Multi-parameter vital sign monitoring methods, devices, electronic equipment and storage media

Abstract

This invention provides a multi-parameter vital sign monitoring method, device, electronic device, and storage medium. The method includes: acquiring monitoring data of multiple vital sign indicators of a monitored object; determining the labeling category of each vital sign indicator of the monitored object based on the monitoring data; selecting relevant vital sign indicators from the multiple vital sign indicators based on the labeling category; calculating the feature value of each relevant vital sign indicator of the monitored object based on the labeling category and the preset original weight of each relevant vital sign indicator; and inputting the feature value of each relevant vital sign indicator of the monitored object into a preset vital sign monitoring model to obtain the health status of the monitored object. This invention can improve the objectivity and accuracy of judging the health status of a monitored object.

Description

Technical Field

[0001] This invention relates to the field of health monitoring technology, and in particular to a method, device, electronic device, and storage medium for monitoring multi-parameter vital signs. Background Technology

[0002] With the development of science and technology, people's living standards are constantly improving, and more and more people are in a sub-healthy or even unhealthy state.

[0003] Monitoring vital signs data allows for the timely detection of potential abnormalities in the monitored individual, which is crucial for saving lives and treating illnesses. Relevant vital sign monitoring methods typically monitor a single vital sign parameter, issuing an alert when the data deviates from a preset normal range. However, relying solely on monitoring a single vital sign parameter results in a low accuracy rate for assessing the health status of the monitored individual, easily leading to misjudgments. Summary of the Invention

[0004] This invention provides a multi-parameter vital sign monitoring method, device, electronic device, and storage medium to improve the objectivity and accuracy of judging the health status of the monitored object.

[0005] In a first aspect, embodiments of the present invention provide a multi-parameter vital sign monitoring method, comprising:

[0006] Obtain monitoring data for multiple vital signs of the monitored subjects;

[0007] Based on the monitoring data, determine the labeling category for each vital sign indicator of the monitored subject;

[0008] Based on the labeling category, relevant vital signs indicators are selected from multiple vital signs indicators;

[0009] Based on the labeling category of each relevant vital sign indicator and the preset original weight of each relevant vital sign indicator, the characteristic value of each relevant vital sign indicator of the monitored object is calculated.

[0010] The characteristic values ​​of each relevant vital sign indicator of the monitored subject are input into the preset vital sign monitoring model to obtain the health status of the monitored subject.

[0011] In one possible implementation, the labeling categories include normal, abnormal, dangerous, and extremely dangerous;

[0012] Based on the monitoring data, determine the labeling category for each vital sign indicator of the monitored subject, including:

[0013] Obtain the data range for each annotation category of each vital sign indicator;

[0014] Based on the monitoring data of each vital sign indicator of the monitored subject and the data range of each labeling category of each vital sign indicator, the labeling category of each vital sign indicator of the monitored subject is determined.

[0015] In one possible implementation, different annotation types correspond to different annotation coefficients, and the degree of danger of the annotation type is positively correlated with the annotation coefficient.

[0016] Based on the labeled category of each relevant vital sign indicator and the preset original weight of each relevant vital sign indicator, the characteristic value of each relevant vital sign indicator of the monitored object is calculated, including:

[0017] The characteristic value of each relevant vital sign indicator is obtained by multiplying the labeled coefficient of each relevant vital sign indicator by the corresponding original weight.

[0018] In one possible implementation, relevant vital signs are selected from multiple vital signs indicators based on the labeling category, including:

[0019] Each vital sign indicator of the monitored subject was checked to determine whether its category was normal.

[0020] Vital signs that are not labeled as normal are identified as relevant vital signs of the monitored subjects.

[0021] In one possible implementation, the process of determining the pre-defined original weights of each vital sign indicator includes:

[0022] Acquire historical monitoring data and corresponding health status for each vital sign indicator of multiple monitored subjects;

[0023] Based on historical monitoring data and corresponding health status, the Pearson correlation coefficient between each vital sign indicator and health status was calculated.

[0024] Calculate the ratio of the Pearson correlation coefficients between each pair of vital signs based on the Pearson correlation coefficient.

[0025] Based on the ratio, the quantitative value between each pair of vital signs is determined, and the quantitative value is positively correlated with the corresponding ratio.

[0026] Calculate the geometric mean of each quantitative value corresponding to each vital sign to obtain the initial weight of each vital sign indicator;

[0027] The initial weights are normalized to obtain the original weights for each vital sign indicator.

[0028] In one possible implementation, the pre-defined vital sign monitoring model is an attention-based neural network model, which includes encoders of multiple different dimensions.

[0029] The characteristic values ​​of each relevant vital sign indicator of the monitored subject are input into a preset vital sign monitoring model to obtain the health status of the monitored subject, including:

[0030] Based on the characteristic values ​​of each relevant vital sign indicator of the monitored object, determine the feature vector and the dimension of the feature vector of the monitored object;

[0031] The health status of the monitored object is obtained by inputting the feature vector into an encoder with the same dimension as the feature vector.

[0032] In one possible implementation, after acquiring monitoring data for multiple vital signs of the monitored object, the method further includes:

[0033] The monitoring data is processed to remove noise and handle outliers.

[0034] Secondly, embodiments of the present invention provide a multi-parameter vital sign monitoring device, comprising:

[0035] The acquisition module is used to acquire monitoring data of multiple vital signs of the monitored object;

[0036] The determination module is used to determine the labeling category of each vital sign indicator of the monitored object based on the monitoring data;

[0037] The filtering module is used to filter relevant vital signs from multiple vital signs indicators based on the labeled category;

[0038] The calculation module is used to calculate the feature value of each relevant vital sign indicator of the monitored object based on the labeling category of each relevant vital sign indicator and the preset original weight of each relevant vital sign indicator.

