CN112957018A

CN112957018A - Heart state detection method and device based on artificial intelligence

Info

Publication number: CN112957018A
Application number: CN202110309837.4A
Authority: CN
Inventors: 李祥; 张闻涛; 冯禹; 万民乐
Original assignee: China Resources Life Sciences Group Co ltd; Tongxintang Health Technology Beijing Co ltd
Current assignee: China Resources Life Sciences Group Co ltd; Tongxintang Health Technology Beijing Co ltd
Priority date: 2021-03-23
Filing date: 2021-03-23
Publication date: 2021-06-15

Abstract

The application relates to a heart state detection method and device based on artificial intelligence. The method comprises the following steps: acquiring pulse wave data of a target object; carrying out characteristic analysis on the pulse wave data of the target object to obtain target characteristic data; the target characteristic data comprises a pulse characteristic of the target subject; sending the target characteristic data to a server, and receiving a heart state detection result fed back by the server; the heart state detection result is obtained by the server performing state detection according to a heart state detection model trained in advance and the target characteristic data, and the heart state detection result is used for indicating whether the heart of the target object is in a normal state. The method can reduce the detection cost and simplify the detection operation.

Description

Heart state detection method and device based on artificial intelligence

Technical Field

The present application relates to the field of artificial intelligence technology, and in particular, to a method and apparatus for detecting a cardiac state based on artificial intelligence.

Background

The heart is one of the most important organs of the human body, and the state of the heart directly affects the health of people. In recent years, the incidence of heart disease is increasing, and the population of the disease is gradually younger, so that it is very necessary to detect the heart state.

Currently, Electrocardiogram (ECG) is generally used for cardiac status detection. However, the electrocardiogram is limited in space and time, and can be detected only when the patient goes to a hospital. Although some home-use electrocardiograph devices can detect the heart state, the home-use electrocardiograph devices have problems of high device cost, complicated operation, and the like, and thus are not widely used.

Disclosure of Invention

In view of the above, there is a need to provide a method and an apparatus for detecting heart condition based on artificial intelligence, which can reduce the detection cost and simplify the detection operation.

The embodiment of the disclosure provides a heart state detection method based on artificial intelligence, which is applied to a terminal and comprises the following steps:

acquiring pulse wave data of a target object;

carrying out characteristic analysis on the pulse wave data of the target object to obtain target characteristic data; the target characteristic data comprises a pulse characteristic of the target object;

sending the target characteristic data to a server, and receiving a heart state detection result fed back by the server; the heart state detection result is obtained by the server according to the pre-trained heart state detection model and the target characteristic data, and the heart state detection result is used for indicating whether the heart of the target object is in a normal state or not.

In one embodiment, the pulse wave data includes pulse wave time domain data and pulse wave frequency domain data; the above-mentioned characteristic analysis of the pulse wave data of the target object to obtain target characteristic data includes:

performing time domain analysis on the pulse wave time domain data of the target object to obtain target time domain characteristic data;

and carrying out frequency domain analysis on the pulse wave frequency domain data of the target object to obtain target frequency domain characteristic data.

In one embodiment, the time domain analyzing the pulse wave time domain data of the target object to obtain the target time domain feature data includes at least one of the following:

calculating the average value of a plurality of RR intervals according to pulse wave time domain data of a target object to obtain an interval average value;

calculating standard deviations of a plurality of RR intervals according to pulse wave time domain data of the target object to obtain interval standard deviations;

calculating the root mean square of a plurality of RR intervals according to pulse wave time domain data of the target object to obtain interval root mean square;

calculating the difference value of every two adjacent RR intervals according to the pulse wave time domain data of the target object, and calculating the standard deviation of a plurality of difference values to obtain the standard deviation of the difference values;

and calculating the difference value of every two adjacent RR intervals according to the pulse wave time domain data of the target object, determining the difference value larger than the preset time length as a target difference value, and calculating the ratio of the number of the target difference values to the total number of the RR intervals to obtain an interval ratio.

In one embodiment, the performing frequency domain analysis on the pulse wave frequency domain data of the target object to obtain target frequency domain feature data includes:

determining a first frequency corresponding to pulse wave frequency domain data with the maximum amplitude according to the pulse wave frequency domain data of the target object, and doubling the first frequency to obtain a second frequency and a third frequency;

and determining a preset amount of pulse wave frequency domain data near the first frequency, the second frequency and the third frequency as target frequency domain characteristic data.

The embodiment of the disclosure provides a heart state detection method based on artificial intelligence, which is applied to a server and comprises the following steps:

receiving target characteristic data sent by a terminal; the target characteristic data is obtained by performing characteristic analysis on the pulse wave data of the target object after the terminal obtains the pulse wave data of the target object; the target characteristic data comprises a pulse characteristic of the target object;

carrying out state detection according to a heart state detection model trained in advance and target characteristic data to obtain a heart state detection result; the heart state detection result is used for indicating whether the heart of the target object is in a normal state or not;

and feeding back the heart state detection result to the terminal.

