Accepted Manuscript

SmartWatch by SmartMonitor: Assessment of Seizure Detection Efficacy for Various

Seizure Types in Children, A Large Prospective Single-Center Study

Amy L. Patterson, MD, Basanagoud Mudigoudar, MD, Stephen Fulton, MD, Amy
McGregor, MD, Kate Van Poppel, MD, Margaret C. Wheless, Lai Brooks, DNP-FNP-
BC, James W. Wheless, MD
PII: S0887-8994(15)00332-X
DOI: 10.1016/j.pediatrneurol.2015.07.002
Reference: PNU 8705

To appear in: Pediatric Neurology

Received Date: 6 June 2015

Revised Date: 6 July 2015
Accepted Date: 7 July 2015

Please cite this article as: Patterson AL, Mudigoudar B, Fulton S, McGregor A, Van Poppel K, Wheless
MC, Brooks L, Wheless JW, SmartWatch by SmartMonitor: Assessment of Seizure Detection Efficacy
for Various Seizure Types in Children, A Large Prospective Single-Center Study, Pediatric Neurology
(2015), doi: 10.1016/j.pediatrneurol.2015.07.002.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

SmartWatch by SmartMonitor: Assessment of Seizure Detection Efficacy for Various Seizure

Types in Children, A Large Prospective Single-Center Study

Amy L. Patterson, MD1,2, Basanagoud Mudigoudar, MD1,2, Stephen Fulton, MD1,2, Amy McGregor,
MD1,2, Kate Van Poppel, MD1,2, Margaret C. Wheless1,2, Lai Brooks, DNP-FNP-BC1,2, James W. Wheless,
MD1,2.
1University of Tennessee Health Science Center. 2Le Bonheur Children’s Hospital.

PT
Corresponding Author:
James W. Wheless, MD
Phone: 901-287-5207
Fax: 901-287-5325
Email: jwheless@uthsc.edu

RI
Abstract

SC
Introduction: Patients with epilepsy and their caregivers are constantly burdened with the
possibility of a seizure and the consequences such as accidents, injuries, and sudden
unexplained death in epilepsy. It is the unpredictable nature of seizures that often affects
both patients with seizures and their caregivers, limits independence, and hinders their

U
quality of life. There are several types of motion detectors on the market, each with varying
degrees of sensitivity.
AN
Material and Methods: We prospectively tested the SmartWatch, a wrist-worn monitor, on
children, adolescents, and young adults with various types of seizures in an Epilepsy
Monitoring Unit. Confirmation of the seizure type and if there was rhythmic upper
extremity jerking associated with the seizure was determined by review of the video EEG.
M

This was compared with the standard detection system of the watch.

Results: This study analyzed total of 191 seizures in 41 patients, aged 5-41 years. Fifty-one
D

of the seizures were generalized tonic-clonic. Forty-seven of the seizures had a rhythmic
arm movement component. The SmartWatch detected 30 seizures (16%) out of the total, 16
(31%) of the generalized tonic-clonic seizures, and 16 (34%) seizures associated with
TE

rhythmic arm movements.

Discussion: Overall, only a minority of generalized tonic-clonic seizures or seizures with

rhythmic movements were detected, highlighting the need for an effective seizure detection
EP

device.

Keywords
C

Seizure, monitor, detection, pediatrics, SmartWatch

AC

Introduction

Epilepsy is the most common childhood neurologic condition with estimated lifetime
prevalence of about 1%.1,2 For patients with epilepsy and their caregivers, fear of a seizure
is ever present, and their quality of life is affected by the constant fear of when a seizure will
occur.3 Accidents and injuries are more frequent in people with epilepsy compared to the
general population, and the majority of accidents occur at home.4 Convulsive seizures in
particular are a risk factor for both seizure-related injuries and sudden unexplained death
in epilepsy.4,5 Additionally, seizures are associated with stigma, headache, and psychiatric
co-morbidities including ADHD and conduct problems. 2,6 Children with epilepsy have more
limitations, were more likely to repeat a school grade, and had poorer social compliance.1 A
 ACCEPTED MANUSCRIPT

device to detect seizures, both during sleep and wakefulness, may be able to help reduce the
risk of injury and sudden unexplained death in epilepsy. Additionally, this type of device
could alleviate some of the concern and fear associated with the constant risk of recurrent
seizures by giving an opportunity for intervention including first aid and medication
administration. This could also allow the child to be more independent, as the parent would
not need to keep him under constant surveillance because of a concern he may have a
seizure. A number of monitor systems have been manufactured and tested with varying

PT
degrees of success.7-13 This study was conducted with children, adolescents, and young
adults in an Epilepsy Monitoring Unit at a tertiary care children’s hospital to assess the
sensitivity and reliability of a commercially available wrist-worn monitor (SmartWatch,
www.smart-monitor.com) to detect various seizure types. It was felt this would be helpful

RI
data to have when parents asked if this monitor would be helpful for their child with
seizures.

