Epileptic seizure sensing and off-bed early warning device and early warning method

Abstract

The invention discloses an epileptic seizure sensing and off-bed early warning device and an early warning method, and relates to the technical field of epilepsy rehabilitation. The system comprises a power supply, a sensor assembly, a data conversion and collection module, a microprocessor, a server, an alarm and a data storage terminal, the sensor assembly transmits data to the microprocessor through the data conversion and collection module, and the microprocessor transmits the data to the server. Through the design of the device and the method, the early warning capability of epileptic seizure is improved, the device can timely discover early signs or omen of the epileptic seizure through real-time monitoring of multi-dimensional physiological signals of a patient, and the early signs or omen of the epileptic seizure can be timely found through a machine learning and data analysis method. The system can efficiently identify abnormal signals and judge the state of a patient, and when the state is lower than a threshold value, early warning is given out in time, so that the injury risk occurring during epileptic seizure is effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of epilepsy rehabilitation, and in particular to an epileptic seizure sensing and bed leaving warning device and a warning method. Background Art

[0002] Epilepsy is a common neurological disease that usually manifests as abnormal electrical activity in the brain, causing patients to experience sudden loss of consciousness, convulsions and other symptoms. Epilepsy attacks are diverse and unpredictable, causing great distress and safety hazards to patients' daily lives. Epilepsy patients often lack the ability to protect themselves during an attack, especially when resting at night. Epilepsy attacks may cause serious injuries or even death. Therefore, it is particularly important to detect epilepsy attacks in a timely manner and provide effective warnings.

[0003] Epileptic seizures usually occur suddenly and without clear warning signs. According to the different types of epileptic seizures, they are clinically divided into partial seizures and generalized seizures. Symptoms of epileptic seizures may include convulsions, full body rigidity, loss of consciousness, vomiting, etc. Especially in the early stages of a seizure, the patient may not be able to realize that the seizure has occurred. Epileptic seizures pose a huge threat to the patient's health, and the duration, frequency and manifestation of each seizure vary greatly.

[0004] At present, the existing epilepsy monitoring technology still has certain limitations, which are mainly reflected in the following aspects:

[0005] Low monitoring accuracy: Although EEG is considered the most reliable method for monitoring epileptic seizures, EEG equipment is bulky, expensive, and requires professional operation, so it is not possible to achieve all-weather, long-term monitoring;

[0006] Lack of comprehensive monitoring: Current monitoring equipment mainly focuses on a single type of physiological signal (such as EEG, motor signal, etc.), but epileptic seizures have various manifestations, and monitoring of a single signal may be difficult to fully and accurately capture the abnormal characteristics of the seizure;

[0007] Therefore, a new solution to the above problems needs to be proposed. Summary of the invention

[0008] The purpose of the present invention is to provide an epileptic seizure sensing and bed leaving warning device and a warning method to solve the technical problems raised in the background technology.

[0009] To achieve the above-mentioned objectives, the present invention provides the following technical solutions: an epileptic seizure sensing and bed leaving warning device, comprising a power supply, a sensor component, a data conversion and collection module, a microprocessor, a server, an alarm and a data storage terminal, wherein the sensor component transmits data to the microprocessor via the data conversion and collection module, and the microprocessor transmits the data to the server, and after the server has identified the warning, data below the warning threshold are not output to the alarm and the data storage terminal, while data above the warning threshold are output to the alarm and the data storage terminal, and the power supply is used to supply power to the entire device.

[0010] Furthermore, the sensor assembly includes but is not limited to a gravity sensor, a heart rate sensor and a breathing sensor.

[0011] Furthermore, the functions of the programs installed in the server include but are not limited to data rendering programs, receiving data sent by the microprocessor and writing it into the database, monitoring and early warning programs and internal audit log storage. The server is also embedded with an AI model to assist in judgment.

[0012] Furthermore, the database in the data storage terminal adopts but is not limited to one or more of MySQL and PostgreSQL, and the database is at least used to store the collected data and internal audit logs.

[0013] Furthermore, the transmission between the sensor component, the data conversion and collection module, the microprocessor, the server, the alarm and the data storage terminal adopts ES-256 encrypted transmission.

[0014] Furthermore, the power supply adopts a dual power supply redundancy design.