[0039] The monitoring module is used to input the feature values ​​of each relevant vital sign indicator of the monitored object into a preset vital sign monitoring model to obtain the health status of the monitored object.

[0040] Thirdly, embodiments of the present invention provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described in the first aspect or any possible implementation of the first aspect.

[0041] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method as described in the first aspect or any possible implementation of the first aspect.

[0042] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0043] This invention, through monitoring data of multiple vital signs of the monitored object, determines the labeling category of each vital sign, enabling a preliminary judgment on each vital sign. By filtering multiple vital signs through labeling categories, vital signs with less impact on the monitored object's health status can be removed, while those with greater impact can be retained, reducing data dimensionality and improving prediction efficiency. By inputting the calculated feature values ​​of relevant vital signs into a preset vital sign monitoring model, the health status of the monitored object is obtained. This allows for a comprehensive determination of the monitored object's health status through multiple vital signs, improving the objectivity and accuracy of monitoring and avoiding misjudgments caused by relying on a single vital sign. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart illustrating the implementation of the multi-parameter vital sign monitoring method provided in this embodiment of the invention.

[0046] Figure 2 This is a schematic diagram of the hierarchical structure of vital signs and health status provided in the embodiments of the present invention;

[0047] Figure 3 This is a flowchart illustrating the implementation of the vital signs monitoring model provided in this embodiment of the invention.

[0048] Figure 4 This is a schematic diagram of the structure of the input matrix of the vital sign monitoring model provided in this embodiment of the invention;

[0049] Figure 5 This is a schematic diagram illustrating the calculation of the attention mechanism of the vital sign monitoring model provided in this embodiment of the invention;

[0050] Figure 6 This is a schematic diagram of the structure of the multi-parameter vital sign monitoring device provided in an embodiment of the present invention;

[0051] Figure 7 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0052] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0053] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0054] Figure 1 The implementation flowchart of the multi-parameter vital sign monitoring method provided in the embodiments of the present invention is described in detail below:

[0055] Step S101: Obtain monitoring data of multiple vital signs of the monitored object.

[0056] In this embodiment, multiple vital signs indicators may include body temperature, pulse rate, blood oxygen, systolic blood pressure, diastolic blood pressure, etc.; the gender and age of the monitored subject may also affect the judgment of health status, so the vital signs indicators may also include the gender and age of the monitored subject. The gender and age information may be actively entered by the monitored subject or other personnel, or it may be obtained through other databases.

[0057] In addition, corresponding vital signs indicators can be set according to the specific type of health status that is desired. For example, when monitoring the heart health status of the monitored subject, vital signs indicators such as the type of chest pain, fasting blood glucose, and whether there is exercise-induced angina can be obtained.

[0058] Step S102: Based on the monitoring data, determine the labeling category for each vital sign indicator of the monitored object.

[0059] In this embodiment, each vital sign indicator corresponds to multiple labeling categories. The labeling categories are determined based on the data range of the vital sign indicator, which can make a preliminary judgment on the corresponding vital sign indicator and determine whether there are any abnormalities in the monitoring data of the corresponding vital sign indicator.

[0060] Step S103: Select relevant vital signs indicators from multiple vital signs indicators according to the labeled categories.

[0061] In this embodiment, the various vital signs indicators of the monitored object are labeled in different categories, and the corresponding vital signs indicators have different effects on the health status of the monitored object. Therefore, the vital signs indicators with greater influence can be selected as relevant vital signs indicators to reduce the dimensionality of the data.

[0062] Step S104: Calculate the characteristic value of each relevant vital sign indicator of the monitored object based on the labeling category of each relevant vital sign indicator and the preset original weight of each relevant vital sign indicator.

[0063] In this embodiment, the feature value is actually the degree of influence of each relevant vital sign indicator of the monitored object on the health status of the monitored object.

[0064] The original weight of each relevant vital sign indicator represents the degree of influence of each relevant vital sign indicator on health status. For example, when monitoring the health status of the heart, the degree of influence of systolic blood pressure and diastolic blood pressure on health status may be the same, and the original weights of systolic blood pressure and diastolic blood pressure are the same. The degree of influence of exercise-induced angina on health status may be greater than the degree of influence of body temperature on health status, and the original weight of exercise-induced angina is greater than the original weight of body temperature.

[0065] The label categories indicate the degree of impact of the abnormality of various relevant vital signs of the monitored object on its health status. For example, when monitoring the health status of the heart, if the monitoring data of systolic blood pressure is normal but the monitoring data of diastolic blood pressure is abnormal, then the degree of impact of the monitored object's systolic blood pressure and diastolic blood pressure on its health status is different.

[0066] Step S105: Input the feature values ​​of each relevant vital sign indicator of the monitored object into the preset vital sign monitoring model to obtain the health status of the monitored object.

[0067] This invention, through monitoring data of multiple vital signs of the monitored object, determines the labeling category of each vital sign, enabling a preliminary judgment on each vital sign. By filtering multiple vital signs through labeling categories, vital signs with less impact on the monitored object's health status can be removed, while those with greater impact can be retained, reducing data dimensionality and improving prediction efficiency. By inputting the calculated feature values ​​of relevant vital signs into a preset vital sign monitoring model, the health status of the monitored object is obtained. This allows for a comprehensive determination of the monitored object's health status through multiple vital signs, improving the objectivity and accuracy of monitoring and avoiding misjudgments caused by relying on a single vital sign.

