Identifying risk levels for seizure activity based on sleep state

Abstract

The computer-implemented method can include receiving data indicative of electroencephalogram (EEG) activity over a particular period of time. The method can also include determining, based on the data, at least one metric associated with at least one sleep state of the subject. Further, the method can include determining, based on the at least one metric, a risk level associated with seizure activity of the subject. The method can further include generating an output indicative of the risk level.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of U.S. Provisional Application No. 63 / 484,570, entitled "Identifying Risk Level For Seizure Activity Based On Sleep States," filed February 13, 2023, which is incorporated herein by reference in its entirety.

[0002] FIELD OF THE DISCLOSURE The present disclosure relates generally to analyzing physiological data, and more particularly (though not necessarily exclusively) to identifying risk levels for seizure activity based on sleep state. [Background technology]

[0003] Electroencephalography (EEG) is a tool used to measure electrical activity generated by the brain. Functional brain activity is collected by electrodes placed on the subject's scalp. Traditional monitoring and diagnostic equipment involves several electrodes attached to the subject, tapping brain signals and transmitting the signals via cables to an amplifier unit. The resulting EEG signals can be used to diagnose and monitor various conditions affecting the brain, such as epilepsy.

[0004] Epilepsy is a brain disorder that causes seizures. There are many types of epilepsy, which can be diagnosed based on the type of seizure a subject suffers from. Examples of types of epilepsy include focal epilepsy, generalized epilepsy, and epilepsy of unknown etiology. A subject may experience more than one type of seizure, and seizures may manifest differently, for example, occurring with varying intensity, duration, symptoms, etc. Treatment for epilepsy may include drug therapy, but subjects may often be resistant to drug therapy. Furthermore, drug therapy may increase the likelihood of sleep deprivation or substance abuse, which may increase the likelihood of seizures. Thus, epilepsy can be a complex disorder that can be difficult to diagnose and treat.

[0005] It can be even more difficult to predict when a person with seizures will experience one. In some subjects, subtle changes in heart rate, electrical skin conductance, or breathing patterns can be indicators used to predict seizure activity just prior to seizure onset. Furthermore, conventional systems apply machine learning techniques to EEG data to extract features that can be used for seizure prediction. However, the complexity of seizure activity can make it difficult to extract features that can be used to accurately predict seizures in a population of subjects (e.g., especially for subjects outside the population used to train the machine learning techniques). Therefore, a holistic and accurate approach to seizure prediction is needed. Furthermore, there is a further need for predictions to occur sooner than just prior to seizure onset to enable preparation, intervention, or treatment. Summary of the Invention

[0006] Aspects of the present disclosure relate to identifying a risk level of seizure activity based on sleep states. One aspect relates to a computer-implemented method that includes receiving data indicative of electroencephalogram (EEG) activity over a period of time, determining at least one metric associated with at least one sleep state of the subject based on the data, determining a risk level associated with seizure activity of the subject based on the at least one metric, and generating an output indicative of the risk level.

[0007] In some embodiments, at least one metric is the amount of time in a particular sleep state, hi some embodiments, the particular sleep state is the rapid eye movement (REM) state of sleep.

[0008] In some embodiments, the data is received from an RF transceiver associated with the multi-electrode device. In some embodiments, the particular time period is a first time period. In some embodiments, the risk level is a prediction of the likelihood that the subject will experience seizure activity during a second time period. In some embodiments, the second time period occurs after the first time period.

[0009] In some embodiments, the output is a first output. In some embodiments, the method further includes identifying a treatment recommendation based on the risk level and generating a second output indicative of the treatment recommendation, the treatment recommendation usable to reduce the risk level.

[0010] In some embodiments, the at least one metric is a first metric. In some embodiments, determining the risk level based on the at least one metric further includes determining a second metric for a healthy population based on historical data, identifying a statistically significant difference between the first metric and the second metric based on a comparison of the first metric and the second metric, and determining the risk level based on the statistically significant difference.

[0011] In some embodiments, generating an output indicative of the risk level further includes providing an output in a first color indicative of a high risk level, providing an output in a second color indicative of a medium risk level, and providing an output in a third color indicative of a low risk level.

[0012] One aspect relates to a system including one or more data processors and a non-transitory computer-readable storage medium. The computer-readable storage medium includes instructions that, when executed on the one or more data processors, cause the one or more data processors to receive data indicative of electroencephalogram activity over a particular period of time, determine at least one metric associated with at least one sleep state of the subject based on the data, determine a risk level associated with seizure activity of the subject based on the at least one metric, and generate an output indicative of the risk level.

[0013] In some embodiments, at least one metric is the amount of time in a particular sleep state, hi some embodiments, the particular sleep state is the rapid eye movement (REM) state of sleep.

[0014] In some embodiments, the data is received from an RF transceiver associated with the multi-electrode device. In some embodiments, the particular time period is a first time period. In some embodiments, the risk level is a prediction of the likelihood that the subject will experience seizure activity during a second time period. In some embodiments, the second time period occurs after the first time period.

[0015] In some embodiments, the output is a first output. In some embodiments, the instructions, when executed on the one or more data processors, cause the one or more data processors to identify a treatment recommendation based on the risk level and generate a second output indicative of the treatment recommendation, the treatment recommendation usable to reduce the risk level.

[0016] In some embodiments, the at least one metric is a first metric. In some embodiments, determining the risk level based on the at least one metric further includes determining a second metric for a healthy population based on historical data, identifying a statistically significant difference between the first metric and the second metric based on a comparison of the first metric and the second metric, and determining the risk level based on the statistically significant difference.

[0017] In some embodiments, generating an output indicative of the risk level further includes providing an output in a first color indicative of a high risk level, providing an output in a second color indicative of a medium risk level, and providing an output in a third color indicative of a low risk level.

[0018] One aspect relates to a computer program product tangibly embodied in a non-transitory machine-readable storage medium, the computer program product including instructions to cause one or more data processors to receive data indicative of electroencephalogram activity over a particular period of time, determine at least one metric associated with at least one sleep state of the subject based on the data, determine a risk level associated with seizure activity of the subject based on the at least one metric, and generate an output indicative of the risk level.

