A pathological early warning method and device based on big data

Abstract

The present invention relates to a pathological early warning method and device based on big data, comprising the following steps: real-time collecting biological signals of a user through a sensor to obtain original acceleration signals of different parts of the user; denoising and segmenting the collected biological signals of the user and generating a physiological signal database of the user; judging the trunk state of the user by extracting data collected by the waist sensor device in the physiological signal database of the user; normalizing the user's trunk feature matrix and the user's upper limb feature matrix, and establishing a trunk state recognition model and an upper limb state recognition model to judge whether tremors occur in the upper limbs of the user in a static state. The present invention can accurately judge whether the user has diseases related to aging according to the physiological signals of the user, and can help the user to give early warnings about diseases accompanied by the aging physique.

Description

Technical Field

[0001] The present invention relates to the field of big data technology, and in particular, to a pathological early warning method and device based on big data. Background Art

[0002] Pathology refers to the process and principle of the occurrence and development of diseases. That is, the causes of diseases, the pathogenesis, and the changes and their laws in the structure, function, and metabolism of cells, tissues, and organs that occur during the disease process.

[0003] Parkinson's disease, also known as paralysis agitans, is a kind of nervous system disease and cannot be cured completely. It can only be controlled by long-term drug treatment. At the same time, Parkinson's disease is an elderly chronic disease, and the average onset age is about 60 years old. The main clinical manifestation is static tremor. Therefore, the motor symptoms of patients are an important basis for the clinical diagnosis of Parkinson's disease.

[0004] At present, in the existing technology, the method for pathological early warning is too cumbersome and cannot accurately identify, which may cause users to miss the best treatment period for Parkinson's disease. Therefore, improvement is urgently needed. Summary of the Invention

[0005] In order to solve the above technical defects existing in the prior art, the present invention provides a pathological early warning method and device based on big data, which can effectively solve the problems in the background art.

[0006] In order to solve the above technical problems, the technical solutions provided by the present invention are as follows:

[0007] An embodiment of the present invention discloses a pathological early warning method based on big data, including the following steps:

[0008] Real-time collect the biological signals of a user through a sensor to obtain the original acceleration signals of different parts of the user;

[0009] Perform noise reduction and segmentation on the collected biological signals of the user, and generate a physiological signal database of the user;

[0010] Judge the trunk state of the user by extracting the data collected by the waist sensor device in the physiological signal database of the user;

[0011] Normalize the user's trunk feature matrix and the user's upper limb feature matrix, and establish a trunk state recognition model and an upper limb state recognition model to judge whether tremors occur in the upper limbs of the user in a static state.

[0012] Preferably, in any of the above solutions, a collection module, a storage module, and a control module are provided in the sensor device. The collection module is used to collect the user's biological signal data, or obtain the acceleration data of the user's movement and posture, and send the collected data to the control module; the control module is connected to the storage module. After the collection module finishes collecting the user's data, the control module stores the data collected by the collection module in the storage module.

[0013] Preferably, in any of the above solutions, the sensor device includes a waist sensor device, a left upper limb sensor device, and a right upper limb sensor device. The waist sensor device is fixed to the user's waist, and the left upper limb sensor device and the right upper limb sensor device are respectively fixed to the ends of the user's left and right upper limbs.

[0014] Preferably, in any of the above solutions, the data collected by the waist sensor device, the left upper limb sensor device, and the right upper limb sensor device are denoised and segmented by a filter, and a physiological signal database of the user is generated.

[0015] Preferably, in any of the above solutions, denoising the collected data by a filter includes the following steps:

[0016] Perform third-order filtering on the collected data;

[0017] Determine the normalized cut-off frequency and order of the filter;

[0018] Calculate the filtering coefficients;

[0019] Filter through the filtering function.

[0020] Preferably, in any of the above solutions, calculate the average value of the absolute values of the accelerations in the X, Y, and Z axes respectively to judge the user's torso state; calculate the standard deviation of the accelerations in the X, Y, and Z axes respectively, and the correlation coefficients between any two of the accelerations in the X, Y, and Z axes; generate a torso feature vector of the user from the absolute value mean, standard deviation, and pairwise correlation coefficients of the acceleration data in the X, Y, and Z axes. Furthermore, combine the torso static state feature matrix of the user and the torso non-static state feature matrix of the user into a torso feature matrix of the user, and create a corresponding two-row N-column 0-1 matrix.