[0068] There are no restrictions on the specific methods for obtaining monitoring data of multiple vital signs of the monitored object; for example, the vital signs data of the monitored object can be obtained through wearable devices.

[0069] Specifically, to obtain body temperature monitoring data, a thermistor can be used to convert temperature changes into resistance changes, a measuring circuit can be used to convert the resistance into voltage, and then the voltage value can be converted into a digital signal to obtain the corresponding temperature value.

[0070] Pulse rate monitoring data can be obtained through reflective measurement, which uses changes in the intensity of reflected light after it enters the blood to obtain information about the heartbeat. The instantaneous pulse rate is obtained by averaging the data from short-term measurements.

[0071] Blood oxygenation data can be obtained through transmissive measurements. Red light has a strong ability to penetrate oxyhemoglobin (O2Hb), while infrared light has a strong ability to penetrate deoxyhemoglobin (HHb). Blood oxygen saturation (SpO2) can be obtained using the Webster linear empirical calibration formula: SpO2 = 110 - 25 × R, where SpO2 represents blood oxygen saturation, approximately 100, and R represents the ratio of red to infrared perfusion coefficients, with R ≈ 0.4. AC represents pulsating signal, DC represents non-pulsating signal, Rd represents red light, IR represents infrared light, and PI = AC / DC represents the perfusion coefficient.

[0072] Blood pressure monitoring data can be obtained by using a barometric pressure sensor. By changing the distance between the positive and negative terminals of a capacitor through air pressure, the capacitance is changed, causing the matching oscillation circuit to generate an oscillation frequency that corresponds one-to-one with the pressure, thereby measuring blood pressure.

[0073] In one possible implementation, after acquiring monitoring data for multiple vital signs of the monitored object, the method further includes:

[0074] The monitoring data is processed to remove noise and handle outliers.

[0075] In this embodiment, outliers are identified by visualizing the monitoring data. Optionally, visualization can be performed by drawing histograms, box plots, or scatter plots.

[0076] Noise reduction can be achieved using Gaussian filtering, which reduces noise by weighting the sample values ​​with Gaussian weights. The specific calculation formula is as follows: Where output represents the filtered output value, and input represents the filtered output value. i w represents the input value of the i-th sample. i Let represent the Gaussian weight of the i-th sample, and N represent the number of samples.

[0077] In one possible implementation, the labeling categories include normal, abnormal, dangerous, and extremely dangerous;

[0078] Based on the monitoring data, determine the labeling category for each vital sign indicator of the monitored subject, including:

[0079] Obtain the data range for each annotation category of each vital sign indicator;

[0080] Based on the monitoring data of each vital sign indicator of the monitored subject and the data range of each labeling category of each vital sign indicator, the labeling category of each vital sign indicator of the monitored subject is determined.

[0081] In this embodiment, each labeling category for each vital sign indicator corresponds to a different data range. For the gender and age of the monitored subject within the vital sign indicators, there is no distinction between normal, abnormal, dangerous, and extremely dangerous information; therefore, labeling category determination is not required.

[0082] Optionally, pulse rate indicators are divided into exercise and non-exercise conditions. The pulse rate under non-exercise conditions is the resting pulse rate, which is generally 70 beats / minute (min). The maximum pulse rate is the highest level of pulse rate reached under maximum load intensity. Gender and age have a certain influence on the maximum pulse rate. The formula for calculating the maximum pulse rate for men is: maximum pulse rate = 220 - age, and the formula for calculating the maximum pulse rate for women is: maximum pulse rate = 224 - age.

[0083] For subjects under non-movement conditions, the data range labeled as normal is (resting pulse rate - 10) to (resting pulse rate + 20). When the resting pulse rate of the monitored subject is within the normal data range, the resting pulse rate can be considered risk-free. The data range labeled as abnormal is (resting pulse rate - 20) to (resting pulse rate - 10) and (resting pulse rate + 20) to (maximum pulse rate × 60%). The data range labeled as dangerous is (maximum pulse rate × 60%) to (maximum pulse rate × 80%). The data range labeled as extremely dangerous is below (resting pulse rate - 20) and above 80% of the maximum pulse rate.

[0084] For monitored subjects in motion, the normal pulse rate range is generally 60% to 70% of the resting pulse rate plus the preserved pulse rate. The preserved pulse rate is the difference between the maximum pulse rate and the resting pulse rate. Accordingly, the data range labeled as normal is (preserved pulse rate × 60% + resting pulse rate) to (preserved pulse rate × 70% + resting pulse rate); the data range labeled as abnormal is (preserved pulse rate × 70% + resting pulse rate) to (preserved pulse rate × 70% + resting pulse rate + 10); the data range labeled as dangerous is (preserved pulse rate × 70% + resting pulse rate + 10) to (preserved pulse rate × 70% + resting pulse rate + 20); and the data range labeled as extremely dangerous is higher than (preserved pulse rate × 70% + resting pulse rate + 20) and lower than the resting pulse rate.

[0085] For example, if the subject being monitored is a 20-year-old male with a resting pulse rate of 70 beats / min, then under non-exercise conditions, the normal data range is 60 to 90 beats / min, the abnormal data range is 50 to 59 beats / min and 91 to 120 beats / min, the dangerous data range is 121 to 160 beats / min, and the extremely dangerous data range is above 160 beats / min and below 50 beats / min. Under exercise conditions, the normal data range is 148 to 161 beats / min, the abnormal data range is 162 to 172 beats / min, the dangerous data range is 173 to 183 beats / min, and the extremely dangerous data range is above 183 beats / min and below 70 beats / min.