[0019] In some embodiments, at least one metric is the amount of time in a particular sleep state, hi some embodiments, the particular sleep state is the rapid eye movement (REM) state of sleep.

[0020] In some embodiments, the data is received from an RF transceiver associated with the multi-electrode device. In some embodiments, the particular time period is a first time period. In some embodiments, the risk level is a prediction of the likelihood that the subject will experience seizure activity during a second time period. In some embodiments, the second time period occurs after the first time period.

[0021] In some embodiments, the output is a first output. In some embodiments, the instructions further cause the one or more data processors to identify a treatment recommendation based on the risk level and generate a second output indicative of the treatment recommendation, the treatment recommendation usable to reduce the risk level.

[0022] In some embodiments, the at least one metric is a first metric. In some embodiments, determining the risk level based on the at least one metric further includes determining a second metric for a healthy population based on historical data, identifying a statistically significant difference between the first metric and the second metric based on a comparison of the first metric and the second metric, and determining the risk level based on the statistically significant difference.

[0023] In some embodiments, generating an output indicative of the risk level further includes providing an output in a first color indicative of a high risk level, providing an output in a second color indicative of a medium risk level, and providing an output in a third color indicative of a low risk level. [Brief explanation of the drawings]

[0024] [Figure 1] FIG. 1 is a block diagram of an example of a system for acquiring physiological data according to an example of the present disclosure. [Figure 2] FIG. 1 is a block diagram of an example system for identifying a risk level of seizure activity based on sleep state, according to an example of the present disclosure. [Figure 3] FIG. 1 is a block diagram of an example of a computing system for identifying a risk level of seizure activity based on sleep state, according to an example of the present disclosure. [Figure 4] 1 is a flowchart of a process for identifying a risk level of seizure activity based on sleep state, according to an example of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0025] Certain aspects and examples of the present disclosure relate to systems and methods for predicting a risk level of seizure activity within a specific time frame based on an analysis of physiological data. The risk level may be the likelihood that a subject will exhibit seizure activity (e.g., within a predetermined period of time). Seizure activity may include neural activity consistent with a seizure (e.g., as collected via one or more electroencephalography (EEG) electrodes) and / or clinical symptoms consistent with a seizure. Various environmental, physiological, or other appropriate factors may be used to determine the risk level. The physiological data used to predict the risk level may include electroencephalography (EEG) data, electrocardiography (ECG) data, electromyography (EMG) data, electrooculography (EOG) data, or other suitable physiological data. The physiological data may be acquired via a physiological data acquisition assembly including at least a single channel of physiological data in close proximity to at least one reference electrode and at least one active electrode. The assembly may be worn by the subject. For example, the assembly may include a patch configurable to be positioned (e.g., adhered) on the user's forehead. Additionally, the patch may have an adhesive film to which electrodes may be attached to collect physiological data.

[0026] In one example of the present disclosure, the risk level of seizure activity can be identified based on EEG data indicative of sleep states. The sleep states can be any distinguishable sleep or wakefulness states that represent behavioral, physical, or signal characteristics. In some cases, the EEG data is processed to infer, for each of a plurality of time intervals, a category indicating a prediction as to whether the subject is awake or asleep, and potentially, if the subject is predicted to be asleep, a specific type or stage of sleep. The inference can be made based on transforming the time-domain electrical signals into frequency-domain intensity or power values ​​for each of the plurality of time intervals. Features may be defined as cumulative or maximum intensity or power values ​​within various frequency bands. The sleep state can then be inferred based on the absolute or relative values ​​of one or more features.

[0027] For example, wakefulness sleep states can be detected or defined by processing EEG data to detect signals within one or more specific frequency bands (e.g., bands spanning from about 13 to about 60 hertz (Hz)) and amplitudes of at least about 30 microvolts (μV) (i.e., beta waves). The frequency and amplitude can be determined by converting the time-domain electrical signals to the frequency domain via a mathematical transform (e.g., Fourier transform) or other appropriate technique. In some examples, additional sleep states can be characterized by stage 1, stage 2, stage 3, and rapid eye movement (REM). Frequency bands for detecting stage 1 sleep from EEG data can be defined to correspond to particular types of waves and / or sleep stages. For example, the frequency band may be defined to span from 3 to 8 Hz. Detection algorithms can be configured in the time domain or the frequency domain to detect signatures that support a prediction as to whether a subject is in a given sleep stage (e.g., stage 1 sleep). For example, if the amplitude of the 3-8 Hz band is 50-100 μV (i.e., theta waves), the subject may be inferred to be in stage 1 sleep. Additional characteristics of sleep states, such as sleep spindles and K-complexes, can be identified through detection algorithms to predict sleep states. For example, a high-frequency band lasting less than 2 seconds in the time domain (e.g., a frequency band of approximately 15 Hz) may be detected as a sleep spindle. Similarly, a low-frequency band lasting approximately 1 second in the time domain (e.g., a frequency band ranging from 1-4 Hz and an amplitude of 100 μV-200 μV) (i.e., delta waves) may be detected as a K-complex. Therefore, if one or more portions of EEG data are detected as sleep spindles and are followed by or otherwise near one or more portions of EEG data detected as K-complexes, the subject may be inferred to be in stage 2 sleep. In another example, a frequency band ranging from 1 to 4 Hz can be detected for significantly longer than 2 seconds (e.g., 20 minutes), which can predict that the subject was in stage 3 sleep. Stage 3 sleep can also be referred to as slow wave or delta sleep. Furthermore, a frequency band ranging from about 13 to about 60 hertz (Hz) and an amplitude of at least about 30 μV (i.e., beta waves) can predict that the subject was in REM sleep. However, beta waves can also be detected during awake sleep. Therefore, additional physiological data, physical or biological indicators, or other appropriate data can be acquired and identified within the detection algorithm to distinguish REM sleep from awake sleep. For example, EMG data can be acquired, and the detection algorithm can detect phasic events (e.g., rapid eye movements and limb twitching) or tonic phenomena (e.g., loss of antigravity muscle tone), both of which can indicate REM sleep. Detection of phasic events or tonic phenomena can be compared or combined with EEG data to distinguish REM sleep from awake sleep or another sleep state.

