Safety diagnosis method for medical monitoring

Abstract

The invention discloses a safety diagnosis method for medical monitoring, and relates to the technical field of intelligent medical monitoring, and the method comprises the steps: collecting the physiological data of a patient through a sensor, carrying out the preprocessing of the physiological data, and constructing a multi-modal data analysis model, inputting the preprocessed physiological data into a multi-modal data analysis model for feature extraction and fusion, outputting physiological state features, constructing a safety diagnosis model by using a deep neural network, inputting the physiological state features into a machine learning model, outputting a health risk value, and determining the health risk value according to the historical physiological data of the patient. An early warning threshold value is obtained through a support vector machine (SVM), and when the health risk value exceeds the early warning threshold value, an alarm is sent to medical staff.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent medical monitoring, and in particular to a safe diagnosis method for medical monitoring. Background Art

[0002] In recent years, with the rapid development of sensor technology and computer science, the field of medical monitoring has made significant progress. Traditional medical monitoring usually relies on manual observation and evaluation by doctors and nurses, which is not only time-consuming and labor-intensive, but also easily affected by human factors, resulting in increased subjectivity and uncertainty in monitoring results. With the rise of Internet of Things technology, various types of sensors have begun to be widely used in medical monitoring, such as heart rate monitors, blood pressure monitors, blood glucose meters, and thermometers. These sensors can collect patients' physiological data in real time, providing doctors with more objective and continuous monitoring methods. In terms of data processing, traditional data processing methods often focus on a single type of physiological signal, such as electrocardiogram (ECG) or blood oxygen saturation (SpO2), which limits doctors' understanding of the patient's overall health status. In recent years, multimodal data analysis technology has gradually become a research hotspot. This method can process multiple types of data at the same time, thereby providing a more comprehensive health assessment. In addition, with the development of machine learning and deep learning technologies, it has become possible to use these technologies for automatic feature extraction and model training, which greatly improves the accuracy and efficiency of disease diagnosis. Summary of the invention

[0003] In view of the above existing problems, the present invention is proposed.

[0004] Therefore, the present invention provides a safe diagnosis method for medical monitoring to solve the problem that in the data preprocessing stage, traditional preprocessing methods are often difficult to effectively remove noise and outliers, especially in the scenario of multimodal data fusion, how to ensure the quality of data is an urgent problem to be solved. Secondly, most of the existing multimodal data analysis methods adopt a simple linear combination method. This processing method may ignore the intrinsic connection and interaction between different physiological data, thereby affecting the accuracy of the final diagnosis result. In addition, the existing diagnostic models often focus on the analysis of a single indicator, while ignoring the complex interaction relationship between multiple physiological parameters, which may lead to the diagnosis result being not comprehensive and accurate. Finally, with regard to the setting of warning thresholds, the existing technology often relies on fixed thresholds or simple statistical methods, and lacks consideration of individual differences, which may lead to insufficient sensitivity and specificity of the warning system.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention provides a safe diagnosis method for medical monitoring, which comprises:

[0007] Collect the patient's physiological data through sensors and pre-process the physiological data;

[0008] Construct a multimodal data analysis model, input the preprocessed physiological data into the multimodal data analysis model for feature extraction and fusion, and output physiological state features;

[0009] Use deep neural networks to build a safety diagnosis model, input physiological status characteristics into the machine learning model, and output health risk values;

[0010] According to the patient's historical physiological data, the warning threshold is obtained through the support vector machine (SVM). When the health risk value exceeds the warning threshold, an alarm is sent to medical staff.

[0011] As a preferred solution of the safety diagnosis method for medical monitoring of the present invention, wherein: the physiological data of the patient is collected by the sensor and the physiological data is preprocessed, and the specific steps are:

[0012] Collect the patient's physiological data using a heart rate monitor, a blood pressure monitor, a blood glucose meter, and a thermometer, remove noise and outliers from the physiological data, perform normalization, and convert the physiological data into a zero-mean unit variance form;

[0013] Assume that the collected physiological data set is x, which can be expressed as x=(x1,x2,...,x4), where x1 represents heart rate, x2 represents blood pressure, x3 represents blood sugar, and x4 represents body temperature.