[0086] Blood oxygen saturation, or blood oxygen, refers to the percentage of oxygen bound to hemoglobin in the blood. The normal range for blood oxygen is 95% to 100%, the abnormal range is 80% to 95%, the dangerous range is 50% to 80%, and the extremely dangerous range is below 50%.

[0087] Blood pressure includes systolic and diastolic pressure. The normal range for systolic blood pressure is below 140 mmHg, the abnormal range is 140 mmHg to 159 mmHg, the dangerous range is 160 mmHg to 179 mmHg, and the extremely dangerous range is above 180 mmHg. The normal range for diastolic blood pressure is below 90 mmHg, the abnormal range is 90 mmHg to 99 mmHg, the dangerous range is 100 mmHg to 109 mmHg, and the extremely dangerous range is above 110 mmHg.

[0088] The normal range for body temperature is below 37.5℃, the abnormal range is 37.5℃ to 39.0℃, the dangerous range is 39.1℃ to 41.0℃, and the extremely dangerous range is above 41.0℃.

[0089] The above are only some of the optional data ranges for vital signs. In practical applications, adjustments can be made accordingly. For other characteristic indicators, the data ranges for the corresponding labeling categories can also be determined according to actual needs.

[0090] In one possible implementation, different annotation types correspond to different annotation coefficients, and the degree of danger of the annotation type is positively correlated with the annotation coefficient.

[0091] Based on the labeled category of each relevant vital sign indicator and the preset original weight of each relevant vital sign indicator, the characteristic value of each relevant vital sign indicator of the monitored object is calculated, including:

[0092] The characteristic value of each relevant vital sign indicator is obtained by multiplying the labeled coefficient of each relevant vital sign indicator by the corresponding original weight.

[0093] In this embodiment, see Figure 2 The diagram illustrates the hierarchical structure of vital signs and health status. Health status is influenced by various vital signs, and the degree of influence varies depending on the labeling type of the vital signs. The labeling types are normal, abnormal, dangerous, and extremely dangerous, with the degree of danger and influence on health status increasing progressively, and the corresponding labeling coefficients also showing an upward trend. Optionally, the labeling coefficient for normal can be 0, for abnormal can be 0.5, for dangerous can be 1, and for extremely dangerous can be 2.

[0094] In addition, there is no distinction between normal, abnormal, dangerous, and extremely dangerous for the gender and age of the monitored subjects, so the labeling coefficient can be set to 1.

[0095] For the original weights of the gender and age of the monitored subjects, gender and age can be directly used as the original weights. For example, the original weight for males can be determined as 1, the original weight for females can be determined as 2, and the original weight for age is the age of the monitored subject.

[0096] The feature value of identity information is actually the data corresponding to the identity information. For example, the feature value corresponding to male is 1×1=1, the feature value corresponding to female is 1×2=2, and the feature value corresponding to age is 1×age of the monitored object =age of the monitored object. That is, the identity information is directly input into the vital signs monitoring model.

[0097] According to T i =k i ×W i The characteristic values ​​of each relevant vital sign indicator of the monitored object are calculated, where T i Let k represent the characteristic value of the i-th relevant vital sign indicator. i W represents the labeled coefficient of the i-th related trait. i This represents the original weight of the i-th related trait.

[0098] By using labeled coefficients and original weights, the impact of relevant vital signs on the health status of the monitored subjects can be quantified into specific numerical values, reducing the dimensionality of the data and improving the speed and accuracy of subsequent predictions using vital sign monitoring models.

[0099] In one possible implementation, relevant vital signs are selected from multiple vital signs indicators based on the labeling category, including:

[0100] Each vital sign indicator of the monitored subject was checked to determine whether its category was normal.

[0101] Vital signs that are not labeled as normal are identified as relevant vital signs of the monitored subjects.

[0102] In this embodiment, if the monitoring data of the vital signs of the monitored object are normal, it indicates that the vital signs have little impact on the health status of the monitored object; if the monitoring data of the vital signs of the monitored object are not normal, it indicates that the vital signs have a significant impact on the health status of the monitored object. Therefore, selection can be based on the labeling category.

[0103] Specifically, normal vital signs can be removed directly, and abnormal vital signs can be identified as relevant vital signs. In addition, the gender and age of the monitored subjects are important for determining their health status and can be directly identified as relevant vital signs.

[0104] Optionally, after calculating the characteristic value of each vital sign indicator, relevant vital sign indicators can be selected. The annotation coefficient of a normal vital sign indicator is 0, so the characteristic value of a normal vital sign indicator is also 0. Vital sign indicators with a characteristic value of 0 can be eliminated, while those with a non-zero characteristic value can be retained and used as relevant vital sign indicators.

[0105] In one possible implementation, the process of determining the pre-defined original weights of each vital sign indicator includes:

[0106] Acquire historical monitoring data and corresponding health status for each vital sign indicator of multiple monitored subjects;

[0107] Based on historical monitoring data and corresponding health status, the Pearson correlation coefficient between each vital sign indicator and health status was calculated.

[0108] Calculate the ratio of the Pearson correlation coefficients between each pair of vital signs based on the Pearson correlation coefficient.

[0109] Based on the ratio, the quantitative value between each pair of vital signs is determined, and the quantitative value is positively correlated with the corresponding ratio.

[0110] Calculate the geometric mean of each quantitative value corresponding to each vital sign to obtain the initial weight of each vital sign indicator;

[0111] The initial weights are normalized to obtain the original weights for each vital sign indicator.