[0028] In some examples, sleep states may be characterized by REM sleep and non-REM sleep. For example, stage 1 sleep, stage 2 sleep, and stage 3 sleep may be combined to form non-REM sleep. Thus, a detection algorithm may detect non-REM sleep by inferring that EEG data outside the 13-60 hertz (Hz) range and having an amplitude above or below approximately 30 μV is non-REM sleep, and may detect REM sleep by inferring that data within the 13-60 Hz range and having an amplitude of approximately 30 μV is REM sleep.

[0029] EEG signals are typically examined temporally in successive increments called epochs. For example, when EEG signals are used to analyze sleep, the sleep can be segmented into one or more epochs for analysis. Epochs can be segmented into different sections using a scanning window, which defines different sections of the time series increments. The code can move the scanning window (incrementally or via a shift) through a sliding or shifting window, with the sections of the sliding window having overlapping or non-overlapping time series sequences. Epochs can alternatively span the entire time series, for example. In some examples, each epoch can be classified to correspond to a predicted sleep state it represents. In some cases, prior to classification, the epochs are normalized or binormalized based on (for example) frequency information, amplitude information, power, intensity, or other suitable features of the EEG data that can be correlated with sleep states. U.S. Patent Application No. 11 / 431,425, filed May 9, 2006, which is incorporated herein by reference for all purposes, discloses exemplary techniques for normalizing biological data.

[0030] In some cases, a given epoch may be classified as REM (or alternatively, non-REM) sleep based on whether the power (or normalized or binormalized power) in a given frequency band (e.g., in an absolute sense or relative to the power in one or more other frequency bands) exceeds a threshold. The threshold may be an absolute threshold, a threshold defined based on data from a population of subjects diagnosed with a given condition, a threshold defined based on data from a population of subjects experiencing a given condition, or a threshold defined based on empirical data associated with the subject.

[0031] Any of the same or similar techniques can be used to predict different types of sleep stages, for example, the intensity in one or more bands of normalized or doubly normalized epochs can be used to predict whether a subject is in stage 1, 2, 3, or REM sleep, whether there are spindles in the sleep data, whether there are k-complexes in the sleep data, etc.

[0032] Any group of epochs initially classified as one sleep state can be divided into multiple sub-classified sleep states according to increasing levels of classification detail. For example, a group of epochs classified as non-REM can be further divided into stage 1, stage 2, stage 3, or a combination thereof.

[0033] In some embodiments, artificial intelligence techniques can be used to predict whether a subject has a given sleep disorder, predict the severity of the sleep disorder, or predict the effectiveness of treating a given sleep disorder. The artificial intelligence techniques may include embedding signal processing (e.g., which may include applying one or more signal transformations) and using one or more models or rules to generate epoch-specific, night-specific, or subject-specific predictions. For example, EEG signals may be collected over a sleep period (e.g., a night). The EEG signals may be separated into epochs corresponding to absolute or relative time increments throughout the period (e.g., 1-minute, 5-minute, or 10-minute time intervals), and a spectrum may be generated for each epoch so that the power or intensity of each of the various frequency bands can be identified for each time increment. Alternatively, a spectrogram may be generated for the period, and the spectrogram identifies the power or intensity values ​​of the various frequency bands for each of multiple time increments within the period (e.g., 1-minute, 5-minute, or 10-minute time increments). The set of features can be defined such that the features indicate or correspond to (1) the power, intensity, or other suitable attributes of a frequency band of a spectrogram or spectrum, and / or (2) the intensity of a spectrum or spectrogram, a derivation of the spectrum (e.g., over time), or a dual derivation of the spectrum or spectrogram (e.g., over time, across frequency, or across both time and frequency). The features can predict (for example) the likelihood of a particular sleep stage or state (e.g., REM or NREM). In some cases, the features are defined based on other epoch-specific features. For example, subject- and / or day- or night-specific features may indicate the percentage of time or percentage of sleep predicted to be in the REM state.

[0034] Artificial intelligence rules can be defined to predict REM sleep deprivation and / or seizure recurrence tendency based on the characteristics. For example, clustering techniques, support vector machines (SVM), principal components, independent components, logistic regression, etc. can be used to predict whether a subject is in REM sleep (vs. non-REM sleep or wakefulness) for each period epoch. In some cases, a likelihood that the subject is in REM sleep for each epoch can be generated and then compared to a predetermined or learned threshold to predict whether the subject is or was in REM sleep.

[0035] A rule can be defined to predict whether a subject is sleep-deprived (or was sleep-deprived) based on REM sleep prediction. For example, the rule can indicate that a subject is sleep-deprived (or was sleep-deprived) if less than a threshold percentage (e.g., 10%, 15%, 20%, 25%, 30%, or 35%) of epochs are predicted to be REM sleep. In another example, the rule can indicate that a subject is sleep-deprived or sleep-deprived by identifying that the length of REM sleep predicted by the detection algorithm for one or more epochs is less than a predetermined threshold for healthy or normal sleep or a learned threshold for the subject. An additional rule can indicate that a subject is sleep-deprived or sleep-deprived if the length or percentage of REM sleep does not increase by at least a threshold amount (e.g., 1 minute, 5 minutes, etc., or 5%, 10%, etc.) for subsequent sleep cycles (i.e., a series of sleep cycles for one night's sleep).

[0036] To identify risk levels, metrics can be extracted from EEG data for sleep states and correlated with their impact on seizure activity. For example, metrics can be the mean, variance, skewness, etc., of frequency bands, amplitudes, or other suitable features of the EEG data. In other examples, metrics can be statistics derived from an epoch or set of epochs, such as the percentage of sleep predicted to be REM sleep. Metrics can also be extracted from EMG data, EOG data, ECG data, or other suitable physiological data obtained during sleep periods. For example, metrics can include heart rate, blood oxygen levels, eye or leg movements, etc.

[0037] Additionally, in some examples, subject characteristics can be further used to correlate the metric with an impact on seizure activity. For example, the metric can be a ratio of REM to non-REM sleep, and the metric can be correlated with an impact on seizure activity based on the subject's age. Additional subject characteristics may include medications or medication dosages prescribed to the subject, previously diagnosed sleep disorders for the subject, the subject's gender, etc.