[0021] Preferably, in any of the above solutions, according to the data collected by the left upper limb sensor device and the right upper limb sensor device in the user's physiological signal database, the coefficients of the autoregressive model are estimated by the least squares method, and five AR coefficients of the left upper limb data and five AR coefficients of the right upper limb data are obtained respectively. Among them, the order of the AR model is set to 5; the signal amplitude range and the five AR coefficients are used as the upper limb feature vectors for identifying the upper limb state, and the upper limb tremor, static and motion state feature matrices are combined into the user's upper limb feature matrix, and a corresponding 0-1 matrix with three rows and N columns is created.

[0022] Preferably, in any of the above solutions, in the trunk state recognition model, the input vector is the user's trunk feature matrix. Among them, the user's trunk feature matrix includes 9 features and the dimension is 9. The number of nodes in the input layer of the trunk state recognition model is set to 9, and the number of nodes in the output layer is set to 2. Then, when the output vector is 0-1, it indicates static, and when it is 1-0, it indicates non-static; in the upper limb state recognition model, the input vector is the user's upper limb feature matrix. Among them, the user's upper limb feature matrix includes 6 features and the dimension is 6. The number of nodes in the input layer of the trunk state recognition model is set to 6, and the number of nodes in the output layer is set to 3. Then, when the output vector is 0-0-1, it indicates tremor, when it is 0-1-0, it indicates static, and when it is 1-0-0, it indicates motion.

[0023] Preferably, in any of the above solutions, when the output vector of the trunk state recognition model is 0-1 and the output vector of the upper limb state recognition model is 0-0-1, it is determined that the user has tremors in the upper limb in the static state, otherwise it does not hold.

[0024] In a second aspect, a pathological early warning device based on big data, the device includes:

[0025] An acquisition module, configured to collect the biological signals of the user in real time through sensors to obtain the original acceleration signals of different parts of the user;

[0026] A processing module, configured to denoise and segment the biological signals collected from the user and generate the user's physiological signal database;

[0027] An extraction module, configured to judge the trunk state of the user by extracting the data collected by the waist sensor device in the user's physiological signal database;

[0028] An identification module, configured to normalize the user's trunk feature matrix and the user's upper limb feature matrix, establish a trunk state recognition model and an upper limb state recognition model, and judge whether the user has tremors in the upper limb in the static state.

[0029] Compared with the prior art, the beneficial effects of the present invention:

[0030] The present invention provides a pathological warning method and device based on big data. By using sensors to collect the biological signals of a user in real time, the original acceleration signals of different parts of the user are obtained; the collected biological signals of the user are denoised and segmented, and a physiological signal database of the user is generated; by extracting the data collected by the waist sensor device in the physiological signal database of the user, the trunk state of the user is judged; the user's trunk feature matrix and upper limb feature matrix are normalized, and a trunk state recognition model and an upper limb state recognition model are established to judge whether tremors occur in the upper limbs of the user in a static state; it can accurately judge whether the user has diseases related to aging based on the physiological signals of the user, and can help the user to give early warnings about the diseases associated with the aging physique. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The drawings are used for further understanding of the present invention, and are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention.

[0032] Figure 1 is a flowchart of the pathological warning method based on big data of the present invention;

[0033] Figure 2 is a schematic diagram of the modules of the pathological warning device based on big data of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0035] It should be noted that when an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or indirectly connected to the other element.

[0036] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention.

[0037] In addition, the terms "first" and "second" are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "a plurality of" means two or more, unless otherwise specifically defined.

[0038] To better understand the above technical solution, the technical solution of the present invention will be described in detail below in conjunction with the accompanying drawings of the specification and specific embodiments.

[0039] The present invention provides a pathological warning method based on big data, as Figure 1 shown, the method includes the following steps:

[0040] Step 1, collect the biological signals of the user in real time through a sensor to obtain the original acceleration signals of different parts of the user.

