CN113558634A

CN113558634A - Data monitoring method and device, electronic equipment and storage medium

Info

Publication number: CN113558634A
Application number: CN202110845813.0A
Authority: CN
Inventors: 胡照埝; 温万惠; 刘光远; 陈实; 胡潇柔; 李闪闪
Original assignee: Southwest University
Current assignee: Southwest University
Priority date: 2021-07-26
Filing date: 2021-07-26
Publication date: 2021-10-29

Abstract

The application provides a data monitoring method, a data monitoring device, electronic equipment and a storage medium, wherein the method comprises the following steps: determining an RR interval of a group of electrocardiosignals according to the occurrence time and the occurrence intensity of each electrocardiosignal in the group of electrocardiosignals of the target user; determining an RR interval sequence according to the RR interval; taking the RR interphase in the RR interphase sequence as input data, and obtaining at least two characteristic values of the group of electrocardiosignals through different preset submodels; and combining at least one characteristic value according to a preset combination mode so as to input at least one combined result into a monitoring model as a group to be detected, marking the group of electrocardiosignal marks corresponding to the group to be detected by the monitoring model for representing the learning state of the student, and monitoring the learning condition of the student according to the marks. Through the method, the learning state of the student can be monitored.

Description

Data monitoring method and device, electronic equipment and storage medium

Technical Field

The application relates to the technical field of education informatization, in particular to a data monitoring method and device, electronic equipment and a storage medium.

Background

The learning state of the students determines the teaching quality of the teachers, and in the actual teaching process, the teachers subjectively judge the learning states of the students according to the performances of the students in the classroom.

The inventor finds that in the prior art, a teacher can not correctly judge the learning state of a student in class, and needs a method for monitoring the learning state of the student in order to comprehensively know the learning state of the student.

Disclosure of Invention

In view of this, embodiments of the present application provide a data monitoring method, an apparatus, an electronic device, and a storage medium, so as to monitor a learning state of a student.

In a first aspect, an embodiment of the present application provides a data monitoring method, including:

after a group of electrocardiosignals of a target user in a target time period are obtained, determining the occurrence time of at least two wave crests of the group of electrocardiosignals according to the occurrence time and the occurrence intensity of each electrocardiosignal in the group of electrocardiosignals;

calculating the difference value between the occurrence moments of every two adjacent wave crests according to the sequence of the occurrence moments, taking the difference value as an interval value of an RR interval, and taking the time period between the occurrence moments of every two adjacent wave crests as the occurrence time period of the RR interval;

after calculating an interval value for at least one RR interval, determining a sequence of RR intervals comprising the RR interval; the RR intervals are arranged according to a calculated sequence, and comprise the interval values and the occurrence time intervals of the RR intervals;

taking the RR intervals in the RR interval sequence as input data, and respectively inputting the RR intervals into at least two preset submodels for calculating characteristic values to obtain at least two characteristic values for expressing the fluctuation characteristics of the group of electrocardiosignals; the different preset submodels are used for calculating characteristic values of the fluctuation characteristics on different dimensions;

combining at least one characteristic value according to a preset combination mode to take at least one combined result as a group to be detected; wherein the preset combination mode is determined by a sequence backward selection method SBS;

after the group to be detected is input into the monitoring model, the electrocardiosignal marks of the group corresponding to the group to be detected are marked with the mark for representing the learning state of the student through the monitoring model, so that the learning condition of the student is monitored according to the mark.

In one possible embodiment, the monitoring model is trained by:

for each learning state, after at least one group of electrocardiosignals of a preset number of target users are acquired, marking a first identification for representing the student in the learning state for each group of electrocardiosignals; wherein the learning state comprises: knowledge learning state and problem solving state, learning stage cognitive load matching state and mismatching state, testing stage cognitive load matching state and mismatching state, knowledge point difficulty and group learning ability matching state and mismatching state, and mental fatigue and fatigue state;

aiming at each group of electrocardiosignals carrying the first identifier, at least two characteristic values of the group of electrocardiosignals in different dimensions are calculated through the preset submodel respectively;

forming at least one training group of the group of electrocardiosignals by the at least two characteristic values according to a fully-combined combination mode; the combination mode of the full combination is a plurality of combination modes formed by taking 1, 2, … and n different characteristic values from n different characteristic values each time; n is the dimension number of the characteristic value;

according to each combination mode of a training set, after training sets of a preset number of target users are obtained, transmitting a training set formed by the preset number of training sets into the monitoring model, and marking each group of electrocardiosignals corresponding to each training set in the training set through a classifier in the monitoring model; the classifier comprises a proximity algorithm KNN, a support vector machine SVM-rbf with a kernel function being a radial basis function rbf and a decision tree DT;

for each classifier, determining the identifier of each group of electrocardiosignals corresponding to each training group in the training set in the learning state by the classifier through a leave-one method, and marking a second identifier for each group of electrocardiosignals carrying a first identifier; wherein the first identifier and the second identifier are both the identifiers; the marks are negative and positive;

and counting the comparison result of the first identifier and the second identifier of each group of electrocardiosignals, and determining the recognition rate Accuracy of each classifier for each training set according to the comparison result.

In a possible embodiment, for each learning state, taking a training group corresponding to Accuracy exceeding a preset threshold as a target training group, and determining a combination mode of one of the target training groups as a preset combination mode in the learning state according to a preset method;

for each preset combination mode, performing weighted calculation on a first recognition rate of a target training group corresponding to the preset combination mode and a second recognition rate of a verification group corresponding to the preset combination mode to obtain a comprehensive score obtained by a target classifier in the preset combination mode; wherein the target classifier is a classifier enabling the target training set to obtain the Accuracy; the combination mode, the classifier and the calculation method of the recognition rate of the verification group and the target training group are the same; each group of electrocardiosignals corresponding to the verification group is different from each group of electrocardiosignals corresponding to the target training group;

when the comprehensive score which can be achieved by a target classifier in the preset combination mode is greater than or equal to a preset score, fixing the target classifier as a classifier for monitoring the group to be detected; wherein, the group to be detected is composed according to the preset combination mode.

In one possible embodiment, the calculation formula of Accuracy is:

wherein N is_FPNumber of false positive samples, N_TPNumber of true positive samples, N_TNNumber of true negative samples, N_FNRepresenting the number of false negative samples; the false positive sample is a training group with the first identification being negative and the second identification being positive; the true positive sample is a training group of which the first identification and the second identification are both positive; the true negative sample is a training group of which the first identification and the second identification are negative; the false negative sample is a training group with the first mark being positive and the second mark being negative.