[0038] In certain examples, sleep deprivation can be detected based on predefined rules, and a metric extracted from the EEG data can be associated with the rule. For example, the predetermined rule can include detecting sleep deprivation if the percentage of REM sleep associated with a set of epochs is below a threshold percentage. Thus, the metric can be the percentage of REM sleep. A classification algorithm (e.g., a k-nearest neighbor algorithm, a decision tree, a support vector machine, etc.) or another appropriate algorithm can then be implemented to predict a risk level based on the percentage of REM sleep. Additional metrics can be extracted from the set of epochs and provided for use in the classification algorithm. The additional metric can be other appropriate indicators of sleep deprivation (e.g., the length of time the subject is predicted to be asleep). The classification algorithm can also accept characteristics of the subject (e.g., age). The classification algorithm can then indicate a risk level of seizure activity for the subject based on the percentage of REM sleep, the additional metric, and / or the subject's characteristics. For example, the classification algorithm can indicate the risk level by classifying the risk level as high, medium, or low.

[0039] Furthermore, a typical sleep cycle of a healthy subject detected and analyzed via EEG data can be determined to last approximately 90 to 110 minutes. For a typical sleep cycle, the chronological order of each sleep state epoch may indicate that the typical sleep cycle occurs as stage 1, stage 2, stage 3, stage 4, and then REM. Furthermore, for EEG data collected over a typical night's sleep of a healthy subject, the EEG data may include four to five sleep cycles, during which the time associated with REM sleep may increase with each subsequent sleep cycle. Furthermore, for a typical sleep cycle, approximately 5% of the acquired EEG data may be associated with stage 1, approximately 45% with stage 2, approximately 25% with stage 3, and approximately 25% with REM. Therefore, approximately 75% of the EEG data for a typical sleep cycle may be associated with non-REM sleep, and approximately 25% with REM sleep.

[0040] Therefore, because abnormal sleep activity (e.g., sleep deprivation) is a common trigger for seizure activity and typical sleep activity can be clearly defined, EEG data associated with sleep states can provide a comprehensive mechanism for predicting the risk level of seizure activity. Sleep states can be predicted based on frequency bands, amplitudes, or other suitable features of the EEG data. Artificial intelligence techniques or rules can then be implemented to predict sleep deprivation for other suitable abnormal sleep activity. Furthermore, metrics can be derived from data associated with the prediction of sleep state and / or sleep deprivation and then classified (i.e., by a classification algorithm) to predict the subject's risk level of seizure activity. Furthermore, abnormal sleep activity may not immediately trigger seizure activity, and therefore the risk level of seizure activity can be predicted based on EEG data for a certain time frame following the abnormal sleep activity. Thus, examples of the present disclosure can provide an accurate and comprehensive method for predicting the risk level of seizure activity, and can further predict the risk level for a specific time frame to enable preparation for or prevention of seizure activity.

[0041] The illustrative examples are provided to introduce the reader to the general subject matter discussed herein and are not intended to limit the scope of the disclosed concepts. The following section describes various additional features and examples with reference to the drawings, in which like numerals indicate like elements and directional descriptions are used to explain the illustrative embodiments, but as with the illustrative embodiments, should not be used to limit the present disclosure.

[0042] FIG. 1 is a block diagram of an example system for acquiring physiological data according to an example of the present disclosure. The system 100 may include a multi-electrode device 104 that may have one or more active electrodes 106a for collecting active signals and one or more reference electrodes 106b that may collect respective reference signals. Additionally, the multi-electrode device 104 may include a ground electrode 106c. The electrodes 106a-106c may be fixed in place within the device (e.g., patch 102) or may be movable (e.g., tethered to the device). The system 100 may further include a processing subsystem 116, a memory subsystem 118, a (radio frequency) RF transceiver 114, a connector interface 112, a power subsystem 108, and an environmental sensor 120, each of which may be communicatively coupled to or a portion of the multi-electrode device 104.

[0043] The processing subsystem 116 can be implemented as one or more integrated circuits, such as one or more single-core or multi-core microprocessors or microcontrollers, examples of which are known in the art. The processing subsystem 116 can control the operation of the multi-electrode device 104 by executing various programs in response to program code and may maintain multiple simultaneously running programs or processes. For example, the processing subsystem 116 can execute code that can control the collection, analysis, application, and / or transmission of physiological data (e.g., electroencephalography (EEG) data, electromyography (EMG) data, etc.). Some or all of the program code can be stored in the processing subsystem 116, or the program code can be stored in a storage medium such as the storage subsystem 118. Additionally, the processing subsystem 116 may amplify, filter, or combine signals detected by the electrodes 106a-106c of the multi-electrode device 104, and further store the signals along with recording details (e.g., recording time or user identifier). In some examples, the processing subsystem 116 can analyze the physiological data or signals to detect physiological correspondence. For example, the recorded signal may reveal frequency characteristics corresponding to sleep stages.

[0044] Additionally, the storage subsystem 118 may be implemented using, for example, magnetic storage media, flash memory, other semiconductor memory (e.g., DRAM, SRAM), or any other non-transitory storage medium or combination of media, and may include volatile and / or non-volatile media. In some examples, the storage subsystem 118 may store physiological data, information about the subject (e.g., identification information or medical history information), or analysis variables derived from the physiological data (e.g., frequency, amplitude, etc.). The storage subsystem 118 may also store one or more programs that may be executed by the processing subsystem 116. The one or more programs may initiate or control the collection, analysis, or transmission of the physiological data.

[0045] The RF transceiver 114 can enable the multi-electrode device 104 to wirelessly communicate with various interfacing devices, such as phones, tablets, laptops, etc. The RF transceiver 114 can include a combination of hardware components, including, for example, driver circuits, antennas, modulators, demodulators, encoders, decoders, and other suitable analog and / or digital signal processing circuitry, and can also include software components. Various wireless communication protocols can be implemented via the RF transceiver 114 using the software components and associated hardware. The RF transceiver component of the RF transceiver 114 can include an antenna and supporting circuitry to enable data communication over a wireless medium, such as Wi-Fi, Bluetooth, or other suitable medium for wireless data communication.