[0041] Specifically, the sensor device is arranged on the user's body. The sensor device is provided with a collection module, a storage module and a control module. The collection module is used to collect the biological signal data of the user, such as the electroencephalogram, electrocardiogram and electromyogram signals of the user, or obtain the acceleration data of the user's movement and posture, and send the collected data to the control module; the control module is connected to the storage module. After the collection module finishes collecting the user's data, the control module stores the data collected by the collection module in the storage module; for the integrity of collecting the overall data of the user, three groups of the sensor devices are provided, namely a waist sensor device, a left upper limb sensor device and a right upper limb sensor device. The waist sensor device is fixed to the user's waist. Among them, when wearing the waist sensor device, the wearing direction is based on the natural coordinate system, so that the Z-axis of the user coordinate system coincides with the Z-axis of the natural coordinate system to eliminate the influence of the gravitational acceleration on the acceleration data of the XY axis; the left upper limb sensor device and the right upper limb sensor device are respectively fixed to the ends of the left and right upper limbs of the user to realize the collection of the original acceleration signals of different parts of the user.

[0042] Step 2, denoise and segment the biological signals collected from the user, and generate a physiological signal database of the user.

[0043] Specifically, since the upper limb morphology of the elderly population does not show tremor symptoms when in a moving state such as walking, running or going up and down stairs, the movement morphology of the user is divided into a trunk movement moment, a trunk static moment, an upper limb tremor moment, an upper limb static moment and an upper limb movement moment, and the physiological signals at the movement morphology of the user are collected respectively.

[0044] Furthermore, since the waist is closer to the center of gravity of the human body when standing, the waist sensor device can identify whether the user's torso is in a moving or stationary state, and the upper limb sensor device can identify whether the user's upper limbs are in a stationary, tremor or moving state, so as to accurately reflect the user's real-time motion state.

[0045] Furthermore, the data collected by the waist sensor device, the left upper limb sensor device and the right upper limb sensor device are denoised and segmented, and a physiological signal database of the user is generated.

[0046] Furthermore, when collecting the user's physiological signals through the sensor device, noise is often generated. To reduce the generation of noise and remove the generated noise and interference, the collected data is denoised by a filter.

[0047] Furthermore, according to the different frequency bands of the signals passed through, the filter includes a low-pass filter, a high-pass filter, a band-pass filter, a band-stop filter and an all-pass filter. Since the frequency of human body acceleration is relatively low, while the frequency of interference signals such as environmental noise mixed in is relatively high, preferably, a low-pass filter is used to denoise the collected data.

[0048] Furthermore, denoising the collected data by a filter includes the following steps:

[0049] Perform third-order filtering on the collected data;

[0050] Determine the normalized cut-off frequency and order of the filter;

[0051] Calculate the filtering coefficients;

[0052] Filter through the filtering function.

[0053] Furthermore, the data within the time period from T0 to Tn collected is divided into n data sequences of equal length with a time window ΔT as the unit, and each data sequence is used as a basic processing unit to complete the segmentation of the data; among them, a time window of ΔT = 2 seconds is selected, and the data collected every 2 seconds is composed into a data sequence.

[0054] Step 3, by extracting the data collected by the waist sensor device in the user's physiological signal database, judge the state of the user's torso.

[0055] Specifically, judge whether the user's torso is in a stationary or non-stationary state through the data collected by the waist sensor device, and use the mean value of the absolute value of the acceleration data collected by the waist sensor device as the feature for identifying the torso state.

[0056] Furthermore, when the user's torso is in a static state, the mean value of the absolute acceleration in the XY-axis direction is close to 0, and the mean value of the absolute acceleration in the Z-axis direction is close to 1; when the user makes random movements, the absolute acceleration in the XY-axis direction is greater than 0 or the mean value of the absolute acceleration in the Z-axis direction is greater than 1; when the user makes regular movements, the mean value of the absolute acceleration in the XY-axis direction is greater than 0, and the mean value of the absolute acceleration in the Z-axis direction is greater than 1. Then, the average values of the absolute accelerations in the X, Y, and Z-axis directions are calculated respectively to determine the state of the user's torso.

[0057] Furthermore, to improve the accuracy of judging the state of the user's torso, the acceleration standard deviations in the X, Y, and Z-axis directions and the correlation coefficients between any two of the accelerations in the X, Y, and Z-axis directions are calculated respectively.

