective for detecting convulsive seizures in patients with uncontrolled epilepsy,” said Dr. Krauss.
EpiWatch App
Dr. Krauss and Nathan Crone, MD, Professor of Neurology at Johns Hopkins University School of Medicine, are developing a seizure-detection app called EpiWatch for use on the Apple Watch. Their objective is to use the Apple Watch’s ability to monitor heart rate, movement, and position to detect seizures. Through the Apple Watch’s user interface, the app asks patients to label seizures and attempts to reduce the number of false detections. During a seizure, the app tests the user’s responsiveness by asking him or her to tap the watch. The test occurs at 20 seconds after seizure onset and at one-minute intervals thereafter. If the user is responsive, the app administers a test of working memory that involves repeating a series of random numbers. At 10 minutes after the seizure ends, the app asks the user whether he or she had had a seizure, whether he or she had had an aura, and whether he or she had lost awareness. These data are logged into a journal that can be transmitted.
Drs. Krauss and Crone are using Apple ResearchKit to study the app’s efficacy. They have enrolled approximately 700 participants in the study, and about 40% of them were able to activate the app during a seizure. The seizure types recorded so far are representative of those of the general US population of patients with epilepsy. Approximately 30% of participants had a 50% increase in heart rate during their seizure. About two-thirds of patients had a 30% increase in heart rate. A significant number of patients with complex partial seizures had decreases in heart rate during their seizure. “Heart rate alone would probably not be a sensitive screening technique for seizures, but it could be one component of an algorithm,” said Dr. Krauss.
Can These Devices Prevent SUDEP?
Many of these devices were designed with the goal of enabling interventions, particularly during severe nocturnal convulsive seizures, to prevent SUDEP. The current understanding of SUDEP provides grounds for optimism that such interventions could be achieved, said Dr. Krauss. Many patients with tonic–clonic seizures have oxygen saturation, postictal generalized EEG suppression, and autonomic alterations. If patients become immobile during a seizure and are lying face down, they may become asphyxiated. “The question is whether we can intervene at that point to prevent the subsequent cardiac arrhythmias which lead to their death,” said Dr. Krauss. A caregiver, for example, could turn a patient on his or her side and stimulate him or her to promote breathing. If the caregiver finds the patient at a later time after seizure onset, he or she could attempt resuscitation. “This is our general concept of perhaps how SUDEP intervention could be performed using mobile device detectors,” Dr. Krauss concluded.
—Erik Greb
Suggested Reading
Hoppe C, Feldmann M, Blachut B, et al. Novel techniques for automated seizure registration: Patients’ wants and needs. Epilepsy Behav. 2015;52(Pt A):1-7.
Lhatoo SD, Nei M, Raghavan M, et al. Nonseizure SUDEP: Sudden unexpected death in epilepsy without preceding epileptic seizures. Epilepsia. 2016;57(7):1161-1168.
Van de Vel A, Smets K, Wouters K, Ceulemans B. Automated non-EEG based seizure detection: Do users have a say? Epilepsy Behav. 2016;62:121-128.
Velez M, Fisher RS, Bartlett V, Le S. Tracking generalized tonic-clonic seizures with a wrist accelerometer linked to an online database. Seizure. 2016;39:13-18.
HOUSTON—Mobile or wearable devices that observe autonomic and motor changes may provide seizure detection, according to an overview presented at the 70th Annual Meeting of the American Epilepsy Society. Devices that monitor heart rate, electrodermal changes, and electromyogram (EMG) changes also may provide accurate detection of convulsive seizures. Many such devices are in development, and some already are available. They may be convenient for patients to use and could provide opportunities for interventions to prevent sudden unexpected death in epilepsy (SUDEP).
What Do Patients Want?