SC
Materials and Methods

The SmartWatch by SmartMonitor (San Jose, CA) is a wristwatch device that works by
continuously monitoring movements and can instantly send alerts to connected caregivers

U
about repetitive, shaking motions. It was developed with the mission to provide an easy-to-
use monitoring device that would immediately notify a caregiver when the user needed
help, specifically for people with chronic health conditions including epilepsy.
AN
SmartMonitor is reported to provide monitoring and tracking solutions with the purpose of
enhancing safety while offering significant peace of mind to users and families. SmartWatch
analyzes activity outside a normal spectrum and also has a user push button notification
option for wearers to indicate their need for assistance (e.g., If a patient has an aura, they
M

could hit this button, and it would notify a pre-identified family member.). It records
movement patterns, time, and duration of unusual occurrences. SmartWatch works in
conjunction with an Android phone and requires the user to carry an Android device, which
D

sends alerts to a family member or caregiver.14 Standard device settings for movement
sensitivity were used during this study.
TE

This study was approved by the university and hospital internal review boards. One
hundred forty-three subjects (ages 2-41 years) were recruited from a population of
children, adolescents, and young adults being admitted to the Epilepsy Monitoring Unit in
EP

Le Bonheur Children’s Hospital for evaluation of ongoing seizures. Patients had standard
video electroencephalography, cardiopulmonary, and nursing monitoring. The SmartWatch
was used with a NEXUS 7 tablet (to record when an event was detected) and was worn
when the patients were undergoing continuous video-electroencephalogram (video-EEG)
C

monitoring. Parents and staff used standard push button marking for the video-EEG record
for any events they felt were consistent with seizures, regardless of whether or not the
AC

SmartWatch was activated. The video EEG record was reviewed to detect any seizures. All
video-EEG events were reviewed retrospectively in 24-hour increments by board-certified
clinical neurophysiologists (and verified by another clinical neurophysiologist) to
determine if events were seizures and were classified as to seizure type and whether or not
they had associated rhythmic arm movements. Rhythmic arm movement was determined
by video, and seizures were classified as such only if there was vigorous movement of the
arm on which the watch was placed. Because of this, not all generalized tonic-clonic and
partial onset hypermotor seizures were classified as having vigorous rhythmic arm
movement. In order to increase the yield of the SmartWatch, in patients with known
unilateral rhythmic movements during seizures, the watch was placed on that arm. In some
 ACCEPTED MANUSCRIPT

instances the watch was even moved if seizures were captured with rhythmic arm
movement of the contralateral arm (in an attempt to increase the detection rate). The
SmartWatch alarm log was reviewed daily and correlated with the video-EEG findings to
ascertain which of the seizures (diagnosed on video-EEG) were captured by the
SmartWatch. We did not record all of the SmartWatch false positive detections, as these are
well known10. Seizure medications were continued or discontinued at the discretion of the
primary epileptologist. The ability of the SmartWatch to detect seizures in general, detect

PT
specific seizure types, and detect seizures with associated rhythmic arm movement was
analyzed.

We have tested several different devices, all of which are available for purchase by

RI
consumers in the United States and in other countries. 8,9 All were tested in the same format
as described with this cohort of patients.

Results

SC
Forty-one patients (ages 5-41 years) experienced a combined total of 197 seizures. One
hundred ninety-one seizures are classified for the purposes of this study; two were not
included because it is unknown if the SmartWatch was activated, two because the watch

U
was disconnected, and two because the video was not available.
AN
We classified seizures by type (generalized tonic-clonic, myoclonic/myoclonic-tonic, partial
onset with minimal motor component, partial onset-hypermotor, and tonic) and looked at
the sensitivity of the SmartWatch in detecting each type individually and seizures as a
whole. At best, the SmartWatch detected 31% of generalized tonic-clonic seizures, and at
M

worst, it detected 0% of myoclonic/myoclonic-tonic seizures. The overall seizure detection

rate was 16%. See Table 1 for seizure types captured and detection yield by SmartWatch.

