WO2019177446A1

WO2019177446A1 - Warning system for epilepsy alarm

Info

Publication number: WO2019177446A1
Application number: PCT/MX2018/000026
Authority: WO
Inventors: Jesús Guillermo SERVÍN AGUILAR; Luis RIZO DOMÍNGUEZ; Jorge Arturo PARDIÑAS MIR
Original assignee: Instituto Tecnológico Y De Estudios Superiores De Occidente, A.C.
Priority date: 2018-03-14
Filing date: 2018-03-15
Publication date: 2019-09-19
Also published as: MX2018003153A

Abstract

The invention relates to a warning system for epilepsy alarm, characterised in that it comprises means for taking samples from electroencephalogram signals (EEG) of a person who suffers from epileptic fits, connected to a biomedical signal amplifier and to a processing card with a microcontroller and a memory where the EEG samples of the patient are stored; a module for low-pass filtering of EEG signals with a cut-off frequency of 40 Hz and of order 20 for suppressing the noise; said microcontroller also includes means for calculating the scattering parameter of the filtered EEG signal, in which the data are sorted and the values of the 25th and 75th quantiles of the filtered EEG signals are calculated, which are associated with the previously defined scattering parameter; wherein the scattering values within the system are entered into a low-pass filtering module with a cut-off frequency of 0.5 Hz and a filtering order with a factor of 20; an analyser which analyses the processed signals to find epileptic fits in real time; said analyser conducts a sample-by-sample comparison of each of the filtered scattering values, having two detection thresholds when an epileptic fit occurs or when the fit has ended, using visual and/or acoustic alarms.

Description

ALERT SYSTEM FOR EPILEPSY NOTICE

FIELD OF THE INVENTION

The present invention is related to the medical industry in general, in particular it relates to the field of medical systems and devices used in the measurement of physiological variables for the detection of problems and for the timely warning of the occurrence of an event, imbalance or physiological involvement. More specifically it refers to an alert system for the announcement of an epilepsy event.

BACKGROUND OF THE INVENTION

The brain activity of a person involves the generation of brain waves derived from the electrical activity of neurons; brain waves can be classified into different types according to their frequency, that is, the time that has a certain periodicity associated with the electrical impulse produced through the neurons, among them we find the Delta waves (from 1 to 3 Hz), the Theta waves (3.5 to 7.5 HZ), Alpha waves (8 to 13 Hz), Beta waves (12 to 33 Hz) and Gamma waves (25 to 100Hz). These waves can be detected by some devices such as the electroencephalograph that generates electroencephalography signals (EEG) that can be collected, processed and interpreted for multiple purposes, such as to provide alerts of neurological events that occur in a subject as epileptic seizures, to detect at least some physiological parameter of a subject while the subject sleeps, among others.

At present, the acquisition of electroencephalographic signals is not yet accessible, in terms of drinkability and low cost.

Currently the main method to detect an epileptic attack is by an electroencephalograph. These devices measure the patient's brain electrical waves. When a patient has an epileptic seizure, the signal changes in amplitude and frequency suddenly. Currently, there are several devices on the market that use EEG to measure brain waves.

According to S Ramgopal et al. ("Seizure detection, seizure p red iction, and close-loop warning systems in epilepsy", Epilepsy & Behavlor Journal, vol. 37, pp. 291-307, June 2014.), a comparison is made between several detection devices for epilepsy, of which some are in the market. This document shows methods based on EEG and methods based on a combination of EEG signals together with some other sensor: electro m achieves fos (E MG), electrocardiog branches (ECG), accelerometers, motion sensors, static measurement of the skin, audio and video recording. Honda et al. (Air Brain - íhe easy telemetric system with smartphone for EEG signa! And human behavior BODYNEST 2013. Boston, MA, US: ICST; 2013. p. 343-6) and Hodson H. (Smartphone EEG to diagnose epilepsy in poor nations. In: New Scientist editor 2014 [February 14, 2018, web: http://www.newcientist.com/article/dn24887-smartphcne-eeg-to- diagnose-epilepsy-in-poor-nations. Html # .U0v -KPIdX76j); show telemetry devices of a portable EEG using a 3G network on a smartphone. Unlike the proposed one, the system requires a cell phone (3G) or a computer with internet access and Bluetooth connection to be able to communicate with the system's processing card, in addition to a specific software for signal reception. On the other hand, the X series device - EEG Wireless moniforing, [February 14, 2018, which can be found on the web: h ttp: // ad vancedbrain monitoring.com] is a portable EEG signal detector, which requires at least have 10 electrodes connected, while e! proposed in this patent only requires 2 electrodes.