[0112] In this embodiment, according to Calculate the Pearson correlation coefficient between each vital sign and health status, where r xyThe Pearson correlation coefficient represents the correlation between vital signs and health status, where x represents a vital sign, y represents health status, σ(x) represents the standard deviation of the vital sign monitoring data, σ(y) represents the standard deviation of the health status data, and cov(x,y) represents the covariance between the vital sign and health status. j This represents the j-th monitoring data point in the vital sign indicator. Y represents the average value of the monitoring data for this vital sign indicator. j This represents the health status data corresponding to the j-th monitoring data. The average value of data representing health status.

[0113] Optional, for example, there are ten vital signs indicators, and the Pearson correlation coefficients between each vital sign indicator and health status are r1, r2, r3, r4, r5, r6, r7, r8, r9 and r 10 The ratios of the Pearson correlation coefficients among the various vital signs were p1:p2:p3:p4:p5:p6:p7:p8:p9:p 10 It can be based on p m =|r m |×10, calculate the corresponding ratio, where m takes values ​​of 1, 2, 3...10.

[0114] For example, the quantitative values ​​between each pair of vital signs can be seen in Table 1 below:

[0115] Table 1. Rules for Quantitative Values ​​Among Vital Sign Indicators

[0116]

[0117] The final formula for calculating the quantized value is a. ij = <p i / p j The symbol ">" indicates the rounded value. Correspondingly, this allows us to obtain the quantified value between any two of the ten vital signs.

[0118] In one specific embodiment, the vital signs indicators include body temperature, pulse rate, blood oxygen saturation, systolic blood pressure, diastolic blood pressure, type of chest pain, fasting blood glucose, and presence of angina, numbered 1-8 in sequence. The corresponding quantitative values ​​are shown in Table 2 below:

[0119] Table 2. Quantitative values ​​between each pair of vital signs.

[0120]

[0121]

[0122] Based on the data in Table 2 above, calculate the geometric mean of the quantified values ​​corresponding to each vital sign indicator to obtain the initial weight of each vital sign indicator. Specifically, this can be done according to... Calculate the initial weight for each vital sign indicator, where w i This represents the initial weight of the i-th vital sign indicator. Therefore, based on... The original weight of each vital sign indicator was calculated.

[0123] In one possible implementation, the pre-defined vital sign monitoring model is an attention-based neural network model, which includes encoders of multiple different dimensions.

[0124] The characteristic values ​​of each relevant vital sign indicator of the monitored subject are input into a preset vital sign monitoring model to obtain the health status of the monitored subject, including:

[0125] Based on the characteristic values ​​of each relevant vital sign indicator of the monitored object, determine the feature vector and the dimension of the feature vector of the monitored object;

[0126] The health status of the monitored object is obtained by inputting the feature vector into an encoder with the same dimension as the feature vector.

[0127] In this embodiment, since the number of relevant vital signs is uncertain, the dimension of the feature vector is also uncertain. Therefore, when inputting the feature values ​​into the preset vital sign monitoring model, it is necessary to consider the dimension of the feature vector and select the corresponding encoder as the input encoder.

[0128] In addition, the gender and age of the monitored subjects also have a certain impact on the assessment of health status. Since the characteristic values ​​of gender and age are determined differently from the characteristic values ​​of other relevant physical signs, it is possible to consider using the characteristic values ​​of gender and age as the first two terms in the characteristic vector, and the characteristic values ​​of other relevant physical signs as the third and subsequent terms, to establish the characteristic vector.

[0129] Optionally, the attention-based neural network model used can be a transformer model. The vital sign monitoring model is trained based on the transformer model. For details, please refer to [link to relevant documentation]. Figure 3 The flowchart shown illustrates the implementation of the vital signs monitoring model, where the transformer model has encoders of multiple different dimensions as inputs.

[0130] The Transformer model employs an attention mechanism to capture dependencies in the input sequence, enabling the model to focus on relevant words and apply weights. It introduces positional encoding to provide positional information of vectors within the sequence. Furthermore, it utilizes multi-head attention to enhance the model's expressive power and allow it to learn different attention subspaces.

[0131] Since the feature vectors defined in this invention may contain data of different dimensions, it is necessary to construct different Transformer models based on data of different dimensions, and to construct an input matrix from data of the same dimension in time series for prediction.

[0132] The four sets of data are constructed into a matrix. If only the first, second, and third sets of data have dimensions that are different from the preceding and following ones, they are not included in the matrix construction. This is because such data may be anomalous. For example, if the blood oxygen saturation of the monitored subject is collected in two very low sets but then immediately returns to normal, and this decrease in blood oxygen does not affect health, then these two sets of data are considered anomalous and do not affect the judgment of health status. In actual observation, even completely healthy monitored subjects often experience a sudden abnormality in a certain indicator followed by a rapid recovery. Therefore, anomalous data can be ignored.

[0133] The Transformer model takes feature vectors as input. For feature vectors of different dimensions, corresponding attention encoders are used, and the vectors are ultimately fed into the same decoder. Taking a ten-dimensional feature vector as an example, the input feature vector consists of ten feature values. Since the maximum labeled coefficient is 2, and the initial weights will not exceed 1, the maximum number of feature values ​​will not exceed 2. Therefore, each feature value is represented by a binary vector. Because the calculated feature values ​​may contain decimals, two decimal places can be retained. Thus, the integer and decimal parts can be represented separately in binary. The maximum two decimal places is 99. Correspondingly, the decimal part occupies a maximum of seven bits. Therefore, the first two bits can represent the integer part (0, 1, or 2), the last seven bits can represent the decimal part (00-99), and the third bit is used to indicate the conversion from integer to decimal. For example, when the feature value is 1.58, the corresponding vector can be represented as (0, 1, 0, 0, 1, 1, 1, 0, 1, 0). T Since the four sets of data form an input matrix, and each set contains ten eigenvalues, the input matrix is ​​a 100×4 matrix. See [link to documentation] for details. Figure 4 The diagram shows the structure of the input matrix of the vital signs monitoring model.