In one possible embodiment, the calculation of the feature values is performed by inputting the input data into a preset model; the preset model comprises preset submodels used for calculating characteristic values of different dimensions;

the preset model is used for:

calculating a first preset ratio of the number of the input data with the interval value exceeding a first preset value to the total number of the input data, and a second preset ratio of the number of the input data with the interval value exceeding a second preset value to the total number of the input data, so as to take the first preset ratio as a first characteristic value and the second preset ratio as a second characteristic value;

calculating a ratio of a standard deviation to an average value of the input data, and taking the ratio of the standard deviation to the average value as a third characteristic value;

calculating approximate entropy ApEn, sample entropy SampEn, semi-major axis SD1 and semi-minor axis SD2 in a Poincare scatter diagram, and the ratio of the SD1 to the SD2 of the RR interval sequence, so that the ApEn serves as a fourth characteristic value, the SampEn serves as a fifth characteristic value, the SD1 serves as a sixth characteristic value, and the ratio of the SD1 to the SD2 serves as a seventh characteristic value; the Poincare scattergram is drawn by using interval values of two continuous input data in the RR interval sequence as an abscissa and an ordinate respectively; ApEn, SampEn, SD1 and SD2 are all calculated by corresponding MATLAB calculation programs;

calculating a first slope alpha of a fitting straight line formed by input data of a first preset part in a Heart Rate Variability (HRV) curve graph by a calculation program of an input Detrending Fluctuation Analysis (DFA)₁A second slope α of a fitted straight line formed by input data of a second predetermined section₂A ratio α of the second slope to the first slope₂/α₁To convert the alpha into₂As an eighth characteristic value, the α is₂/α₁As a ninth eigenvalue; wherein, the HRV curve chart is a curve chart which is drawn according to the input data and is used for describing the variation situation of the heartbeat speed;

calculating an average value of the input data, an average value of an absolute value of a first derivative of the input data, and a power of the input data at a preset frequency portion in the HRV graph, so as to take the average value of the input data as a tenth characteristic value, take the average value of the absolute value of the first derivative of the input data as an eleventh characteristic value, and take the power as a twelfth characteristic value.

In a second aspect, an embodiment of the present application further provides a data monitoring apparatus, including:

the analysis unit is used for determining the occurrence time of at least two wave crests of a group of electrocardiosignals according to the occurrence time and the occurrence intensity of each electrocardiosignal in the group of electrocardiosignals after the group of electrocardiosignals of a target user in a target time period are acquired;

the first calculating unit is used for calculating the difference value between the occurrence moments of every two adjacent wave crests according to the sequence of the occurrence moments, taking the difference value as an interval value of an RR interval, and taking the time period between the occurrence moments of every two adjacent wave crests as the occurrence time period of the RR interval;

a data unit for determining an RR interval sequence comprising at least one RR interval after calculating an interval value of the RR interval; the RR intervals are arranged according to a calculated sequence, and comprise the interval values and the occurrence time intervals of the RR intervals;

the second calculation unit is used for inputting the RR intervals in the RR interval sequence as input data into at least two preset submodels for calculating characteristic values respectively to obtain at least two characteristic values for expressing the fluctuation characteristics of the group of electrocardiosignals; the different preset submodels are used for calculating characteristic values of the fluctuation characteristics on different dimensions;

the first combination unit is used for combining at least one characteristic value according to a preset combination mode so as to take at least one combined result as a group to be detected; wherein the preset combination mode is determined by a sequence backward selection method SBS;

and the first marking unit is used for marking the group of electrocardiosignals corresponding to the group to be detected with an identifier for representing the learning state of a student through the monitoring model after the group to be detected is input into the monitoring model, so as to monitor the learning condition of the student according to the identifier.

In one possible embodiment, the monitoring model in the labeling unit is trained by:

the second marking unit is used for marking a first identifier used for representing the student in the learning state for each group of electrocardiosignals after acquiring at least one group of electrocardiosignals of a preset number of target users for each learning state; wherein the learning state comprises: knowledge learning state and problem solving state, learning stage cognitive load matching state and mismatching state, testing stage cognitive load matching state and mismatching state, knowledge point difficulty and group learning ability matching state and mismatching state, and mental fatigue and fatigue state;

the third calculation unit is used for calculating at least two characteristic values of each group of electrocardiosignals carrying the first identifier in different dimensions through the preset submodel;

the second combination unit is used for combining the at least two characteristic values into at least one training group of the group of electrocardiosignals according to a fully combined combination mode; the combination mode of the full combination is a plurality of combination modes formed by taking 1, 2, … and n different characteristic values from n different characteristic values each time; n is the dimension number of the characteristic value;

the data transmission unit is used for transmitting a training set formed by a preset number of training groups into the monitoring model after the training groups of a preset number of target users are obtained according to each combination mode of the training groups, so that each group of electrocardiosignals corresponding to each training group in the training set is marked through a classifier in the monitoring model; the classifier comprises a proximity algorithm KNN, a support vector machine SVM-rbf with a kernel function being a radial basis function rbf and a decision tree DT;

a third marking unit, configured to determine, for each classifier, an identifier of each group of electrocardiographic signals corresponding to each training group in the training set in the learning state by a leave-one-out method, so as to mark a second identifier for each group of electrocardiographic signals carrying the first identifier; wherein the first identifier and the second identifier are both the identifiers; the marks are negative and positive;

and the counting unit is used for counting the comparison result of the first identifier and the second identifier of each group of electrocardiosignals so as to determine the recognition rate Accuracy of each classifier for each training set according to the comparison result.

In one possible embodiment, the apparatus further comprises:

setting a combination unit, which is used for regarding each learning state, taking a training group corresponding to the Accuracy exceeding a preset threshold value as a target training group, and determining a combination mode of one target training group as a preset combination mode in the learning state according to a preset method;

the score calculation unit is used for carrying out weighted calculation on the first recognition rate of the target training group corresponding to each preset combination mode and the second recognition rate of the verification group corresponding to the preset combination mode so as to obtain a comprehensive score obtained by the target classifier in the preset combination mode; wherein the target classifier is a classifier enabling the target training set to obtain the Accuracy; the combination mode, the classifier and the calculation method of the recognition rate of the verification group and the target training group are the same; each group of electrocardiosignals corresponding to the verification group is different from each group of electrocardiosignals corresponding to the target training group;

the classifier determining unit is used for fixing the target classifier as the classifier for monitoring the group to be detected when the comprehensive score which can be reached by the target classifier in the preset combination mode is greater than or equal to a preset score; wherein, the group to be detected is composed according to the preset combination mode.

In one possible embodiment, the calculation formula of Accuracy is:

the preset model is used for:

In a third aspect, an embodiment of the present application further provides an electronic device, including: a processor, a storage medium and a bus, the storage medium storing machine-readable instructions executable by the processor, the processor and the storage medium communicating over the bus when the electronic device is operated, the processor executing the machine-readable instructions to perform the steps of the method according to any one of the first aspect.

In a fourth aspect, this application further provides a computer-readable storage medium, on which a computer program is stored, which, when executed by a processor, performs the steps of the method according to any one of the first aspect.