In 2015, Hoppe and colleagues published the results of their survey of patients’ preferences for automated seizure-detection devices. The authors found that patients want convenient devices such as wearable wristbands, but do not want devices that are conspicuous. Patients want devices that perform automatic seizure registration and make emergency calls to family members or caregivers. They do not, however, want devices that notify their doctors automatically about seizures, preferring to maintain control of their own medical data. On the other hand, patients do want to get help during medical appointments with controlling and using the devices. Finally, patients want devices that predict seizures, but this task “probably requires intracranial recording,” said Gregory Krauss, MD, Professor of Neurology at Johns Hopkins University School of Medicine in Baltimore.
Another survey asked patients about the accuracy that they wanted from these devices. One survey suggested that patients would accept a false-positive or false-negative rate of less than 25%. Seizure-free patients would accept a false-positive rate of less than one per week. Patients with recurring seizures would accept a higher false-positive rate. “Most patients said [that] as long as the ratio of false detections to true detections was not more than 1:1, they would find that acceptable,” said Dr. Krauss.
Patients also want rapid detection of seizures so that they can protect themselves or quickly get help. “The majority of patients would like detection to occur with alerting within one minute” of seizure onset, said Dr. Krauss. Some patients want detection to occur within 30 seconds of seizure onset, and about 14% of patients said that detection within three minutes of onset would be sufficient.
Investigators have pursued various methods for seizure detection. The major methods are physiologic (eg, heart rate, pulse oximetry, and electrodermal activity), motion-related (eg, accelerometers, mattress sensors, EMG, and video motion detection), and responsive testing (eg, testing memory and responsiveness). The best technique may be to combine several of these methods to minimize the rate of false positives, said Dr. Krauss.
SmartWatch
One device that currently is available is SmartWatch, which is manufactured by SmartMonitor. The SmartWatch device is a wrist accelerometer that predominantly detects convulsive seizures. Data from epilepsy monitoring units (EMUs) on the device’s efficacy have been mixed. In a Stanford University study, SmartWatch recorded 62 seizures in 27 patients. Nine patients had 13 tonic–clonic seizures, and the device detected 12 of them. The device had a false positive rate of 87%. In a University of Tennessee study involving 41 patients who had 191 seizures, the SmartWatch device detected 31% of tonic–clonic seizures. These studies suggest that the device has “somewhat limited sensitivity without individualizing settings,” and its accuracy during real-world use is unclear, said Dr. Krauss.
Embrace Watch
The device furthest along in development is Embrace Watch, which is being designed by Empatica. Embrace Watch is a smart watch that performs electrodermal and accelerometer measurements and is connected to a smart phone. It uses an app that graphically displays information about the user’s activity, stress level, and sleep. The watch is intended to detect convulsive seizures and send alerts to caregivers.
One reason that Embrace Watch measures electrodermal activity is that the latter correlates with postictal EEG suppression. Research indicates that postictal generalized EEG suppression corresponds with the risk of SUDEP. “That [finding] raises the possibility that the Empatica device may be useful for intervening in SUDEP,” said Dr. Krauss.
An ongoing study is evaluating Embrace Watch’s efficacy in an EMU. Investigators collected video EEG data for 69 patients and monitored electrodermal activity and movement with the Embrace Watch. The investigators reviewed the detection of 55 convulsive seizures in 22 patients. They tested two classifier models, which were different combinations of electrodermal activity and heart rate changes. One classifier model had 100% sensitivity, but a moderately high rate of false alarms (1.26/day). The other classifier model had a sensitivity of 95% and a rate of 0.2 false alarms per day, which may be acceptable for a patient with uncontrolled seizures, said Dr. Krauss. A separate case study suggests that Embrace Watch accurately detects major seizures in patients with Dravet syndrome.