[0046] The connector interface 112 can enable the multi-electrode device 104 to communicate with various interface devices over a wired communication path, for example, using a universal serial bus (USB), a universal asynchronous receiver / transmitter (UART), or other protocols for wired data communication. In some examples, the connector interface 112 can provide a power port to allow the multi-electrode device 104 to receive power. The connector interface 112 may also provide a connection for transmitting or receiving physiological data. For example, the physiological data can be transmitted to or from another device, such as another multi-electrode device, in analog or digital format.

[0047] The environmental sensors 120 may include a variety of electronic, mechanical, electromechanical, optical, or other devices that provide information about external conditions surrounding the multi-electrode device 104 or regarding the subject. Any type and combination of environmental sensors 120 may be used. For example, an accelerometer may be used to estimate whether the user is sleeping or attempting to sleep, or to otherwise estimate activity state. In another example, an electro-oculogram sensor may be used to detect eye movement and assist in identifying rapid eye movement (REM) sleep stages.

[0048] Additionally, the power subsystem 108 can provide power and power management functions for the multi-electrode device 104. For example, the power subsystem 108 can include the battery 110 and associated circuitry that distributes power to other components of the system 100 that may require power from the battery 110.

[0049] It will be understood that system 100 is illustrative and that variations and modifications are possible. In one example, processing subsystem 116 can execute code from memory subsystem 118 to analyze sleep states based on EEG data and predict a risk level of seizure activity based on the analysis. Accordingly, system 100 can further include a user interface that allows a user to directly interact with the device to, for example, receive a risk level. The risk level can be the likelihood that a subject will experience seizure activity for a particular time frame. The risk level can be displayed in the user interface using a color that indicates, for example, whether the risk level is high, medium, or low, or the risk level can be displayed as a percentage or another suitable format. Furthermore, while system 100 is described with reference to particular blocks, it should be understood that these blocks are defined for convenience of explanation and are not intended to imply a particular physical arrangement of components.

[0050] 2 is a block diagram of an example system 200 for identifying a risk level 202 for seizure activity based on a sleep state 216, according to an example of the present disclosure. The risk level 202 may be the likelihood (e.g., within a predetermined time period) that a subject will experience seizure activity. The system 200 may include a computing device 201 that can be communicatively coupled to a display device 220 and a multi-electrode device 204 for identifying the risk level 202 based on physiological data indicative of the sleep state 216 and providing the risk level 202 to the subject, a physician, or another suitable user. The physiological data may include electroencephalogram (EEG) data 232, electromyogram (EMG) data, electrocardiogram (ECG) data, electrooculogram (EOG) data, or other suitable physiological data. The computing device 201 may communicate with the display device 220 and the multi-electrode device 204 via a network 230, such as a local area network (LAN) or the Internet.

[0051] In some examples, the computing device 201 can receive physiological data from the multi-electrode device 204 or other suitable devices or sensors. In one example, the multi-electrode device 204 can correspond to the multi-electrode device 104 of FIG. 1 . Accordingly, the computing device 201 can receive physiological data, such as EEG data 232, from the RF transceiver 114 associated with the multi-electrode device 204. The EEG data 232 can indicate the subject's brainwave activity 238. The multi-electrode device 204 can also be associated with environmental sensors, such as an accelerometer, ECG, EOG, EMG, etc., to collect additional physiological data. Additionally, the computing device 201 can receive EEG data 232 for a first period 222 a, such as a sleep cycle, a series of sleep cycles (e.g., four subsequent cycles), etc. In addition to recording EEG data 232, the multi-electrode device 204 may filter, amplify, transmit, or otherwise perform operations on the EEG data 232 or additional physiological data to improve uptake of the EEG data 232 or additional physiological data at the computing device 201, thereby improving the efficiency and accuracy of identifying the risk level 202.

[0052] Additionally, the computing device 201 may analyze the EEG data 232 or additional physiological data to detect or otherwise distinguish between sleep states 216. For example, the computing device 201 may collect EEG data 232 for a first time period 222a (i.e., 8 hours of sleep time) via the multi-electrode device 204 and segment the EEG data 232 for the first time period 222a into epochs, for example, via a scanning window. The epochs may be analyzed by the computing device 201 to predict a sleep state associated with each epoch. The epochs may first be segmented into many shorter time segments (e.g., 1-minute, 3-minute, or 5-minute time segments) and then reorganized based on the predicted sleep states.

[0053] To analyze the epochs, the computing device 201 may further perform a fast Fourier transform (FFT) or other suitable mathematical transform, algorithm, or technique to transform the EEG data 232 of each epoch into the frequency domain. Once the EEG data 232 is transformed into the frequency domain, additional features such as peak amplitude, peak frequency, median frequency, etc. can be extracted. In some cases, the epochs may be further normalized or double-normalized based on (for example) frequency information, amplitude information, power, intensity, or other suitable features of the EEG data. Additionally, the computing device 201 may perform a short-time Fourier transform (STFT) or other suitable technique to identify time-frequency domain features from the EEG data 232, which may depend on both time-domain and frequency-domain characteristics. For example, the computing device 201 may perform an STFT to extract time-frequency domain features such as average power, entropy, and partial energy.

[0054] After analysis, frequency information, amplitude information, time-frequency domain features, etc. associated with each epoch can be correlated with sleep states to predict the sleep state of each epoch. For example, the sleep states 216 can be characterized or defined as wakefulness, stage 1, stage 2, stage 3, and rapid eye movement (REM), or the sleep states 216 can be characterized or defined as REM and non-REM. The computing device 201 can detect the sleep states 216 based on time-domain features, frequency-domain features, time-frequency domain features, other suitable features, or combinations thereof identified by the computing device 201 based on the EEG data 232. The features of the EEG data 232 can indicate characteristics or changes in brainwave activity 228 that can correspond to the sleep states 216.