[0134] See Figure 5The diagram shown illustrates the calculation of the attention mechanism in the vital signs monitoring model. The calculation process of the attention mechanism is as follows:

[0135] Step 1, using the linear transformation matrix W q W k and W v Multiply each input vector by the corresponding attention weight (Query, Q), weight index (Key, K), and word vector (Value, V) obtained after training, where Q, K, and V are matrices composed of q, k, and v, respectively.

[0136] Step two: Calculate the attention score (attention source, α) based on the attention weights and weight indices. The specific formula for α is... ij =q i ·k j In the formula, α ij Let q represent the attention scores obtained from the i-th and j-th data sets. i k represents the attention weight of the i-th data set. j This represents the weight index of the j-th data group.

[0137] Step 3: Map all attention scores from the four data sets to the interval (0, 1) using a normalized exponential function (softmax) layer. The calculation formula for the softmax layer is as follows: In the formula, α i ′ j This represents the attention scores of the i-th and j-th data groups after mapping. α represents the base of the natural logarithm e. ij Power of 1.

[0138] Step four: Obtain the final output based on the mapped attention score and the word vectors obtained after training. The specific calculation formula is as follows: In the formula, y i v represents the output matrix corresponding to the i-th data set. i Let represent the word vector obtained after training the i-th set of data.

[0139] In summary, the complete formula for transforming the input matrix X into the output matrix Y is as follows:

[0140]

[0141] The Transformer model incorporates positional encoding to provide positional information of vectors within a sequence. The corresponding calculation formula is:

[0142]

[0143] Where PE(pos,2i) represents the positional encoding of even-positioned eigenvectors in the feature matrix, PE(pos,2i+1) represents the expression for odd-positioned eigenvectors in the feature matrix, i represents the positional information of the eigenvector in the feature matrix, pos represents the position of the eigenvector in the feature matrix, and d... model This represents the dimension of the feature vector itself.

[0144] For example, if vector x2 has a dimension of 3, then its position encoding is:

[0145]

[0146] See Figure 3 The flowchart shown illustrates the implementation process of the vital signs monitoring model, and also shows the training process of the transformer model, which includes an encoder and a decoder.

[0147] For the encoder part, the obtained feature values ​​are used as input. An input embedding layer transforms the feature vectors into an input matrix, using the same feature vector processing method described earlier. This transforms the input feature vectors into an input matrix composed of 0s and 1s. Different encoders are then used to construct positional encodings based on the different dimensions of the input matrix. After processing by the Attention model, the input matrix and the aforementioned positional encodings yield the output matrix Y. This is followed by a residual and normalization layer (Add & Norm). The residual (Add) adds each vector y in the output matrix Y of the Attention model to the corresponding vector x in the input matrix X, resulting in a new vector y′. The normalization layer (Norm) normalizes each vector y′ individually, yielding a new output matrix Y′. Finally, the output matrix Y′ passes through a fully connected feedforward neural network layer and an Add & Norm layer to obtain the final result processed by the encoder.

[0148] For the decoder, the output is transformed into a vector through an output embedding layer, then passes through a masked attention layer and an Add&Norm layer to obtain matrix O. Different types of Y′ outputs from different encoders are concatenated and then used as input to the Multi-Head Cross Attention layer along with matrix O. This input then passes through the Add&Norm, Feed Forward, and Add&Norm layers before a softmax operation is performed in the Hazard classification layer. This transforms the corresponding output vector into the input result, completing the classification of the hazard level for the data. Finally, steps for calculating loss and accuracy are added during training.

[0149] The solution of this invention pertains to multi-class classification tasks. It employs multi-class cross-entropy loss (Loss) as the loss function to calculate the negative log-likelihood of the predicted probabilities of the true health status being normal, abnormal, dangerous, and extremely dangerous. The corresponding calculation formula is as follows:

[0150]

[0151] Where LOSS represents the loss function, y ic This indicates whether the health status predicted by the vital signs monitoring model for sample i is the same as the actual health status c of sample i (if the predicted health status of sample i is the same as the actual health status c, then y ic Select 1, otherwise y ic Take 0), Let M represent the probability that the health status of sample i belongs to c, as predicted by the vital signs monitoring model. Let M represent the number of health status categories, c represent the specific category of health status, and N represent the set of sample i.

[0152] In one specific embodiment, the accuracy of the trained vital sign monitoring model was tested, and the accuracy reached above 0.86.

[0153] Optionally, after inputting the feature values ​​of each relevant vital sign indicator of the monitored object into the preset vital sign monitoring model to obtain the health status of the monitored object, the feature values ​​and corresponding health status of the monitored object can be added to the training set to continue training the vital sign monitoring model.

[0154] Optional, health status includes normal, abnormal, dangerous, and extremely dangerous.

[0155] After inputting the feature values ​​of each relevant vital sign indicator of the monitored subject into a preset vital sign monitoring model to obtain the health status of the monitored subject, it may also include:

[0156] If the health status is "abnormal", an alert message will be sent to the monitored object;

[0157] If the health status is "dangerous", an alarm message will be sent to the monitored person, the associated guardian, and the hospital monitoring center;

[0158] If the health status is "extremely dangerous", obtain the geographical location information of the monitored object and send the geographical location information to the nursing and rescue center.