[0015] An epileptic seizure sensing and bed leaving early warning method comprises at least the following steps:

[0016] S1: Use sensor components to collect multiple types of data, see the following formula:

[0017] X={x i , i{x 1 ,x 2 ,…,x n}}

[0018] Represents the collected multi-dimensional sensor data, X is a data set containing multiple time points and sensors;

[0019] where x 1 ,x 2 ,…,x n Represents data from various types of sensors; x i Indicates that at time t i Data points from multiple types of sensor data;

[0020] S2: Use the data conversion and collection module to normalize the data and remove high-frequency noise;

[0021] S3: Use microprocessor to perform time domain feature analysis and frequency domain feature extraction;

[0022] S4: The AI ​​model uses the fused time domain and frequency domain features to judge the events of seizure, leaving the bed and turning over;

[0023] Based on the time domain and frequency domain features, classification is performed using a trained machine learning model, wherein the machine learning model is not limited to one of a support vector machine, a random forest, and a neural network;

[0024] Assume F is the extracted feature vector:

[0025] y=f(F)

[0026] Among them, y represents the model output, and at least the categories covered are "seizure", "leaving bed" and "turning over";

[0027] Bed leaving detection: judge whether you have left the bed by the angle change of the gravity sensor;

[0028]

[0029] Assume θ is the angle of the bed head, and the condition for leaving the bed is θ>θ threshold , where θ threshold is the preset threshold;

[0030] S5: Alarm triggering and data storage. When the AI ​​model determines an attack or bed leaving event, an alarm is triggered and the data is stored.

[0031] Furthermore, the S2 at least includes the following steps:

[0032] First, normalize the data as follows:

[0033]

[0034] in, represents the data after normalization; μ i represents the mean value of the i-th sensor data; σ i represents the standard deviation of the i-th sensor data;

[0035] Then, weighted averaging is performed to remove high-frequency noise and retain the low-frequency components of the signal, ensuring that subsequent analysis is more accurate and reliable:

[0036]

[0037] is the signal after low-pass filtering, τ is the time constant of the filter, t′ is the integral variable, which represents a small time point before time t, used to calculate the filter response; e is the base of the natural logarithm, which is approximately equal to 2.71828, and e appears in the exponential function e -(t-t′) / τ The function is an exponential decay process; dt′ represents a small increase in time.

[0038] Furthermore, the time domain features in S3 at least include the following steps:

[0039] S3.1.1: Determine the time domain characteristics of the heart rate sensor, referring to the following equation:

[0040]

[0041] Among them, μ HR is the mean of the heart rate data; x HR (t i ) is at time t i The collected heart rate data; n is the total number of signal points; σ HR is the standard deviation of the heart rate data, indicating the degree of fluctuation of the heart rate data;

[0042] S3.1.2: Determine the time domain characteristics of the gravity sensor, referring to the following equation:

[0043]

[0044] Among them, μ gravity is the mean value of the gravity sensor data, x gravity (t i ) represents the gravity signal at the i-th moment; n is the total number of signal points; σ gravity is the standard deviation of the gravity sensor data, indicating the volatility of the signal;

[0045] S3.1.3: Determine the time domain characteristics of the respiratory signal and analyze the changes in the respiratory signal by calculating the mean and standard deviation of the respiratory signal, as shown in the following formula:

[0046]

[0047] Among them, μ respiration is the mean of the respiratory signal; x respiration (t i ) is at time t i The collected respiratory signal data; n is the total number of signal points; σ respiration Indicates the fluctuation degree or variation amplitude of the respiratory signal.

[0048] Furthermore, the frequency domain feature extraction in S3 at least includes the following steps:

[0049] S3.2.1: Convert the signal from the time domain to the frequency domain using the Fast Fourier Transform, see the following equation:

[0050]

[0051] Among them, X(f) is the representation of the signal in the frequency domain; x(t) is the original signal in the time domain; U represents the total number of data points; e -i2πft / U It is a complex exponential function, which converts the time domain signal into the frequency domain;

[0052] S3.2.2: Frequency domain feature extraction of epileptic seizures, refer to the following formula:

[0053] The characteristics of epileptic seizures are determined by calculating the integral of the spectral amplitude within a specific frequency range:

[0054]