Seizures were also evaluated by whether or not they had repetitive shaking movement of
D

the arm, and the efficacy of the SmartWatch was evaluated based on that movement
independent of seizure type (Table 2). The SmartWatch detected 34% of seizures with
TE

vigorous rhythmic arm movement and 10% of seizures without vigorous rhythmic arm
movement.

Additionally, we compared the SmartWatch to three other seizure detection systems that
EP

we previously evaluated (Table 3).8,9 The Emfit monitor had the highest detection rate for
nocturnal generalized tonic-clonic seizures at 75%. The SmartWatch rates were
substantially lower at 31% but included seizures that occurred during both wakefulness
and sleep. The Medpage systems identified even fewer nocturnal seizures.
C

Discussion
AC

While epilepsy is the most common neurological condition in childhood, there remains a
stigma associated with seizures affecting the quality of life of those affected by it. 1,2,6
Additionally, there are social limitations on people with epilepsy secondary to the
unpredictability of seizure recurrence. In fact, a vast majority of patients with epilepsy
consider this unpredictability to play an important role in their everyday lives and express
interest in the ability to detect seizures. 15 The concern of injuries and even death associated
with unwitnessed seizures is a real concern of people with epilepsy and their caregivers. 10
A seizure detection device could alleviate concern by notifying caregivers, provide a greater
degree of independence to those with epilepsy, and possibly prevent injuries associated
with seizures. An ideal device would be one that is portable and detects seizures in mobile,
 ACCEPTED MANUSCRIPT

potentially independent individuals and provides a means to notify caregivers to facilitate

the necessary interventions including first aid and rescue medications. An additional useful
feature of SmartWatch is the push button notification option, which in patients with reliable
auras could allow them notify a family member during times when they are not closely
supervised. This would allow the family to assist the patient, even if the watch did not
activate during the seizure.

PT
Lockman, et al. reported false positives of the SmartWatch, and these included, but were not
limited to, brushing teeth, arranging bed, nail filing, changing watch, and tapping pen. In
their study, the detection rate for non-seizure events was high, but they instructed patients
not to cancel the alarm for non-seizure events while knowing that the cancellation feature

RI
would actually decrease false positives.10 While we know that the rate of false positive
alarms with the SmartWatch is high, our study goal was not to evaluate the false positive
rate, which may be an additional useful piece of information for providers to give families

SC
inquiring about seizure detection devices.

When compared with previously evaluated seizure detection devices, the SmartWatch
seizure detection rates were substantially lower, but this is the only device currently

U
available for seizure detection of ambulatory patients during the day. The SmartWatch
sensitivity proved to be less than ideal in all types of seizures, even in seizures with a
significant rhythmic arm movement component.
AN
There continues to be an unmet need for a device that can detect seizures (both partial
onset and convulsive seizures) in ambulatory patients. This would be welcomed by
caregivers and patients.
M

Acknowledgment

This study was conducted at Le Bonheur Children’s Hospital in Memphis, Tennessee. We

D

acknowledge the generous support of the Shainberg Neuroscience Fund to complete this
project. SmartWatches and the tablets were supplied by SmartMonitor at no cost.
TE

Author Contributions

BM, SF, KVP, AM, and JWW enrolled patients and reviewed video EEGs for seizure type.
EP

MCW consented patients and set up data collection system. LB enrolled patients, confirmed
alarms were working, and checked patients in the hospital each day. ALP compiled and
analyzed data and drafted the manuscript. JWW conceived the project, obtained IRB
approval, and reviewed the manuscript.
C

Declaration of Conflicting Interests

AC

The authors declared no potential conflicts of interest with respect to the research,
authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication
of this article.

Ethical Approval
 ACCEPTED MANUSCRIPT

This study was approved by the University of Tennessee Health Science Center and Le
Bonheur Children’s Hospital internal review boards.

