The goal of this observational study is to understand heart rhythm changes in adults with epilepsy. It will also explore whether computer programs can use heart recordings to detect or predict seizures. The study will include 20 adults with temporal lobe epilepsy and 20 healthy volunteers. Participants with epilepsy will have ongoing seizures despite treatment with at least two suitable medicines. All participants will be 18 years or older. In temporal lobe epilepsy, seizures start in a specific region of the brain. These seizures may spread to both sides of the brain, causing muscle stiffening and rhythmic jerking. Researchers will focus on these spreading seizures when developing seizure detection and prediction methods. The main questions are: * How do heart rate and heartbeat timing differ between people with and without epilepsy? * How do these measures change before, during, and after seizures? * Can computer programs learn patterns in heart recordings to detect seizures or predict them before they start? Researchers will compare heart rate and heart rate variability between the two groups. Heart rate variability describes changes in the time between one heartbeat and the next. They will compare recordings during rest, wakefulness, and sleep. Participants will: * Wear a fabric chest band that records the heart's electrical activity. * Have brain activity recorded through sensors on the scalp, alongside video recording. * Rest quietly while awake for one hour during the recording period. Participants with epilepsy will have recordings during their usual hospital monitoring. Healthy volunteers will have 24 hours of recording at the same hospital. Researchers will use the brain recordings and video to identify when seizures happen. They will develop and test computer programs that learn patterns in the heart recordings.
This prospective, single-center observational study will characterize heart rate and heart rate variability (HRV) in adults with drug-resistant temporal lobe epilepsy and explore their use in detecting and predicting focal to bilateral tonic-clonic seizures. The study combines comparisons between epilepsy participants and healthy volunteers with analyses of physiological changes surrounding seizures and development of computational models. The planned sample comprises 20 adults with epilepsy and 20 healthy volunteers. Recruitment will use non-probability sampling. Epilepsy participants will be recruited from patients admitted for clinically indicated inpatient video-electroencephalography (video-EEG) monitoring. Healthy volunteers will be recruited through the researchers' university and professional networks, with age and sex distributions as similar as feasible to those of the epilepsy group. Recordings will take place at Sancaktepe Şehit Prof. Dr. İlhan Varank Training and Research Hospital. A textile-based wearable chest band will record electrocardiography (ECG) simultaneously with video-EEG. Epilepsy participants will be recorded during their routine inpatient monitoring, while healthy volunteers will undergo 24 hours of recording at the same hospital. Similar equipment and recording conditions will be used in both groups to reduce methodological differences related to the recording environment, physical burden and comfort. Both groups will complete a standardized one-hour awake resting recording on the first morning, between 09:00 and 12:00, following a 15-minute adaptation period. The timing of the last meal and caffeine and nicotine use will be recorded. Neurologists will review video-EEG recordings from both epilepsy participants and healthy volunteers to identify sleep and wakefulness. In epilepsy participants, they will evaluate interictal epileptiform discharges, seizure onset and offset, seizure classification and postictal EEG changes. In healthy volunteers, they will verify the absence of electrographic seizure activity during the recording. Analyses by sleep stage will be limited to periods that can be reliably classified. Heart rate and HRV will be compared between groups during standardized rest, sleep and wakefulness. Within epilepsy participants, analyses will examine interictal, preictal, ictal and postictal changes while accounting for repeated observations from the same participant. Medication type, dose and timing relative to the recordings will also be considered in the analysis and interpretation of heart rate and HRV findings. Effect sizes and confidence intervals will accompany the comparisons. Textile ECG will be the primary data source for seizure detection and prediction models. Synchronized EEG and video will provide reference annotations and may also contribute complementary information through transfer learning and multimodal modelling. Conventional machine learning and deep learning approaches will be compared, including transfer learning, adaptation using limited examples from a new patient (few-shot learning), evaluation in previously unseen patients without adaptation (zero-shot evaluation) and anomaly detection. Where appropriate, models using ECG, EEG and video separately and in combination will be compared to assess the contribution of each modality. Suitable external datasets may support model pretraining. Model performance will be evaluated offline using separate test recordings. Patient-independent evaluations will separate participants between development and testing, while patient-specific evaluations will preserve the chronological separation of training or adaptation data from later test recordings. Related segments from the same seizure will not be split between training and testing. Recordings will be assigned to training, validation and test sets before being divided into analysis windows. Model development and tuning will exclude test recordings. Results will be summarized across cross-validation folds or repeated runs. The sample size reflects the exploratory nature of the study and experience from related research. Statistical power to detect moderate effects may be limited. The study is intended to characterize physiological patterns, estimate effect sizes and obtain preliminary model performance estimates to inform subsequent studies. Routine clinical care will remain unchanged, and study recordings and model outputs will not guide medical decisions.