[0055] In some examples, a detection algorithm may be configured by the computing device 201 in the time or frequency domain to detect a signature supporting a sleep state prediction. For example, the brain wave activity 228 of a stage 3 sleep state may include delta waves, which may exhibit low frequency (i.e., a frequency band ranging from 1 to 4 Hz) and high amplitude (i.e., an amplitude ranging from 100 μV to 200 μV) in the frequency domain of the EEG data 232. Thus, if a first epoch includes a frequency band from 1 to 4 Hz and an amplitude from 100 μV to 200 μV, the first sleep state of the first epoch may be predicted to be stage 3.

[0056] In another example, a classification algorithm (i.e., K-nearest neighbors, SVM, or other suitable classification algorithm) can be configured to predict the sleep state of each epoch by classifying the epoch based on a signature that supports the sleep state prediction. The classification algorithm can be trained to predict sleep states by inputting epochs labeled with sleep states, or the classification algorithm can classify sleep states based on predetermined rules. Thus, for example, the classification algorithm may classify a second epoch having an amplitude in the 3-8 Hz frequency band of 50-100 μV (i.e., theta waves) as stage 1 sleep.

[0057] After the sleep state for each epoch is predicted, the computing device 201 may further determine metrics 206a-206b associated with the sleep state 216. In some examples, the metrics 206a-206b may be determined based on frequency patterns, amplitudes, or other suitable features extractable by the computing device 201 from the EEG data 232 and corresponding to the sleep state 216. For example, the metrics 206a-206b may be frequency band power, intensity, or other suitable attributes. The metrics 206a-206b may be averaged, summed, or otherwise determined for one or more epochs associated with a particular sleep state. Furthermore, in some examples, the metrics 206a-206b may be statistics obtained based on the predicted sleep state. For example, the metrics 206a-206b may be the ratio of REM to non-REM sleep, the percentage of time spent in a particular sleep state, etc. In a particular example, the first metric 206a may be the cumulative amount of time the subject was predicted to be in REM sleep over the first time period 222a. The first metric 206a may be determined by adding the duration associated with each epoch in which the subject was predicted to be in REM sleep. The first metric 206a may be an absolute amount of time (e.g., 20 minutes) or a relative amount of time (e.g., 8% over the first time period 222a).

[0058] In another example, the first metric 206a may be an estimated amount of REM sleep for a sleep cycle or series of sleep cycles. Thus, the epochs of the first period 222a may be further grouped or organized by sleep cycle, and for each sleep cycle, the predicted duration of the subject's REM sleep state may be determined. The first metric 206a may also be the mean, median, mode, etc. of REM sleep over the series of sleep cycles.

[0059] Additionally, the second metric 206b may be the amount of time spent sleeping (e.g., accumulation of stage 1, stage 2, stage 3, and REM). In some examples, the metrics 206a-206b may further include the amount of time spent in stage 1, stage 2, stage 3, or a combination thereof. Additionally, the metrics 206a-206b may include characteristics of the subject's sleep cycle (e.g., the length of the sleep cycle or the percentage of time spent in each sleep state of the sleep cycle) or other suitable metrics that can be calculated, estimated, or otherwise identified based on the EEG data 232 or other suitable physiological data.

[0060] The computing device 201 can further determine the risk level 202 based on the metrics 206a-206b. In some examples, the computing device 201 can implement artificial intelligence techniques that can be used to predict the risk level 202 based on the metrics 206a-206b. The artificial intelligence techniques can include using one or more models or rules to generate subject-specific predictions based on the metrics 206a-206b derived from one or more epochs of the first period 222a. The rules can be defined to predict REM sleep deprivation and / or seizure recurrence tendency based on the metrics 206a-206b. For example, a rule can be defined to predict whether a subject is (or was) sleep-deprived. For example, the rule can indicate that a subject is (or was) sleep-deprived if less than a threshold percentage of epochs (e.g., 10%, 15%, 20%, 25%, 30%, or 35%) are predicted to be REM sleep. In another example, the rule may indicate that a subject has been sleep deprived or is sleep deprived by identifying that the length of REM sleep predicted by the detection algorithm for one or more epochs is below a predetermined threshold for healthy or normal sleep or for a learned threshold for the subject. Artificial intelligence techniques may further make subject-specific predictions based on additional characteristics of the subject (i.e., age, previous diagnosis of a sleep or seizure-related disorder, prescribed medications, etc.).

[0061] In a particular example, the first metric 206a may be the relative amount of time the subject is predicted to be in REM sleep over the first time period 222a. The threshold amount of REM sleep may be predefined based on an age group. The threshold amount of REM sleep may be a gradient such that multiple threshold amounts of REM sleep correspond to different risk levels. For example, for an age group between 30 and 50, the threshold amounts of REM sleep may be 25%, 20%, and 15% relative time. Thus, to illustrate risk level prediction, if the subject is 35 years old and the subject's predicted relative time of REM sleep over the first time period 222a is 14%, the computing device 201 may predict a high risk level for the subject.

[0062] Additionally or alternatively, the computing device 201 may determine the risk level 202 based on historical data 236. For example, the historical data 236 may be previous EEG data associated with the subject's sleep (e.g., while the subject was undergoing different medications), or EEG data collected for a healthy population, a population of subjects with a seizure history, a population of subjects diagnosed with epilepsy, or another suitable population for which physiological data can be obtained and for which metrics 208a-208b can be determined. Metrics 208a-208b may be the same metrics as metrics 206a-206b or may otherwise reflect the same characteristics of the EEG data 232. Metrics 208a-208b may be labeled with seizure occurrences for a predetermined period of time or may otherwise include indications of seizure activity. Thus, metrics 208a-208b, historical data 236, or a combination thereof, may be used to train artificial intelligence techniques or to generate predefined rules for artificial intelligence techniques.

[0063] In response to determining the risk level 202, the computing device 201 may generate a first output 224a indicative of the risk level 202. The first output 224a may be output for display on the display device 220. The first output 224a may indicate a high risk level, a medium risk level, or a low risk level. In some examples, a likelihood associated with the risk level may also be displayed. For example, a low risk level may be associated with a likelihood of less than 33%, a medium risk level may be associated with a likelihood of greater than 33% and less than 66%, and a high risk level may be associated with a likelihood of greater than 66%. The likelihood may be a likelihood 210 of seizure activity over a second period 222b. For example, the second period 222b may be an eight-hour period occurring after the first period 222a. Further, in some examples, the first output 224a may provide a high risk level in a first color, a medium risk level in a second color, and a low risk level in a third color. In other examples, the first output 224 a may include a percentage of the likelihood 210 of seizure activity or otherwise indicate the risk level 202 on the display device 220 .