In one embodiment, before performing state detection according to the pre-trained cardiac state detection model and the target feature data to obtain a cardiac state detection result, the method further includes:

obtaining a sample set; the sample set comprises a plurality of sample characteristic data and labels of the sample characteristic data; the sample characteristic data comprises pulse characteristics of the training subject; labeling to indicate whether the heart of the training subject is in a normal state;

and training the neural network model based on the sample set to obtain a heart state detection model.

In one embodiment, the acquiring a sample set includes:

acquiring pulse wave data and electrocardiogram data of a plurality of training objects;

for each training object, performing characteristic analysis on the pulse wave data of the training object to obtain sample characteristic data, and performing characteristic analysis on the electrocardiogram data of the training object to obtain a label of whether the heart of the training object is in a normal state;

and obtaining a sample set according to the sample characteristic data and the labels of the training objects.

In one embodiment, the sample set comprises a training sample set, a validation sample set, and a test sample set; the training of the neural network model based on the sample set to obtain the cardiac state detection model comprises:

constructing a plurality of initial neural network models according to the number range of the layers of the neural network models and the number range of neurons of each layer in the neural network models;

training the initial neural network models based on a training sample set to obtain a plurality of candidate neural network models;

verifying the candidate neural network models based on the verification sample set to obtain the detection accuracy of each candidate neural network model;

determining the candidate neural network model with the highest detection accuracy as a target neural network model, testing the target neural network model based on the test sample set, adjusting adjustable parameters in the target neural network model to continue training if the detection accuracy of the target neural network model is smaller than a preset threshold, stopping training until the detection accuracy of the target neural network model is larger than or equal to the preset threshold, and determining the target neural network model when the training is stopped as a heart state detection model.

The embodiment of the disclosure provides a heart state detection device based on artificial intelligence, which is applied to a terminal and comprises:

the pulse wave acquisition module is used for acquiring pulse wave data of the target object;

the characteristic analysis module is used for carrying out characteristic analysis on the pulse wave data of the target object to obtain target characteristic data; the target characteristic data comprises a pulse characteristic of the target object;

the result receiving module is used for sending the target characteristic data to the server and receiving a heart state detection result fed back by the server; the heart state detection result is obtained by the server according to the pre-trained heart state detection model and the target characteristic data, and the heart state detection result is used for indicating whether the heart of the target object is in a normal state or not.

In one embodiment, the pulse wave data includes pulse wave time domain data and pulse wave frequency domain data; the above-mentioned feature analysis module includes:

the time domain analysis submodule is used for carrying out time domain analysis on the pulse wave time domain data of the target object to obtain target time domain characteristic data;

and the frequency domain analysis submodule is used for carrying out frequency domain analysis on the pulse wave frequency domain data of the target object to obtain target frequency domain characteristic data.

In one embodiment, the time domain analysis submodule is specifically configured to calculate an average value of a plurality of RR intervals according to pulse wave time domain data of the target object to obtain an interval average value; calculating standard deviations of a plurality of RR intervals according to pulse wave time domain data of the target object to obtain interval standard deviations; calculating the root mean square of a plurality of RR intervals according to pulse wave time domain data of the target object to obtain interval root mean square; calculating the difference value of every two adjacent RR intervals according to the pulse wave time domain data of the target object, and calculating the standard deviation of a plurality of difference values to obtain the standard deviation of the difference values; and calculating the difference value of every two adjacent RR intervals according to the pulse wave time domain data of the target object, determining the difference value larger than the preset time length as a target difference value, and calculating the ratio of the number of the target difference values to the total number of the RR intervals to obtain an interval ratio.

In one embodiment, the frequency domain analysis sub-module is specifically configured to determine, according to pulse wave frequency domain data of the target object, a first frequency corresponding to the pulse wave frequency domain data with the largest amplitude, and perform doubling processing on the first frequency to obtain a second frequency and a third frequency; and determining a preset amount of pulse wave frequency domain data near the first frequency, the second frequency and the third frequency as target frequency domain characteristic data.

The embodiment of the present disclosure provides a heart state detection device based on artificial intelligence, which is applied to a server, and the device includes:

the data receiving module is used for receiving target characteristic data sent by the terminal; the target characteristic data is obtained by performing characteristic analysis on the pulse wave data of the target object after the terminal obtains the pulse wave data of the target object; the target characteristic data comprises a pulse characteristic of the target object;

the state detection module is used for carrying out state detection according to a heart state detection model trained in advance and target characteristic data to obtain a heart state detection result; the heart state detection result is used for indicating whether the heart of the target object is in a normal state or not;

and the result sending module is used for feeding back the heart state detection result to the terminal.

In one embodiment, the apparatus further comprises:

the sample acquisition module is used for acquiring a sample set; the sample set comprises a plurality of sample characteristic data and labels of the sample characteristic data; the sample characteristic data comprises pulse characteristics of the training subject; labeling to indicate whether the heart of the training subject is in a normal state;

and the training module is used for training the neural network model based on the sample set to obtain the heart state detection model.

In one embodiment, the sample acquiring module is specifically configured to acquire pulse wave data and electrocardiogram data of a plurality of training subjects; for each training object, performing characteristic analysis on the pulse wave data of the training object to obtain sample characteristic data, and performing characteristic analysis on the electrocardiogram data of the training object to obtain a label of whether the heart of the training object is in a normal state or not; and obtaining a sample set according to the sample characteristic data and the labels of the training objects.