Study Type
OBSERVATIONAL
Enrollment
40
A textile-based wearable chest band will record electrocardiography (ECG) simultaneously with video-EEG. Epilepsy participants will wear the device during their clinically indicated inpatient monitoring, without changes to routine monitoring duration or clinical care. Healthy volunteers will undergo 24 hours of recording at the same hospital. Both groups will complete a standardized one-hour awake resting recording. ECG data will be used for heart rate and heart rate variability analyses and development and offline evaluation of seizure detection and prediction models. Study recordings and model outputs will not guide participants' medical care.
Sancaktepe Şehit Prof. Dr. İlhan Varank Training and Research Hospital
Istanbul, Turkey (Türkiye)
Mean heart rate by recording state
Mean heart rate will be calculated in beats per minute for each analyzed recording period. Textile-based wearable electrocardiography (ECG) recordings will be used. Values from the standardized one-hour awake resting recording will be compared between the epilepsy and healthy volunteer groups. Sleep and wakefulness will also be compared between groups, with sleep stage analyses limited to reliably classified periods. Within epilepsy participants, interictal, preictal, ictal and postictal values will be compared using video-EEG reference annotations and accounting for repeated observations. Equal-length, technically usable segments will be analyzed under prespecified rules. Effect sizes and confidence intervals will be reported.
Time frame: During one monitoring admission: approximately 48-72 hours for epilepsy participants and 24 hours for healthy volunteers, including the one-hour awake resting recording on the first morning (09:00-12:00).
Heart rate variability: SDNN by recording state
The standard deviation of normal-to-normal heartbeat intervals (SDNN) will be calculated in milliseconds for each analyzed recording period. Textile-based wearable electrocardiography (ECG) recordings will be used. Values from the standardized one-hour awake resting recording will be compared between the epilepsy and healthy volunteer groups. Sleep and wakefulness will also be compared between groups, with sleep stage analyses limited to reliably classified periods. Within epilepsy participants, interictal, preictal, ictal and postictal values will be compared using video-EEG reference annotations and accounting for repeated observations. Equal-length, technically usable segments will be analyzed under prespecified rules. Effect sizes and confidence intervals will be reported.
Time frame: During one monitoring admission: approximately 48-72 hours for epilepsy participants and 24 hours for healthy volunteers, including the one-hour awake resting recording on the first morning (09:00-12:00).
Heart rate variability: RMSSD by recording state
The root mean square of successive differences between normal-to-normal heartbeat intervals (RMSSD) will be calculated in milliseconds for each analyzed recording period. Textile-based wearable electrocardiography (ECG) recordings will be used. Values from the standardized one-hour awake resting recording will be compared between the epilepsy and healthy volunteer groups. Sleep and wakefulness will also be compared between groups, with sleep stage analyses limited to reliably classified periods. Within epilepsy participants, interictal, preictal, ictal and postictal values will be compared using video-EEG reference annotations and accounting for repeated observations. Equal-length, technically usable segments will be analyzed under prespecified rules. Effect sizes and confidence intervals will be reported.
Time frame: During one monitoring admission: approximately 48-72 hours for epilepsy participants and 24 hours for healthy volunteers, including the one-hour awake resting recording on the first morning (09:00-12:00).
Heart rate variability: pNN50 by recording state
pNN50 is the percentage of successive normal-to-normal heartbeat interval pairs that differ by more than 50 milliseconds. It will be calculated for each analyzed recording period. Textile-based wearable electrocardiography (ECG) recordings will be used. Values from the standardized one-hour awake resting recording will be compared between the epilepsy and healthy volunteer groups. Sleep and wakefulness will also be compared between groups, with sleep stage analyses limited to reliably classified periods. Within epilepsy participants, interictal, preictal, ictal and postictal values will be compared using video-EEG reference annotations and accounting for repeated observations. Equal-length, technically usable segments will be analyzed under prespecified rules. Effect sizes and confidence intervals will be reported.
Time frame: During one monitoring admission: approximately 48-72 hours for epilepsy participants and 24 hours for healthy volunteers, including the one-hour awake resting recording on the first morning (09:00-12:00).