Brain Sentinel
The FDA is reviewing the Brain Sentinel seizure-monitoring system, which uses EMG recordings to detect convulsive seizures. The device takes advantage of a characteristic firing pattern on EMG that is associated with convulsive seizures. Research has indicated that, when used properly, the system detects seizures with near-perfect accuracy. The system’s accuracy was slightly higher in adults than in a group of adults and children. The average time between seizure onset and alarm was 5.34 seconds in adults and 7.70 seconds in adults and children. The rate of false positives per eight-hour period was 0.51 in adults and 0.48 in adults and children. Most false alarms occurred in a minority of participants. “This device is probably going to be effective for detecting convulsive seizures in patients with uncontrolled epilepsy,” said Dr. Krauss.
EpiWatch App
Dr. Krauss and Nathan Crone, MD, Professor of Neurology at Johns Hopkins University School of Medicine, are developing a seizure-detection app called EpiWatch for use on the Apple Watch. Their objective is to use the Apple Watch’s ability to monitor heart rate, movement, and position to detect seizures. Through the Apple Watch’s user interface, the app asks patients to label seizures and attempts to reduce the number of false detections. During a seizure, the app tests the user’s responsiveness by asking him or her to tap the watch. The test occurs at 20 seconds after seizure onset and at one-minute intervals thereafter. If the user is responsive, the app administers a test of working memory that involves repeating a series of random numbers. At 10 minutes after the seizure ends, the app asks the user whether he or she had had a seizure, whether he or she had had an aura, and whether he or she had lost awareness. These data are logged into a journal that can be transmitted.
Drs. Krauss and Crone are using Apple ResearchKit to study the app’s efficacy. They have enrolled approximately 700 participants in the study, and about 40% of them were able to activate the app during a seizure. The seizure types recorded so far are representative of those of the general US population of patients with epilepsy. Approximately 30% of participants had a 50% increase in heart rate during their seizure. About two-thirds of patients had a 30% increase in heart rate. A significant number of patients with complex partial seizures had decreases in heart rate during their seizure. “Heart rate alone would probably not be a sensitive screening technique for seizures, but it could be one component of an algorithm,” said Dr. Krauss.
Can These Devices Prevent SUDEP?
Many of these devices were designed with the goal of enabling interventions, particularly during severe nocturnal convulsive seizures, to prevent SUDEP. The current understanding of SUDEP provides grounds for optimism that such interventions could be achieved, said Dr. Krauss. Many patients with tonic–clonic seizures have oxygen saturation, postictal generalized EEG suppression, and autonomic alterations. If patients become immobile during a seizure and are lying face down, they may become asphyxiated. “The question is whether we can intervene at that point to prevent the subsequent cardiac arrhythmias which lead to their death,” said Dr. Krauss. A caregiver, for example, could turn a patient on his or her side and stimulate him or her to promote breathing. If the caregiver finds the patient at a later time after seizure onset, he or she could attempt resuscitation. “This is our general concept of perhaps how SUDEP intervention could be performed using mobile device detectors,” Dr. Krauss concluded.
—Erik Greb
Suggested Reading
Hoppe C, Feldmann M, Blachut B, et al. Novel techniques for automated seizure registration: Patients’ wants and needs. Epilepsy Behav. 2015;52(Pt A):1-7.
Lhatoo SD, Nei M, Raghavan M, et al. Nonseizure SUDEP: Sudden unexpected death in epilepsy without preceding epileptic seizures. Epilepsia. 2016;57(7):1161-1168.
Van de Vel A, Smets K, Wouters K, Ceulemans B. Automated non-EEG based seizure detection: Do users have a say? Epilepsy Behav. 2016;62:121-128.
Velez M, Fisher RS, Bartlett V, Le S. Tracking generalized tonic-clonic seizures with a wrist accelerometer linked to an online database. Seizure. 2016;39:13-18.
HOUSTON—Mobile or wearable devices that observe autonomic and motor changes may provide seizure detection, according to an overview presented at the 70th Annual Meeting of the American Epilepsy Society. Devices that monitor heart rate, electrodermal changes, and electromyogram (EMG) changes also may provide accurate detection of convulsive seizures. Many such devices are in development, and some already are available. They may be convenient for patients to use and could provide opportunities for interventions to prevent sudden unexpected death in epilepsy (SUDEP).
