Health assessment system and method based on wearable device multi-source data fusion analysis

Abstract

The invention discloses a health assessment system and method based on wearable device multi-source data fusion analysis. Wearable equipment health assessment features are obtained through the hearing feature vectors obtained through processing and the heart rate and blood pressure features based on a multi-source data fusion analysis method, and a wearable equipment health assessment model is constructed through the wearable equipment health assessment features and an adopted wearable health assessment result. According to the method, the convolutional neural network model is adjusted based on the auditory features acquired and processed by the wearable device in combination with the physiological parameters, so that data with highly structured features are efficiently processed, and the health assessment features of the wearable device, which are more suitable for fusion analysis of the multi-source data of the wearable device, are obtained by imaging the multi-source data, so that the health assessment accuracy of the wearable device is improved. And the trained model has less parameter quantity and stronger generalization ability, and the health assessment of the wearable device on the wearer is more objective under the multi-source processing of the health assessment features generated by fusion on the model.

Description

Technical Field

[0001] The present invention belongs to the technical field of artificial intelligence, and particularly relates to a health assessment system and method based on multi-source data fusion analysis of wearable devices. Background Art

[0002] Hearing refers to the ability to activate the auditory organs and receive speech information. Hearing is crucial for a person, and people use this ability to communicate quickly. Therefore, people have gradually improved the detection methods for hearing. Moreover, hearing ability is an important indicator for health assessment, and many literatures evaluate the health status of people based on hearing data. For example, the patent with the publication number CN107358034A comprehensively monitors the health of children through hearing indicators.

[0003] The prior art shows that, on the one hand, currently, for the detection of hearing, it is through manual operation of equipment to detect hearing according to the equipment. However, when it comes to analyzing the hearing tested by the equipment, it is all analyzed by personnel, which is time-consuming, reduces work efficiency, and is prone to errors. On the other hand, when using a model for health assessment, the hearing ability data has not been detected and expressed, and there is no detailed improvement direction of using wearable devices to comprehensively process basic physiological parameters for adaptive characterization to correct the model, so as to obtain a more objective portable health assessment result of the wearable device.

[0004] Therefore, how to use the existing hearing detection data for processing methods adapted to input into the traditional CNN neural network model, how to image-process the feature vector values of the traditional input model to correct the CNN neural network model, so as to obtain highly structured feature data, and how to obtain multi-source image data, and obtain health assessment features of wearable devices that are more suitable for multi-source data fusion analysis of wearable devices, so that the trained model has fewer parameters and stronger generalization ability. Under the multi-source processing of the health assessment features generated by fusion, the wearable device can more objectively assess the health of the wearer. Therefore, the present invention proposes a health assessment system and method based on multi-source data fusion analysis of wearable devices. Summary of the Invention

[0005] To solve the above technical problems, the present invention proposes a health assessment system and method based on multi-source data fusion analysis of wearable devices.

[0006] In the first aspect of the present invention, a health assessment method based on multi-source data fusion analysis of wearable devices is provided. The method includes: S1. Collect and process to obtain the auditory feature vector of the wearer, acquire the heart rate and blood pressure features of the wearer based on the wearable device, obtain the health assessment features of the wearable device based on the multi-source data fusion analysis method, and obtain the health assessment result of the wearer; S2. The multi-source data fusion analysis method is to image the multi-source data of the auditory feature vector, the heart rate and the blood pressure feature, and obtain the image health assessment feature of the wearable device with image format features; S3. Input the health assessment features of the wearable device and the health assessment result into a neural network model improved based on the image health assessment features of the wearable device to construct a health assessment model for the wearable device; S4. Perform personnel health assessment on the personnel wearing the wearable device again or newly based on the health assessment model of the wearable device.

[0007] Further, the health assessment model of the wearable device is a CNN neural network model improved based on the image health assessment features of the wearable device.

[0008] Further, the auditory feature vector is calculated from the detection difference and the analysis difference and age data, detection distance data, noise data, analysis influence factors, detection influence factors, age influence factors, noise influence factors and distance influence factors.