In one embodiment, the sample set comprises a training sample set, a validation sample set, and a test sample set; the training module is used for constructing a plurality of initial neural network models according to the number range of the layers of the neural network models and the number range of neurons of each layer in the neural network models; training the initial neural network models based on a training sample set to obtain a plurality of candidate neural network models; verifying the candidate neural network models based on the verification sample set to obtain the detection accuracy of each candidate neural network model; determining the candidate neural network model with the highest detection accuracy as a target neural network model, testing the target neural network model based on the test sample set, adjusting adjustable parameters in the target neural network model to continue training if the detection accuracy of the target neural network model is smaller than a preset threshold, stopping training until the detection accuracy of the target neural network model is larger than or equal to the preset threshold, and determining the target neural network model when the training is stopped as a heart state detection model.

The embodiment of the present disclosure provides a computer device, including a memory and a processor, where the memory stores a computer program, and the processor implements the following steps when executing the computer program:

acquiring pulse wave data of a target object;

The disclosed embodiments provide a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:

acquiring pulse wave data of a target object;

The heart state detection method and device based on artificial intelligence acquire pulse wave data of a target object; carrying out characteristic analysis on the pulse wave data of the target object to obtain target characteristic data; and sending the target characteristic data to a server, and receiving a heart state detection result fed back by the server. In the embodiment of the disclosure, the terminal acquires the pulse wave of the target object and sends the target characteristic data representing the pulse characteristic of the target object to the server for cardiac state detection.

Drawings

FIG. 1 is a diagram of an exemplary environment in which an artificial intelligence based cardiac state detection method may be implemented;

FIG. 2 is a schematic flow chart of an artificial intelligence based cardiac state detection method in one embodiment;

FIG. 3 is a flowchart illustrating the steps of performing feature analysis on the pulse wave data of the target object to obtain target feature data according to an embodiment;

FIG. 4a is a diagram illustrating time domain data of a pulse wave according to an embodiment;

FIG. 4b is a diagram illustrating the pulse wave frequency domain data according to an embodiment;

FIG. 5 is a schematic flow chart of a method for artificial intelligence based cardiac state detection in another embodiment;

FIG. 6 is a schematic flow chart of another embodiment of training a cardiac state detection model;

FIG. 7 is a second flowchart illustrating a method for training a cardiac state detection model according to another embodiment;

FIG. 8 is a schematic flow chart illustrating an artificial intelligence based cardiac state detection method according to yet another embodiment;

FIG. 9 is a block diagram of an artificial intelligence based cardiac state detection apparatus according to an embodiment;

FIG. 10 is a block diagram showing the structure of an artificial intelligence based heart state detecting apparatus according to another embodiment;

FIG. 11 is a diagram illustrating an internal structure of a computer device in one embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

The artificial intelligence based heart state detection method provided by the application can be applied to the application environment shown in fig. 1. The application environment includes a terminal 102 and a server 104; wherein the terminal 102 communicates with the server 104 via a network. The terminal 102 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, smart bands, smart watches, and other portable wearable devices, and the server 104 may be implemented by an independent server or a server cluster formed by a plurality of servers.

In one embodiment, as shown in fig. 2, there is provided an artificial intelligence based cardiac state detection method, which is illustrated by applying the method to the terminal in fig. 1, and includes the following steps:

step 201, obtaining pulse wave data of a target object.

The pulse wave is formed by the propagation of the heart along the artery and the blood flow to the periphery, and the pulse wave data can reflect the heart state of the measured object to a certain extent. The pulse wave data includes pulse wave time domain data and pulse wave frequency domain data.

The terminal can acquire pulse wave data of a target object; the pulse wave acquisition equipment can be connected externally in various connection modes, and the pulse wave time domain data of the target object can be acquired from the pulse wave acquisition equipment. And then, the terminal performs Fourier transform on the pulse wave time domain data to obtain pulse wave frequency domain data.

The pulse wave time domain data can be obtained in real time or according to a preset period. For example, the terminal acquires the pulse wave time domain data of the target object from the externally connected pulse wave acquisition device every 1 minute. The embodiment of the present disclosure does not limit the preset period.

The connection mode may include at least one of a Serial port, a bluetooth and a USB (Universal Serial Bus) interface; the pulse wave collecting device can be a photoelectric fingertip photoelectric volume pulse wave collecting device and can also be a pressure type radial pulse wave collecting device. The embodiments of the present disclosure do not limit this.

Step 202, performing feature analysis on the pulse wave data of the target object to obtain target feature data.

Wherein the target characteristic data comprises a pulse characteristic of the target subject.

After the terminal acquires the pulse wave data of the target object, the characteristic analysis is carried out on the pulse wave data of the target object. For example, time domain analysis and frequency domain analysis are performed on the pulse wave data of the target object to obtain target characteristic data. Or inputting the pulse wave data of the target object into a pre-trained feature extraction model to obtain target feature data output by the feature extraction model. The analysis mode is not limited in the embodiment of the disclosure, and can be selected according to actual conditions.