References:

1. Russ SA, Larson K, Halfon N. A national profile of childhood epilepsy and seizure
disorder. Pediatrics. Feb 2012;129(2):256-264.
2. Jones JE, Austin JK, Caplan R, Dunn D, Plioplys S, Salpekar JA. Psychiatric disorders in

PT
children and adolescents who have epilepsy. Pediatrics in review / American
Academy of Pediatrics. Feb 2008;29(2):e9-14.
3. Arthurs S, Zaveri HP, Frei MG, Osorio I. Patient and caregiver perspectives on

RI
seizure prediction. Epilepsy & behavior : E&B. Nov 2010;19(3):474-477.
4. Tomson T, Beghi E, Sundqvist A, Johannessen SI. Medical risks in epilepsy: a review
with focus on physical injuries, mortality, traffic accidents and their prevention.
Epilepsy research. Jun 2004;60(1):1-16.

SC
5. Tomson T, Walczak T, Sillanpaa M, Sander JW. Sudden unexpected death in epilepsy:
a review of incidence and risk factors. Epilepsia. 2005;46 Suppl 11:54-61.
6. Jacoby A. Stigma, epilepsy, and quality of life. Epilepsy & behavior : E&B. Dec

U
2002;3(6s2):10-20.
7. Nijsen TM, Arends JB, Griep PA, Cluitmans PJ. The potential value of three-
dimensional accelerometry for detection of motor seizures in severe epilepsy.
AN
Epilepsy & behavior : E&B. Aug 2005;7(1):74-84.
8. Poppel KV, Fulton SP, McGregor A, Ellis M, Patters A, Wheless J. Prospective Study of
the Emfit Movement Monitor. Journal of child neurology. Nov 2013;28(11):1434-
1436.
M

9. Fulton S, Poppel KV, McGregor A, Ellis M, Patters A, Wheless J. Prospective Study of 2

Bed Alarms for Detection of Nocturnal Seizures. Journal of child neurology. Nov
2013;28(11):1430-1433.
D

10. Lockman J, Fisher RS, Olson DM. Detection of seizure-like movements using a wrist
accelerometer. Epilepsy & behavior : E&B. Apr 2011;20(4):638-641.
TE

11. Poh MZ, Loddenkemper T, Reinsberger C, et al. Convulsive seizure detection using a
wrist-worn electrodermal activity and accelerometry biosensor. Epilepsia. May
2012;53(5):e93-97.
12. Poh MZ, Loddenkemper T, Swenson NC, Goyal S, Madsen JR, Picard RW. Continuous
EP

monitoring of electrodermal activity during epileptic seizures using a wearable

sensor. Conference proceedings : ... Annual International Conference of the IEEE
Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and
Biology Society. Annual Conference. 2010;2010:4415-4418.
C

13. Kramer U, Kipervasser S, Shlitner A, Kuzniecky R. A novel portable seizure detection

alarm system: preliminary results. Journal of clinical neurophysiology : official
AC

publication of the American Electroencephalographic Society. Feb 2011;28(1):36-38.

14. Monitor S. 2015. Accessed March 27, 2015.
15. Schulze-Bonhage A, Sales F, Wagner K, et al. Views of patients with epilepsy on
seizure prediction devices. Epilepsy & behavior : E&B. Aug 2010;18(4):388-396.
 ACCEPTED MANUSCRIPT

Seizure type Total Detected % Detected

Generalized tonic-clonic 51 16 31

Myoclonic/myoclonic-tonic 11 0 0

PT
Partial onset with minimal 63 1 2
motor component

RI
Partial onset-hypermotor 45 11 24

Tonic 21 2 10

SC
Total 191 30 16

Table 1. Seizure Types and Detection Rates of SmartWatch

U
AN
M
D
TE
CEP
AC
 ACCEPTED MANUSCRIPT

Total Detected % Detected

Arm movement 47 16 34

No arm movement 144 14 10

PT
Table 2. SmartWatch Detection Rate for Rhythmic Arm Movement

RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

SmartWatch Emfit Medpage MP5 Medpage ST-2

Generalized tonic 16/51 (31%) 12/16 (75%) 1/13 (8%) 0/4 (0%)

PT
clonic seizures

Myoclonic, tonic, 3/32 (6%) 4/34 (12%) 0/6 (0%) 1/34 (3%)
myoclonic-tonic

RI
seizures

Partial onset 11/45 (24%) 7/19 (37%) 0/4 (0%) 1/8 (13%)

SC
seizures with motor
component

Table 3. Seizures Types and Detection Rates for Various Devices Tested

U
AN
M
D
TE
C EP
AC