[0009] Further, the heart rate and blood pressure features of the wearer are composed of a two-dimensional feature vector consisting of the maximum heart rate, minimum heart rate, minimum blood pressure and maximum blood pressure of the wearer.

[0010] Further, the multi-source data imaging specifically refers to filling the four-dimensional feature vector with the auditory feature vector based on the heart rate and blood pressure features of the wearer.

[0011] Further, the detection difference is obtained through the difference between the start time and end time of the wear detection of the wearer, and the analysis difference is obtained through the difference between the start time data and end time data of the wear assessment analysis.

[0012] Further, the health assessment model of the wearable device uses the filling vector value based on the auditory feature vector to correct the activation function of the CNN neural network model.

[0013] There is also provided a health assessment system based on multi-source data fusion analysis of a wearable device, including an auditory feature processing and generating module, a wearable device heart rate and blood pressure feature acquisition module, a multi-source data imaging and generating module, a wearable device health assessment model construction module, and a personnel health assessment module: The auditory feature processing and generation module: used to collect and process to obtain the auditory feature vector of the wearer; The heart rate and blood pressure feature acquisition module of the wearable device: obtain the heart rate and blood pressure features of the wearer according to the wearable device; The multi-source data imaging generation module: obtain the health assessment features of the wearable device based on the multi-source data fusion analysis method, and the multi-source data fusion analysis method is to perform multi-source data imaging on the auditory feature vector and the heart rate and blood pressure features to obtain the wearable device image health assessment features with image format features; The wearable device health assessment model construction module: construct a wearable device health assessment model based on the wearable device health assessment features and the health assessment results input into a neural network model improved based on the wearable device image health assessment features; The personnel health assessment module: perform personnel health assessment on the personnel wearing the wearable device again or newly wearing the wearable device according to the wearable device health assessment model.

[0014] Further, the wearable device health assessment model is a CNN neural network model improved based on the wearable device image health assessment features.

[0015] Further, the auditory feature vector is calculated from the detection difference, the analysis difference, age data, detection distance data, noise data, analysis influence factors, detection influence factors, age influence factors, noise influence factors and distance influence factors.

[0016] On the one hand, when analyzing the hearing of the personnel wearing the wearable device in the present invention, it is all through simply collecting the hearing-related data, which improves the work efficiency and is not easy to make mistakes; on the other hand, using the model for health assessment is to obtain the wearable device health assessment features based on the processed auditory feature vector and heart rate and blood pressure features through the multi-source data fusion analysis method, and construct a wearable device monitoring and assessment model by using the wearable device health assessment features and the taken wearable health assessment results to obtain a more objective portable health assessment result of the wearable device.

[0017] Therefore, the beneficial effects of the present invention are as follows: using the existing hearing detection data to fuse multi-modal basic physiological parameters, processing them with an improved CNN neural network model adapted to the input based on the feature image format features, that is, performing image processing on the feature vector values of the traditional input model to correct the CNN neural network model, thereby obtaining highly structured feature data, and obtaining multi-source image data based on the mode of feature convolution by referring to and using the auditory vector filling values, and obtaining wearable device health assessment features more suitable for the multi-source data fusion analysis of wearable devices, making the trained model have fewer parameters and stronger generalization ability. Under the multi-source processing of the health assessment features generated by the fusion, the wearable device's health assessment of the wearer is more objective.

[0018] More embodiments and improvement effects of the present invention will be further introduced in combination with the drawings and specific embodiments. Brief Description of the Drawings

[0019] Figure 1 is a flowchart of a health assessment method based on multi-source data fusion analysis of wearable devices according to the present invention; Figure 2 is a schematic diagram of a health assessment system based on multi-source data fusion analysis of wearable devices according to the present invention; Figure 3 is an example diagram of a conventional padding convolution in an embodiment of the present invention; Figure 4 is a schematic diagram of the activation function principle of the CNN neural network model in an embodiment of the present invention; Figure 5 is a schematic diagram of the structure of an electronic device for implementing the method of the present invention in an embodiment of the present invention. Detailed Embodiments

[0020] Next, in combination with the drawings and specific embodiments, a further description of the invention will be made. The model used in the present invention is an improved version of the CNN neural network model.