It should be considered that all the aforementioned references are devices that are found as an end product. There are also algorithms which are intended to detect epileptic seizures; however, these are only theoretical and have no physical implementation. According to Webber WR et al. (An approach to seizure detection using an artificial neural neíwork (ANN).

Eiectroencephalogr Clin Neurophysiol 1996; 98: 250-72); Pradhan N, et ai. Detection of seizure activity in EEG by an artificial neural neíwork: a preíiminary study. Comput Biomed Res 1996; 29: 303-13.) And Alkan A, et al. (Automatic seizure detection in EEG using logistic regression and artificial neural network. J Neurosci Methods 2005; 148: 167-76.), Use a technique called “artificial neural network” (ANN) to detect attacks based solely on EEG

Pradhan N, et al. Detection heard seizure activiíy in EEG by an artificial neural network: a preliminary study Compui omiomed Res 1998; 29: 303-13.), Casson AJ, et al. (Algorithm for AEEG data selection leading to wireless and long term epi! Epsy monitoring. Conf Proc IEEE Eng Med Bio! Soc 2007; 2007: 2456-9.), Petersen EB, et al (Generic single-channel detection of absence seizures. Conf Proc IEEE Eng Med Biol Soc 2011; 2011: 4820-3.), Liu Y, ef al. (Automatic seizure detection using wavelet transform and SVM in ¡ong-term intracranial EEG. IEEE Trans Neural Syst Rehabil Eng 2012; 20 (8): 749-55.) And SV Mehfa, et al. (Wavelet Anaíysis as a potential tool for seizure detection ”, IEEE-SP! Nt. Symp., Philadelphia, PA, Oct. 1994.), use the Wavelet transform to detect epileptic seizures; being that this transformed is calculated using a large number of multiplications and sums, which entails a processing complex and expensive computationally.

Meanwhile, Wilson SB (A neural network method íor automatic and increment! Learning applied to paien-dependent seizure detection ClinNeurophysio! 2005; 116: 1785-95.) And D'ASessandro M, et al. (A mu líi-featu re and multi-channel univariate selection process for seizure prediction. Clin Neurophysio! 2005: 116: 506— 16.), use a neuron network! of probability for the detection of epileptic seizures.

Other authors use a technique called “supporí vector machine” (SVM).

On the other hand, according to Z. Jiang (Detecting mint! EEG properties using detrended fluctuation analysis ”, lEMBS Con!., Shanghai, China, Jan 2006.), they use a technique called“ Dentrended Fluctuation ”.

On the other hand, Y. Wang, et al. (A Cauchy-based staie-space model for seizure detection in EEG monitoring sysíems ”, IEEE Inteiiigent Systems, pp. 6-12. January 2015), compares EEG signals with noise with particular distributions.

This last article uses statistical comparison methods to detect epileptic seizures.

A search was conducted to determine the closest state of the art, the following documents being found They cite as a reference.

It was found e! US9643019 B2 (which has been called document D1) by Higgins Jason A, et ai. of February 14, 2011, which reveals a system to provide alerts of neurological events that occur in a human subject. The system includes: a monitoring module adapted to detect and sample a neurological signal; an event detection module coupled to the monitoring module to detect one or more types of predetermined notifiable events based on the neurological signal detected; and an alert module coupled to the event detection module, in which to detect an event notified by the event detection module, said alert module selects a first alert contact from a plurality of contacts contained in a distance of contacts and generates a first alert communication to the first alert contact.

Although this patent describes a methodology for sampling neurological signals, it describes the blocks or modules to be used, type of communication between modules, the use of a CPLD as a programmable logic device for storage and possible processing is detailed. However, said Di document does not disclose, nor does it suggest that the means for sampling the electroencephalography signals are obtained by means of electrodes arranged in leads F p 1 - F7 (or in their counterpart leads F p 2 - F8), where sayings electrodes are connected to a biomedical signal amplifier and an NXP Freedom K64 processing card. It does not disclose that the system includes a microcontroller where the dispersion parameter calculation is executed and where the patient's EEG samples are stored in a buffer.