In one embodiment, the terminal performs noise reduction and filtering on the pulse wave data of the target object, and then performs feature analysis on the processed pulse wave data.

And step 203, sending the target characteristic data to a server, and receiving a heart state detection result fed back by the server.

The heart state detection result is obtained by the server according to the pre-trained heart state detection model and the target characteristic data, and the heart state detection result is used for indicating whether the heart of the target object is in a normal state or not.

And after the terminal obtains the target characteristic data, the terminal sends the target characteristic data to the server. Correspondingly, the server receives the target characteristic data sent by the terminal, and inputs the target characteristic data into a heart state detection model trained in advance to obtain a heart state detection result output by the heart state detection model.

For example, the terminal sends target characteristic data of the target object a to the server, the server receives the target characteristic data and inputs the target characteristic data into the heart state detection model, and the heart state detection model outputs a heart state detection result that the heart of the target object a is in a normal state. After a period of time, the terminal sends the target characteristic data of the target object A to the server, the server inputs the target characteristic data into the heart state detection model, and the heart state detection model outputs a heart state detection result that the heart of the target object A is in an abnormal state.

And after obtaining the heart state detection result, the server feeds the heart state detection result back to the terminal. Correspondingly, the terminal receives the heart state detection result fed back by the server.

In the heart state detection method based on artificial intelligence, a terminal acquires pulse wave data of a target object; carrying out characteristic analysis on the pulse wave data of the target object to obtain target characteristic data; and sending the target characteristic data to a server, and receiving a heart state detection result fed back by the server. In the embodiment of the disclosure, the terminal acquires the pulse wave of the target object and sends the target characteristic data representing the pulse characteristic of the target object to the server for cardiac state detection. Further, after the detection cost is reduced, the application range of the heart state detection can be expanded.

In one embodiment, the step of analyzing the pulse wave data of the target object to obtain the target characteristic data may include:

step 301, performing time domain analysis on the pulse wave time domain data of the target object to obtain target time domain characteristic data.

The target time domain feature data comprises at least one of RR intervals, interval mean NNVGR, interval standard deviation SDNN, interval root mean square RMSSD, difference standard deviation SDSD and interval ratio pNN 50. The RR interval is the duration between two peaks in the pulse wave.

The terminal performs time domain analysis on the pulse wave time domain data of the target object to obtain target time domain characteristic data, and the target time domain characteristic data may include at least one of the following: and calculating the average value of a plurality of RR intervals according to the pulse wave time domain data of the target object to obtain an interval average value NNVGR. Calculating standard deviations of a plurality of RR intervals according to pulse wave time domain data of a target object to obtain interval standard deviation SDNN; calculating the root mean square of a plurality of RR intervals according to pulse wave time domain data of a target object to obtain an interval root mean square RMSSD; calculating the difference value of every two adjacent RR intervals according to the pulse wave time domain data of the target object, and calculating the standard deviation of a plurality of difference values to obtain a standard deviation SDSD; and calculating the difference value of every two adjacent RR intervals according to the pulse wave time domain data of the target object, determining the difference value larger than the preset time length as a target difference value, and calculating the ratio of the number of the target difference values to the total number of the RR intervals to obtain an interval ratio pNN 50.

For example, as shown in fig. 4a, the pulse wave time domain data of the target object can determine that there are 6 RR intervals. And calculating the average value of 6 RR intervals to obtain an interval average value NNVGR. And calculating standard deviations of 6 RR intervals to obtain an interval standard deviation SDNN. And calculating the root mean square between the 1 st RR interval and the 2 nd RR interval, calculating the root mean square between the 2 nd RR interval and the 3 rd RR interval, and the like until the root mean square between the 5 th RR interval and the 6 th RR interval is calculated, and obtaining the 5 interval root mean square interval standard deviation SDNN. Calculating the difference between the 1 st RR interval and the 2 nd RR interval, calculating the difference between the 2 nd RR interval and the 3 rd RR interval, and so on, until calculating the difference between the 5 th RR interval and the 6 th RR interval, obtaining 5 differences; the standard deviation of these 5 differences is then calculated to yield the standard deviation of difference SDSD. And (3) setting the preset time length to be 50ms, determining the difference value larger than 50ms in the 5 calculated difference values as a target difference value, counting that the number of the target difference values is 2, and the total number of RR intervals is 6, and calculating the ratio of the number of the target difference values 2 to the total number of RR intervals 6 to obtain an interval ratio pNN 50. The embodiment of the present disclosure does not limit the preset duration.

Step 302, performing frequency domain analysis on the pulse wave frequency domain data of the target object to obtain target frequency domain characteristic data.

The terminal determines a first frequency corresponding to pulse wave frequency domain data with the maximum amplitude according to the pulse wave frequency domain data of the target object, and doubles the first frequency to obtain a second frequency and a third frequency; and then, determining a preset number of pulse wave frequency domain data near the first frequency, the second frequency and the third frequency as target frequency domain characteristic data.