According to the embodiment of the application, the RR interval sequence of the group of electrocardiosignals is determined through the obtained electrocardiosignals of the target user, the characteristic values of multiple dimensions of the group of electrocardiosignals corresponding to the RR interval sequence are obtained through the RR interval sequence, and the fluctuation condition of the current group of electrocardiosignals is reflected through the sizes of different characteristic values. Therefore, according to the fluctuation condition of the characteristic value reaction, the group of electrocardiosignals are marked with the marks for representing the learning state of the student through the monitoring model. For the monitoring model, the standard of the mark for each group of electrocardiosignal marks is the same, and compared with the method in the prior art that a teacher subjectively judges the learning state of students according to the expressions, expressions and forms of the students, the mark of the monitoring model has more objectivity and authenticity. By the method, when the target user is set to be a class-wide classmate or a specific classmate, the teacher can know the learning state of the class-wide classmate or the specific classmate according to the identification marked by the monitoring model, the learning state of the student can be monitored, and the problem that the teacher cannot correctly judge the learning state of the student in class in the prior art is solved.

In order to make the aforementioned objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.

Fig. 1 shows a flowchart of a data monitoring method provided in an embodiment of the present application.

Fig. 2 shows a flowchart of another data monitoring method provided in an embodiment of the present application.

Fig. 3 shows a schematic structural diagram of a data monitoring device provided in an embodiment of the present application.

Fig. 4 shows a schematic structural diagram of an electronic device provided in an embodiment of the present application.

Detailed Description

In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it should be understood that the drawings in the present application are for illustrative and descriptive purposes only and are not used to limit the scope of protection of the present application. Additionally, it should be understood that the schematic drawings are not necessarily drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of the present application. It should be understood that the operations of the flow diagrams may be performed out of order, and steps without logical context may be performed in reverse order or simultaneously. One skilled in the art, under the guidance of this application, may add one or more other operations to, or remove one or more operations from, the flowchart.

In addition, the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.

It should be noted that in the embodiments of the present application, the term "comprising" is used to indicate the presence of the features stated hereinafter, but does not exclude the addition of further features.

It should be noted that the apparatuses, electronic devices, and the like according to the embodiments of the present application may be executed on a single server or may be executed in a server group. The server group may be centralized or distributed. In some embodiments, the server may be local or remote to the terminal. For example, the server may access information and/or data stored in the service requester terminal, the service provider terminal, or the database, or any combination thereof, via the network. As another example, the server may be directly connected to at least one of the service requester terminal, the service provider terminal and the database to access the stored information and/or data. In some embodiments, the server may be implemented on a cloud platform; by way of example only, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud (community cloud), a distributed cloud, an inter-cloud, a multi-cloud, and the like, or any combination thereof.

In addition, the apparatus or the electronic device related to the embodiment of the present application may be implemented on an access device or a third-party device, and specifically may include: a mobile device, a tablet computer, a laptop computer, or a built-in device in a motor vehicle, etc., or any combination thereof. In some embodiments, the mobile device may include a smart home device, a wearable device, a smart mobile device, a virtual reality device, an augmented reality device, or the like, or any combination thereof. In some embodiments, the smart home devices may include a control device of a smart electrical device, a smart monitoring device, a smart television, a smart camera, or an intercom, or the like, or any combination thereof. In some embodiments, the wearable device may include a smart bracelet, a smart helmet, a smart watch, a smart accessory, and the like, or any combination thereof. In some embodiments, the smart mobile device may include a smartphone, a Personal Digital Assistant (PDA), a gaming device, a navigation device, or the like, or any combination thereof. In some embodiments, the virtual reality device and/or the augmented reality device may include a virtual reality helmet, an augmented reality helmet, or the like, or any combination thereof. For example, the virtual reality device and/or augmented reality device may include various virtual reality products and the like.

Example one

Fig. 1 is a flowchart of a data monitoring method according to an embodiment of the present application, and as shown in fig. 1, the method is implemented by the following steps:

step 101, after a group of electrocardiosignals of a target user in a target time period is obtained, determining the occurrence time of at least two wave crests of the group of electrocardiosignals according to the occurrence time and the occurrence intensity of each electrocardiosignal in the group of electrocardiosignals.

Specifically, a detection device capable of detecting an electrocardiosignal is worn for a target user, and the detection device sends the detected electrocardiosignal to a data processing center in a wireless or wired mode such as bluetooth. The target user can be one or a plurality of users; when a plurality of target users are available, one set of detection equipment is worn for each target user, and meanwhile, a plurality of groups of electrocardiosignals of the target users in a target time period are obtained through each set of detection equipment. In the embodiment of the application, the data processing center completes data processing based on a computer. The duration of the target period is preset and can be adjusted.

For each group of acquired electrocardiosignals, a series of processing such as denoising and the like needs to be performed on the electrocardiosignals, and the electrocardiosignals are subjected to smoothing processing. And determining the wave crest of the group of electrocardiosignals and the time when the wave crest occurs according to the occurrence time and the occurrence intensity of each electrocardiosignal in the group of electrocardiosignals. The peak can be determined by the electrocardiogram of the electrocardiosignal or by the calculation of each electrocardiosignal data in the group of electrocardiosignals.

And 102, calculating a difference value between the occurrence moments of every two adjacent wave crests according to the sequence of the occurrence moments, taking the difference value as an interval value of an RR interval, and taking a time period between the occurrence moments of every two adjacent wave crests as an occurrence time period of the RR interval.

Specifically, the RR interval is the time length between two consecutive R waves in the QRS complex (the combination of waveforms reflecting the depolarization potentials and time variation of the left and right ventricles) on the electrocardiogram; the interval value of the RR interval is the duration between two R waves corresponding to the RR interval. The wave crest is the wave crest where the R wave is located. For each group of acquired electrocardiosignals, each electrocardiosignal in the group of electrocardiosignals is arranged in the group of electrocardiosignals from small to large according to the occurrence time of the electrocardiosignal, so that the calculation of the interval value of the group of electrocardiosignals is also carried out according to the sequence of the occurrence time. After the occurrence time of at least two peaks of the set of electrocardiographic signals is determined in step 101, the following data can be acquired.

For example, assume that the peaks and the occurrence time of the peaks of a set of processed electrocardiographic signals are:

number of wave crests	Time(s) of occurrence of peak of R wave	Peak size of R wave (mv)
			1	0.2	0.08
2	0.8	0.085
			3	1.2	0.08
4	2.2	0.09

Then, after determining the data, the RR intervals of the set of cardiac signals are:

number of RR intervals	RR interval occurrence time(s)	Interval value(s) of RR interval
			1	0.2-0.8	0.6
2	0.8-1.2	0.4
			3	1.2-2.2	1

Step 103, after calculating an interval value of at least one RR interval, determining an RR interval sequence containing the RR interval; the RR intervals are arranged according to the calculated sequence, and comprise the interval values and the occurrence time intervals of the RR intervals.