[0064] Additionally or alternatively, the computing device 201 may generate a second output 224b that may include a treatment recommendation 226. The computing device 201 may determine the treatment recommendation 226 based on the subject's EEG data 232, additional physiological data, the risk level 202, an electronic health record, other suitable data, or a combination thereof. The treatment recommendation 226 may be a medication, a dosage of a medication, or another appropriate treatment recommendation that may reduce the risk level 202 for the subject.

[0065] 3 is a block diagram of an example computing system 300 for identifying a risk level of seizure activity based on sleep state, according to an example of the present disclosure. Computing system 300 includes a processor 303 communicatively coupled to a memory device 305. In some examples, processor 303 and memory device 305 can be part of the same computing device, such as a server 310. In other examples, processor 303 and memory device 305 can be distributed (e.g., remotely) from one another.

[0066] The processor 303 may include one processor or multiple processors. Non-limiting examples of the processor 303 include a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or a microprocessor. The processor 303 may execute instructions 307 stored in the memory device 305 to perform operations. The instructions 307 may include processor-specific instructions generated by a compiler or interpreter from code written in any suitable computer programming language, such as C, C++, C#, Java, or Python.

[0067] The memory device 305 may include one memory or multiple memories. The memory device 305 may be volatile or nonvolatile. Nonvolatile memory includes any type of memory that retains stored information when power is off. Examples of the memory device 305 include electrically erasable programmable read-only memory (EEPROM) or flash memory. At least a portion of the memory device 305 may include a non-transitory computer-readable medium from which the processor 303 can read instructions 307. The non-transitory computer-readable medium may include electronic, optical, magnetic, or other storage devices that can provide computer-readable instructions or other program code to the processor 303. Examples of non-transitory computer-readable media include magnetic disks, memory chips, ROM, random access memory (RAM), ASICs, configured processors, and optical storage devices.

[0068] The processor 303 may execute the instructions 307 to perform operations. For example, the processor 303 may receive data 308 indicative of brainwave activity over a period of time 312. The processor 303 may also determine a metric 314 associated with the subject's sleep state 316 based on the data 308. Further, the processor 303 may determine a risk level 318 associated with the subject's seizure activity 320 based on the metric 314. The processor 303 may further generate an output 322 for display on the display device 304. The output 322 may indicate the risk level 318.

[0069] 4 is a flowchart of a process for identifying a risk level 202 of seizure activity based on a sleep state 216, according to an example of the present disclosure. In some examples, the processor 303 may implement some or all of the steps shown in FIG. 4. Other examples may include more, fewer, different, or differently ordered steps than those shown in FIG. 4. The steps of FIG. 4 are described below with reference to the components described above in connection with FIGS. 2 and 3.

[0070] In block 402, the processor 303 may receive data indicative of brainwave activity 228 over a particular period of time. The data may be electroencephalogram (EEG) data obtained via the multi-electrode device 204. The data may be received by the processor 303 from an RF transceiver associated with the multi-electrode device 204. In some examples, the particular period of time may be a first period 222a that is a predetermined period of time (e.g., an 8-hour period) or may be defined by a number of sleep cycles (e.g., four sleep cycles). Additionally, the processor 303 may temporally segment the data into epochs. An epoch may be any suitable length of time (e.g., 3 minutes, 5 minutes, 1 hour, etc.).

[0071] In block 404, the processor 303 can determine at least one metric 206a-206b associated with at least one sleep state of the subject based on the data. For example, the sleep states 216 can be characterized or defined by wakefulness, stage 1, stage 2, stage 3, and rapid eye movement (REM), or the sleep states 216 can be characterized or defined by REM and non-REM. The sleep states 216 can be predicted via a detection algorithm, classification algorithm, or other suitable algorithm executed by the processor 303 based on frequency patterns, amplitude, or other suitable characteristics of the EEG data 232. In some examples, a sleep state can be predicted for each epoch of the first period 222a. The epochs can then be reorganized based on the sleep states to which they are associated.

[0072] After predicting the sleep states 216, the processor 203 can derive metrics 206a-206b from the EEG data 232 associated with the sleep states 216. For example, the first metric 206a can be the amount of time in a particular sleep state (e.g., REM). The amount of time in a particular sleep state can be accumulated over a predetermined period of time. In some examples, the amount of time in a particular sleep state can be the average amount of time in a particular sleep state per sleep cycle based on a series of sleep cycles. Additionally, the metrics 206a-206b may include the predicted sleep states or other suitable statistics derived from the EEG data 232. For example, the metrics 206a-206b can be the ratio of REM to non-REM sleep, the amount of time spent sleeping (e.g., stage 1, stage 2, stage 3, and REM), the average frequency of brain waves during a particular sleep stage, the average amplitude of brain waves during a particular sleep stage, etc.

[0073] In block 406, the processor 303 may determine a risk level 202 associated with the subject's seizure activity based on at least one metric 206a-206b. In some examples, the processor 303 may implement artificial intelligence techniques to predict the risk level 202 based on the metrics 206a-206b. The artificial intelligence techniques may include using one or more models or rules to generate a subject-specific prediction based on the metrics 206a-206b derived from one or more epochs of the first period 222a. The rules may be defined to predict REM sleep deprivation, other suitable abnormal sleep patterns, and / or seizure recurrence propensity based on the metrics 206a-206b. For example, a rule may be defined to predict whether the subject is (or has been) sleep-deprived. For example, the rule may indicate that the subject is (or has been) sleep-deprived if less than a threshold percentage (e.g., 10%, 15%, 20%, 25%, 30%, or 35%) of the epochs are predicted to be REM sleep. The rules can further define thresholds based on the subject's age or other appropriate characteristics. The rules can also include gradients (i.e., multiple thresholds at various levels) that can correspond to risk levels. Thus, the rules can enable the processor to predict the risk level 202 based on the first metric 206a.