For example, the pulse wave frequency domain data of the target object is as shown in fig. 4b, wherein the first frequency corresponding to the pulse wave frequency domain data with the largest amplitude is 85Hz, 2 times of 85Hz is 170Hz, and 3 times of 85Hz is 255Hz, then 170Hz is determined as the second frequency, and 255Hz is determined as the third frequency. Determining 10 pulse wave frequency domain data corresponding to the first frequency, namely X1 pulse wave frequency domain data and X1 pulse wave frequency domain data, as target frequency domain characteristic data; determining 10 pulse wave frequency domain data about the pulse wave frequency domain data X2 and the pulse wave frequency domain data X2 with the maximum amplitude near the second frequency as target frequency domain characteristic data; and determining 10 pulse wave frequency domain data of the left and right sides of the pulse wave frequency domain data X3 and the pulse wave frequency domain data X3 with the maximum amplitude near the third frequency as target frequency domain characteristic data to obtain 63 target frequency domain characteristic data in total. The preset number is not limited in the embodiment of the present disclosure.

The order of

steps

301 and 302 is not limited.

In the step of performing characteristic analysis on the pulse wave data of the target object to obtain target characteristic data, the terminal performs time domain analysis on the pulse wave time domain data of the target object to obtain target time domain characteristic data; and carrying out frequency domain analysis on the pulse wave frequency domain data of the target object to obtain target frequency domain characteristic data. In the embodiment of the disclosure, the terminal performs time domain analysis and frequency domain analysis on the pulse wave data to obtain target characteristic data which can reflect the pulse characteristics of the target object, so that the heart state can be detected according to the target characteristic data in the following process. Compared with the heart state detection by utilizing the electrocardiogram in the prior art, the embodiment of the disclosure utilizes the pulse wave to detect the heart state, so that the detection cost can be reduced, and the detection operation is simplified.

In one embodiment, as shown in fig. 5, there is provided an artificial intelligence based cardiac state detection method, which is illustrated by applying the method to the server in fig. 1, and includes the following steps:

step 401, receiving target characteristic data sent by a terminal.

The target characteristic data is obtained by performing characteristic analysis on the pulse wave data of the target object after the terminal obtains the pulse wave data of the target object; the target characteristic data includes a pulse characteristic of the target subject.

The terminal can acquire the pulse wave data of the target object and can also acquire the pulse wave data of the target object from an external pulse wave acquisition device. And then, the terminal performs characteristic analysis on the pulse wave data of the target object to obtain target characteristic data and sends the target characteristic data to the server. Correspondingly, the server receives the target characteristic data sent by the terminal.

And step 402, performing heart state detection according to the pre-trained heart state detection model and the target characteristic data to obtain a heart state detection result.

Wherein, the heart state detection result is used for indicating whether the heart of the target object is in a normal state.

The server trains a heart state detection model in advance, and after target characteristic data are received, the target characteristic data are input into the heart state detection model to obtain a heart state detection result output by the heart state detection model.

And step 403, feeding back the heart state detection result to the terminal.

And after obtaining the heart state detection result, the server sends the heart state detection result to the terminal. Correspondingly, the terminal receives the heart state detection result sent by the server.

In one embodiment, the server stores target characteristic data and cardiac state detection results of the target object for subsequent review.

In the heart state detection method based on artificial intelligence, target characteristic data sent by a terminal is received; carrying out state detection according to a heart state detection model trained in advance and target characteristic data to obtain a heart state detection result; and feeding back the heart state detection result to the terminal. In the embodiment of the disclosure, the server performs the heart state detection according to the target characteristic data which is sent by the terminal and reflects the pulse characteristic of the target object, compared with the prior art that the heart state detection is performed by using an electrocardiogram, the pulse wave acquisition equipment is relatively cheap, the detection cost can be reduced by performing the heart state detection by using the pulse wave, and meanwhile, the detection operation is simplified. Further, after the detection cost is reduced, the application range of the heart state detection can be expanded.

In one embodiment, as shown in fig. 6, on the basis of the above embodiment, the embodiment of the present disclosure may further include a process of training the cardiac state detection model:

step 501, a sample set is obtained.

The sample set comprises a plurality of sample characteristic data and labels of the sample characteristic data; the sample characteristic data comprises pulse characteristics of the training subject; the callout is used to indicate whether the heart of the training subject is in a normal state.

The method comprises the steps that a server obtains pulse wave data and electrocardiogram data of a plurality of training objects; for each training object, performing characteristic analysis on the pulse wave data of the training object to obtain sample characteristic data, and performing characteristic analysis on the electrocardiogram data of the training object to obtain a label of whether the heart of the training object is in a normal state; and obtaining a sample set according to the sample characteristic data and the labels of the training objects.

In practical applications, the number of labels for the heart of the training subject to be in a normal state is equal to the number of labels for the heart of the training subject to be in an abnormal state.

In one embodiment, the pulse wave data includes pulse wave time domain data and pulse wave frequency domain data; the pulse wave time domain data can be acquired by the server through the pulse wave acquisition equipment, and the pulse wave frequency domain data can be obtained by performing Fourier transform on the pulse wave time domain data. The server performs feature analysis on the pulse wave data of the training object, and may refer to the process of performing feature analysis on the pulse wave data of the target object by the terminal in the above embodiment.