[0159] This invention, through monitoring data of multiple vital signs of a monitored subject, determines the labeling category of each vital sign, enabling a preliminary assessment of each vital sign. By filtering multiple vital signs through labeling categories, indicators with minimal impact on the monitored subject's health status are removed, while those with a greater impact are retained, reducing data dimensionality and improving prediction efficiency. Specifically, normal vital signs are removed, while abnormal vital signs are identified as relevant, avoiding the influence of normal indicators on health status assessment. The feature values ​​of relevant vital signs are calculated using labeling coefficients and original weights, enabling the labeling of relevant vital signs... The impact of categories on health status is quantified, reducing the dimensionality of the data and improving the speed and accuracy of subsequent predictions using vital sign monitoring models. Specifically, by calculating the Pearson coefficient between each vital sign indicator and health status, the correlation between the vital sign indicator and health status can be accurately obtained, thus ensuring the accuracy of the original weights of each vital sign indicator. By inputting the feature values ​​of the calculated relevant vital sign indicators into the preset vital sign monitoring model, the health status of the monitored object can be obtained. The health status of the monitored object can be comprehensively determined through multiple vital sign indicators, improving the objectivity and accuracy of monitoring and avoiding misjudgments caused by judging based on a single vital sign indicator.

[0160] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0161] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0162] Figure 6 A schematic diagram of the multi-parameter vital sign monitoring device provided in an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below:

[0163] like Figure 6 As shown, the multi-parameter vital sign monitoring device 60 includes:

[0164] The acquisition module 61 is used to acquire monitoring data of multiple vital signs of the monitored object;

[0165] The determination module 62 is used to determine the labeling category of each vital sign indicator of the monitored object based on the monitoring data;

[0166] The filtering module 63 is used to filter relevant vital signs from multiple vital signs indicators based on the labeled category;

[0167] The calculation module 64 is used to calculate the feature value of each relevant vital sign indicator of the monitored object based on the labeling category of each relevant vital sign indicator and the preset original weight of each relevant vital sign indicator.

[0168] The monitoring module 65 is used to input the feature values ​​of each relevant vital sign indicator of the monitored object into the preset vital sign monitoring model to obtain the health status of the monitored object.

[0169] In one possible implementation, the labeling categories include normal, abnormal, dangerous, and extremely dangerous;

[0170] Module 62 is specifically used for:

[0171] Obtain the data range for each annotation category of each vital sign indicator;

[0172] Based on the monitoring data of each vital sign indicator of the monitored subject and the data range of each labeling category of each vital sign indicator, the labeling category of each vital sign indicator of the monitored subject is determined.

[0173] In one possible implementation, different annotation types correspond to different annotation coefficients, and the degree of danger of the annotation type is positively correlated with the annotation coefficient.

[0174] Calculation module 64 is specifically used for:

[0175] The characteristic value of each relevant vital sign indicator is obtained by multiplying the labeled coefficient of each relevant vital sign indicator by the corresponding original weight.

[0176] In one possible implementation, the filtering module 63 is specifically used for:

[0177] Each vital sign indicator of the monitored subject was checked to determine whether its category was normal.

[0178] Vital signs that are not labeled as normal are identified as relevant vital signs of the monitored subjects.

[0179] In one possible implementation, the process of determining the pre-defined original weights of each vital sign indicator includes:

[0180] Acquire historical monitoring data and corresponding health status for each vital sign indicator of multiple monitored subjects;

[0181] Based on historical monitoring data and corresponding health status, the Pearson correlation coefficient between each vital sign indicator and health status was calculated.

[0182] Calculate the ratio of the Pearson correlation coefficients between each pair of vital signs based on the Pearson correlation coefficient.

[0183] Based on the ratio, the quantitative value between each pair of vital signs is determined, and the quantitative value is positively correlated with the corresponding ratio.

[0184] Calculate the geometric mean of each quantitative value corresponding to each vital sign to obtain the initial weight of each vital sign indicator;

[0185] The initial weights are normalized to obtain the original weights for each vital sign indicator.

[0186] In one possible implementation, the pre-defined vital sign monitoring model is an attention-based neural network model, which includes encoders of multiple different dimensions.

[0187] Monitoring module 65 is specifically used for:

[0188] Based on the characteristic values ​​of each relevant vital sign indicator of the monitored object, determine the feature vector and the dimension of the feature vector of the monitored object;

[0189] The health status of the monitored object is obtained by inputting the feature vector into an encoder with the same dimension as the feature vector.

[0190] In one possible implementation, the multi-parameter vital sign monitoring device 60 further includes a preprocessing module, which is used for:

[0191] The monitoring data is processed to remove noise and handle outliers.

[0192] Figure 7 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Figure 7 As shown, the electronic device 70 of this embodiment includes: a processor 71, a memory 72, and a computer program 73 stored in the memory 72 and executable on the processor 71. When the processor 71 executes the computer program 73, it implements the steps in the various multi-parameter vital sign monitoring method embodiments described above, for example... Figure 1Steps S101 to S105 are shown. Alternatively, when processor 71 executes computer program 73, it implements the functions of each module in the above-described device embodiments, for example... Figure 6 The functions of modules 61 to 65 are shown.

[0193] For example, computer program 73 can be divided into one or more modules / units, one or more of which are stored in memory 72 and executed by processor 71 to complete the present invention. One or more modules / units can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 73 in electronic device 70. For example, computer program 73 can be divided into... Figure 6 Modules 61 to 65 are shown.

[0194] Electronic device 70 may include, but is not limited to, processor 71 and memory 72. Those skilled in the art will understand that... Figure 7 This is merely an example of electronic device 70 and does not constitute a limitation on electronic device 70. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device may also include input / output devices, network access devices, buses, etc.