The Di document also does not reveal that the stored signal goes through a low-pass filtering process with a cut-off frequency of 40 Hz and order 20 for noise elimination. Document D1 also does not reveal that the filtered signal is passed to a third block of calculation of the dispersion parameter through calculation means in the same microcontroller, which allows the data to be sorted and the values of quantiles 25 and 75 of the filtered EEG signal, which are associated with the previously defined scatter parameter. The statistical quantiles are averaged and used in the estimation of the dispersion parameter that are again filtered to analyze them by means of analysis of the signals processed in order to look for epileptic seizures in real time. Document D1 does not disclose that a sample-by-sample comparison of each of the filtered dispersion values is executed, having two detection thresholds. The first threshold compares the current sample with the previous sample (each sample represents 0.03 second of signal), if the current sample is 7 times larger than the previous one, then the system determines that the patient is suffering an epileptic attack and sends an alert to the last block that is the attack notifier, where an LED is lit red color and an audible alarm sounds. And where the second threshold is the same as the previous one, by amplitudes, where the current sample is analyzed with the previous one. If the current sample is 7 times smaller than the previous sample then the system determines that the patient left the seizure and is in normal state or recovering from the attack, sending a signal to! Notifier block e! which turns off the audible alarm and changes the color of the LED from red to green.

Therefore, our system is new to the subject matter disclosed in document D1

Document US8679034 B2 was also located (which has been called document D 2) of Avner Halperin et al of January 25, 2013, which discloses devices and methods that include detecting at least one parameter of a subject while the subject is sleeping. parameter is analyzed, and a condition of the subject is determined at least in part in response to the analysis The subject is alerted to the condition only after the subject wakes up Other applications are also described.

In this patent three sensors are proposed which will be in the patient's bed while he sleeps. The sensors are: motion, acoustic and temperature. With the combination of the three sensors they detect the following parameters of the patient: breathing, heartbeat, cough, whether he is excited or restless and his blood pressure. Said document D2 did not reveal, nor suggest sampling of the electroencephalography signals, nor that the electrocardiogram signals can be obtained by means of electrodes arranged in leads F p 1 - F7 (or in their counterpart leads F p 2 - F8, in where said electrodes are connected to a biomedical signal amplifier and an NXP Freedom K64 processing card It does not disclose that the system includes a microcontroller where the dispersion parameter calculation is executed and where the patient's EEG samples are stored in a buffer .

Document D2 also did not reveal the stored signal going through a low-pass filtering process with a cut-off frequency of 40 Hz and order 20 for noise elimination. Document D2 also does not disclose that the filtered signal is passed to a third block of calculation of the dispersion parameter through calculation means in the same microcontroller. That allows the data to be sorted and the values of quantiles 25 and 75 of the filtered EEG signal are calculated, which are associated with the previously defined dispersion parameter. The statistical quantiles are averaged and used in the estimation of the dispersion parameter that are again filtered to analyze them by means of analysis of the signals processed in order to look for epileptic seizures in real time. Document D2 does not disclose that a sample-by-sample comparison of each of the filtered dispersion values is executed, having two detection thresholds. The first threshold compares the sample current with the previous sample (each sample represents 0.Q3 second of signal), if the current sample is 7 times greater than the previous one, then the system determines that the patient is suffering an epileptic attack and sends an alert to the last block that is the attack notifier, where a red LED turns on and an audible alarm sounds. And where the second threshold is the same as the previous one, by amplitudes, where the current sample is analyzed with the previous one. If the current sample is 7 times smaller than the previous sample then the system determines that the patient left the seizure and is in normal state or recovering from the attack, sending a signal to the notification block e! which turns off the audible alarm and changes the color of the LED from red to green. The main difference in our request is the type of signal that enters our device. While document D2 requires motion, acoustic and temperature sensors, the proposed patent is based on the patient's electroencephalography (EEG) signal. Since the signals considered are different, the processing is different in the own proposal compared to the one mentioned in the state of the art. Thus our invention is new on document D2.

