Preictal state detection using prodromal symptoms: A machine learning approach

Seizure Detection, Prediction, and Forecasting

SUMMARY

Among the many fears associated with seizures, patients with epilepsy are greatly frustrated and distressed over seizure's apparent unpredictable occurrence. However, increasing evidence have emerged over the years to support that seizure occurrence is not a random phenomenon as previously presumed; it has a cyclic rhythm that oscillates over multiple timescales. The pattern in rises and falls of seizure rate that varies over 24 hours, weeks, months, and years has become a target for the development of innovative devices that intend to detect, predict, and forecast seizures. This article will review the different tools and devices available or that have been previously studied for seizure detection, prediction, and forecasting, as well as the associated challenges and limitations with the utilization of these devices. Although there is strong evidence for rhythmicity in seizure occurrence, very little is known about the mechanism behind this oscillation. This article concludes with early insights into the regulations that may potentially drive this cyclical variability and future directions.

Differential patterns of very high-frequency oscillations in two seizure types of the pilocarpine-induced TLE model

AIMS

Very high-frequency oscillations (VHFOs, >500 Hz) are considered a highly sensitive biomarker of seizures. We hypothesized that VHFOs may exhibit specificity towards hypersynchronous (HYP) seizures and low-voltage fast (LVF) seizures in temporal lobe epilepsy (TLE).

METHODS

Local field potentials were recorded from the hippocampal network in TLE mice induced by pilocarpine. Subsequently, we analyzed the VHFO features, including their temporal-frequency characteristics and VHFO/theta coupling, during three states: baseline, preictal, and postictal for both HYP- and LVF-seizure groups.

RESULTS

Significant changes in most of the VHFO features were observed during the preictal state in both seizure groups. In the postictal state, VHFO features in the HYP-seizure group exhibited inverse alterations and appeared to align with those observed during baseline conditions. However, such phenomena were not observed after TLE seizures in the LVF-seizure group.

CONCLUSION

Our findings highlight distinct patterns of VHFO feature changes across different states of HYP seizures and LVF seizures. These results suggest that VHFOs could serve as indicative biomarkers for seizure alterations specifically associated with HYP-seizure states.

The cost-effectiveness of a real-time seizure detection application for people with epilepsy

OBJECTIVES

Automated seizure detection modalities can increase safety among people with epilepsy (PWE) and reduce seizure-related anxiety. We evaluated the potential cost-effectiveness of a seizure detection mobile application for PWE in Singapore.

METHODS

We used a Markov cohort model to estimate the expected changes to total costs and health outcomes from a decision to adopt the seizure detection application versus the current standard of care from the health provider perspective. The time horizon is ten years and cycle duration is one month. Parameter values were updated from national databases and published literature. As we do not know the application efficacy in reducing seizure-related injuries, a conservative estimate of 1% reduction was used. Probabilistic sensitivity analysis, scenario analyses, and value of information analysis were performed.

RESULTS

At a willingness-to-pay of $45,000/ quality-adjusted life-years (QALY), the incremental cost-effectiveness ratio was $1,096/QALY, and the incremental net monetary benefit was $13,656. Probabilistic sensitivity analyses reported that the application had a 99.5% chance of being cost-effective. In a scenario analysis in which the reduction in risk of seizure-related injury was 20%, there was a 99.8% chance that the application was cost-effective. Value of information analysis revealed that health utilities was the most important parameter group contributing to model uncertainty.

CONCLUSIONS

This early-stage modeling study reveals that the seizure detection application is likely to be cost-effective compared to current standard of care. Future prospective trials will be needed to demonstrate the real-world impact of the application. Changes in health-related quality of life should also be measured in future trials.