[0014] As a preferred solution of the safety diagnosis method for medical monitoring of the present invention, wherein: a multimodal data analysis model is constructed, and the pre-processed physiological data is input into the multimodal data analysis model for feature extraction, and the specific steps are:

[0015] The nonlinear feature mapping is used to extract the features of the collected physiological data x, and the expression is:

[0016]

[0017] where f k represents the characteristic value of the kth physiological data, x k,i represents the value of the i-th sampling point in the k-th physiological data, n represents the number of sampling points, α, β, γ represent positive real number parameters, log represents logarithmic function, and exp represents exponential function.

[0018] As a preferred solution of the safety diagnosis method for medical monitoring of the present invention, wherein: the feature vectors of different physiological data are fused using a multimodal data analysis model, and the specific steps are:

[0019] The features of different physiological data are fused into a comprehensive feature vector, which is expressed as:

[0020]

[0021] Where F represents the integrated feature vector after fusion, w k represents the weight of the kth type of physiological data, m represents the number of types of physiological data, and w m f m Represents the product of the eigenvalue of the mth physiological data and its corresponding weight.

[0022] As a preferred solution of the safety diagnosis method for medical monitoring of the present invention, the output of physiological state characteristics comprises the following specific steps:

[0023] Based on the fused feature vector, the physiological state feature is output, and the expression is:

[0024]

[0025] Wherein, S represents the physiological state characteristic, η represents the nonlinear transformation degree parameter of the physiological state characteristic S, and θ represents the scaling degree parameter of the comprehensive characteristic vector F in the exponential function.

[0026] As a preferred solution of the safety diagnosis method for medical monitoring described in the present invention, wherein: the safety diagnosis model is constructed using a deep neural network, the physiological state characteristics are input into the machine learning model, and the health risk value is output, the specific steps are:

[0027] Based on the physiological state characteristics S, a safety diagnosis model is constructed using a deep neural network. The physiological state characteristics are input into the safety diagnosis model, and the health risk value R is output, which is expressed as:

[0028]

[0029] Among them, sin represents the sine function, π represents pi, t represents the integral variable, cos represents the cosine function, and dt represents a small change in t.

[0030] As a preferred solution of the safety diagnosis method for medical monitoring of the present invention, the safety diagnosis model is trained according to the patient's historical physiological data, and the specific steps are as follows:

[0031] Collect the patient's historical physiological data, input into the safety diagnosis model, output the historical physiological state characteristics, and find the best hyperplane by solving the optimization problem. The expression is:

[0032] w·S+b=0;

[0033] Among them, w is the normal vector of the hyperplane of the safety diagnosis model, and b is the intercept of the hyperplane of the multimodal data analysis model.

[0034] As a preferred solution of the safety diagnosis method for medical monitoring of the present invention, wherein: the warning threshold is obtained by the support vector machine, and when the health risk value exceeds the warning threshold, an alarm is sent to the medical staff, and the specific steps are:

[0035] The score is predicted by the safety diagnosis model, and the warning threshold T is determined based on this score. The expression is:

[0036]

[0037] Among them, α i represents the Lagrange multiplier, C represents the penalty parameter, N represents the number of samples, y i represents the label of the i-th sample, S i represents the physiological state feature vector of the i-th sample, w and b represent key parameters;

[0038] When R>T, there is a health risk, and the system immediately sends an alert to medical staff;

[0039] When r≤T, there is no health risk and the system remains in the monitoring state.

[0040] In a second aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the safe diagnostic method for medical monitoring as described in the first aspect of the present invention is implemented.

[0041] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the safe diagnostic method for medical monitoring as described in the first aspect of the present invention is implemented.

[0042] The beneficial effects of the present invention are as follows: the present invention realizes the efficient collection, preprocessing, feature extraction, fusion and health risk assessment of patients' physiological data, and ultimately notifies medical staff in a timely manner through a dynamic early warning mechanism, thereby improving the accuracy and timeliness of diagnosis, and also significantly improving the quality and efficiency of medical services through intelligent means. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0044] Figure 1 Flow chart of the safety diagnostic method for medical monitoring in Example 1.