Finally, CN1Q1583311 B (which has been called document D3 for reference) was found by Uri Kramer et al. of September 19, 2006, which unveils a device and a procedure to detect and alert an epileptic seizure. The detector is portable by an active user and does not interfere with normal daily movement. The detector relies on at least one motion sensor and performs a computerized analysis to determine if a seizure is occurring. The parameters of the motion signal are compared with the non-epileptic epileptic movement signal parameters; and the epileptic parameters, and a decision is reached as to whether or not to indicate an alert. In a preferred embodiment, the analysis is based on one or more of the following movement signal parameters: the duration of the movement, the frequency of the movement, the amplitude of the signal, the direction of the movement and the ratio of the amplitude over the movement frequency

Said document D3 does not disclose, nor does it suggest that the sampling means of the electroencephalography signals are obtained by means of electrodes arranged in the leads F p 1 - F7 (or in their counterpart the leads F p 2 - F8), wherein said electrodes are connected to a biomedical signal amplifier and an NXP Freedom K84 processing card. It does not disclose that the system includes a microcontroller where the dispersion parameter calculation is executed and where the patient's EEG samples are stored in a buffer.

Document D3 also does not reveal the signal stored for a low-pass filtering process with a cut-off frequency of 40 Hz and Order 20 for noise elimination. Document D3 also does not disclose that the filtered signal is passed to a third calculation block of the dispersion parameter through calculation means in the same microcontroller. That allows the data to be sorted and the values of quanile ios 25 and 75 of the filtered EEG signal are calculated, which are associated with the previously defined dispersion parameter. The statistical quaníiles are averaged and are used in the estimation of the dispersion parameter that are again filtered to analyze them by means of analysis of the signals processed in order to look for epileptic seizures in real time. Document D3 does not disclose that a sample-by-sample comparison of each of the filtered dispersion values is executed, having two detection thresholds. The first threshold compares the current sample with the previous sample (each sample represents 003 second of signal), if the current sample is 7 times larger than the previous one, then the system determines that the patient is suffering an epileptic attack and sends an alert to the last block that is the notifier of attacks, where a red LED is lit and an audible alarm sounds And where the second threshold is the same as the previous one, by amplitudes, where the current sample is analyzed with the previous one. If the current sample is 7 times smaller than the previous sample then the system determines that the patient left the epileptic attack and is in normal state or recovering from the attack, sending a signal to the notifying block which turns off the audible alarm and changes the color of the LED from red to green. Although the invention of said document D3 also detects epileptic events (such as that proposed in our invention), it is based on the detection of sudden movements of the patient by a portable device. Our invention is an alert system for epilepsy warning based on the detection of the patient's EEG signal and the quantification of statistical signal parameters. Thus our invention is new on document D3. Given the need for an alert system for the highly reliable epilepsy warning of rapid real-time detection of epileptic seizures using the dispersion parameter, the present invention was developed.

OBJECTIVES OF THE INVENTION

The main objective of the present invention is to make available a system integrated by a device whose function is to alert for the warning of an epileptic attack, both auditively (with a sound), and visually (with a red LED ) in real time; where e! The system can be implemented in any person who suffers epileptic seizures, having as a priority those whose trigger factor is visually.

Another object of the invention is to make available a system of alert for the epilepsy warning that also allows to detect reliably, quickly and in real time the occurrence of epilepsy using the dispersion parameter. Another objective of the invention is to provide such an alert system for epilepsy warning which, in addition, allows the processing of information and search for epileptic seizures through an esocostatic analysis of the signal in question, making a! most robust system to! noise, quickly and efficiently.

Another objective of the invention is to provide such an alert system for epilepsy warning that, in addition, is less sensitive to noise by the stochastic processing of the EEG signal, which is also robust and that guarantees to alert effectively and in real time when a patient suffers an epileptic seizure,

And all those qualities and objectives that will become apparent when making a general and detailed description of the present invention supported by the illustrated modalities.

OF THE INVENTION

In general, the alert system for epilepsy warning, in accordance with the present invention, consists of means for sampling electroencephalography (EEG) signals to a person suffering from epileptic seizures, through preferably gold electrodes arranged in leads F p 1 - F7 (or in their counterpart leads F p 2 - F8); where the EEG signal has little amplitude (in the range of micro volts), so a conductive gel should be used before placing the gold electrodes preferably on the skin. Said electrodes are connected to a biomedical signal amplifier and to a processing card preferably an NXP Freedom K64 card. The system includes a microcontroller where the dispersion parameter calculation is executed and where the patient's EEG samples are stored in a buffer. These samples will be stored in an internal memory, the internal buffer (approximately 0.39 seconds or 1000 samples being the equivalent).

