ECMO intelligent monitoring system and method based on multi-modal data fusion

Abstract

The invention discloses an ECMO intelligent monitoring system and method based on multi-modal data fusion, and relates to the technical field of severe recovery monitoring. Acquiring patient basic information of the critical patient, wherein the patient basic information comprises personal information, medical record information and physiological parameter information of a plurality of recent continuous time periods; taking the basic information of the patient as input of a severe recovery condition analysis model for analysis to obtain risk assessment information; the severe recovery condition analysis model comprises a static feature processing branch, a dynamic feature processing branch and a fusion prediction branch, and the static feature processing branch is used for performing feature extraction on personal information and medical record information to obtain a static feature vector; the dynamic feature processing branch is used for performing feature extraction on the physiological parameter information to obtain a dynamic feature vector; and the fusion prediction branch is used for fusing the static feature vector and the dynamic feature vector to obtain risk assessment information. According to the invention, the risk condition of the critical patient can be accurately analyzed in real time while the consumption of medical resources is reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of severe recovery monitoring, and particularly relates to an ECMO intelligent monitoring system and method based on multi-modal data fusion. BACKGROUND

[0002] Severe patient physiological parameter monitoring is one of the core links of clinical medical treatment. By monitoring various physiological parameter indexes of severe patients, the risks that may occur to the severe patients can be evaluated, so as to perform early warning and optimize treatment strategies.

[0003] For risk monitoring of severe patients, although the existing technology has introduced the data set function of the electronic medical record system (EMR) and part of the monitoring equipment, the monitoring means still mainly depends on medical staff to make comprehensive judgments based on the time collection of vital signs (such as heart rate, blood pressure, and blood oxygen saturation) combined with clinical experience. It is highly dependent on the professional knowledge of medical staff and is easily affected by factors such as personal experience and certification bias. At the same time, since it needs to be analyzed and judged at regular intervals, it leads to the consumption of a large amount of medical resources.

[0004] Therefore, how to provide an effective scheme to facilitate the accurate analysis of the risk situation of severe patients in real time while reducing the consumption of medical resources has become a difficult problem to be solved in the prior art. SUMMARY

[0005] The purpose of the application is to provide an ECMO intelligent monitoring system and method based on multi-modal data fusion to solve the above problems existing in the prior art.

[0006] In order to achieve the above purpose, the application adopts the following technical scheme: In a first aspect, the application provides an ECMO intelligent monitoring method based on multi-modal data fusion, comprising: obtaining patient basic information of a severe patient, the patient basic information including personal information, medical record information, and physiological parameter information detected in a plurality of continuous time periods; performing severe recovery condition analysis on the patient basic information as the input of a pre-trained severe recovery condition analysis model to obtain risk assessment information of the severe patient at present; The pre-trained severe recovery condition analysis model includes a static feature processing branch, a dynamic feature processing branch, and a fusion prediction branch. The static feature processing branch is used for feature extraction on the personal information and the medical record information to obtain a static feature vector. The dynamic feature processing branch is used for feature extraction on the physiological parameter information detected in the plurality of continuous time periods to obtain a dynamic feature vector. The fusion prediction branch is configured to perform feature fusion on the static feature vector and the dynamic feature vector to obtain the current risk assessment information of the critical patient.

[0007] In a possible design, the static feature processing branch comprises an embedding layer and a normalization layer, the embedding layer is configured to map the categorical data in the personal information and the medical record information into a low-dimensional vector, and the normalization layer is configured to perform normalization processing on the numerical data in the personal information and the medical record information, so as to splice the low-dimensional vector and the numerical data after the normalization processing to obtain the static feature vector. The dynamic feature processing branch comprises an LSTM encoder and an attention mechanism module, the LSTM encoder is configured to extract the time sequence dependency between the physiological parameter information of the recent continuous multiple time periods and generate a feature representation sequence containing global context, and the attention mechanism module is configured to learn the weight of each feature in the feature representation sequence and integrate the feature representation sequence into the dynamic feature vector through weighted summation. The fusion prediction branch comprises a feature splicing layer, a full connection layer and an output layer, the feature splicing layer is configured to splice the static feature vector and the dynamic feature vector to obtain a comprehensive feature vector, the full connection layer is configured to perform nonlinear transformation processing on the comprehensive feature vector, and the output layer is configured to activate the comprehensive feature vector after the nonlinear transformation processing through a softmax activation function to obtain the current risk assessment information of the critical patient.