Specifically, according to the RR intervals calculated in step 102, it may be determined that the RR interval sequence of the set of cardiac signals is: 0.6s, 0.4s, 1 s. It can be known that the sequence of RR intervals in the RR interval sequence is arranged according to the sequence of the occurrence periods of the RR intervals.

Step 104, inputting the RR intervals in the RR interval sequence as input data into at least two preset submodels for calculating characteristic values respectively to obtain at least two characteristic values for representing the fluctuation characteristics of the group of electrocardiosignals; wherein, different preset submodels are used for calculating the characteristic values of the fluctuation characteristics on different dimensions.

Specifically, the RR interval sequence is input as input data into the preset submodel according to the RR interval sequence determined in step 103. Each preset submodel can correspondingly calculate a characteristic value of one dimension, and the preset submodel comprises a calculation formula or a calculation program which can calculate the characteristic value. The calculated characteristic values can reflect the time domain, frequency domain and nonlinear characteristics of the set of electrocardiosignals from different dimensions.

Step 105, combining at least one characteristic value according to a preset combination mode to take at least one combined result as a group to be detected; wherein the preset combination is determined by SBS (Sequential Backward Selection).

Specifically, the SBS selects a plurality of the feature values, calculates a group with the highest recognition rate among the remaining feature value combinations after reducing the number of the feature values, and repeats this until reducing the number of the feature values in the preset combination manner to the preset combination number, thereby reducing the amount of calculation in the operation by reducing the number of the feature values in the preset combination manner. The preset combination number is preset and can be adjusted. The preset combination mode comprises a characteristic value of preset combination quantity. In the embodiment of the application, a plurality of learning states are set, and a corresponding preset combination mode is determined for each learning state through the SBS. And combining the characteristic values according to the characteristic values of multiple dimensions calculated in the step 104, and taking multiple combination results of the characteristic values formed according to multiple preset combination modes as multiple groups to be detected of the electrocardiosignals.

And 106, after the group to be detected is input into the monitoring model, marking an identifier for representing the learning state of the student on the group of electrocardiosignals corresponding to the group to be detected through the monitoring model, so as to monitor the learning condition of the student according to the identifier.

Specifically, each group of electrocardiosignals corresponds to a plurality of learning states, in each learning state, the monitoring model can mark one group of electrocardiosignals corresponding to the group to be detected, and by marking the group of electrocardiosignals with an identifier for indicating the learning state of the student, when only one group of electrocardiosignals is monitored, the learning state of the student corresponding to the group of electrocardiosignals can be determined according to the identifier; when multiple groups of electrocardiosignals are monitored, the comprehensive learning state of multiple students corresponding to the multiple groups of electrocardiosignals can be determined according to the statistical condition of the marks marked on the multiple groups of electrocardiosignals. In the embodiment of the application, after the learning identifiers of the students in different learning states are determined, the learning states, the identifiers and the statistical results of the identifiers of the students can be displayed.

In a possible implementation, fig. 2 is a flow chart of another data monitoring method provided in the present application, and as shown in fig. 2, it can be seen that, when step 106 is executed, the monitoring model in step 106 is trained by the following steps:

step 201, for each learning state, after at least one group of electrocardiosignals of a preset number of target users is acquired, marking a first identification used for representing the student in the learning state for each group of electrocardiosignals; wherein the learning state comprises: knowledge learning state and problem solving state, learning stage cognition load matching state and unmatched state, testing stage cognition load matching state and unmatched state, knowledge point difficulty and group learning ability matching state and unmatched state, and mental fatigue and fatigue state.

In particular, teaching tasks based on cognitive load theory are more efficient than traditional teaching tasks because they require less training time and less mental effort to achieve the same or better learning and migration results. The learners' information processing ability is restricted by their mental fatigue state in addition to the working memory. When the learner is under a high-intensity cognitive load for a long time, the learner is easy to have a psychophysiological state of mental fatigue, and the cognitive resources cannot be effectively utilized, so that the information processing efficiency is reduced. The following three groups of states of the monitoring learning process are significant for improving the learning effect, and are respectively: (1) information input/processing and processing/extraction states, (2) states of cognitive load matching and mismatching, (3) states of mental fatigue and non-fatigue.

Therefore, the plurality of learning states are divided according to the cognitive load theory, and in the embodiment of the application, the learning states are divided into: knowledge learning state and problem solving state, learning stage cognition load matching state and unmatched state, testing stage cognition load matching state and unmatched state, knowledge point difficulty and group learning ability matching state and unmatched state, and mental fatigue and fatigue state.

And training the monitoring model to monitor a plurality of learning states of each group of electrocardiosignals through the trained monitoring model. Therefore, in order to improve the accuracy of monitoring the monitoring model for each learning state, it is necessary to train the monitoring model for each learning state using a plurality of sets of electrocardiographic signals in the learning state. In the embodiment of the application, the learning states of students are divided according to an educational theory, 5 learning states are defined, each learning state corresponds to two learning conditions, five binary problems are modeled according to the five learning states, a rule of a calibration identifier in each learning state is defined, an identifier is marked for student data corresponding to each learning state according to the defined calibration rule, and a monitoring model is trained through the student data containing the identifier in a supervised learning mode. The marks are divided into negative marks and positive marks. The negative and positive respectively represent different learning conditions under the same learning state, the positive represents that the learning condition of the student is ideal, and the negative represents that the learning condition of the student is not ideal. The negative and positive representatives have different meanings for different learning states. When labeling was performed, 0 represents positive and 1 represents negative.

The data source of each learning state and the calibration rule of the data in the learning state are introduced as follows:

acquiring at least one group of electrocardiosignals of a preset number of target users in a knowledge learning state and a problem solving state. The electrocardiosignals in the knowledge learning state and the problem solving state are electrocardiosignals acquired when students listen to the class (knowledge learning state) and participate in class tests (problem solving state). For example, when the preset number is 5, 5 groups of electrocardiosignals corresponding to 5 classmates in the knowledge learning state and the problem solving state are respectively collected; when the preset number is 10, 10 groups of electrocardiosignals corresponding to 10 classmates in the knowledge learning state and the problem solving state are respectively acquired. Wherein, the acquisition duration of each group of electrocardiosignals is preset and can be modified. In the knowledge learning state and the problem solving state, the first identifier of each group of electrocardiosignals (data of the knowledge learning state) acquired during the class listening is marked as 0, and the first identifier of each group of electrocardiosignals (data of the problem solving state) acquired during the class testing is marked as 1.

In the learning stage, the electrocardiosignals in the cognitive load matching state and the non-matching state are electrocardiosignals collected by students when the students practice classroom test questions; under the cognitive load matching state and the non-matching state in the learning stage, marking a first identification mark of electrocardiosignals (data of the cognitive load matching state in the learning stage) of the students with the scores of the class test questions being larger than or equal to a first preset score as 0; the first identification mark of the electrocardiosignal (data of the state of mismatching of the cognitive load in the learning stage) of the student with the score lower than the preset score is 1.