[0045] Figure 2 This is a flow chart for obtaining physiological state characteristics in Example 1. DETAILED DESCRIPTION

[0046] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0048] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0049] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a safe diagnosis method for medical monitoring, comprising the following steps:

[0050] S1 collects the patient's physiological data through sensors and preprocesses the physiological data. The specific steps are as follows:

[0051] Collect the patient's physiological data using a heart rate monitor, a blood pressure monitor, a blood glucose meter, and a thermometer, remove noise and outliers from the physiological data, perform normalization, and convert the physiological data into a zero-mean unit variance form;

[0052] Assume that the collected physiological data set is x, which can be expressed as x=(x1,x2,...,x4), where x1 represents heart rate, x2 represents blood pressure, x3 represents blood sugar, and x4 represents body temperature;

[0053] By removing noise and outliers, the accuracy of subsequent feature extraction is ensured and misdiagnosis due to data quality issues is avoided. The physiological data is converted into a zero-mean unit variance form so that different types of physiological data can be compared on the same scale, which improves the generalization ability of the model, ensures the quality of the data, and provides reliable basic data for subsequent feature extraction and fusion.

[0054] S2 builds a multimodal data analysis model and inputs the preprocessed physiological data into the multimodal data analysis model for feature extraction. The specific steps are as follows:

[0055] The nonlinear feature mapping is used to extract the features of the collected physiological data x, and the expression is:

[0056]

[0057] where f k represents the characteristic value of the kth physiological data, x k,i represents the value of the i-th sampling point in the k-th physiological data, n represents the number of sampling points, α, β, γ represent positive real number parameters, log represents logarithmic function, and exp represents exponential function;

[0058] By extracting features through nonlinear mapping, we can capture the complex relationship between physiological data and enhance the model's ability to recognize physiological states.

[0059] The S3 model fuses the feature vectors of different physiological data. The specific steps are as follows:

[0060] The features of different physiological data are fused into a comprehensive feature vector, which is expressed as:

[0061]

[0062] Where F represents the integrated feature vector after fusion, w k represents the weight of the kth physiological data, m represents the number of types of physiological data, and w m f m represents the product of the characteristic value of the mth physiological data and its corresponding weight;

[0063] The features of different physiological data are integrated into a comprehensive feature vector, which makes full use of the information of multimodal data, improves the comprehensive judgment ability of the model, can more accurately characterize the patient's physiological state, and provides a rich information basis for subsequent risk assessment.

[0064] S4 outputs physiological status characteristics, the specific steps are:

[0065] Based on the fused feature vector, the physiological state characteristics are output, and the expression is:

[0066]

[0067] Wherein, S represents the physiological state characteristic, η represents the nonlinear transformation degree parameter of the physiological state characteristic S, and θ represents the scaling degree parameter of the comprehensive feature vector F in the exponential function.

[0068] S5 uses a deep neural network to build a safety diagnosis model, inputs physiological status characteristics into the machine learning model, and outputs health risk values. The specific steps are as follows:

[0069] Based on the physiological state characteristics S, a safety diagnosis model is constructed using a deep neural network. The physiological state characteristics are input into the safety diagnosis model, and the health risk value R is output, which is expressed as:

[0070]

[0071] Among them, sin represents the sine function, π represents the circumference of a circle, t represents the integral variable, cos represents the cosine function, and dt represents a small change in t;

[0072] Using deep neural networks to build a safety diagnosis model can automatically learn and extract complex features, improve the model's prediction accuracy, and by outputting health risk values, it can provide medical staff with quantitative assessments, helping to detect potential health problems early.

[0073] S6 trains a safety diagnosis model based on the patient's historical physiological data. The specific steps are:

[0074] Collect the patient's historical physiological data, input into the safety diagnosis model, output the historical physiological state characteristics, and find the best hyperplane by solving the optimization problem. The expression is:

[0075] w·S+b=0;

[0076] Wherein, w is the normal vector of the hyperplane of the multimodal data analysis model, and b is the intercept of the hyperplane of the multimodal data analysis model;

[0077] The warning threshold is determined by support vector machine, which solves the problem that fixed threshold cannot adapt to individual differences.