This system also includes a low-pass filtering module of EEG signals with a cut-off frequency of 40 Hz and order 20, which will help eliminate noise caused by electrodes, environmental factors or voltages caused by the human body at the time Take sample.

Said microcontroller also includes means for calculating the dispersion parameter of the filtered EEG signal where the data is ordered and the values of quantiles 25 and 75 of the signal are calculated! of filtered EEG, which are associated with the previously defined dispersion parameter. Statistical quantiles are averaged and used in the estimation of the dispersion parameter. This process is repeated consecutively, until complete the sign! EEG or until the device is removed from the user. The following equation 1 is associated with the calculation of the dispersion estimator d.

Equation 1. Characteristic formula to calculate the dispersion parameter.

The dispersion values within! The system is entered into a low-pass filter module with a cut-off frequency of 0.5 Hz and a filter order with a factor of 20, which will leave the dispersion values ready to be analyzed reliably.

Finally, the system performs a signal analysis through means of analysis of the processed signals, in order to look for epileptic seizures in real time. The analyzer makes a sample-by-sample comparison of each of the filtered dispersion values, having two detection thresholds. The first threshold compares the current sample with the previous sample (each sample represents 0.03 second of signal), if the current sample is 7 times larger than the previous one, then the system determines that the user is suffering an epileptic attack and sends an alert to attack notifier, where a red LED turns on and an audible alarm sounds. The second threshold is the same as the previous one, by amplitudes, where analyze the current sample with the previous one If the current sample is

7 times lower than the previous sample then the system determines that the user left the epileptic attack and is in normal state or recovering from the attack, sending a signal to the notifying block which turns off the audible alarm and changes the color of the red LED green color

The process is repeated constantly until the system is turned off or the electrodes are removed from the user.

The alert system for the epilepsy warning analyzes and processes EEG signals in order to find epileptic seizures using the parameter of the analyzed EEG signal.

The main advantage of the system of the present invention is that it is less sensitive to noise by the stochastic processing of the EEG signal, having a robust device, ensuring that it will be alerted effectively and in real time when a user suffers an epileptic attack.

The computational expenditure consumed by the system is minimal and the amount of memory required to function properly is reduced, using only 0.39 seconds or 1,000 samples of the EEG signal for processing.

If the system has only one analysis channel, the user does not take much time to put on and take off the electrode that will be censoring the EEG signal, in addition to being more comfortable for the patient.

In case the system must process stored information, it is able to process 30 minutes of information in 0.17 seconds, with 100% alerts of epileptic attacks, making the system fast and efficient.

In order to better understand the characteristics of the present invention, the present description is attached, as an integral part thereof, to drawings with illustrative but not limitative character, which are described below. BRIEF DESCRIPTION OF THE FIGURES Figure 1 shows a block diagram of! device that integrates the alert system for the epilepsy warning, in accordance with the present invention.

Figures 2 and 3 illustrate a side view and a top view, respectively, of the head of a user showing the placement and arrangement of the electrodes for taking the EEG signals, in accordance with the alert system for the epilepsy warning of The present invention. Figure 4 shows a flow chart of the process that follows the alert system for the epilepsy warning, in accordance with the present invention. Figure 5 illustrates a block diagram of the epilepsy event detector that integrates the alert system for the epilepsy warning, in accordance with the present invention. Figure 8 shows a block diagram and its graphs of the process of sampling of the electroencephalography (EEG) signals, their processing, conditioning and analysis for the alert of the occurrence of an epilepsy attack. Figure 7 shows a graph illustrating the spikes of the dispersion values associated with an epileptic attack.

For a better understanding of the invention, a detailed description of some of the modalities thereof will be shown, shown in the drawings which, for illustrative but non-limiting purposes, are attached to this description.

DETAILED DESCRIPTION OF THE INVENTION

The characteristic details of! Alert system for epilepsy warning, are clearly shown in the following description and in the accompanying illustrative drawings, serving the same reference signs to indicate the same parts.