[0021] As Figure 2 shown, the system of the present invention belongs to the field of artificial intelligence such as the development of application software such as biometric recognition software, so it belongs to artificial intelligence and also belongs to the next-generation information network industry.

[0022] In the first aspect of the present invention, a health assessment method based on multi-source data fusion analysis of wearable devices is provided. The method includes: S1. Collect and process the auditory feature vectors of the wearer, obtain the heart rate and blood pressure characteristics of the wearer based on the wearable device, obtain wearable device health assessment features based on the multi-source data fusion analysis method, and obtain the health assessment result of the wearer; S2. The multi-source data fusion analysis method is to image the auditory feature vector, the heart rate and blood pressure features into multi-source data to obtain the wearable device image health assessment features with image format features; S3. Input the wearable device health assessment features and the health assessment results into a neural network model improved based on the wearable device image health assessment features to construct a wearable device health assessment model; S4. Conduct a personnel health assessment on the personnel wearing the re-worn or newly worn wearable device according to the wearable device health assessment model.

[0023] Furthermore, the wearable device health assessment model is a CNN neural network model improved based on the wearable device image health assessment features. The CNN neural network is a commonly used model for processing image features. By using the image format features to improve the convolution padding of the CNN neural network model and further correcting the calculation method of the activation function, the classification and processing generalization ability of the model for multi-source fusion monitoring data is improved, thereby enhancing the accurate and objective health assessment accuracy of the model.

[0024] In another embodiment of the present invention, the auditory feature vector is calculated from the detection difference, the analysis difference, age data, detection distance data, noise data, analysis influencing factors, detection influencing factors, age influencing factors, noise influencing factors, and distance influencing factors:

[0025] Among them, is denoted as the auditory feature vector, is a specific feature vector value, J is denoted as the detection distance data, Z is denoted as the noise data, G is denoted as the age data, D is denoted as the detection difference, Y is denoted as the analysis difference, J, Z, G, D, and Y are all one-dimensional feature vector values, , , , , are respectively denoted as the distance influencing factor, the noise influencing factor, the age influencing factor, the detection influencing factor, and the analysis influencing factor. The above influencing factors are used to remove the dimension influence. e is denoted as the calculation deviation adjustment factor of the auditory feature vector. The value of e generally takes a value between 1.05 and 1.1. This is to ensure that the deviation of the auditory feature vector will not be too large, resulting in a large result deviation in the input model calculation and increasing the stability of the trained model.

[0026] Among them, the calculation of each influencing factor is obtained through certain innovative calculations by the wearer of the wearable device. Each influencing factor affects the calculation and acquisition of the auditory feature vector. By calculating with controlled variables for the same wearer of the wearable device, characteristic values that have a more obvious impact on the auditory feature vector can be obtained, thereby contributing more accurate input feature vector data for subsequent training of the CNN neural network model.