In the testing stage, the electrocardiosignals in the cognitive load matching state and the non-matching state are electrocardiosignals collected by students when the students take examination test questions; under the cognitive load matching state and the non-matching state in the testing stage, marking a first identification mark of an electrocardiosignal of a student with the score of an examination test question being greater than or equal to a second preset score (the cognitive load matching state in the testing stage) as 0; the first identification of the electrocardiosignals of the students with scores lower than the preset score (the state of mismatching of the cognitive load in the test stage) is marked as 1.

The first preset fraction and the second preset fraction may be specific preset values, or may be (X) according to the formula D_h-X_l) Obtained as/W. Wherein, W is the total score of the test questions, X_hRank average score of top A% student's score for test question score, X_lThe average score of A% of student results after the score of the test questions is ranked. When the first preset score is calculated, taking W as the total score of the class test questions; when calculating the second preset score, taking W as the total score of the test questions; a is a discrimination value set in advance, and for example, a — 27 may be set; by making use ofVarying the size of A, X_h、X_lChanges occur to adjust the value of D. And according to the actual situation, determining an ideal numerical value of D, and taking A which enables D to reach the ideal numerical value as an actual distinguishing value. When the test questions are classroom test questions, taking the scores of the students at A% as first preset scores; and when the test question is an examination test question, taking the score of the student at the A% position as a second preset score.

The electrocardiosignals in the state of matching the knowledge point difficulty with the group learning ability and the state of mismatching are electrocardiosignals of the students when the students learn new knowledge; under the matching state and the mismatching state of the knowledge point difficulty and the group learning ability, marking each group of electrocardiosignals (data of the matching state of the knowledge point difficulty and the group learning ability) corresponding to the students with the excellent rate of the student test results of 30-50% as 0 according to the knowledge point difficulty investigation result and the test question results submitted by the students; marking each group of electrocardiosignals (data of the state that the difficulty of the knowledge points is not matched with the group learning ability) corresponding to students with the excellent rate of the test results of 30-50% as 1.

According to the electrocardiosignals of each group of the same student in different learning states and the fatigue condition investigation result of the student, marking the electrocardiosignals (the data of mental fatigue) of each group corresponding to the student with the fatigue investigation result as 1; each group of electrocardiographic signals (mental fatigue data) corresponding to the students whose investigation results are non-fatigue is marked as 0.

Step 202, for each group of electrocardiosignals carrying the first identifier, calculating at least two characteristic values of the group of electrocardiosignals in different dimensions through the preset submodel respectively.

Specifically, after marking a first identifier for indicating that the student is in the learning state for each group of electrocardiographic signals in step 201, an RR interval corresponding to each group of electrocardiographic signals is calculated, and an RR interval sequence corresponding to each group of electrocardiographic signals is obtained. And inputting the RR interval sequence of the group of electrocardiosignals into a monitoring model, and calculating characteristic values of multiple dimensions through a plurality of preset submodels in the monitoring model. The method for calculating the characteristic value of each group of electrocardiographic signals carrying the first identifier is the same as the method for calculating the characteristic value in step 104.

Step 203, forming at least one training group of the group of electrocardiosignals by the at least two characteristic values according to a fully-combined combination mode; the combination mode of the full combination is a plurality of combination modes formed by taking 1, 2, … and n different characteristic values from n different characteristic values each time; n is the dimension number of the eigenvalue.

Specifically, a combination of all combinations is exemplified, for example, the feature values are a and b; the combination modes of the full combination respectively comprise: a; b; a. b three combination modes. Similarly, the characteristic values are a, b and c; the combination mode of the full combination includes: a; b; c; a. b; a. c; b. c; a. b, c; seven combination modes. By analogy, 2 composed of n eigenvalues can be obtainedⁿ-1 combination.

After step 202 is executed, at least two characteristic values of each group of electrocardiographic signals are acquired, and for each group of electrocardiographic signals, the at least two characteristic values are combined into at least one training group of the group of electrocardiographic signals in a fully combined manner. The combination mode of the training set is any one of the full combination modes, and a training set under the electrocardiosignal of the set is generated for each combination mode. The dimensionality of the characteristic values calculated by each group of electrocardiosignals is the same, namely the number of the combination modes of the training groups corresponding to each group of electrocardiosignals is also the same. That is, assuming that the dimension n of the feature value is 2, the combination methods of the feature values are 3 in total, and the training group corresponding to each group of the electrocardiographic signals is 3 groups.

204, after acquiring training groups of a preset number of target users according to each combination mode of the training groups, transmitting a training set formed by the preset number of training groups into the monitoring model, so as to mark each group of electrocardiosignals corresponding to each training group in the training set through a classifier in the monitoring model; wherein, the classifier comprises KNN (K-nearest neighbor, proximity algorithm), SVM-rbf (support vector machine with kernel function of radial basis function rbf), DT (Decision Tree).

Specifically, the monitoring model includes a classifier for labeling each group of electrocardiographic signals corresponding to each combination of feature values. In step 203, a corresponding number of training sets has been generated for each group of electrocardiographic signals, for example, each group of electrocardiographic signals corresponds to 3 training sets. Since each group of electrocardiographic signals is composed in the 3 manners, for each combination manner of the training set, a training set composed of training sets of a plurality of target users composed in the same combination manner is used as a training set in the combination manner. That is, assuming that the preset number of target users is 30, for each target user, each group of electrocardiographic signals corresponding to each target user corresponds to 3 training groups; the 90 training groups corresponding to the 30 target users are classified according to the combination mode of each training group, and the training groups with the same combination mode are put into the same training set, so that 3 training sets are generated, wherein each training set corresponds to one combination mode of the training groups, each training set comprises 30 training groups in the combination mode, and each training group corresponds to one group of electrocardiosignals of one target user. After the grouping of the training sets is completed, for each training set, the training set is passed into the monitoring model to perform step 205.

Step 205, for each classifier, determining, by a leave-one-out method, an identifier of each group of electrocardiographic signals in the learning state, corresponding to each training group in the training set, so as to mark a second identifier for each group of electrocardiographic signals carrying a first identifier; wherein the first identifier and the second identifier are both the identifiers; the mark is divided into negative and positive.

Specifically, the leave-one method is that, assuming that the training set includes 30 training groups carrying the first identifier, 29 training groups are used as samples to train the classifier each time training is performed, the remaining 1 training group is used as a test sample, and the training results of the classifier are tested by the first 29 training groups.