[0055] Among them, f 1 and f 2 is the frequency range related to epilepsy; |X(f)| is the amplitude of the signal in the frequency domain, that is, the intensity of the frequency component, which removes the phase information of the signal and retains only the amplitude information; df is the differential element of the frequency variable, which means that the amplitude within the frequency range is integrated to obtain the total intensity of the frequency band.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The present invention improves the early warning capability of epileptic seizures through the design of devices and methods. The device can promptly detect early signs or precursors of epileptic seizures through real-time monitoring of the patient's multi-dimensional physiological signals. Through machine learning and data analysis methods, the system can efficiently identify abnormal signals and determine the patient's state. When the state is below the threshold, an early warning is issued in time, thereby effectively reducing the risk of injury during epileptic seizures. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0059] Figure 1 It is a schematic diagram of the overall structure and flow chart of the present invention. DETAILED DESCRIPTION

[0060] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0061] Embodiment 1:

[0062] See also Figure 1 , an epileptic seizure sensing and bed leaving warning device, including a power supply, a sensor component, a data conversion and collection module, a microprocessor, a server, an alarm and a data storage terminal. The sensor component transmits data to the microprocessor through the data conversion and collection module, and the microprocessor transmits the data to the server. After the server has been identified by the warning, the data below the warning threshold will not be output to the alarm and the data storage terminal, and the data above the warning threshold will be output to the alarm and the data storage terminal. The power supply is used to supply power to the entire device.

[0063] The sensor components include, but are not limited to, gravity sensors, heart rate sensors, and breathing sensors.

[0064] The functions of the programs installed in the server include but are not limited to data rendering programs, receiving data sent by the microprocessor and writing it to the database, monitoring and early warning programs, and internal audit log storage. The server also has an embedded AI model to assist in judgment.

[0065] The database in the data storage terminal adopts but is not limited to one or more of MySQL and PostgreSQL, and the database is at least used to store the collected data and internal audit logs.

[0066] The transmission between sensor components, data conversion and collection modules, microprocessors, servers, alarms and data storage terminals adopts ES-256 encryption transmission.

[0067] The power supply adopts dual power supply redundant design.

[0068] Embodiment 2:

[0069] This embodiment proposes an epileptic seizure sensing and bed leaving warning method based on the device in the above embodiment 1, which at least includes the following steps:

[0070] S1: Use sensor components to collect multiple types of data, see the following formula:

[0071] X={x i , i{x 1 ,x 2 ,…,x n}}

[0072] Represents the collected multi-dimensional sensor data, X is a data set containing multiple time points and sensors;

[0073] where x 1 ,x 2 ,…,x n Represents data from various types of sensors; x i Indicates that at time t i Data points from multiple types of sensor data;

[0074] S2: Use the data conversion and collection module to normalize the data and remove high-frequency noise;

[0075] S3: Use microprocessor to perform time domain feature analysis and frequency domain feature extraction;

[0076] S4: The AI ​​model uses the fused time domain and frequency domain features to judge the events of seizure, leaving the bed and turning over;

[0077] Based on the time domain and frequency domain features, classification is performed using a trained machine learning model, where the machine learning model is not limited to one of a support vector machine, a random forest, and a neural network;

[0078] Assume F is the extracted feature vector:

[0079] y=f(F)

[0080] Among them, y represents the model output, and at least the categories covered are "seizure", "leaving bed" and "turning over";

[0081] Bed leaving detection: judge whether you have left the bed by the angle change of the gravity sensor;

[0082]

[0083] Assume θ is the angle of the bed head, and the condition for leaving the bed is θ>θ threshold , where θ threshold is the preset threshold;

[0084] S5: Alarm triggering and data storage. When the AI ​​model determines an attack or bed leaving event, an alarm is triggered and the data is stored.

[0085] S2 at least includes the following steps:

[0086] First, normalize the data as follows:

[0087]

[0088] in, represents the data after normalization; μ i represents the mean value of the i-th sensor data; σ i represents the standard deviation of the i-th sensor data;

[0089] Then, weighted averaging is performed to remove high-frequency noise and retain the low-frequency components of the signal, ensuring that subsequent analysis is more accurate and reliable:

[0090]

[0091] is the signal after low-pass filtering, τ is the time constant of the filter, t′ is the integral variable, which represents a small time point before time t, used to calculate the filter response; e is the base of the natural logarithm, which is approximately equal to 2.71828, and e appears in the exponential function e -(t-t′) / τ The function is an exponential decay process; dt′ represents a small increase in time.