[0074] In some examples, the historical data 236 may be EEG data of a healthy population, a population with a history of seizure activity, etc., or the historical data 236 may be EEG data previously obtained for the subject. Accordingly, the processor 303 may further use the historical data 236 to train artificial intelligence techniques or to generate predetermined rules, thresholds, etc. for determining the risk level 202.

[0075] In block 408, the processor 303 may generate an output 322 indicating the risk level 202. The output 322 may indicate a high risk level, a medium risk level, or a low risk level. In some examples, a low risk level may be indicated by an output of a first color, a medium risk level may be indicated by an output of a second color, and a high risk level may be indicated by an output of a third color. The output 322 may also include a metric 206a-206b associated with the predicted sleep state, a confidence value for the risk level, a likelihood value (i.e., the percentage chance of seizure activity), or other suitable indicator of the risk level 202. Further, in some examples, the processor 303 may generate a second output indicating the second time period. Additionally or alternatively, the processor 303 may identify a treatment recommendation based on the risk level 202 and may include the treatment recommendation in the output 322. The treatment recommendation can be used to reduce the subject's risk level 202.

[0076] The foregoing description of specific examples, including the illustrated examples, has been presented for purposes of illustration and description only and is not intended to be exhaustive or to limit the disclosure to the precise form disclosed. Numerous modifications, adaptations, and uses thereof will be apparent to those skilled in the art without departing from the scope of the disclosure.

Claims

1. receiving data indicative of brainwave activity over a specified period of time; determining at least one metric associated with at least one sleep state of the subject based on the data; determining a risk level associated with seizure activity of the subject based on at least one metric; generating an output indicative of the risk level.

2. The computer-implemented method of claim 1 , wherein the at least one metric is the amount of time in a particular sleep state.

3. The computer-implemented method of claim 2 , wherein the particular sleep state is a rapid eye movement (REM) state of sleep.

4. The computer-implemented method of any one of claims 1 to 3, wherein the data is received from an RF transceiver associated with a multi-electrode device.

5. 5. The computer-implemented method of claim 1, wherein the particular time period is a first time period and the risk level is a prediction of the likelihood that the subject will experience the seizure activity for a second time period, the second time period occurring after the first time period.

6. the output is a first output; identifying a treatment recommendation based on said risk level; 6. The computer-implemented method of claim 1, further comprising: generating a second output indicative of the treatment recommendation, the treatment recommendation usable to reduce the risk level.

7. the at least one metric is a first metric, and determining the risk level based on the at least one metric; determining a second metric for the healthy population based on historical data; identifying a statistically significant difference between the first metric and the second metric based on a comparison of the first metric and the second metric; The computer-implemented method of claim 1 , further comprising: determining the risk level based on the statistically significant difference.

8. generating the output indicative of the risk level further comprises: providing said output in a first color indicating a high risk level; providing said output in a second color indicating a medium risk level; and providing the output in a third color, the third color indicating a low risk level.

9. one or more data processors; When executed on the one or more data processors, it causes the one or more data processors to: receiving data indicative of brainwave activity over a specified period of time; determining at least one metric associated with at least one sleep state of the subject based on the data; determining a risk level associated with seizure activity for the subject based on the at least one metric; generating an output indicative of the risk level. and a non-transitory computer-readable storage medium containing:

10. The system of claim 9 , wherein the at least one metric is the amount of time in a particular sleep state.

11. The system of claim 10 , wherein the particular sleep state is a rapid eye movement (REM) state of sleep.

12. The system of any one of claims 9 to 11, wherein the data is received from an RF transceiver associated with a multi-electrode device.

13. 13. The system of claim 9, wherein the particular time period is a first time period and the risk level is a prediction of the likelihood that the subject will experience the seizure activity for a second time period, the second time period occurring after the first time period.

14. the output is a first output; identifying a treatment recommendation based on said risk level; 14. The system of claim 9, further comprising: generating a second output indicative of the treatment recommendation, the treatment recommendation usable to reduce the risk level.

15. the at least one metric is a first metric, and determining the risk level based on the at least one metric; determining a second metric for the healthy population based on historical data; identifying a statistically significant difference between the first metric and the second metric based on a comparison of the first metric and the second metric; The system of any one of claims 9 to 14, further comprising: determining the risk level based on the statistically significant difference.

16. generating the output indicative of the risk level further comprises: providing said output in a first color indicating a high risk level; providing said output in a second color indicating a medium risk level; and providing the output in a third color to indicate a low risk level.

17. one or more data processors, receiving data indicative of brainwave activity over a specified period of time; determining at least one metric associated with at least one sleep state of the subject based on the data; determining a risk level associated with seizure activity for the subject based on the at least one metric; A computer program product tangibly embodied in a non-transitory machine-readable storage medium comprising instructions configured to cause generation of an output indicative of said risk level.

18. 18. The computer program of claim 17, wherein the at least one metric is the amount of time in a particular sleep state.

19. 20. The computer program of claim 18, wherein the particular sleep state is the rapid eye movement (REM) state of sleep.

20. A computer program according to any one of claims 17 to 19, wherein the data is received from an RF transceiver associated with a multi-electrode device.

21. 21. The computer program of claim 17, wherein the particular time period is a first time period and the risk level is a prediction of the likelihood that the subject will experience the seizure activity for a second time period, the second time period occurring after the first time period.

22. the output is a first output; identifying a treatment recommendation based on said risk level; 22. The computer program of claim 17, further comprising generating a second output indicative of the treatment recommendation, the treatment recommendation usable to reduce the risk level.

23. the at least one metric is a first metric, and determining the risk level based on the at least one metric; determining a second metric for the healthy population based on historical data; identifying a statistically significant difference between the first metric and the second metric based on a comparison of the first metric and the second metric; The computer program of any one of claims 17 to 22, further comprising determining the risk level based on the statistically significant difference.

24. generating the output indicative of the risk level further comprises: providing said output in a first color indicating a high risk level; providing said output in a second color indicating a medium risk level; and providing the output in a third color, the third color indicating a low risk level.

Images