After step 204 is executed, a training set formed by a plurality of training groups carrying first identifiers is transmitted into the monitoring model, assuming that the number of the training groups in the training set is 30, for each training set, applying a leave-one method, taking the 30 training groups as test samples in sequence, training the classifier 30 times by using the training set, and after the training of 30 times is finished, marking a second identifier for each training group as a test sample by the classifier.

Step 206, counting the comparison result between the first identifier and the second identifier of each group of electrocardiographic signals, so as to determine Accuracy (recognition rate) of each classifier for each training set according to the comparison result.

Specifically, after the training is finished in step 205, each group of electrocardiographic signals corresponding to each training group in each training set carries a first identifier and a second identifier. Comparing the first identification and the second identification of each group of electrocardiosignals, judging whether the negative and positive of the first identification and the second identification are the same correspondingly, counting the comparison result of each group of electrocardiosignals, and calculating the Accuracy of each classifier to each training set according to the counted result. As can be seen from step 204, the combination manner of the training sets in each training set is different, and therefore, the Accuracy obtained by each classifier for the training sets formed by the training sets in different combination manners is also different.

In a possible embodiment, after the step 206 is executed, according to the obtained Accuracy of each classifier for each training set, the preset combination mode needs to be determined by the following steps:

step 301, regarding each learning state, taking a training group corresponding to Accuracy exceeding a preset threshold as a target training group, and determining a combination mode of one of the target training groups as a preset combination mode in the learning state according to a preset method.

Specifically, the preset threshold is set according to actual conditions, and the higher the preset threshold is, the higher the Accuracy calculated by the classifier is required to be. According to the step 204, the monitoring model includes 3 classifiers of KNN, SVM-rbf and DT, the 3 training sets in the step 204 are trained respectively for the 3 classifiers, and a combination mode of 9 classifiers and training sets is obtained in total, that is, 9 Accuracy are obtained, and each Accuracy is obtained by a training result of one classifier on one training set. Comparing the 9 accuracys with a preset threshold respectively, taking the training groups in the training sets corresponding to the accuracys exceeding the preset threshold as target training groups, wherein for each training set, the combination mode of all the training groups in the training set is the same, so that any one training group in the training set can be taken as a target training group. Therefore, assuming that all of the 9 accuracys exceed the preset threshold, the 9 accuracys correspond to nine target training sets. And determining the combination mode of one target training set as a preset combination mode in the learning state. For example, 9 accuracys corresponding to the combination mode of 9 classifiers and training sets may select a training group in the combination of the classifier and the training set corresponding to the highest Accuracy as a target training group; or, the selection may be performed according to the dimension of the feature values in the training set and the size of the 9 accuacy.

For example, the 3 classifiers are: KNN, SVM-rbf, DT. Setting 3 dimensionality eigenvalues which are respectively an eigenvalue a, an eigenvalue b and an eigenvalue c; then, according to step 203, the combination of the training sets includes: a; b; a. b three combination modes. Wherein, the size of the characteristic value of each group of electrocardiosignals aiming at the same dimension is different.

Then 9 combination modes of the classifier and the training set and Accuracy (assumed value) corresponding to each combination mode are:

the combination group is a training group, the classifier is KNN, and a; accuracy: 60 percent.

A second combination, wherein the classifier is KNN, and the training set comprises b; accuracy: 61 percent.

Thirdly, the classifier is KNN, and the training set comprises a and b; accuracy: 62 percent.

Fourthly, the classifier is an SVM-rbf, and a training set comprises a; accuracy: and 63 percent.

Fifthly, the classifier is SVM-rbf, and a training set comprises b; accuracy: and 64 percent.

Combining a sixth classifier into SVM-rbf, wherein the training set comprises a and b; accuracy: 60 percent.

A seventh combination, wherein the classifier is DT, and the training set comprises a; accuracy: and 64 percent.

The combination eight, the classifier is DT, and the training set comprises b; accuracy: 65 percent.

A ninth combination, wherein the classifier is DT, and the training set comprises a and b; accuracy: 66 percent.

Therefore, the target training set may be selected in a manner that if the preset threshold is 65%, the combination manner in which Accuracy exceeds the preset threshold is combination eight and combination nine.

The method for selecting the target training set is not limited in the embodiment of the application, for example, for Accuracy exceeding a preset threshold, the training set corresponding to the largest Accuracy is selected as the target training set, that is, one of the nine training sets is selected as the target training set, and the combination modes (a, b) of the target training set are used as the preset combination modes in the learning state; the number of features in the training set may also be determined according to the number of dimensions of the features in the training set, for example, for Accuracy exceeding a preset threshold, a training set with a small number of dimensions of features in a training set corresponding to Accuracy is selected as a target training set, that is, one of eight training sets in combination is selected as a target training set, and the combination mode (b) of the target training set is used as a preset combination mode in the learning state.

Step 302, for each preset combination mode, performing weighted calculation on a first recognition rate of a target training group corresponding to the preset combination mode and a second recognition rate of a verification group corresponding to the preset combination mode to obtain a comprehensive score of a target classifier in the preset combination mode; wherein the target classifier is a classifier enabling the target training set to obtain the Accuracy; the combination mode, the classifier and the calculation method of the recognition rate of the verification group and the target training group are the same; each group of electrocardiosignals corresponding to the verification group is different from each group of electrocardiosignals corresponding to the target training group.

Specifically, the first recognition rate is obtained by different classifiers according to a training set formed by a target training set; the second recognition rate is derived for a verification set of verification groups by different classifiers. The difference between the verification set and the training set is that target users corresponding to the electrocardiosignals acquired by the verification set and the training set are different;

according to the example in step 301, when the training set in combination nine is selected as the target training set, then the classifier DT in combination nine is the target classifier; when the training set in the combination eight is selected as the target training set, the classifier DT in the combination eight is the target classifier. Or, if the training set in the first combination is selected as the target training set, the classifier KNN in the first combination is taken as the target classifier. The combined score of the first combination, the second combination, … …, and the ninth combination is a combined score of the recognition rates obtained by weighting the first recognition rate and the second recognition rate for each of the first recognition rate and the second recognition rate obtained by the combination. For example, when the weights of the first recognition rate and the second recognition rate are the same, for a combination, the verification set and the training set are respectively put into the model corresponding to the combination, so that the first recognition rate of the KNN classifier for the training set is 60%, the second recognition rate of the KNN classifier for the verification set is 80%, and the comprehensive score is 70%.

After the preset combination modes in different learning states are obtained by executing step 301, as shown in an example in step 301, the combination modes of 9 classifiers and training sets correspond to the combination modes of 9 classifiers and verification sets. And obtaining a second identification rate of the new electrocardiosignal group corresponding to the verification set based on the verification set in the same method for obtaining the first identification rate by using the new electrocardiosignal group as the verification set in the same combination mode. According to the method for calculating the comprehensive score, the weight of the first recognition rate and the weight of the second recognition rate to the comprehensive score can be adjusted according to actual conditions, and the comprehensive score of the target classifier in a preset combination mode is determined according to the set actual weight.