Heart rate variability: low-frequency power by recording state
Low-frequency (LF) power of normal-to-normal heartbeat interval variability will be calculated in milliseconds squared for each analyzed recording period. Textile-based wearable electrocardiography (ECG) recordings will be used. Values from the standardized one-hour awake resting recording will be compared between the epilepsy and healthy volunteer groups. Sleep and wakefulness will also be compared between groups, with sleep stage analyses limited to reliably classified periods. Within epilepsy participants, interictal, preictal, ictal and postictal values will be compared using video-EEG reference annotations and accounting for repeated observations. Equal-length, technically usable segments will be analyzed under prespecified rules. Effect sizes and confidence intervals will be reported.
Time frame: During one monitoring admission: approximately 48-72 hours for epilepsy participants and 24 hours for healthy volunteers, including the one-hour awake resting recording on the first morning (09:00-12:00).
Heart rate variability: high-frequency power by recording state
High-frequency (HF) power of normal-to-normal heartbeat interval variability will be calculated in milliseconds squared for each analyzed recording period. Textile-based wearable electrocardiography (ECG) recordings will be used. Values from the standardized one-hour awake resting recording will be compared between the epilepsy and healthy volunteer groups. Sleep and wakefulness will also be compared between groups, with sleep stage analyses limited to reliably classified periods. Within epilepsy participants, interictal, preictal, ictal and postictal values will be compared using video-EEG reference annotations and accounting for repeated observations. Equal-length, technically usable segments will be analyzed under prespecified rules. Effect sizes and confidence intervals will be reported.
Time frame: During one monitoring admission: approximately 48-72 hours for epilepsy participants and 24 hours for healthy volunteers, including the one-hour awake resting recording on the first morning (09:00-12:00).
Heart rate variability: LF/HF ratio by recording state
The ratio of low-frequency power to high-frequency power (LF/HF) of normal-to-normal heartbeat interval variability will be calculated for each analyzed recording period. The ratio is dimensionless. Textile-based wearable electrocardiography (ECG) recordings will be used. Values from the standardized one-hour awake resting recording will be compared between the epilepsy and healthy volunteer groups. Sleep and wakefulness will also be compared between groups, with sleep stage analyses limited to reliably classified periods. Within epilepsy participants, interictal, preictal, ictal and postictal values will be compared using video-EEG reference annotations and accounting for repeated observations. Equal-length, technically usable segments will be analyzed under prespecified rules. Effect sizes and confidence intervals will be reported.
Time frame: During one monitoring admission: approximately 48-72 hours for epilepsy participants and 24 hours for healthy volunteers, including the one-hour awake resting recording on the first morning (09:00-12:00).
Heart rate variability: SD1 by recording state
Poincare plot SD1 is the standard deviation of normal-to-normal heartbeat interval pairs perpendicular to the line of identity. It will be calculated in milliseconds for each analyzed recording period. Textile-based wearable electrocardiography (ECG) recordings will be used. Values from the standardized one-hour awake resting recording will be compared between the epilepsy and healthy volunteer groups. Sleep and wakefulness will also be compared between groups, with sleep stage analyses limited to reliably classified periods. Within epilepsy participants, interictal, preictal, ictal and postictal values will be compared using video-EEG reference annotations and accounting for repeated observations. Equal-length, technically usable segments will be analyzed under prespecified rules. Effect sizes and confidence intervals will be reported.
Time frame: During one monitoring admission: approximately 48-72 hours for epilepsy participants and 24 hours for healthy volunteers, including the one-hour awake resting recording on the first morning (09:00-12:00).
Heart rate variability: SD2 by recording state
Poincare plot SD2 is the standard deviation of normal-to-normal heartbeat interval pairs along the line of identity. It will be calculated in milliseconds for each analyzed recording period. Textile-based wearable electrocardiography (ECG) recordings will be used. Values from the standardized one-hour awake resting recording will be compared between the epilepsy and healthy volunteer groups. Sleep and wakefulness will also be compared between groups, with sleep stage analyses limited to reliably classified periods. Within epilepsy participants, interictal, preictal, ictal and postictal values will be compared using video-EEG reference annotations and accounting for repeated observations. Equal-length, technically usable segments will be analyzed under prespecified rules. Effect sizes and confidence intervals will be reported.
Time frame: During one monitoring admission: approximately 48-72 hours for epilepsy participants and 24 hours for healthy volunteers, including the one-hour awake resting recording on the first morning (09:00-12:00).