[0195] The processor 71 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0196] The memory 72 can be an internal storage unit of the electronic device 70, such as a hard disk or RAM of the electronic device 70. The memory 72 can also be an external storage device of the electronic device 70, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the electronic device 70. Furthermore, the memory 72 can include both internal and external storage units of the electronic device 70. The memory 72 is used to store computer programs and other programs and data required by the electronic device. The memory 72 can also be used to temporarily store data that has been output or will be output.

[0197] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0198] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0199] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0200] In the embodiments provided by this invention, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0201] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0202] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0203] If integrated modules / units are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0204] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for monitoring multi-parameter vital signs, characterized in that, include: Obtain monitoring data for multiple vital signs of the monitored subjects; Based on the monitoring data, determine the labeling category for each vital sign indicator of the monitored object; Based on the labeled categories, relevant vital signs indicators are selected from the multiple vital signs indicators; Based on the labeling category of each relevant vital sign indicator and the preset original weight of each relevant vital sign indicator, the characteristic value of each relevant vital sign indicator of the monitored object is calculated. The characteristic values ​​of each relevant vital sign indicator of the monitored object are input into a preset vital sign monitoring model to obtain the health status of the monitored object. The process of determining the original weights of each pre-defined vital sign indicator includes: Acquire historical monitoring data and corresponding health status for each vital sign indicator of multiple monitored subjects; Based on the historical monitoring data and the corresponding health status, calculate the Pearson correlation coefficient between each vital sign indicator and the health status; Based on the Pearson correlation coefficient, calculate the ratio of the Pearson correlation coefficients between each pair of vital signs. Based on the ratio, a quantitative value is determined for each pair of vital signs, and the quantitative value is positively correlated with the corresponding ratio. Calculate the geometric mean of each quantitative value corresponding to each vital sign to obtain the initial weight of each vital sign indicator; The initial weights are normalized to obtain the original weights for each vital sign indicator.

2. The multi-parameter vital sign monitoring method according to claim 1, characterized in that, The labeling categories include normal, abnormal, dangerous, and extremely dangerous; The step of determining the labeling category of each vital sign indicator of the monitored object based on the monitoring data includes: Obtain the data range for each annotation category of each vital sign indicator; Based on the monitoring data of each vital sign indicator of the monitored subject and the data range of each labeling category of each vital sign indicator, the labeling category of each vital sign indicator of the monitored subject is determined.

3. The multi-parameter vital sign monitoring method according to claim 2, characterized in that, Different annotation types correspond to different annotation coefficients, and the degree of danger of an annotation type is positively correlated with the annotation coefficient; The step of calculating the feature value of each relevant vital sign indicator of the monitored object based on the labeling category of each relevant vital sign indicator and the preset original weight of each relevant vital sign indicator includes: The characteristic value of each relevant vital sign indicator is obtained by multiplying the labeled coefficient of each relevant vital sign indicator by the corresponding original weight.

4. The multi-parameter vital sign monitoring method according to claim 2, characterized in that, The step of selecting relevant vital signs from the multiple vital signs indicators based on the labeled categories includes: Each vital sign indicator of the monitored object was checked to see if its labeled category was normal. Vital signs that are not labeled as normal are identified as relevant vital signs of the monitored object.

5. The multi-parameter vital sign monitoring method according to any one of claims 1-4, characterized in that, The preset vital sign monitoring model is a neural network model based on an attention mechanism, and the neural network model based on an attention mechanism includes encoders of multiple different dimensions. The step of inputting the feature values ​​of each relevant vital sign indicator of the monitored object into a preset vital sign monitoring model to obtain the health status of the monitored object includes: Based on the characteristic values ​​of each relevant vital sign indicator of the monitored object, determine the feature vector of the monitored object and the dimension of the feature vector; The feature vector is input into an encoder with the same dimension as the feature vector to obtain the health status of the monitored object.

6. The multi-parameter vital sign monitoring method according to claim 5, characterized in that, After acquiring monitoring data of multiple vital signs of the monitored object, the process also includes: The monitoring data is processed to remove noise and handle outliers.

7. A multi-parameter vital sign monitoring device, characterized in that, include: The acquisition module is used to acquire monitoring data of multiple vital signs of the monitored object; The determination module is used to determine the labeling category of each vital sign indicator of the monitored object based on the monitoring data; The filtering module is used to filter relevant vital signs indicators from the multiple vital signs indicators according to the labeled categories; The calculation module is used to calculate the feature value of each relevant vital sign indicator of the monitored object based on the labeling category of each relevant vital sign indicator and the preset original weight of each relevant vital sign indicator. The monitoring module is used to input the feature values ​​of each relevant vital sign indicator of the monitored object into a preset vital sign monitoring model to obtain the health status of the monitored object; The process of determining the original weights of each pre-defined vital sign indicator includes: Acquire historical monitoring data and corresponding health status for each vital sign indicator of multiple monitored subjects; Based on the historical monitoring data and the corresponding health status, calculate the Pearson correlation coefficient between each vital sign indicator and the health status; Based on the Pearson correlation coefficient, calculate the ratio of the Pearson correlation coefficients between each pair of vital signs. Based on the ratio, a quantitative value is determined for each pair of vital signs, and the quantitative value is positively correlated with the corresponding ratio. Calculate the geometric mean of each quantitative value corresponding to each vital sign to obtain the initial weight of each vital sign indicator; The initial weights are normalized to obtain the original weights for each vital sign indicator.

8. An electronic device comprising a memory and a processor, the memory for storing a computer program, the processor for calling and running the computer program stored in the memory, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6 above.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 6 above.

Images