[0078] S7 obtains the warning threshold through the support vector machine. When the health risk value exceeds the warning threshold, it sends an alarm to medical staff. The specific steps are as follows:

[0079] The score is predicted by the safety diagnosis model, and the warning threshold T is determined based on this score. The expression is:

[0080]

[0081] Among them, α i represents the Lagrange multiplier, C represents the penalty parameter, N represents the number of samples, y i represents the label of the i-th sample, S i represents the physiological state feature vector of the i-th sample, w and b represent key parameters;

[0082] When R>T, there is a health risk, and the system immediately sends an alert to medical staff;

[0083] When R≤T, there is no health risk and the system remains in monitoring state;

[0084] When the health risk value exceeds the warning threshold, an alarm can be sent immediately to medical staff, ensuring that timely measures can be taken when health risks occur.

[0085] This embodiment also provides a computer device suitable for the case of a safe diagnosis method for medical monitoring, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to implement the safe diagnosis method for medical monitoring proposed in the above embodiment.

[0086] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0087] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the safe diagnosis method for medical monitoring proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0088] In summary, the present invention achieves efficient collection, preprocessing, feature extraction, fusion and health risk assessment of patients' physiological data, and finally notifies medical staff in a timely manner through a dynamic early warning mechanism, thereby improving the accuracy and timeliness of diagnosis, and significantly improving the quality and efficiency of medical services through intelligent means.

[0089] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of a safety diagnosis method for medical monitoring is provided.

[0090] This embodiment aims to verify the effectiveness and superiority of the proposed safety diagnostic method for medical monitoring, especially in improving the accuracy of health risk assessment. This embodiment was tested on a group of volunteers by comparing the traditional method with the method proposed in the present invention.

[0091] Ten healthy adults aged between 25 and 55 were selected as test subjects. Each participant wore a smart bracelet equipped with a heart rate monitor, blood pressure monitor, blood glucose meter and thermometer. These devices can continuously collect physiological data of heart rate, blood pressure, blood glucose and body temperature. The data was collected once a minute for 24 hours.

[0092] The pandas and scikit-learn libraries in the Python programming language were used to denoise the raw data, remove outliers, and standardize them to zero mean and unit variance. Nonlinear feature mapping was used to extract features, and the features of different physiological data were fused into a comprehensive feature vector by weighted averaging. A deep neural network model was built using the TensorFlow library and trained using the above fused feature vector to output health risk values. The physiological data of each participant for the past week were collected and input into the multimodal data analysis model to output historical physiological status features, find the optimal hyperplane, and determine the warning threshold. When the health risk value exceeds the warning threshold, the system will immediately send an alert to medical staff.

[0093] For comparison, the traditional health risk assessment method based on a single physiological indicator was also used. In addition, a model based on a shallow neural network was used as another comparison benchmark.

[0094] The details are shown in Table 1:

[0095] Table 1 Experimental record table

[0096]

[0097]

[0098] This embodiment compares three different health risk assessment methods to verify the advantages of the safety diagnosis method of the present invention in improving assessment accuracy.

[0099] The method of the present invention: the safety diagnosis model constructed by the multimodal data analysis model and the deep neural network has a lower output health risk value, indicating that the method can more accurately assess health risks.

[0100] Comparative Method 1: The traditional method using a single physiological indicator generally has a higher health risk value, which means that this method may lead to false alarms and increase unnecessary consumption of medical resources.

[0101] Comparative method 2: Although the method based on shallow neural network is improved over the traditional method, it is still inferior to the method of the present invention in terms of the accuracy of health risk assessment.

[0102] It can be seen from the table data that the health risk value of the method of the present invention is generally lower than that of the other two methods, which shows that the present invention has significant advantages in the following aspects:

[0103] More accurate health risk assessment: Due to the use of multimodal data analysis and deep neural networks, the method of the present invention can more comprehensively consider the interactions between multiple physiological data, thereby improving the accuracy of the assessment.

[0104] Lower false alarm rate: Compared with the other two methods, the health risk value of the method of the present invention is lower, which means that it can reduce unnecessary alarms and reduce the waste of medical resources.

[0105] More timely warnings: The warning thresholds obtained through support vector machines can more accurately identify situations that require attention, thereby achieving more timely warnings.

[0106] In summary, the method of the present invention has shown significant advantages in improving the accuracy and timeliness of health risk assessment. Compared with existing technologies, the method of the present invention can significantly improve the quality and efficiency of medical services in practical applications.