Heart rate variability: sample entropy by recording state
Sample entropy, a dimensionless measure of the irregularity of normal-to-normal heartbeat interval sequences, will be calculated for each analyzed recording period. Textile-based wearable electrocardiography (ECG) recordings will be used. Values from the standardized one-hour awake resting recording will be compared between the epilepsy and healthy volunteer groups. Sleep and wakefulness will also be compared between groups, with sleep stage analyses limited to reliably classified periods. Within epilepsy participants, interictal, preictal, ictal and postictal values will be compared using video-EEG reference annotations and accounting for repeated observations. Equal-length, technically usable segments will be analyzed under prespecified rules. Effect sizes and confidence intervals will be reported.
Time frame: During one monitoring admission: approximately 48-72 hours for epilepsy participants and 24 hours for healthy volunteers, including the one-hour awake resting recording on the first morning (09:00-12:00).
Event-level sensitivity for focal to bilateral tonic-clonic seizure detection
The percentage of eligible focal to bilateral tonic-clonic seizures confirmed by video-EEG that are correctly detected by each model: correctly detected seizures divided by all eligible reference seizures in the test recordings, multiplied by 100. Reference events will be annotated independently of ECG changes and model outputs. Alarm-to-event matching rules will be specified before outcome analysis. Performance will be evaluated by internal validation on held-out recordings, with participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Results will be reported separately by model approach, with variability across validation folds or repeated runs.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
False alarms per hour for seizure detection
The number of detection alarms not matched to a reference seizure, divided by the total evaluable monitoring time in hours. Reference seizures will be established by independent video-EEG annotation. Alarm-to-event matching and alarm-merging rules will be specified before outcome analysis and applied consistently to held-out recordings. Performance will be evaluated by internal validation on held-out recordings, with participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Results will be reported separately by model approach, with variability across validation folds or repeated runs.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Event-level sensitivity for focal to bilateral tonic-clonic seizure prediction
The percentage of eligible lead focal to bilateral tonic-clonic seizures correctly preceded by a model warning within the prespecified prediction timing rules. Correctly predicted lead seizures will be divided by all eligible lead seizures in the test recordings and multiplied by 100. Independent targets must be preceded by at least two seizure-free hours. The prediction horizon, seizure occurrence period and alarm-to-event matching rules will be specified before outcome analysis. Performance will be evaluated by internal validation on held-out recordings, with participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Results will be reported separately by model approach, with variability across validation folds or repeated runs.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
False alarms per hour for seizure prediction
The number of prediction alarms not followed by an eligible reference seizure within the prespecified seizure occurrence period, divided by the total evaluable monitoring time in hours. Lead seizure eligibility, prediction horizon, seizure occurrence period and alarm-matching rules will be specified before outcome analysis. Video-EEG annotations will provide the reference events. Performance will be evaluated by internal validation on held-out recordings, with participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Results will be reported separately by model approach, with variability across validation folds or repeated runs.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Window-level classification sensitivity
The percentage of reference-positive analysis windows classified as positive: true-positive windows divided by all reference-positive windows, multiplied by 100. Reference-positive windows represent seizure activity for detection and the prespecified preictal target for prediction. Calculated separately for seizure detection and seizure prediction classification tasks and for each model approach, using prespecified reference window labels derived from video-EEG. Evaluation will use held-out recordings, preserving participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Values will be summarized across validation folds or repeated runs, with standard deviations or appropriate confidence intervals.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Window-level classification specificity
The percentage of reference-negative analysis windows classified as negative: true-negative windows divided by all reference-negative windows, multiplied by 100. Calculated separately for seizure detection and seizure prediction classification tasks and for each model approach, using prespecified reference window labels derived from video-EEG. Evaluation will use held-out recordings, preserving participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Values will be summarized across validation folds or repeated runs, with standard deviations or appropriate confidence intervals.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Window-level positive predictive value
The percentage of positive model classifications that are reference-positive: true-positive windows divided by all predicted-positive windows, multiplied by 100. This measure is also called precision. Calculated separately for seizure detection and seizure prediction classification tasks and for each model approach, using prespecified reference window labels derived from video-EEG. Evaluation will use held-out recordings, preserving participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Values will be summarized across validation folds or repeated runs, with standard deviations or appropriate confidence intervals.