Step 303, when the comprehensive score which can be achieved by the target classifier in the preset combination mode is greater than or equal to a preset score, fixing the target classifier as a classifier for monitoring the group to be detected; wherein, the group to be detected is composed according to the preset combination mode.

Specifically, after step 302 is executed, according to the example in step 302, the comprehensive scores corresponding to the combination one, the combination two, … …, and the combination nine may be determined respectively. And determining a fixed monitoring combination between the classifier and the group to be detected formed according to a preset combination mode by determining a preset value. For example, the highest comprehensive score of the combination from the first combination to the ninth combination is set as a preset score, and the matching result of the classifier and the training set in the combination with the highest comprehensive score is used as a fixed monitoring combination for monitoring the electrocardiosignals in the learning state.

In the embodiment of the present application, the fixed monitoring combinations determined for each learning state are:

for knowledge learning states and problem solving states: the classifier is SVM-rbf, and the characteristic values in the preset combination mode are as follows: a first eigenvalue, a third eigenvalue, and an eleventh eigenvalue.

For learning stage cognitive load matching and mismatching states: the classifier is KNN, and the characteristic values in the preset combination mode are as follows: the first characteristic value, the fourth characteristic value and the fifth characteristic value.

Aiming at the cognitive load matching state and the non-matching state in the testing stage: the classifier is SVM-rbf, and the characteristic values in the preset combination mode are as follows: a second eigenvalue, a sixth eigenvalue, and an eleventh eigenvalue.

Aiming at the matching state and the mismatching state of the difficulty of the knowledge points and the group learning ability: the classifier is DT, and the characteristic values in the preset combination mode are as follows: a seventh eigenvalue, an eighth eigenvalue, a ninth eigenvalue, and a tenth eigenvalue.

For mental weariness and fatigue state: the classifier is DT, and the characteristic values in the preset combination mode are as follows: a fifth eigenvalue and a twelfth eigenvalue.

In one possible embodiment, the calculation formula of Accuracy is:

Specifically, the first recognition rate and the second recognition rate both adopt the calculation formula of Accuracy, and the training set is replaced with the verification set, so that the obtained Accuracy is the second recognition rate.

In one possible embodiment, the calculation of the feature values is performed by inputting the input data into a preset model; wherein the preset model comprises preset submodels for calculating feature values of different dimensions.

The preset model is used for:

and calculating a first preset ratio of the number of the input data with the interval value exceeding a first preset value to the total number of the input data, and a second preset ratio of the number of the input data with the interval value exceeding a second preset value to the total number of the input data, so as to take the first preset ratio as a first characteristic value and the second preset ratio as a second characteristic value.

Specifically, the first preset value is 20ms, and the second preset value is 30 ms; the first preset ratio is the ratio of the number of RR intervals with the interval value exceeding 20 milliseconds to the total number of RR intervals in an RR interval sequence; the second preset ratio is the ratio of the number of RR intervals with the interval value exceeding 30 milliseconds to the total number of RR intervals in the RR interval sequence.

And calculating the ratio of the standard deviation to the average value of the input data to use the ratio of the standard deviation to the average value as a third characteristic value.

Calculating ApEn (approximate entropy), SampEn (sample entropy), semi-major axis SD1, semi-minor axis SD2 in a Poincare scattergram, and the ratio of the SD1 to the SD2 of the RR interval sequences, wherein the ApEn is used as a fourth characteristic value, the SampEn is used as a fifth characteristic value, the SD1 is used as a sixth characteristic value, and the ratio of the SD1 to the SD2 is used as a seventh characteristic value; the Poincare scattergram is drawn by using interval values of two continuous input data in the RR interval sequence as an abscissa and an ordinate respectively; the ApEn, SampEn, SD1, SD2 are all calculated by corresponding MATLAB (software for performing data analysis, wireless communication, deep learning, image processing, etc.) calculation programs.

Specifically, the ApEn is approximate entropy, which is a non-linear dynamic parameter for quantifying regularity and unpredictability of time series fluctuation; the SampEn is sample entropy, which is an improved method for measuring time series complexity based on approximate entropy. The SD1 is the distance between the longest two points in the scattergram region in the direction perpendicular to the X-Y direction in the poincare scattergram. The SD2 is the distance between the two shortest points in the poincare scattergram perpendicular to the X-Y direction.

Calculating a first slope alpha of a fitting straight line formed by input data of a first preset part in a Heart Rate Variability (HRV) curve graph through a calculation program of input DFA (Detrended fluctuation analysis)₁A second slope α of a fitted straight line formed by input data of a second predetermined section₂A ratio α of the second slope to the first slope₂/α₁To convert the alpha into₂As an eighth characteristic value, the α is₂/α₁As a ninth eigenvalue; wherein, the HRV graph is a graph which is drawn according to the input data and is used for describing the variation situation of the heartbeat speed.

Specifically, in the embodiment of the application, the first preset part is 4 th to 16 th points in the HRV curve detrending fluctuation analysis; the second preset part is 16 th to 32 th points in the HRV curve detrending fluctuation analysis.

Calculating an average value of the input data, an average value of an absolute value of a first derivative of the input data, and a power of the input data at a preset frequency portion in the HRV graph, so as to take the average value of the input data as a tenth characteristic value, take the average value of the absolute value of the first derivative of the input data as an eleventh characteristic value, and take the power as a twelfth characteristic value. Wherein the preset frequency is: 0.0033 to 0.04 Hz.

Example two

Fig. 3 is a schematic structural diagram of a data monitoring apparatus according to an embodiment of the present application, and as shown in fig. 3, the apparatus includes: an analysis unit 311, a first calculation unit 312, a data unit 313, a second calculation unit 314, a first combination unit 315, and a first labeling unit 316.

The analysis unit 311 is configured to, after acquiring a set of electrocardiographic signals of the target user in the target time period, determine occurrence times of at least two peaks of the set of electrocardiographic signals according to an occurrence time and an occurrence intensity of each electrocardiographic signal in the set of electrocardiographic signals.

The first calculating unit 312 is configured to calculate a difference between the occurrence times of two adjacent peaks according to the sequence of the occurrence times, so as to use the difference as an interval value of an RR interval, and use a time period between the occurrence times of two adjacent peaks as an occurrence time period of the RR interval.

A data unit 313 for determining an RR interval sequence comprising at least one RR interval after calculating an interval value of the RR interval; the RR intervals are arranged according to the calculated sequence, and comprise the interval values and the occurrence time intervals of the RR intervals.

A second calculating unit 314, configured to input RR intervals in the RR interval sequence as input data into at least two preset submodels for calculating characteristic values, respectively, to obtain at least two characteristic values for representing fluctuation characteristics of the group of electrocardiographic signals; wherein, different preset submodels are used for calculating the characteristic values of the fluctuation characteristics on different dimensions.