[0107] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A safe diagnostic method for medical monitoring, characterized in that: include, Collect the patient's physiological data through sensors and pre-process the physiological data; Construct a multimodal data analysis model, input the preprocessed physiological data into the multimodal data analysis model for feature extraction and fusion, and output physiological state features; Use deep neural networks to build a safety diagnosis model, input physiological status characteristics into the machine learning model, and output health risk values; According to the patient's historical physiological data, the warning threshold is obtained through the support vector machine (SVM). When the health risk value exceeds the warning threshold, an alarm is sent to medical staff.

2. The safe diagnostic method for medical monitoring according to claim 1, characterized in that: The method of collecting the patient's physiological data through sensors and preprocessing the physiological data includes the following specific steps: Collect the patient's physiological data using a heart rate monitor, a blood pressure monitor, a blood glucose meter, and a thermometer, remove noise and outliers from the physiological data, perform normalization, and convert the physiological data into a zero-mean unit variance form; Assume that the collected physiological data set is x, which can be expressed as x=(x1,x2,...,x4), where x1 represents heart rate, x2 represents blood pressure, x3 represents blood sugar, and x4 represents body temperature.

3. The safe diagnostic method for medical monitoring according to claim 2, characterized in that: The method of constructing a multimodal data analysis model and inputting the preprocessed physiological data into the multimodal data analysis model for feature extraction comprises the following specific steps: The nonlinear feature mapping is used to extract the features of the collected physiological data x, and the expression is: where f k represents the characteristic value of the kth physiological data, x k,i represents the value of the i-th sampling point in the k-th physiological data, n represents the number of sampling points, α, β, γ represent positive real number parameters, log represents logarithmic function, and exp represents exponential function.

4. The safe diagnostic method for medical monitoring according to claim 3, characterized in that: The multimodal data analysis model is used to fuse the feature vectors of different physiological data, and the specific steps are as follows: The features of different physiological data are fused into a comprehensive feature vector, which is expressed as: Where F represents the integrated feature vector after fusion, w k represents the weight of the kth type of physiological data, m represents the number of types of physiological data, and w m f m Represents the product of the eigenvalue of the mth physiological data and its corresponding weight.

5. The safe diagnostic method for medical monitoring according to claim 4, characterized in that: The outputting of physiological state characteristics comprises the following specific steps: Based on the fused feature vector, the physiological state feature is output, and the expression is: Wherein, S represents the physiological state characteristic, η represents the nonlinear transformation degree parameter of the physiological state characteristic S, and θ represents the scaling degree parameter of the comprehensive characteristic vector F in the exponential function.

6. The safe diagnostic method for medical monitoring according to claim 5, characterized in that: The method uses a deep neural network to build a safety diagnosis model, inputs physiological state characteristics into a machine learning model, and outputs a health risk value. The specific steps are as follows: Using the deep neural network model as the basis of the safety diagnosis model, the physiological state feature S is input into the safety diagnosis model, and the health risk value R is output, which is expressed as: Among them, sin represents the sine function, π represents pi, t represents the integral variable, cos represents the cosine function, and dt represents a small change in t.

7. The safe diagnostic method for medical monitoring according to claim 6, characterized in that: The specific steps of training the safety diagnosis model based on the patient's historical physiological data are as follows: Collect the patient's historical physiological data, input into the safety diagnosis model, output the historical physiological state characteristics, and find the best hyperplane by solving the optimization problem. The expression is: w·S+b=0; Among them, w represents the normal vector of the hyperplane of the safety diagnosis model, and b represents the intercept of the hyperplane of the multimodal data analysis model.

8. The safe diagnostic method for medical monitoring according to claim 6, characterized in that: The support vector machine is used to obtain the warning threshold, and when the health risk value exceeds the warning threshold, an alarm is sent to medical staff. The specific steps are: The score is predicted by the safety diagnosis model, and the warning threshold T is determined based on this score. The expression is: Among them, α i represents the Lagrange multiplier, C represents the penalty parameter, N represents the number of samples, y i represents the label of the i-th sample, S i represents the physiological state feature vector of the i-th sample, w and b represent key parameters; When R>T, there is a health risk, and the system immediately sends an alert to medical staff; When r≤T, there is no health risk and the system remains in the monitoring state.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the safe diagnosis method for medical monitoring described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the safe diagnosis method for medical monitoring described in any one of claims 1 to 7 are implemented.

Images