[0027] S1: Select wearable device wearers with the same detection distance data, detection difference, analysis difference, noise data, age data, and actual hearing data. Among them, if the detection distance data is different, obtain the corresponding hearing data, and respectively label the two different hearing data as T1 and T2, and label the two different detection distance data as J1 and J2. Calculate the difference between the two detection distance data, and substitute J1, J2, T1, and T2 into the calculation formula: (J1 - J2) * = (T1 - T2), and calculate the value of , which is expressed as the influencing factor of the detection distance data on the hearing detection, and label it as the distance influencing factor; S2: Select wearable device wearers with the same detection distance data, detection difference, analysis difference, noise data, age data, and actual hearing data. Among them, if the noise data is different, obtain the corresponding hearing data, and respectively label the two different hearing data as T3 and T4, and label the two different noise data as Z1 and Z2. Substitute T3, T4, Z1, and Z2 into the calculation formula: (Z1 - Z2) * = (T3 - T4), and calculate the value of , which is expressed as the influencing factor of the noise data on the hearing detection, and label it as the noise influencing factor; S3: Select wearable device wearers with the same detection distance data, detection difference, analysis difference, noise data, and actual hearing data. Among them, if the age data is different, obtain the corresponding hearing data, and respectively label the two different hearing data as T6 and T5, and label the two different age data as N1 and N2. Substitute N1 and N2, T6 and T5 into the calculation formula: (N1 - N2) * = (T5 - T6), and calculate the value of , which is expressed as the influencing factor of the age data on the hearing detection, and label it as the age influencing factor; S4: Select wearable device wearers with the same detection distance data, analysis difference, noise data, age data, and actual hearing data. Among them, if the detection differences are different, obtain the corresponding hearing data, label the two different hearing data as T7 and T8 respectively, label the two different detection differences as C1 and C2 respectively, and substitute C1 and C2, and T7 and T8 into the calculation formula: (C1 - C2) * = (T7 - T8), and calculate the value of , which is expressed as the influence factor of the detection difference on hearing detection, and label it as the detection influence factor; S5: Select wearable device wearers with the same detection distance data, detection difference, noise data, age data, and actual hearing data. Among them, if the analysis differences are different, obtain the corresponding hearing data, label the two different hearing data as T9 and T10 respectively, label the two different analysis differences as F1 and F2 respectively, calculate the difference between the two detection distance data, and substitute F1 and F2, and T9 and T10 into the calculation formula: (F1 - F2) * = (T9 - T10), and calculate the value of , which is expressed as the influence factor of the analysis difference on hearing detection, and label it as the analysis influence factor; Furthermore, the heart rate and blood pressure characteristics of the wearer consist of a two-dimensional feature vector composed of the maximum heart rate, minimum heart rate, minimum blood pressure, and maximum blood pressure of the wearer. [[ID=IS=18]]

[0028] Furthermore, the multi-source data is visualized, specifically referring to filling the auditory feature vector with a four-dimensional feature vector based on the heart rate and blood pressure characteristics of the wearer. The filling operation can keep the spatial size of the two-dimensional feature vector unchanged, which helps to increase hearing health-related information and can better retain edge information at the same time. The calculation method of its filling value is:

[0029] In the formula, represents the filling vector value, represents the auditory feature vector, D represents the detection difference, and Y represents the analysis difference. The specific four-dimensional feature vector is the image health assessment feature of the wearable device, that is, the four-dimensional feature vector is .

[0030] Furthermore, the detection difference is obtained through the difference between the start time and end time of the wear detection of the wearer, and the analysis difference is obtained through the difference between the start time data and end time data of the wear assessment analysis.

[0031] Further, for the wearable device health assessment model, the activation function of the CNN neural network model is corrected by using the filling vector value based on the auditory feature vector, and the specific expression of the activation function is:

[0032] In the formula, is the activation function value, is the filling vector value, 、 、 、 、 respectively represent the distance influence factor, the noise influence factor, the age influence factor, the detection influence factor and the analysis influence factor. e represents the calculation deviation adjustment factor of the auditory feature vector, and the value of e generally ranges from 1.05 to 1.1.

[0033] The present invention adopts the supplementary input of the input feature by using the filling vector value of the input feature, which can be more suitable for the processing efficiency of the present invention to input the input feature as an image feature, and the filling method of the feature vector matrix based on the convolutional filling can generalize the activation function of the CNN neural network, so as to obtain a more objective classification for the subsequent output health assessment result, thereby improving the applicability of the wearable device, increasing the robustness of the health assessment of the wearable device, and making the trained model have a stable and reliable endorsement for the classification result.