A first combining unit 315, configured to combine at least one of the feature values according to a preset combining manner, so as to use at least one result of the combination as a group to be tested; wherein the preset combination is determined by the SBS.

The first marking unit 316 is configured to mark, by the monitoring model, an identifier for indicating a learning state of a student on the set of electrocardiographic signals corresponding to the set to be tested after the set to be tested is input to the monitoring model, so as to monitor a learning condition of the student according to the identifier.

the second marking unit is used for marking a first identifier used for representing the student in the learning state for each group of electrocardiosignals after acquiring at least one group of electrocardiosignals of a preset number of target users for each learning state; wherein the learning state comprises: knowledge learning state and problem solving state, learning stage cognition load matching state and unmatched state, testing stage cognition load matching state and unmatched state, knowledge point difficulty and group learning ability matching state and unmatched state, and mental fatigue and fatigue state.

And the third calculating unit is used for calculating at least two characteristic values of each group of electrocardiosignals carrying the first identifier in different dimensions through the preset submodel.

The second combination unit is used for combining the at least two characteristic values into at least one training group of the group of electrocardiosignals according to a fully combined combination mode; the combination mode of the full combination is a plurality of combination modes formed by taking 1, 2, … and n different characteristic values from n different characteristic values each time; n is the dimension number of the eigenvalue.

The data transmission unit is used for transmitting a training set formed by a preset number of training groups into the monitoring model after the training groups of a preset number of target users are obtained according to each combination mode of the training groups, so that each group of electrocardiosignals corresponding to each training group in the training set is marked through a classifier in the monitoring model; wherein the classifier comprises KNN, SVM-rbf and DT.

A third marking unit, configured to determine, for each classifier, an identifier of each group of electrocardiographic signals corresponding to each training group in the training set in the learning state by a leave-one-out method, so as to mark a second identifier for each group of electrocardiographic signals carrying the first identifier; wherein the first identifier and the second identifier are both the identifiers; the mark is divided into negative and positive.

And the counting unit is used for counting the comparison result of the first identifier and the second identifier of each group of electrocardiosignals so as to determine the Accuracy of each classifier aiming at each training set according to the comparison result.

In one possible embodiment, the apparatus further comprises:

and the setting and combining unit is used for taking the training group corresponding to the Accuracy exceeding a preset threshold as a target training group aiming at each learning state, and determining the combination mode of one target training group as the preset combination mode in the learning state according to a preset method.

The score calculation unit is used for carrying out weighted calculation on the first recognition rate of the target training group corresponding to each preset combination mode and the second recognition rate of the verification group corresponding to the preset combination mode so as to obtain a comprehensive score obtained by the target classifier in the preset combination mode; wherein the target classifier is a classifier enabling the target training set to obtain the Accuracy; the combination mode, the classifier and the calculation method of the recognition rate of the verification group and the target training group are the same; each group of electrocardiosignals corresponding to the verification group is different from each group of electrocardiosignals corresponding to the target training group.

In one possible embodiment, the calculation formula of Accuracy is:

The preset model is used for:

Calculating ApEn, SampEn, semi-major axis SD1 and semi-minor axis SD2 in a Poincare scattergram of the RR interval sequences, and taking the ratio of the SD1 to the SD2 as a fourth characteristic value, the ApEn as a fifth characteristic value, the SD1 as a sixth characteristic value, and the ratio of the SD1 to the SD2 as a seventh characteristic value; the Poincare scattergram is drawn by using interval values of two continuous input data in the RR interval sequence as an abscissa and an ordinate respectively; ApEn, SampEn, SD1, SD2 are all calculated by corresponding MATLAB calculation programs.

Calculating a first slope alpha of a fitting straight line formed by input data of a first preset part in a HRV curve graph of the heart rate variability by a calculation program of the input DFA₁A second slope α of a fitted straight line formed by input data of a second predetermined section₂A ratio α of the second slope to the first slope₂/α₁To convert the alpha into₂As an eighth characteristic value, the α is₂/α₁As a ninth eigenvalue; wherein, the HRV graph is a graph which is drawn according to the input data and is used for describing the variation situation of the heartbeat speed.

EXAMPLE III

Fig. 4 is a schematic structural diagram of an electronic device according to a third embodiment of the present application, including: a processor 401, a storage medium 402 and a bus 403, wherein the storage medium 402 stores machine-readable instructions executable by the processor 401, when the electronic device executes the method according to the first embodiment, the processor 401 communicates with the storage medium 402 via the bus 403, and the processor 401 executes the machine-readable instructions to perform the steps according to the first embodiment.

In this embodiment of the application, the storage medium 402 may further execute other machine-readable instructions to execute other methods as described in the first embodiment, and for the specific steps and principles of the executed method, reference is made to the description of the first embodiment, which is not described in detail herein.

Example four

A fourth embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and the computer program is executed by a processor when the computer program is executed to perform the steps in the first embodiment.

In the embodiment of the present application, when being executed by a processor, the computer program may further execute other machine-readable instructions to perform other methods as described in the first embodiment, and for the specific method steps and principles to be performed, reference is made to the description of the first embodiment, which is not described in detail herein.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. The above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical division, and there may be other divisions in actual implementation, and for example, a plurality of modules or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or modules through some communication interfaces, and may be in an electrical, mechanical or other form.

The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a U disk, a removable hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method of data monitoring, comprising:

2. The method of claim 1, wherein the monitoring model is trained by:

3. The method according to claim 2, wherein for each learning state, the training set corresponding to Accuracy exceeding a preset threshold is used as a target training set, and a combination mode of one of the target training sets is determined as a preset combination mode in the learning state according to a preset method;

4. The method of claim 2, wherein the calculation formula of Accuracy is:

wherein N is_FPNumber of false positive samples, N_TPNumber of true positive samples, N_TNNumber of true negative samples, N_FNRepresenting the number of false negative samples; the false positive sample is a training group with the first identification being negative and the second identification being positive; the true positive sample is a training group of which the first identification and the second identification are both positive; the true negative sample is the first markA training set that is negative for the second marker; the false negative sample is a training group with the first mark being positive and the second mark being negative.

5. The method of claim 1, wherein the calculation of the eigenvalues is done by inputting the input data into a preset model; the preset model comprises preset submodels used for calculating characteristic values of different dimensions;

the preset model is used for:

6. A data monitoring device, comprising:

7. The apparatus of claim 6, wherein the monitoring model in the labeling unit is trained by:

8. The apparatus of claim 7, further comprising:

9. An electronic device, comprising: a processor, a storage medium and a bus, the storage medium storing machine-readable instructions executable by the processor, the processor and the storage medium communicating over the bus when the electronic device is operating, the processor executing the machine-readable instructions to perform the steps of the method according to any one of claims 1 to 5.

10. A computer-readable storage medium, having stored thereon a computer program which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 5.