According to figures 1 and 5, the device that is integrated into the alert system for the epilepsy warning, in accordance with The present invention consists of a sampling module of the electroencephalography (EEG) signals (1) where sampling means of the electroencephalography (EEG) signals (2) are integrated consisting of preferably gold electrodes arranged in the leads F p 1 - F7 or in their counterparts leads F p 2 - F8 (see figures 2 and 3) of the head of a user (3); where the EEG signal has little amplitude (in the range of micro volts), so a conductive gel should be used before placing the gold electrodes preferably on the skin.

Said device also includes a module for treatment, conditioning and processing (4) of the electroencephalography (EEG) signals, defined by a dispersion estimator (5) consisting of a biomedical signal amplifier (8) where said means of connection are connected sampling of the electroencephalography (EEG) signals (2) and a processing card (7) preferably an NXP Freedom K64 card that integrates a microcontroller (8) where the scatter parameter calculation is executed and where it is stored in a buffer the EEG samples of the patient, these samples will be stored in an internal memory (9) of the same device, the internal buffer (approximately 0.39 seconds or 1000 samples being the equivalent).

Said treatment, conditioning and processing modules (4) also includes a pass-through filtering module. low (10) sampling means of the EEG electroencephalogram signals (2) with cut-off frequency of 40 Hz and order 20, which will help eliminate noise caused by electrodes, environmental factors or voltages caused by the body human when taking the sample.

Said treatment, conditioning and processing module (4) also includes means for calculating the dispersion parameter of the filtered EEG signal where the data is ordered and the values of quantile ios 25 and 75 of the filtered EEG signal are calculated, which are associated with the previously defined dispersion parameter. Statistical quantiles are averaged and used in the estimation of the dispersion parameter. This process is repeated consecutively, until the EEG signal is completed or until the device is removed from the user. The following equation 1 is associated with the calculation of! scatter parameter.

Equation 1. Characteristic formula to calculate the dispersion parameter.

The dispersion values within the system are entered into a low-pass filter module (11) with a cutoff frequency of 0.5 Hz and a filtering order with a factor of 20, which will leave the dispersion values smooth to be analyzed in a manner trustworthy. Said treatment, conditioning and processing module (4) also includes an epileptic attack detector (12) that performs a signal analysis through means of analyzing the processed signals, with the aim of searching for epileptic attacks in real time. The analyzer makes a sample-by-sample comparison of each of the filtered dispersion values, having two detection thresholds. The first threshold compares the current sample with the previous sample (each sample represents 0.03 second of signal), if the current sample is 7 times larger than the previous one, then the system determines that the user is suffering an epileptic attack and sends an alert to attack notifier

In accordance with said figures 1 and 5, the device includes an epileptic attack notification module (13) that notifies the occurrence of an epileptic attack on the user's head (3) through an LED (14) that is switched on at red color and also sounds an audible alarm (15) integrated in the device The second threshold is the same as the previous one, by amplitudes, where the current sample is analyzed with the previous one. If the current sample is 7 times smaller than the previous sample then the system determines that the user left the epileptic attack and is in normal state or recovering from the attack, sending a signal to the notifying block which turns off the audible alarm (15) and Change the color of the LED (14) from red to green. With reference to Figures 4 and 6, the process followed by the system to detect and report the occurrence of an epileptic seizure consisting of a first stage (a) where "samples of an electroencephalography (EEG) signal are taken" a a person suffering from epileptic seizures, through preferably gold electrodes arranged in leads F p 1 - F7 or Fp2 - F8 where the user must be awake and relaxed, and using a conductive gel before placing the electrodes preferably gold on the skin These electrodes are connected to a biomedical signal amplifier and a processing card (7) with a microcontroller (8) and an internal memory (9) where the patient's EEG samples are stored.

Once the signal is stored, it goes to the second stage (b), which is the “filtering of EEG signals” through a low-pass filtering module (10) with a cut-off frequency of 40 Hz and order 20, which will help eliminate the noise caused by the electrodes, environmental factors or voltages caused by the human body when taking the sample.

The filtered EEG signal passes to a third stage (c) which is the "calculation of the dispersion parameter" where the data is sorted and the values of quantiles 25 and 75 of the filtered EEG signal are calculated, which are associated with the previously defined dispersion parameter. Statistical quantiles are averaged and used in the estimation of the dispersion parameter. This process is repeated consecutively, until the signal is completed. EEG or until the device is removed from! patient.

In the next step (d) "dispersion parameter filtering", the dispersion values within the system are entered into a low-pass filter module (11) with a cutoff frequency of 05 Hz and a filtering order with a factor of 20, which will leave the dispersion values ready to be analyzed reliably.