[0034] A health assessment system based on multi-source data fusion analysis of wearable devices is also provided, including an auditory feature processing and generation module, a wearable device heart rate and blood pressure feature acquisition module, a multi-source data image generation module, a wearable device health assessment model construction module, and a personnel health assessment module; The auditory feature processing and generation module: is used to collect and process the auditory feature vector of the wearing person; The wearable device heart rate and blood pressure feature acquisition module: obtains the heart rate and blood pressure features of the wearing person according to the wearable device; The multi-source data image generation module: obtains the wearable device health assessment feature based on the multi-source data fusion analysis method. The multi-source data fusion analysis method is to perform multi-source data imaging on the auditory feature vector and the heart rate and blood pressure features to obtain the wearable device image health assessment feature with image format features; The wearable device health assessment model construction module: constructs a wearable device health assessment model based on the wearable device health assessment feature and the health assessment result input into the neural network model improved based on the wearable device image health assessment feature; The personnel health assessment module: conducts personnel health assessment on the personnel wearing the re - worn or newly - worn wearable device according to the wearable device health assessment model.

[0035] Further, the wearable device health assessment model is a CNN neural network model improved based on the health assessment features of the wearable device image. The CNN neural network is a commonly used model for processing image features. By using the image format features to improve the convolution padding of the CNN neural network model and further correcting the calculation method of the activation function, the classification and processing generalization ability of the model for multi - source fusion monitoring data is improved, thereby enhancing the accurate and objective health assessment accuracy of the model.

[0036] Further, the auditory feature vector is calculated from the detection difference and the analysis difference and age data, detection distance data, noise data, analysis influence factor, detection influence factor, age influence factor, noise influence factor, and distance influence factor:

[0037] Among them, is denoted as the auditory feature vector, J is denoted as the detection distance data, Z is denoted as the noise data, G is denoted as the age data, D is denoted as the detection difference, Y is denoted as the analysis difference, , , , , are respectively denoted as the distance influence factor, noise influence factor, age influence factor, detection influence factor, and analysis influence factor, and e is denoted as the calculation deviation adjustment factor of the auditory feature vector.

[0038] On the one hand, when analyzing the hearing of the personnel wearing the wearable device, this invention always uses the simply - collected hearing - related data, which improves work efficiency and is not prone to errors. On the other hand, when using the model for health assessment, it obtains the health assessment features of the wearable device by using the processed auditory feature vector, heart rate, and blood pressure features based on the multi - source data fusion analysis method, and constructs a wearable device monitoring and assessment model by using the health assessment features of the wearable device and the obtained wearable health assessment results to obtain a more objective portable health assessment result of the wearable device.

[0039] Therefore, the beneficial effect of the present invention is to use the existing hearing detection data to fuse basic physiological parameters in a multi-modal manner and process them using a CNN neural network model improved based on the feature image format adapted to the input, that is, to image the feature vector values of the traditional input model to correct the CNN neural network model, so as to obtain highly structured feature data, and use the reference of the auditory vector filling value based on the feature convolution mode to obtain multi-source image data, and obtain wearable device health assessment features more suitable for multi-source data fusion analysis of wearable devices, so that the trained model has fewer parameters and stronger generalization ability. Under the multi-source processing of the model by the health assessment features generated by fusion, the wearable device can more objectively assess the health of the wearer.

[0040] Of course, it can be understood that each embodiment of the present invention can achieve one of the effects alone, and the combination of multiple embodiments of the present invention can achieve all the above effects. However, it is not required that each embodiment of the present invention achieve all the above advantages and effects, because each embodiment of the present invention can constitute an independent technical solution and make one or more contributions to the prior art. Each embodiment of the present invention is not affected, and deleting one of the embodiments, the solution can still be implemented.

[0041] For the part of the module structure not specifically defined in the present invention, it shall be subject to the content recorded in the prior art. The prior art mentioned in the foregoing background art part and specific embodiment part of the present invention can be used as a part of the present invention to understand the meaning of some technical features or parameters.