Systemic inflammation as a biomarker of seizure propensity and a target for treatment to reduce seizure propensity

People with diabetes can wear a device that measures blood glucose and delivers just the amount of insulin needed to return the glucose level to within bounds. Currently, people with epilepsy do not have access to an equivalent wearable device that measures a systemic indicator of an impending seizure and delivers a rapidly acting medication or other intervention (e.g., an electrical stimulus) to terminate or prevent a seizure. Given that seizure susceptibility is reliably increased in systemic inflammatory states, we propose a novel closed-loop device where release of a fast-acting therapy is governed by sensors that quantify the magnitude of systemic inflammation. Here, we review the evidence that patients with epilepsy have raised levels of systemic indicators of inflammation than controls, and that some anti-inflammatory drugs have reduced seizure occurrence in animals and humans. We then consider the options of what might be incorporated into a responsive anti-seizure system.

Daily resting-state intracranial EEG connectivity for seizure risk forecasts

Forecasting seizure risk aims to detect proictal states in which seizures would be more likely to occur. Classical seizure prediction models are trained over long-term electroencephalographic (EEG) recordings to detect specific preictal changes for each seizure, independently of those induced by shifts in states of vigilance. A daily single measure-during a vigilance-controlled period-to estimate the risk of upcoming seizure(s) would be more convenient. Here, we evaluated whether intracranial EEG connectivity (phase-locking value), estimated from daily vigilance-controlled resting-state recordings, could allow distinguishing interictal (no seizure) from preictal (seizure within the next 24 h) states. We also assessed its relevance for daily forecasts of seizure risk using machine learning models. Connectivity in the theta band was found to provide the best prediction performances (area under the curve ≥ .7 in 80% of patients), with accurate daily and prospective probabilistic forecasts (mean Brier score and Brier skill score of .13 and .72, respectively). More efficient ambulatory clinical application could be considered using mobile EEG or chronic implanted devices.

Decrease in wearable-based nocturnal sleep efficiency precedes epileptic seizures

Introduction

While it is known that poor sleep is a seizure precipitant, this association remains poorly quantified. This study investigated whether seizures are preceded by significant changes in sleep efficiency as measured by a wearable equipped with an electrocardiogram, respiratory bands, and an accelerometer.

Methods

Nocturnal recordings from 47 people with epilepsy hospitalized at our epilepsy monitoring unit were analyzed (304 nights). Sleep metrics during nights followed by epileptic seizures (24 h post-awakening) were compared to those of nights which were not.

Results

Lower sleep efficiency (percentage of sleep during the night) was found in the nights preceding seizure days (p < 0.05). Each standard deviation decrease in sleep efficiency and increase in wake after sleep onset was respectively associated with a 1.25-fold (95 % CI: 1.05 to 1.42, p < 0.05) and 1.49-fold (95 % CI: 1.17 to 1.92, p < 0.01) increased odds of seizure occurrence the following day. Furthermore, nocturnal seizures were associated with significantly lower sleep efficiency and higher wake after sleep onset (p < 0.05), as well as increased odds of seizure occurrence following wake (OR: 5.86, 95 % CI: 2.99 to 11.77, p < 0.001).

Discussion

Findings indicate lower sleep efficiency during nights preceding seizures, suggesting that wearable sensors could be promising tools for sleep-based seizure-day forecasting in people with epilepsy.

Epileptic Prodromes: Insights from Surveying 196 Patients and 150 Caregivers

OBJECTIVE

Uncontrolled epilepsy creates a constant source of worry for patients and puts them at a high risk of injury. Identifying recurrent "premonitory" symptoms of seizures and using them to recalibrate seizure prediction algorithms may improve prediction performances. This study aimed to investigate patients' ability to predict oncoming seizures based on preictal symptoms.

METHODS

Through an online survey, demographics and clinical characteristics (e.g., seizure frequency, epilepsy duration, and postictal symptom duration) were collected from people with epilepsy and caregivers across Canada. Respondents were asked to answer questions regarding their ability to predict seizures through warning symptoms. A total of 196 patients and 150 caregivers were included and were separated into three groups: those who reported warning symptoms within the 5 minutes preceding a seizure, prodromes (symptoms earlier than 5 minutes before seizure), and no warning symptoms.

RESULTS

Overall, 12.2% of patients and 12.0% of caregivers reported predictive prodromes ranging from 5 minutes to more than 24 hours before the seizures (median of 2 hours). The most common were dizziness/vertigo (28%), mood changes (26%), and cognitive changes (21%). Statistical testing showed that respondents who reported prodromes also reported significantly longer postictal recovery periods compared to those who did not report predictive prodromes (P < 0.05).