Finally, in the next step (e) the system performs a “signal analysis to look for epileptic attacks” through means of analysis of the processed signals, with the aim of searching for epileptic attacks in real time.

The analyzer makes a sample-by-sample comparison of each of the filtered dispersion values, having two detection thresholds. The first threshold {e 1} compares the current sample with the previous sample (each sample represents 003 second of signal) if the current sample is 7 times larger than the previous one, then the system determines that the patient is suffering an epileptic attack (e 1 ') and sends an alert to the last block that is the attack notifier (f), where a red LED turns on and an audible alarm sounds.

The second threshold (e 2) is the same as the previous one, by amplitudes, where the current sample is analyzed with the previous one. If the current sample is 7 times smaller than the previous sample then the system determines that the user left the epileptic seizure (e 2 ') and is in normal state or recovering from! attack by sending a signal to the notifying block and which turns off the audible alarm and changes the color of the LED from red to green (g).

AND! The process is repeated constantly until the system is switched off or the electrodes are removed from the patient.

The alert system for the epilepsy warning analyzes and processes EEG signals in order to find epileptic seizures using the dispersion parameter of the analyzed EEG signal.

The main advantage of the system of the present invention is that it is less sensitive to noise by stochastic processing of the EEG signal, having a robust device, ensuring that it will be alerted effectively and in real time when a patient suffers an epileptic attack.

The expense computes! that consumes the system is minimal and the amount of memory required to function properly is reduced, using only 0.39 seconds or 1,000 samples of the EEG signal for processing.

If the system has only one analysis channel, the patient does not take much time to put on and take off the electrode that will be sensing the EEG signal, in addition to being more comfortable for the patient. If the system must process stored information, it is able to process 30 minutes of information in 0.17 seconds, with 100% alerts of epileptic attacks, making the system fast and efficient.

Figure 7 shows a graph illustrating the peaks of the gamma values associated with an epileptic attack.

The invention has been described sufficiently that a person with average knowledge in the field can reproduce and obtain the results mentioned in the present invention. However, any skilled person in the field of the art that is in charge of the present invention may be able to make modifications not described in the present application, however, if for the application of these modifications in a given structure or in the manufacturing process thereof, the matter claimed in the following claims is required, said structures must be included within the scope of the invention.

Claims

REINVINDICATIONS Having sufficiently described the invention, what is contained in the following claims clauses is claimed as property.

1.- An alert system for epilepsy warning, characterized in that it includes means for sampling electroencephalogram (EEG) signals to a person suffering from epileptic seizures, connected to a biomedical signal amplifier and a processing card with a microcontroller and a memory where the patient's EEG samples are stored; a low-pass filtering module of EEG signals with a cut-off frequency of 40 Hz and order 20 to eliminate noise; said microcontroller also includes means for calculating the dispersion parameter of the filtered EEG signal in which the data are ordered and the values of the 25 and 75 signals of the signal are calculated! of filtered EEG, which are associated with! previously defined dispersion parameter; where the dispersion values within the system are entered into a low-pass filtering module with a cut-off frequency of 0.5 Hz and an order of fixture with a factor of 20; an analyzer that analyzes the signals processed to look for epileptic seizures in real time; said analyzer makes a sample-by-sample comparison of each of the filtered dispersion values, having two detection thresholds, where one compares the current sample with the previous sample and if the current sample is 7 times greater than the previous, then e! The user is suffering from an epileptic attack and sends a visual and / or auditory alert to the attack notifier and the second analyzes the current sample with the previous one and if the current sample is 7 times smaller than the previous sample then the user leaves the epileptic attack by turning off the visual and / or auditory alarm.

2 - The alert system for epilepsy warning, according to claim 1, characterized in that said sampling means of the electroencephalogram (EEG) signals consist preferably of gold electrodes arranged in the leads F p 1 --- F7 or in its counterpart the leads Fp2 - F8.

3. The alert system for epilepsy warning, according to claim 1, characterized in that said dispersion parameter is calculated by the formula.

The alert system for epilepsy warning, according to claim 1, characterized in that the calculation of the dispersion parameter of the EEG signal orders the data and the values of the quantiles 25 and 75 of the filtered EEG signal are calculated. .