[0008] In a possible design, the personal information comprises gender, age, height and / or weight, the medical record information comprises disease type and / or surgery type, and the physiological parameter information comprises blood flow, average anticoagulation index, average pre- and post-membrane pressure difference, average oxygenation index, average heart rate, average respiratory rate, average blood pressure, average blood glucose, average blood oxygen saturation and / or average body temperature.

[0009] In a possible design, the method further comprises: obtaining a sample data set, the sample data in the sample data set comprising patient basic information of a sample critical patient and risk assessment information of the sample critical patient evaluated by a doctor according to the patient basic information of the sample critical patient; training the critical recovery condition analysis model by taking the patient basic information of the sample critical patient in the sample data as sample input and taking the risk assessment information of the sample critical patient in the sample data as sample output to obtain the pre-trained critical recovery condition analysis model.

[0010] In a possible design, after obtaining the patient basic information of the critical patient, the method further comprises: Outlier rejection and missing value imputation are performed on the patient basic information.

[0011] In one possible design, after obtaining the current risk assessment information of the critical patient, the method further includes: generating a risk assessment report of the critical patient based on the current risk assessment information of the critical patient.

[0012] In a second aspect, the present application provides an ECMO intelligent monitoring system based on multi-modal data fusion, comprising: An acquisition unit is configured to acquire patient basic information of a critical patient, wherein the patient basic information comprises personal information, medical record information, and physiological parameter information detected in a plurality of continuous time periods; An analysis unit is configured to perform critical recovery condition analysis on the patient basic information as input of a pre-trained critical recovery condition analysis model to obtain current risk assessment information of the critical patient. The pre-trained critical recovery condition analysis model comprises a static feature processing branch, a dynamic feature processing branch, and a fusion prediction branch. The dynamic feature processing branch is configured to perform feature extraction on the physiological parameter information detected in the plurality of continuous time periods to obtain a dynamic feature vector. The fusion prediction branch is configured to perform feature fusion on the static feature vector and the dynamic feature vector to obtain the current risk assessment information of the critical patient.

[0013] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a transceiver connected in sequence and in communication, wherein the memory is configured to store a computer program, the transceiver is configured to receive and send messages, and the processor is configured to read the computer program and execute the ECMO intelligent monitoring method based on multi-modal data fusion according to the first aspect or any possible design of the first aspect.

[0014] In a fourth aspect, the present application provides a computer readable storage medium having instructions stored thereon, wherein when the instructions are run on a computer, the ECMO intelligent monitoring method based on multi-modal data fusion according to the first aspect or any possible design of the first aspect is executed.

[0015] In a fifth aspect, the present application provides a computer program product comprising instructions, wherein when the instructions are run on a computer, the computer is caused to execute the ECMO intelligent monitoring method based on multi-modal data fusion according to the first aspect or any possible design of the first aspect.

[0016] Advantages: The present application obtains patient basic information of a critical patient, the patient basic information including personal information, medical record information and physiological parameter information detected in a plurality of continuous time periods, then performs critical recovery condition analysis on the patient basic information as an input of a pre-trained critical recovery condition analysis model to obtain current risk assessment information of the critical patient. The pre-trained critical recovery condition analysis model includes a static feature processing branch, a dynamic feature processing branch and a fusion prediction branch, the static feature processing branch is used for feature extraction on the personal information and the medical record information to obtain a static feature vector, the dynamic feature processing branch is used for feature extraction on the physiological parameter information detected in the plurality of continuous time periods to obtain a dynamic feature vector, and the fusion prediction branch is used for feature fusion on the static feature vector and the dynamic feature vector to obtain the current risk assessment information of the critical patient. In this way, through the double-branch design and the feature fusion mechanism, the static features in heterogeneous data can be effectively captured, and the key dynamic patterns in time series can also be captured, so that the model can better understand and analyze the data, improve the accuracy and efficiency of critical patient risk (such as complication risk) condition analysis, and can be analyzed without relying on professional medical staff, reduce medical resource consumption, and facilitate practical application and promotion. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A flowchart of an ECMO intelligent monitoring method based on multi-modal data fusion provided by the embodiments of the present application is shown in the figure; Figure 2 A structural diagram of a critical recovery condition analysis model provided by the embodiments of the present application is shown in the figure; Figure 3 A block diagram of an ECMO intelligent monitoring system based on multi-modal data fusion provided by the embodiments of the present application is shown in the figure; Figure 4 A block diagram of an electronic device provided by the embodiments of the present application is shown in the figure. DETAILED DESCRIPTION

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the present application will be briefly introduced below in combination with the drawings and the descriptions of the embodiments or the prior art. Obviously, the following description of the drawings structure is only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor. It should be noted that the description of these embodiments is used to help understand the present application, but does not constitute a limitation on the present application.

[0019] It should be understood that although the terms first, second, etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit, without departing from the scope of the exemplary embodiments of the invention.

[0020] It should be understood that the term "and / or" that may appear in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, and A and B exist simultaneously. The term " / and" that may appear in this document describes another relationship between related objects, indicating that two relationships can exist. For example, A / and B can mean: A exists alone, and A and B exist alone. In addition, the character " / " that may appear in this document generally indicates that the related objects before and after it are in an "or" relationship.

[0021] It should be understood that specific details are provided in the following description to facilitate a complete understanding of the exemplary embodiments. However, those skilled in the art will understand that the exemplary embodiments can be implemented without these specific details. For example, the system may be shown in block diagrams to avoid obscuring the example with unnecessary details. In other instances, well-known processes, structures, and techniques may be shown without unnecessary details to avoid obscuring the exemplary embodiments.

[0022] To analyze the recovery status of critically ill patients, this application provides an intelligent monitoring system and method for ECMO (extracorporeal life support) based on multimodal data fusion. This intelligent ECMO monitoring system and method based on multimodal data fusion can accurately analyze the risk status of critically ill patients in real time while reducing the consumption of medical resources.

[0023] like Figure 1 The diagram shown is a flowchart of an ECMO intelligent monitoring method based on multimodal data fusion provided in the first aspect of the embodiments of this application. The ECMO intelligent monitoring method based on multimodal data fusion may include, but is not limited to, the following steps S101-S102.

[0024] Step S101. Obtain basic patient information for critically ill patients.

[0025] The patient basic information can include, but is not limited to, personal information of the patient, medical record information, and physiological parameter information detected in a plurality of continuous time periods. The personal information can include, but is not limited to, gender, age, height, weight, and the like of the patient, the medical record information can include, but is not limited to, disease type and / or surgery type of the patient, and the physiological parameter information can include, but is not limited to, blood flow, average anticoagulation index, average pre- and post-membrane pressure difference, average oxygenation index, average heart rate, average respiratory rate, average blood pressure, average blood glucose, average blood oxygen saturation, and / or average body temperature of the patient. For example, the average heart rate of the patient can be measured by a dynamic electrocardiogram monitor, the average respiratory rate of the patient can be measured by a chest movement sensor or an impedance respiration monitor, the average blood pressure of the patient can be measured by a non-invasive blood pressure meter or an invasive blood pressure monitoring system, the average blood glucose of the patient can be measured by a fingertip blood glucose meter or a continuous blood glucose monitoring system, the average blood oxygen saturation of the patient can be measured by a pulse oximeter or a blood oxygen saturation monitoring system, and the average body temperature of the patient can be measured by an electronic thermometer.

[0026] The time length corresponding to each time period in the plurality of continuous time periods can be set according to actual conditions, and the number of time periods can also be set according to actual conditions. For example, the plurality of continuous time periods include four time periods, and the time length corresponding to each time period is three hours.

[0027] In one or more embodiments, after obtaining the patient basic information of the critical patient, the patient basic information can also be subjected to outlier rejection and missing value filling processing.

[0028] Step S102. Perform critical recovery condition analysis on the patient basic information as input of the pre-trained critical recovery condition analysis model to obtain the current risk assessment information of the critical patient.

[0029] