A study in Nature by Mayo Clinic researchers and international collaborators found patterns could be identified in patients who wear a special wristwatch monitoring device for six to 12 months, allowing about 30 minutes of warning before a seizure occurred. This worked well most of the time for five of six patients studied.
“Just as a reliable weather forecast helps people plan their activities, so, too, could seizure forecasting help patients living with epilepsy adjust their plans if they knew a seizure were imminent,” says Dr Benjamin Brinkmann, an epilepsy scientist at Mayo Clinic and the senior author. “This study using a wrist-worn device shows that providing reliable seizure forecasts is possible without directly measuring brain activity.”
In the study, patients with drug-resistant epilepsy and an implanted neurostimulation device that monitors electrical brain activity were given two wrist-worn recording devices and a tablet computer to upload data daily to cloud storage. Patients were instructed to wear one wristband while the other charged. They switched devices at a set time each day. They used the devices while participating in their normal activities, providing unique long-term data for the study.
Information collected from the wearable device included electrical characteristics of the skin, body temperature, blood flow, heart rate and accelerometry data that tracks movement. Data were analysed with a deep learning neural network approach to artificial intelligence, using an algorithm for time series and frequency analysis. Because the research participants already had an implanted deep brain stimulation device to treat their epilepsy, those neurostimulation devices were used to confirm seizures, allowing the team to measure the accuracy of forecasting by the
wrist-worn devices.
While the ability to forecast seizures previously has been shown using implanted brain devices, many patients don't want an invasive implant, Brinkmann noted. "We hope this research with wearable devices paves the way toward integrating seizure forecasting into clinical practice in the future," he added, saying this was a preliminary study and additional patients are recording data to expand this test.
This study is part of the Epilepsy Foundation of America’s Epilepsy Innovation Institute, and the My Seizure Gauge project, an international collaboration aimed at using wearable devices for seizure detection and forecasting in epilepsy. Additional support was provided by the Mayo Clinic Neurology Artificial Intelligence Program.
Study details
Ambulatory seizure forecasting with a wrist-worn device using long-short term memory deep learning
Mona Nasseri, Tal Pal Attia, Boney Joseph, Nicholas Gregg, Ewan Nurse, Pedro Viana, Gregory Worrell, Matthias Dümpelmann, Mark Richardson, Dean Freestone & Benjamin Brinkmann.
Published in Nature Scientific on 9 November 2021
Abstract
The ability to forecast seizures minutes to hours in advance of an event has been verified using invasive EEG devices, but has not been previously demonstrated using noninvasive wearable devices over long durations in an ambulatory setting. In this study we developed a seizure forecasting system with a long short-term memory (LSTM) recurrent neural network (RNN) algorithm, using a noninvasive wrist-worn research-grade physiological sensor device, and tested the system in patients with epilepsy in the field, with concurrent invasive EEG confirmation of seizures via an implanted recording device. The system achieved forecasting performance significantly better than a random predictor for 5 of 6 patients studied, with mean AUC-ROC of 0.80 (range 0.72–0.92). These results provide the first clear evidence that direct seizure forecasts are possible using wearable devices in the ambulatory setting for many patients with epilepsy.
Introduction
Despite optimised medication therapy, resective surgery, and neuromodulation therapy, many people with epilepsy continue to experience seizures. Half or more of patients who undergo resective surgery for epilepsy have eventual recurrence of seizures, and devices for neuromodulation rarely achieve long-term seizure freedom. People living with epilepsy consistently report the unpredictability of seizures to be the most limiting aspect of their condition. Reliable seizure forecasts could potentially allow people living with recurrent seizures to modify their activities, take a fast-acting medication, or increase neuromodulation therapy to prevent or manage impending seizures. Accurate seizure forecasts have been demonstrated using invasively sampled ultralong-term EEG in ambulatory canine and human subjects, including a prospective study with a dedicated device. However, invasive devices may not be acceptable for some patients with epilepsy, and no clinically available invasive device currently has the capability to sample and telemeter data needed for seizure forecasting. Hence there is presently great interest in forecasting seizures using wearable or minimally invasive devices. Deep learning approaches have shown promising performance for a variety of difficult applications, including seizure forecasting.
In particular these “end-to-end learning” methods are attractive for seizure forecasting given the challenges of identifying salient features in ultra-long term time-series data, and the heterogeneity in time series data characteristics between different patients. The power and capability of deep learning algorithms trained on very large datasets hold promise to enable applications not previously believed possible, and may open the door to seizure forecasting with noninvasive sampling devices.
Many challenges exist in designing a reliable system for forecasting seizures from noninvasively recorded data. Training, testing, and validating a forecasting algorithm requires ultra-long duration recordings with an adequate number of seizures. Additionally, concurrent video and/or EEG validation of seizures in an ambulatory setting over months to years is logistically difficult, and is not possible using conventional in-hospital monitoring methods. Self-reported seizure diaries are the most accessible validation, but the poor reliability of such diaries is widely recognised.
Performing device studies on in-hospital patients with concurrent video-EEG validation is logistically feasible, but such studies are expensive, and limited in duration, and restrict normal daily activities which could produce false alarms, such as exercise, brushing teeth, or other activities. Because of these challenges an ILAE-IFCN working group recently published guidelines for seizure detection studies with non-invasive wearable devices, but few studies achieve phase 3–4 evidence in an ambulatory setting. In studies of seizure forecasting it is imperative that ambulatory data, including the full range of normal activities, be included in the training, testing, and validation sets.
Seizure prediction with wearable devices was recently investigated in a cohort of in-hospital patients using a cross-patient deep learning algorithm on data recorded from Empatica E4 devices. The dataset comprised multiday recordings from 69 epilepsy patients (28 female, duration 2311.4 h, 452 seizures). In a leave-one-patient-out cross-validation approach, they achieved better than chance prediction in 43% of patients, with no difference in performance between generalised and focal seizure types. It has also been shown that seizure occurrence can be modelled as circadian or multiday patterns of seizure risk over long periods, and these patterns may be used to forecast seizures.
Using a mobile electronic seizure diary application seizure forecasts calculated based on circadian and multiday seizure cycles using data from 50 application users produced accurate forecasts for approximately half the cohort. Long-term cycles of seizure risk offer complementary information to direct forecasting of seizures, and signals from wearable fitness trackers have been shown to have value in identifying circadian and multidian cycles of seizure risk. This study aimed to develop a wearable seizure forecasting system for ambulatory use, and to evaluate the forecasting performance relative to seizures identified with concurrent chronic intracranial EEG (iEEG).
Conclusions
This preliminary study in a small cohort has demonstrated seizure forecasting using a noninvasive wrist-worn multimodal sensor significantly better than a random predictor in ambulatory ultra-long-term recordings of patients with epilepsy for the majority of patients studied. Wearable data was recorded in an ambulatory setting during normal activity with concurrent EEG validation of seizure events. Five of six patients analyzed achieved seizure forecasts significantly more accurate than a chance predictor, and seizure alerts in these five patients provided ample warning time to administer fast-acting medication or to increase neuromodulation therapy. This is the first study reporting successful seizure forecasting with noninvasive devices in ultra-long-term recordings in freely-behaving humans outside the clinical environment.
See more from MedicalBrief archives:
Epilepsy: Seizures not forecastable as expected
New epilepsy seizure-detecting device to cut nocturnal fatalities
FDA approves epileptic seizure monitoring watch