CONCLUSION

Findings suggest that patients who present predictive seizure prodromes may be characterized by longer patient-reported postictal recovery periods. Studying the correlation between seizure severity and predictability and investigating the electrical activity underlying prodromes may improve our understanding of preictal mechanisms and ability to predict seizures.

An Explainable Statistical Method for Seizure Prediction Using Brain Functional Connectivity from EEG

Background. Epilepsy is a group of chronic neurological disorders characterized by recurrent and abrupt seizures. The accurate prediction of seizures can reduce the burdens of this disorder. Now, existing studies use brain network features to classify patients’ preictal or interictal states, enabling seizure prediction. However, most predicting methods are based on deep learning techniques, which have weak interpretability and high computational complexity. To address these issues, in this study, we proposed a novel two-stage statistical method that is interpretable and easy to compute. Methods. We used two datasets to evaluate the performance of the proposed method, including the well-known public dataset CHB-MIT. In the first stage, we estimated the dynamic brain functional connectivity network for each epoch. Then, in the second stage, we used the derived network predictor for seizure prediction. Results. We illustrated the results of our method in seizure prediction in two datasets separately. For the FH-PKU dataset, our approach achieved an AUC value of 0.963, a prediction sensitivity of 93.1%, and a false discovery rate of 7.7%. For the CHB-MIT dataset, our approach achieved an AUC value of 0.940, a prediction sensitivity of 93.0%, and a false discovery rate of 11.1%, outperforming existing state-of-the-art methods. Significance. This study proposed an explainable statistical method, which can estimate the brain network using the scalp EEG method and use the net-work predictor to predict epileptic seizures. Availability and Implementation. R Source code is available at https://github.com/HaoChen1994/Seizure-Prediction.

Outliers in clinical symptoms as preictal biomarkers

Previous findings have suggested that a preictal state might precede the epileptic seizure onset, which is the basis for seizure prediction attempts. Preictal states can be apprehended as outliers that differ from an interictal baseline and display clinical changes. We collected daily clinical scores from patients with epilepsy who underwent continuous video-EEG and assessed the ability of several outlier detection methods to identify preictal states. Results from 24 patients suggested that outlying clinical features were suggestive of preictal states and can be identified by statistical methods: AUC = 0.71, 95 % CI = [0.63 - 0.79]; PPV = 0.77, 95 % CI = [0.70 - 0.84]; FPR = 0.31, 95 % CI = [0.21 - 0.44]); and F1 score = 0.74, 95 % CI = [0.64 - 0.81]. Such algorithms could be straightforwardly implemented in a mobile device (e.g., tablet or smartphone), which would allow a longer data collection that could improve prediction performances. Additional clinical - and even multimodal - parameters could identify more subtle physiological modifications.

Learning from machine learning: prediction of age-related athletic performance decline trajectories

Factors that determine individual age-related decline rates in physical performance are poorly understood and prediction poses a challenge. Linear and quadratic regression models are usually applied, but often show high prediction errors for individual athletes. Machine learning approaches may deliver more accurate predictions and help to identify factors that determine performance decline rates. We hypothesized that it is possible to predict the performance development of a master athlete from a single measurement, that prediction by a machine learning approach is superior to prediction by the average decline curve or an individually shifted decline curve, and that athletes with a higher starting performance show a slower performance decline than those with a lower performance. The machine learning approach was implemented using a multilayer neuronal network. Results showed that performance prediction from a single measurement is possible and that the prediction by a machine learning approach was superior to the other models. The estimated performance decline rate was highest in athletes with a high starting performance and a low starting age, as well as in those with a low starting performance and high starting age, while the lowest decline rate was found for athletes with a high starting performance and a high starting age. Machine learning was superior and predicted trajectories with significantly lower prediction errors compared to conventional approaches. New insights into factors determining decline trajectories were identified by visualization of the model outputs. Machine learning models may be useful in revealing unknown factors that determine the age-related performance decline.