In one embodiment, before performing feature analysis on the pulse wave data, the server performs noise reduction and filtering on the pulse wave data, and then performs feature analysis on the processed pulse wave data.

In one embodiment, the electrocardiographic data comprises electrocardiographic time domain data and electrocardiographic frequency domain data; the electrocardiogram time domain data can be acquired by the server through the electrocardiogram acquisition equipment, and the electrocardiogram frequency domain data can be acquired by the server through Fourier transform on the electrocardiogram time domain data. The server can receive a label of whether the heart of the training object is in a normal state or not, wherein the label is input by a user according to electrocardiogram time domain data of the training object; the electrocardiogram data of the training object can also be subjected to characteristic analysis to obtain a label of whether the heart of the training object is in a normal state.

The server performs feature analysis on the electrocardiogram data of the training object, and the feature analysis may include the following processes: the server compares the electrocardiogram data of the training object with the electrocardiogram data of which the heart is in a normal state, and determines that the heart of the training object is in a normal state if the electrocardiogram data of the training object is similar to the electrocardiogram data of which the heart is in a normal state.

In one example, the server performs noise reduction and filtering on the electrocardiographic data before analyzing the electrocardiographic data, and then analyzes the processed electrocardiographic data.

Step 502, training a neural network model based on a sample set to obtain a heart state detection model.

The activation function of the neural network model may adopt a Linear rectification function (ReLU) function, the loss function for training the neural network model may adopt a quadratic cost function (quadratic cost), and the learning rate of the neural network model may be set to 0.01. The embodiment of the present disclosure does not limit the activation function, the loss function, and the learning rate.

The sample set comprises a training sample set, a verification sample set and a test sample set, wherein the proportion of sample characteristic data in the training sample set, the verification sample set and the test sample set can be 8:1: 1. The embodiments of the present disclosure do not limit this. As shown in fig. 7, the training process may include the following steps:

step 5021, a plurality of initial neural network models are built according to the number range of the layers of the neural network models and the number range of neurons of each layer in the neural network models.

For example, the number of layers of the neural network model ranges from 3 to 8, the number of neurons in each layer ranges from 8 to 12, and 30 initial neural network models can be constructed by combining the number of layers and the number of neurons in pairs.

Step 5022, training the initial neural network models based on the training sample set to obtain candidate neural network models.

For each initial neural network model, the server inputs the sample characteristic data in the training sample set into the initial neural network model to obtain a training result output by the initial neural network model. And then, calculating a loss value between the training result and the label by using a loss function, if the loss value does not accord with a preset convergence condition, adjusting adjustable parameters in the initial neural network model to continue training until the loss value accords with the preset convergence condition, stopping training, and determining the initial neural network model when the training is stopped as a candidate neural network model.

For example, 30 initial neural network models are trained respectively to obtain 30 candidate neural network models.

And 5023, verifying the candidate neural network models based on the verification sample set to obtain the detection accuracy of each candidate neural network model.

For each candidate neural network model, the server inputs a plurality of sample characteristic data in the verification sample set into the candidate neural network model to obtain a verification result corresponding to each sample characteristic data output by the candidate neural network model. And the server judges whether the verification result of each sample characteristic data is consistent with the label of the sample characteristic data or not, and determines the detection accuracy of the candidate neural network model according to the judgment result.

For example, inputting m sample feature data into the candidate neural network model 1 to obtain m verification results; and determining the detection accuracy of the candidate neural network model as P (n/m) if n verification results are consistent with the labels and m-n verification results are inconsistent with the labels. By analogy, the detection accuracy of other candidate neural network models can be obtained.

Step 5024, determining the candidate neural network model with the highest detection accuracy as a target neural network model, testing the target neural network model based on the test sample set, adjusting adjustable parameters in the target neural network model to continue training if the detection accuracy of the target neural network model is smaller than a preset threshold, stopping training until the detection accuracy of the target neural network model is larger than or equal to the preset threshold, and determining the target neural network model with the training stopped as a heart state detection model.

And after the server obtains the detection accuracy of each candidate neural network model, determining the candidate neural network model with the highest detection accuracy as the target neural network. And then, the server inputs a plurality of sample characteristic data in the test sample set into the target neural network model to obtain a test result corresponding to each sample characteristic data output by the target neural network model. And the server determines the detection accuracy of the target neural network model according to the test result of each sample characteristic data and the label of the sample characteristic data. And if the detection accuracy is smaller than the preset threshold, adjusting the adjustable parameters in the target neural network model to continue training until the detection accuracy is larger than or equal to the preset threshold, and determining the target neural network model when the training is stopped as the heart state detection model.

For example, the preset threshold is 98%, if the server determines that the detection accuracy of the target neural network model is less than 98%, the adjustable parameters in the target neural network model are adjusted to continue training until the server determines that the detection accuracy is greater than or equal to 98%.

In one embodiment, the server receives the target characteristic data sent by the plurality of terminals and updates the heart state detection model according to the plurality of target characteristic data.

In the process of training the heart state detection model, the server acquires a sample set; and training the neural network model based on the sample set to obtain a heart state detection model. In the embodiment of the disclosure, the server performs training of the heart state detection model by using the sample characteristic data representing the pulse characteristics and the label indicating whether the heart is in a normal state or not obtained according to the electrocardiogram data, and the trained heart state detection model can convert heart state detection from electrocardiogram to pulse wave, so that not only can the detection accuracy be ensured, but also the detection cost can be reduced and the detection operation can be simplified.