[0058] Furthermore, the mean value of the absolute values, the standard deviation, and the correlation coefficients between any two of the X, Y, and Z-axis acceleration data are used to generate the user's torso feature vector. Then, the user's torso static state feature matrix and the user's torso non-static state feature matrix are combined into the user's torso feature matrix, and a corresponding 0-1 matrix with two rows and N columns is created.

[0059] Furthermore, by extracting the data collected by the left upper limb sensor device and the right upper limb sensor device in the user's physiological signal database, the state of the user's upper limb is judged, where the state of the user's upper limb includes a tremor state, a static state, and a movement state.

[0060] Furthermore, since the acceleration signal of human movement has continuity, randomness, and autocorrelation in the time series, it is impossible to accurately analyze tremors, static states, or other states only based on features such as the mean value, standard deviation, and correlation coefficient of the acceleration in three axes. When tremors occur in the human upper limb, the data collected by the left upper limb sensor device and the right upper limb sensor device have specific regularities, and for the tremor data in different time periods, the coefficients of the autoregressive models estimated by the same method are also similar.

[0061] Furthermore, according to the data collected by the left upper limb sensor device and the right upper limb sensor device in the user's physiological signal database, the coefficients of the autoregressive model are estimated by the least squares method, and five AR coefficients of the left upper limb data and five AR coefficients of the right upper limb data are obtained respectively, where the order of the AR model is set to 5.

[0062] Furthermore, since the signal amplitude range can reflect the intensity of the user's actions, the larger the SMA value, the faster the frequency of the user's action changes or the faster the change in their action speed during this period. During the time period when upper limb tremor symptoms occur, the value of its signal amplitude range fluctuates within a stable range and the value is relatively large; when the upper limb is at rest, its value is also relatively stable but relatively small; when the upper limb is moving, various different actions also have their relatively fixed signal amplitude ranges. Furthermore, the signal amplitude range and five AR coefficients are used as the upper limb feature vectors for identifying the upper limb state, and the upper limb tremor, rest, and movement state feature matrices are combined into the user's upper limb feature matrix, and a corresponding 0-1 matrix with three rows and N columns is created.

[0063] Step 4: Normalize the user's torso feature matrix and the user's upper limb feature matrix, and establish a torso state recognition model and an upper limb state recognition model to determine whether tremors occur in the upper limb in the static state.

[0064] Specifically, in the torso state recognition model, the input vector is the user's torso feature matrix. Among them, the user's torso feature matrix includes 9 features and the dimension is 9. Set the number of nodes in the input layer of the torso state recognition model to 9 and the number of nodes in the output layer to 2. Then, when the output vector is 0-1, it represents rest, and when it is 1-0, it represents non-rest.

[0065] Furthermore, in the upper limb state recognition model, the input vector is the user's upper limb feature matrix. Among them, the user's upper limb feature matrix includes 6 features and the dimension is 6. Set the number of nodes in the input layer of the torso state recognition model to 6 and the number of nodes in the output layer to 3. Then, when the output vector is 0-0-1, it represents tremor, when it is 0-1-0, it represents rest, and when it is 1-0-0, it represents movement.

[0066] Furthermore, when the output vector of the torso state recognition model is 0-1 and the output vector of the upper limb state recognition model is 0-0-1, it is determined that tremors occur in the upper limb in the static state, otherwise it does not hold.

[0067] The present invention also provides a pathological early warning device based on big data, as Figure 2 shown, the device includes:

[0068] An acquisition module, configured to collect the user's biological signals in real time through sensors to obtain the original acceleration signals of different parts of the user;

[0069] A processing module, configured to denoise and segment the collected biological signals of the user and generate a physiological signal database of the user;

[0070] An extraction module, configured to determine the trunk state of a user by extracting the data collected by the waist sensor device in the user's physiological signal database;

[0071] An identification module, configured to normalize the user's trunk feature matrix and the user's upper limb feature matrix, establish a trunk state identification model and an upper limb state identification model, and determine whether tremors occur in the user's upper limbs in a static state.

[0072] Compared with the prior art, the beneficial effects of the present invention are:

[0073] The present invention provides a pathological early warning method and device based on big data. By collecting the biological signals of a user in real time through sensors, the original acceleration signals of different parts of the user are obtained; the biological signals collected from the user are denoised and segmented, and a physiological signal database of the user is generated; the trunk state of the user is determined by extracting the data collected by the waist sensor device in the user's physiological signal database; the user's trunk feature matrix and the user's upper limb feature matrix are normalized, and a trunk state identification model and an upper limb state identification model are established to determine whether tremors occur in the user's upper limbs in a static state; it can accurately determine whether the user has diseases related to aging based on the user's physiological signals, and can help the user to give early warnings about the diseases associated with the aging physique.