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Window-level negative predictive value
The percentage of negative model classifications that are reference-negative: true-negative windows divided by all predicted-negative windows, multiplied by 100. Calculated separately for seizure detection and seizure prediction classification tasks and for each model approach, using prespecified reference window labels derived from video-EEG. Evaluation will use held-out recordings, preserving participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Values will be summarized across validation folds or repeated runs, with standard deviations or appropriate confidence intervals.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Window-level F1 score
The F1 score is the harmonic mean of window-level precision and sensitivity. It is dimensionless, ranges from 0 to 1, and is higher when classification performance is better. Calculated separately for seizure detection and seizure prediction classification tasks and for each model approach, using prespecified reference window labels derived from video-EEG. Evaluation will use held-out recordings, preserving participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Values will be summarized across validation folds or repeated runs, with standard deviations or appropriate confidence intervals.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Window-level balanced accuracy
Balanced accuracy is the arithmetic mean of window-level sensitivity and specificity, expressed as a percentage. It gives equal weight to performance on positive and negative reference classes. Calculated separately for seizure detection and seizure prediction classification tasks and for each model approach, using prespecified reference window labels derived from video-EEG. Evaluation will use held-out recordings, preserving participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Values will be summarized across validation folds or repeated runs, with standard deviations or appropriate confidence intervals.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Window-level receiver operating characteristic area under the curve
The area under the receiver operating characteristic curve (ROC-AUC) will summarize discrimination across classification thresholds using window-level prediction scores. It is dimensionless and ranges from 0 to 1. Calculated separately for seizure detection and seizure prediction classification tasks and for each model approach, using prespecified reference window labels derived from video-EEG. Evaluation will use held-out recordings, preserving participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Values will be summarized across validation folds or repeated runs, with standard deviations or appropriate confidence intervals.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Window-level precision-recall area under the curve
The area under the precision-recall curve (AUPRC) will summarize the relationship between precision and sensitivity across classification thresholds using window-level prediction scores. It is dimensionless and ranges from 0 to 1. Calculated separately for seizure detection and seizure prediction classification tasks and for each model approach, using prespecified reference window labels derived from video-EEG. Evaluation will use held-out recordings, preserving participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Values will be summarized across validation folds or repeated runs, with standard deviations or appropriate confidence intervals.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Window-level classification accuracy
The percentage of all analysis windows classified correctly: the sum of true-positive and true-negative windows divided by all evaluated windows, multiplied by 100. Accuracy will be interpreted alongside the other metrics because seizures are much less frequent than non-seizure periods. Calculated separately for seizure detection and seizure prediction classification tasks and for each model approach, using prespecified reference window labels derived from video-EEG. Evaluation will use held-out recordings, preserving participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Values will be summarized across validation folds or repeated runs, with standard deviations or appropriate confidence intervals.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Mean seizure detection latency
For correctly detected focal to bilateral tonic-clonic seizures, the interval from the clinical seizure onset annotated using video-EEG to the corresponding model alarm will be calculated in seconds. Mean latency will be reported for each detection model, with variability or confidence intervals. Alarm matching will follow prespecified rules. Timing will be assessed offline from time-stamped recordings and algorithm outputs; it does not represent prospective deployment of a clinical alarm. Performance will be evaluated by internal validation on held-out recordings, with participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Results will be reported separately by model approach, with variability across validation folds or repeated runs.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Mean seizure prediction lead time
For correctly predicted eligible lead focal to bilateral tonic-clonic seizures, the interval from the corresponding model warning to reference seizure onset will be calculated in minutes. Mean lead time will be reported for each prediction model, with variability or confidence intervals. Reference onset, lead seizure eligibility, prediction horizon, seizure occurrence period and the rule for selecting the matched warning will be specified before outcome analysis. Performance will be evaluated by internal validation on held-out recordings, with participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Results will be reported separately by model approach, with variability across validation folds or repeated runs.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.
Percentage of evaluable recording time spent in seizure warning
The total duration of the model-defined seizure warning state divided by the total evaluable recording duration, multiplied by 100. Overlapping warning intervals will contribute only once to total warning time. The duration and timing rules defining the warning state will be specified before outcome analysis and applied consistently. This measure will be reported separately for each prediction model alongside event-level sensitivity and false alarms per hour. Performance will be evaluated by internal validation on held-out recordings, with participant-level separation for patient-independent models and separation by seizure event and time for patient-specific models. Results will be reported separately by model approach, with variability across validation folds or repeated runs.
Time frame: During clinically indicated video-EEG monitoring, ordinarily 48-72 hours per epilepsy participant; performance will be assessed offline using held-out portions of these recordings.