In one embodiment, as shown in fig. 8, there is provided an artificial intelligence based cardiac state detection method comprising the steps of:

step 601, the terminal acquires pulse wave data of the target object.

Step 602, the terminal performs feature analysis on the pulse wave data of the target object to obtain target feature data.

In one embodiment, the terminal graphically displays the target feature data.

Step 603, the terminal sends the target characteristic data to the server.

In step 604, the server receives the target feature data sent by the terminal.

And step 605, the server performs state detection according to the pre-trained heart state detection model and the target characteristic data to obtain a heart state detection result.

Step 606, the server feeds back the heart state detection result to the terminal.

In step 607, the terminal receives the heart state detection result fed back by the server.

In one embodiment, the terminal graphically displays the heart state detection result.

And step 608, the terminal determines that the heart state detection result meets a preset alarm condition and outputs alarm information.

Wherein, the preset alarm condition may include: a heart state detection result that the heart of the target object is in an abnormal state is continuously acquired. For example, a cardiac state detection result that the heart of the target object is in an abnormal state, which is transmitted by one server, is received every 1 minute.

After the terminal determines that the heart state detection result meets the preset alarm condition, voice alarm information can be output, and vibration alarm information can also be output. The alarm information can be continuously output until the user triggers to stop alarming, or can be output only once. The alarm mode is not limited in the embodiment of the disclosure.

In the above disclosed embodiment, the terminal and the server perform heart state detection alternately, and compared with the prior art that cardiac state detection is performed by using an electrocardiogram, the pulse wave acquisition device is relatively cheap, and the detection cost can be reduced by using the pulse wave to perform heart state detection, and meanwhile, the detection operation is simplified. Further, after the detection cost is reduced, the application range of the heart state detection can be expanded.

It should be understood that, although the steps in the flowcharts of fig. 2 to 8 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 2 to 8 may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of performing the steps or stages is not necessarily sequential, but may be performed alternately or alternately with other steps or at least some of the other steps or stages.

In one embodiment, as shown in fig. 9, there is provided an artificial intelligence based heart state detecting apparatus applied to a terminal, the apparatus including:

a pulse wave obtaining module 701, configured to obtain pulse wave data of a target object;

a feature analysis module 702, configured to perform feature analysis on the pulse wave data of the target object to obtain target feature data; the target characteristic data comprises a pulse characteristic of the target object;

a result receiving module 703, configured to send the target feature data to the server, and receive a cardiac state detection result fed back by the server; the heart state detection result is obtained by the server according to the pre-trained heart state detection model and the target characteristic data, and the heart state detection result is used for indicating whether the heart of the target object is in a normal state or not.

In one embodiment, the pulse wave data includes pulse wave time domain data and pulse wave frequency domain data; the feature analysis module 702 includes:

In one embodiment, as shown in fig. 9, there is provided an artificial intelligence based heart state detecting apparatus applied to a server, the apparatus comprising:

a data receiving module 801, configured to receive target feature data sent by a terminal; the target characteristic data is obtained by performing characteristic analysis on the pulse wave data of the target object after the terminal obtains the pulse wave data of the target object; the target characteristic data comprises a pulse characteristic of the target object;

the state detection module 802 is configured to perform cardiac state detection according to a pre-trained cardiac state detection model and target feature data to obtain a cardiac state detection result; the heart state detection result is used for indicating whether the heart of the target object is in a normal state or not;

and a result sending module 803, configured to feed back the heart state detection result to the terminal.

In one embodiment, the apparatus further comprises:

In one embodiment, the sample acquiring module is specifically configured to acquire pulse wave data and electrocardiogram data of a plurality of training subjects; for each training object, performing characteristic analysis on the pulse wave data of the training object to obtain sample characteristic data; performing characteristic analysis on electrocardiogram data of the training object to obtain a label of whether the heart of the training object is in a normal state; and obtaining a sample set according to the sample characteristic data and the labels of the training objects.

For specific limitations of the artificial intelligence based cardiac state detection apparatus, reference may be made to the above limitations of the artificial intelligence based cardiac state detection method, which are not described herein again. The modules in the artificial intelligence based heart state detecting device can be wholly or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.

In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 10. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless communication can be realized through WIFI, an operator network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement an artificial intelligence based cardiac state detection method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.

Those skilled in the art will appreciate that the architecture shown in fig. 10 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:

acquiring pulse wave data of a target object;

sending the target characteristic data to a server, and receiving a heart state detection result fed back by the server; the heart state detection result is obtained by the server according to the heart state detection model trained in advance and the target characteristic data, and the heart state detection result is used for indicating whether the heart of the target object is in a normal state or not.

In one embodiment, the pulse wave data includes pulse wave time domain data and pulse wave frequency domain data; the processor, when executing the computer program, further performs the steps of:

In one embodiment, the processor, when executing the computer program, further performs the steps of:

In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:

acquiring pulse wave data of a target object;

In one embodiment, the pulse wave data includes pulse wave time domain data and pulse wave frequency domain data; the computer program when executed by the processor further realizes the steps of:

In one embodiment, the computer program when executed by the processor further performs the steps of:

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.