[0092] The time domain features in S3 include at least the following steps:

[0093] S3.1.1: Determine the time domain characteristics of the heart rate sensor, referring to the following equation:

[0094]

[0095] Among them, μ HR is the mean of the heart rate data; x HR (t i ) is at time t i The collected heart rate data; n is the total number of signal points; σ HR is the standard deviation of the heart rate data, indicating the degree of fluctuation of the heart rate data;

[0096] S3.1.2: Determine the time domain characteristics of the gravity sensor, referring to the following equation:

[0097]

[0098] Among them, μ gravity is the mean value of the gravity sensor data, x gravity (t i ) represents the gravity signal at the i-th moment; n is the total number of signal points; σ gravity is the standard deviation of the gravity sensor data, indicating the volatility of the signal;

[0099] S3.1.3: Determine the time domain characteristics of the respiratory signal and analyze the changes in the respiratory signal by calculating the mean and standard deviation of the respiratory signal, as shown in the following formula:

[0100]

[0101]

[0102] Among them, μ respiration is the mean of the respiratory signal; xrespiration (t i ) is at time t i The collected respiratory signal data; n is the total number of signal points; σ respiration Indicates the fluctuation degree or variation amplitude of the respiratory signal.

[0103] Frequency domain feature extraction in S3 includes at least the following steps:

[0104] S3.2.1: Convert the signal from the time domain to the frequency domain using the Fast Fourier Transform, see the following equation:

[0105]

[0106] Among them, X(f) is the representation of the signal in the frequency domain; x(t) is the original signal in the time domain; U represents the total number of data points; e -i2πft / U It is a complex exponential function, which converts the time domain signal into the frequency domain;

[0107] S3.2.2: Frequency domain feature extraction of epileptic seizures, refer to the following formula:

[0108] The characteristics of epileptic seizures are determined by calculating the integral of the spectral amplitude within a specific frequency range:

[0109]

[0110] Among them, f 1 and f 2 is the frequency range related to epilepsy; |X(f)| is the amplitude of the signal in the frequency domain, that is, the intensity of the frequency component, which removes the phase information of the signal and retains only the amplitude information; df is the differential element of the frequency variable, which means that the amplitude within the frequency range is integrated to obtain the total intensity of the frequency band.

[0111] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. An epileptic seizure sensing and bed leaving warning device, characterized in that: It includes a power supply, a sensor component, a data conversion and collection module, a microprocessor, a server, an alarm and a data storage terminal. The sensor component transmits data to the microprocessor through the data conversion and collection module, and the microprocessor transmits the data to the server. After the server has been warned and identified, data below the warning threshold will not be output to the alarm and the data storage terminal, while data above the warning threshold will be output to the alarm and the data storage terminal. The power supply is used to power the entire device.

2. The epileptic seizure sensing and bed leaving warning device according to claim 1, characterized in that: The sensor components include but are not limited to a gravity sensor, a heart rate sensor and a breathing sensor.

3. The epilepsy attack sensing and bed leaving warning device according to claim 1, characterized in that: The functions of the programs installed in the server include but are not limited to data rendering programs, receiving data sent by the microprocessor and writing it into the database, monitoring and early warning programs and internal audit log storage. The server is also embedded with an AI model for auxiliary judgment.

4. The epilepsy attack sensing and bed leaving warning device according to claim 1, characterized in that: The database in the data storage terminal adopts but is not limited to one or more of MySQL and PostgreSQL, and the database is at least used to store the collected data and internal audit logs.

5. The epileptic seizure sensing and bed leaving warning device according to claim 1, characterized in that: The transmission between the sensor components, data conversion and collection modules, microprocessors, servers, alarms and data storage terminals adopts ES-256 encrypted transmission.

6. The epileptic seizure sensing and bed leaving warning device according to claim 1, characterized in that: The power supply adopts a dual power supply redundancy design.