[0074] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A pathological warning method based on big data, characterized in that: It includes the following steps: Real-time collect the biological signals of the user through the sensor device to obtain the original acceleration signals of different parts of the user; Denoise and segment the collected biological signals of the user, and generate a physiological signal database of the user; By extracting the data collected by the waist sensor device in the user's physiological signal database, calculate the average value of the absolute values of the accelerations in the X, Y, and Z axes respectively to judge the trunk state of the user; calculate the standard deviation of the accelerations in the X, Y, and Z axes respectively, and the correlation coefficients between the accelerations in the X, Y, and Z axes pairwise; generate a trunk feature vector of the user from the absolute value mean, standard deviation, and pairwise correlation coefficients of the X, Y, and Z axis acceleration data, and then merge the user trunk static state feature matrix and the user trunk non-static state feature matrix into a user trunk feature matrix, and create a corresponding 0-1 matrix with two rows and N columns; and according to the data collected by the left upper limb sensor device and the right upper limb sensor device in the user's physiological signal database, estimate the coefficients of the autoregressive model by the least squares method, and obtain five AR coefficients of the left upper limb data and five AR coefficients of the right upper limb data respectively, where the order of the AR model is set to 5; use the signal amplitude range and the five AR coefficients as the upper limb feature vector for identifying the upper limb state, and merge the upper limb tremor, static, and motion state feature matrices into a user upper limb feature matrix, and create a corresponding 0-1 matrix with three rows and N columns; Normalize the user trunk feature matrix and the user upper limb feature matrix, and establish a trunk state recognition model and an upper limb state recognition model to judge whether tremors occur in the upper limb when the user is in a static state.

2. The pathological warning method based on big data according to claim 1, characterized in that: The sensor device is provided with a collection module, a storage module, and a control module. The collection module is used to collect the biological signal data of the user, or obtain the acceleration data of the user's movement and posture, and send the collected data to the control module; the control module is connected to the storage module, and after the collection module finishes collecting the user's data, the control module stores the data collected by the collection module in the storage module.

3. The pathological warning method based on big data according to claim 2, characterized in that: The sensor device includes a waist sensor device, a left upper limb sensor device, and a right upper limb sensor device. The waist sensor device is fixed to the user's waist, and the left upper limb sensor device and the right upper limb sensor device are respectively fixed to the ends of the user's left and right upper limbs.

4. The pathological warning method based on big data according to claim 3, characterized in that: Denoise and segment the data collected by the waist sensor device, the left upper limb sensor device, and the right upper limb sensor device through a filter, and generate a physiological signal database of the user.

5. The pathological warning method based on big data according to claim 4, characterized in that: The steps of denoising the collected data through a filter include the following: Perform third-order filtering on the collected data; Determine the normalized cut-off frequency and order of the filter; Calculate the filtering coefficients; Filter through the filtering function.

6. The big data-based pathological early warning method according to claim 5, characterized in that: In the trunk state recognition model, the input vector is the user's trunk feature matrix. Among them, the user's trunk feature matrix includes 9 features, with a dimension of 9. It is set that the number of nodes in the input layer of the trunk state recognition model is 9, and the number of nodes in the output layer is 2. Then, when the output vector is 0-1, it represents static, and when it is 1-0, it represents non-static; in the upper limb state recognition model, the input vector is the user's upper limb feature matrix. Among them, the user's upper limb feature matrix includes 6 features, with a dimension of 6. It is set that the number of nodes in the input layer of the trunk state recognition model is 6, and the number of nodes in the output layer is 3. Then, when the output vector is 0-0-1, it represents tremor, when it is 0-1-0, it represents static, and when it is 1-0-0, it represents movement.

7. The big data-based pathological early warning method according to claim 6, characterized in that: When the output vector of the trunk state recognition model is 0-1 and the output vector of the upper limb state recognition model is 0-0-1, it is determined that the user has tremors in the upper limb in a static state, otherwise it does not hold.

Images