What Do Patients Want?
In 2015, Hoppe and colleagues published the results of their survey of patients’ preferences for automated seizure-detection devices. The authors found that patients want convenient devices such as wearable wristbands, but do not want devices that are conspicuous. Patients want devices that perform automatic seizure registration and make emergency calls to family members or caregivers. They do not, however, want devices that notify their doctors automatically about seizures, preferring to maintain control of their own medical data. On the other hand, patients do want to get help during medical appointments with controlling and using the devices. Finally, patients want devices that predict seizures, but this task “probably requires intracranial recording,” said Gregory Krauss, MD, Professor of Neurology at Johns Hopkins University School of Medicine in Baltimore.
Another survey asked patients about the accuracy that they wanted from these devices. One survey suggested that patients would accept a false-positive or false-negative rate of less than 25%. Seizure-free patients would accept a false-positive rate of less than one per week. Patients with recurring seizures would accept a higher false-positive rate. “Most patients said [that] as long as the ratio of false detections to true detections was not more than 1:1, they would find that acceptable,” said Dr. Krauss.
Patients also want rapid detection of seizures so that they can protect themselves or quickly get help. “The majority of patients would like detection to occur with alerting within one minute” of seizure onset, said Dr. Krauss. Some patients want detection to occur within 30 seconds of seizure onset, and about 14% of patients said that detection within three minutes of onset would be sufficient.
Investigators have pursued various methods for seizure detection. The major methods are physiologic (eg, heart rate, pulse oximetry, and electrodermal activity), motion-related (eg, accelerometers, mattress sensors, EMG, and video motion detection), and responsive testing (eg, testing memory and responsiveness). The best technique may be to combine several of these methods to minimize the rate of false positives, said Dr. Krauss.
SmartWatch
One device that currently is available is SmartWatch, which is manufactured by SmartMonitor. The SmartWatch device is a wrist accelerometer that predominantly detects convulsive seizures. Data from epilepsy monitoring units (EMUs) on the device’s efficacy have been mixed. In a Stanford University study, SmartWatch recorded 62 seizures in 27 patients. Nine patients had 13 tonic–clonic seizures, and the device detected 12 of them. The device had a false positive rate of 87%. In a University of Tennessee study involving 41 patients who had 191 seizures, the SmartWatch device detected 31% of tonic–clonic seizures. These studies suggest that the device has “somewhat limited sensitivity without individualizing settings,” and its accuracy during real-world use is unclear, said Dr. Krauss.
Embrace Watch
The device furthest along in development is Embrace Watch, which is being designed by Empatica. Embrace Watch is a smart watch that performs electrodermal and accelerometer measurements and is connected to a smart phone. It uses an app that graphically displays information about the user’s activity, stress level, and sleep. The watch is intended to detect convulsive seizures and send alerts to caregivers.
One reason that Embrace Watch measures electrodermal activity is that the latter correlates with postictal EEG suppression. Research indicates that postictal generalized EEG suppression corresponds with the risk of SUDEP. “That [finding] raises the possibility that the Empatica device may be useful for intervening in SUDEP,” said Dr. Krauss.
An ongoing study is evaluating Embrace Watch’s efficacy in an EMU. Investigators collected video EEG data for 69 patients and monitored electrodermal activity and movement with the Embrace Watch. The investigators reviewed the detection of 55 convulsive seizures in 22 patients. They tested two classifier models, which were different combinations of electrodermal activity and heart rate changes. One classifier model had 100% sensitivity, but a moderately high rate of false alarms (1.26/day). The other classifier model had a sensitivity of 95% and a rate of 0.2 false alarms per day, which may be acceptable for a patient with uncontrolled seizures, said Dr. Krauss. A separate case study suggests that Embrace Watch accurately detects major seizures in patients with Dravet syndrome.