The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.

The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims

1. A heart state detection method based on artificial intelligence is applied to a terminal, and comprises the following steps:

acquiring pulse wave data of a target object;

carrying out characteristic analysis on the pulse wave data of the target object to obtain target characteristic data; the target characteristic data comprises a pulse characteristic of the target subject;

sending the target characteristic data to a server, and receiving a heart state detection result fed back by the server; the heart state detection result is obtained by the server performing state detection according to a heart state detection model trained in advance and the target characteristic data, and the heart state detection result is used for indicating whether the heart of the target object is in a normal state.

2. The method of claim 1, wherein the pulse wave data comprises pulse wave time domain data and pulse wave frequency domain data; the step of performing feature analysis on the pulse wave data of the target object to obtain target feature data includes:

3. The method of claim 2, wherein the time domain analyzing the pulse wave time domain data of the target object to obtain target time domain feature data comprises at least one of:

calculating the average value of a plurality of RR intervals according to pulse wave time domain data of the target object to obtain an interval average value;

and calculating the difference value of every two adjacent RR intervals according to the pulse wave time domain data of the target object, determining the difference value larger than a preset time length as a target difference value, and calculating the ratio of the number of the target difference values to the total number of the RR intervals to obtain an interval ratio.

4. The method of claim 2, wherein the frequency domain analyzing the pulse wave frequency domain data of the target object to obtain target frequency domain feature data comprises:

determining a first frequency corresponding to the pulse wave frequency domain data with the maximum amplitude according to the pulse wave frequency domain data of the target object, and performing doubling processing on the first frequency to obtain a second frequency and a third frequency;

and determining a preset amount of pulse wave frequency domain data around the first frequency, the second frequency and the third frequency as the target frequency domain characteristic data.

5. A heart state detection method based on artificial intelligence is applied to a server, and comprises the following steps:

receiving target characteristic data sent by a terminal; the target characteristic data is obtained by performing characteristic analysis on the pulse wave data of the target object after the terminal obtains the pulse wave data of the target object; the target characteristic data comprises a pulse characteristic of the target subject;

carrying out state detection according to a heart state detection model trained in advance and the target characteristic data to obtain a heart state detection result; the heart state detection result is used for indicating whether the heart of the target object is in a normal state or not;

and feeding back the heart state detection result to the terminal.

6. The method of claim 5, wherein before the performing the state detection based on the pre-trained cardiac state detection model and the target feature data to obtain the cardiac state detection result, the method further comprises:

obtaining a sample set; the sample set comprises a plurality of sample characteristic data and an annotation of each sample characteristic data; the sample feature data comprises a pulse feature of a training subject; the label is used for indicating whether the heart of the training object is in a normal state or not;

and training a neural network model based on the sample set to obtain the heart state detection model.

7. The method of claim 6, wherein obtaining the sample set comprises:

acquiring pulse wave data and electrocardiogram data of a plurality of training subjects;

for each training object, performing feature analysis on the pulse wave data of the training object to obtain sample feature data, and performing feature analysis on the electrocardiogram data of the training object to obtain a label of whether the heart of the training object is in a normal state;

and obtaining the sample set according to the sample characteristic data and the labels of the training objects.

8. The method of claim 6, wherein the sample set comprises a training sample set, a validation sample set, and a test sample set; the training of the neural network model based on the sample set to obtain the cardiac state detection model comprises:

training the plurality of initial neural network models based on the training sample set to obtain a plurality of candidate neural network models;

determining the candidate neural network model with the highest detection accuracy as a target neural network model, testing the target neural network model based on the test sample set, if the detection accuracy of the target neural network model is smaller than a preset threshold, adjusting adjustable parameters in the target neural network model to continue training until the detection accuracy of the target neural network model is larger than or equal to the preset threshold, stopping training, and determining the target neural network model when training is stopped as the heart state detection model.

9. A heart state detection device based on artificial intelligence is characterized in that the device is applied to a terminal and comprises:

the characteristic analysis module is used for carrying out characteristic analysis on the pulse wave data of the target object to obtain target characteristic data; the target characteristic data comprises a pulse characteristic of the target subject;

the result receiving module is used for sending the target characteristic data to a server and receiving a heart state detection result fed back by the server; the heart state detection result is obtained by the server performing state detection according to a heart state detection model trained in advance and the target characteristic data, and the heart state detection result is used for indicating whether the heart of the target object is in a normal state.

10. An artificial intelligence based heart condition detection device, applied to a server, the device comprising:

the data receiving module is used for receiving target characteristic data sent by the terminal; the target characteristic data is obtained by performing characteristic analysis on the pulse wave data of the target object after the terminal obtains the pulse wave data of the target object; the target characteristic data comprises a pulse characteristic of the target subject;

the state detection module is used for carrying out state detection according to a heart state detection model trained in advance and the target characteristic data to obtain a heart state detection result; the heart state detection result is used for indicating whether the heart of the target object is in a normal state or not;