Claims

1. A health assessment method based on multi-source data fusion analysis of wearable devices, characterized in that The method includes: S1. Collect and process to obtain the auditory feature vector of the wearer, and obtain the heart rate and blood pressure characteristics of the wearer based on the wearable device. Based on the multi-source data fusion analysis method, obtain the health assessment features of the wearable device, and obtain the health assessment result of the wearer; S2. The multi-source data fusion analysis method is to image the multi-source data of the auditory feature vector, the heart rate and the blood pressure characteristics to obtain the image health assessment features of the wearable device with image format characteristics; S3. Input the health assessment features of the wearable device and the health assessment result into a neural network model improved based on the image health assessment features of the wearable device to construct a health assessment model of the wearable device; S4. According to the health assessment model of the wearable device, conduct a health assessment on the personnel wearing the wearable device again or newly.

2. The health assessment method based on multi-source data fusion analysis of the wearable device according to claim 1, wherein: The health assessment model of the wearable device is a CNN neural network model improved based on the image health assessment features of the wearable device.

3. The health assessment method based on multi-source data fusion analysis of the wearable device according to claim 1 or 2, wherein: The auditory feature vector is calculated from the detection difference and the analysis difference and age data, detection distance data, noise data, analysis influence factor, detection influence factor, age influence factor, noise influence factor and distance influence factor.

4. The health assessment method based on multi-source data fusion analysis of the wearable device according to claim 3, wherein: The heart rate and blood pressure characteristics of the wearer are composed of a two-dimensional feature vector consisting of the maximum heart rate, minimum heart rate, minimum blood pressure and maximum blood pressure of the wearer.

5. The health assessment method based on multi-source data fusion analysis of the wearable device according to claim 4, wherein: The multi-source data imaging specifically refers to filling the four-dimensional feature vector with the auditory feature vector based on the heart rate and blood pressure characteristics of the wearer.

6. The health assessment method based on multi-source data fusion analysis of the wearable device according to claim 3, wherein: The detection difference is obtained through the difference between the start time and end time of the wear detection of the wearer, and the analysis difference is obtained through the difference between the start time data and end time data of the wear assessment analysis.

7. The health assessment method based on multi-source data fusion analysis of the wearable device according to claim 5, wherein: The health assessment model of the wearable device uses the filling vector value based on the auditory feature vector to correct the activation function of the CNN neural network model.

8. A health assessment system based on multi-source data fusion analysis of the wearable device, which implements the method according to claim 1, including an auditory feature processing and generation module, a wearable device heart rate and blood pressure feature acquisition module, a multi-source data imaging and generation module, a wearable device health assessment model construction module, and a personnel health assessment module, wherein: The auditory feature processing and generation module: for collecting and processing to obtain the auditory feature vector of the wearer; The heart rate and blood pressure feature acquisition module of the wearable device: obtaining the heart rate and blood pressure features of the wearer according to the wearable device; The multi-source data imaging generation module: obtaining the health assessment features of the wearable device based on the multi-source data fusion analysis method, and the multi-source data fusion analysis method is to perform multi-source data imaging on the auditory feature vector and the heart rate and blood pressure features to obtain the wearable device image health assessment features with image format features; The wearable device health assessment model construction module: constructing a wearable device health assessment model based on the wearable device health assessment features and the health assessment results input into a neural network model improved based on the wearable device image health assessment features; The personnel health assessment module: performing personnel health assessment on the personnel wearing the wearable device again or newly according to the wearable device health assessment model.

9. The health assessment system based on multi-source data fusion analysis of a wearable device according to claim 8, wherein: The wearable device health assessment model is a CNN neural network model improved based on the wearable device image health assessment features.

10. The health assessment system based on multi-source data fusion analysis of a wearable device according to claim 9, wherein: The auditory feature vector is calculated from the detection difference and the analysis difference and age data, detection distance data, noise data, analysis influence factors, detection influence factors, age influence factors, noise influence factors and distance influence factors.

Images