Brain Sentinel
The FDA is reviewing the Brain Sentinel seizure-monitoring system, which uses EMG recordings to detect convulsive seizures. The device takes advantage of a characteristic firing pattern on EMG that is associated with convulsive seizures. Research has indicated that, when used properly, the system detects seizures with near-perfect accuracy. The system’s accuracy was slightly higher in adults than in a group of adults and children. The average time between seizure onset and alarm was 5.34 seconds in adults and 7.70 seconds in adults and children. The rate of false positives per eight-hour period was 0.51 in adults and 0.48 in adults and children. Most false alarms occurred in a minority of participants. “This device is probably going to be effective for detecting convulsive seizures in patients with uncontrolled epilepsy,” said Dr. Krauss.
EpiWatch App
Dr. Krauss and Nathan Crone, MD, Professor of Neurology at Johns Hopkins University School of Medicine, are developing a seizure-detection app called EpiWatch for use on the Apple Watch. Their objective is to use the Apple Watch’s ability to monitor heart rate, movement, and position to detect seizures. Through the Apple Watch’s user interface, the app asks patients to label seizures and attempts to reduce the number of false detections. During a seizure, the app tests the user’s responsiveness by asking him or her to tap the watch. The test occurs at 20 seconds after seizure onset and at one-minute intervals thereafter. If the user is responsive, the app administers a test of working memory that involves repeating a series of random numbers. At 10 minutes after the seizure ends, the app asks the user whether he or she had had a seizure, whether he or she had had an aura, and whether he or she had lost awareness. These data are logged into a journal that can be transmitted.
Drs. Krauss and Crone are using Apple ResearchKit to study the app’s efficacy. They have enrolled approximately 700 participants in the study, and about 40% of them were able to activate the app during a seizure. The seizure types recorded so far are representative of those of the general US population of patients with epilepsy. Approximately 30% of participants had a 50% increase in heart rate during their seizure. About two-thirds of patients had a 30% increase in heart rate. A significant number of patients with complex partial seizures had decreases in heart rate during their seizure. “Heart rate alone would probably not be a sensitive screening technique for seizures, but it could be one component of an algorithm,” said Dr. Krauss.
Can These Devices Prevent SUDEP?
Many of these devices were designed with the goal of enabling interventions, particularly during severe nocturnal convulsive seizures, to prevent SUDEP. The current understanding of SUDEP provides grounds for optimism that such interventions could be achieved, said Dr. Krauss. Many patients with tonic–clonic seizures have oxygen saturation, postictal generalized EEG suppression, and autonomic alterations. If patients become immobile during a seizure and are lying face down, they may become asphyxiated. “The question is whether we can intervene at that point to prevent the subsequent cardiac arrhythmias which lead to their death,” said Dr. Krauss. A caregiver, for example, could turn a patient on his or her side and stimulate him or her to promote breathing. If the caregiver finds the patient at a later time after seizure onset, he or she could attempt resuscitation. “This is our general concept of perhaps how SUDEP intervention could be performed using mobile device detectors,” Dr. Krauss concluded.
—Erik Greb
Suggested Reading
Hoppe C, Feldmann M, Blachut B, et al. Novel techniques for automated seizure registration: Patients’ wants and needs. Epilepsy Behav. 2015;52(Pt A):1-7.
Lhatoo SD, Nei M, Raghavan M, et al. Nonseizure SUDEP: Sudden unexpected death in epilepsy without preceding epileptic seizures. Epilepsia. 2016;57(7):1161-1168.
Van de Vel A, Smets K, Wouters K, Ceulemans B. Automated non-EEG based seizure detection: Do users have a say? Epilepsy Behav. 2016;62:121-128.
Velez M, Fisher RS, Bartlett V, Le S. Tracking generalized tonic-clonic seizures with a wrist accelerometer linked to an online database. Seizure. 2016;39:13-18.