7. An epileptic seizure sensing and bed leaving warning method, used in an epileptic seizure sensing and bed leaving warning device as claimed in any one of claims 1 to 6, characterized in that: At least the following steps are included: S1: Use sensor components to collect multiple types of data, see the following formula: X={x i ,i{x1,x2,…,x n }} Represents the collected multi-dimensional sensor data, X is a data set containing multiple time points and sensors; where x1, x2, …, x n Represents data from various types of sensors; x i Indicates that at time t i Data points from multiple types of sensor data; S2: Use the data conversion and collection module to normalize the data and remove high-frequency noise; S3: Use microprocessor to perform time domain feature analysis and frequency domain feature extraction; S4: The AI ​​model uses the fused time domain and frequency domain features to judge the events of seizure, leaving the bed and turning over; Based on the time domain and frequency domain features, classification is performed using a trained machine learning model, wherein the machine learning model is not limited to one of a support vector machine, a random forest, and a neural network; Assume F is the extracted feature vector: y=f(F) Among them, y represents the model output, and at least the categories covered are "seizure", "leave bed" and "turn over"; Bed leaving detection: judge whether you have left the bed by the angle change of the gravity sensor; Assume θ is the angle of the bed head, and the condition for leaving the bed is θ>θ threshold , where θ threshold is the preset threshold; S5: Alarm triggering and data storage. When the AI ​​model determines an attack or bed leaving event, an alarm is triggered and the data is stored.

8. The method for sensing epileptic seizures and leaving bed early warning according to claim 7, characterized in that: The S2 at least comprises the following steps: First, normalize the data as follows: in, represents the data after normalization; μ i represents the mean value of the i-th sensor data; σ i represents the standard deviation of the i-th sensor data; Then, weighted averaging is performed to remove high-frequency noise and retain the low-frequency components of the signal, ensuring that subsequent analysis is more accurate and reliable: is the signal after low-pass filtering, τ is the time constant of the filter, t′ is the integral variable, which represents a small time point before time t, used to calculate the filter response; e is the base of the natural logarithm, which is approximately equal to 2.71828, and e appears in the exponential function e -(t-t′) / τ The function is an exponential decay process; dt′ represents a small increase in time.

9. The method for sensing epileptic seizures and leaving bed early warning according to claim 8, characterized in that: The time domain features in S3 at least include the following steps: S3.1.1: Determine the time domain characteristics of the heart rate sensor, referring to the following equation: Among them, μ HR is the mean of the heart rate data; x HR (t i ) is at time t i The collected heart rate data; n is the total number of signal points; σ HR is the standard deviation of the heart rate data, indicating the degree of fluctuation of the heart rate data; S3.1.2: Determine the time domain characteristics of the gravity sensor, referring to the following equation: Among them, μ gravity is the mean value of the gravity sensor data, x gravity (t i ) represents the gravity signal at the i-th moment; n is the total number of signal points; σ gravity is the standard deviation of the gravity sensor data, indicating the volatility of the signal; S3.1.3: Determine the time domain characteristics of the respiratory signal and analyze the changes in the respiratory signal by calculating the mean and standard deviation of the respiratory signal, as shown in the following formula: Among them, μ respiration is the mean of the respiratory signal; x respiration (t i ) is at time t i The collected respiratory signal data; n is the total number of signal points; σ respiration Indicates the fluctuation degree or variation amplitude of the respiratory signal.

10. The epileptic seizure sensing and bed leaving early warning method according to claim 9, characterized in that: The frequency domain feature extraction in S3 at least includes the following steps: S3.2.1: Convert the signal from the time domain to the frequency domain using the Fast Fourier Transform, see the following equation: Among them, X(f) is the representation of the signal in the frequency domain; x(t) is the original signal in the time domain; U represents the total number of data points; e -i2πft / U It is a complex exponential function, which converts the time domain signal into the frequency domain; S3.2.2: Frequency domain feature extraction of epileptic seizures, refer to the following formula: The characteristics of epileptic seizures are determined by calculating the integral of the spectral amplitude within a specific frequency range: Among them, f1 and f2 are the frequency ranges related to epilepsy; |X(f)| is the amplitude of the signal in the frequency domain, that is, the intensity of the frequency component, which removes the phase information of the signal and retains only the amplitude information; df is the differential element of the integral, which means that the amplitude within the frequency range is integrated to obtain the total intensity of the frequency band.

Images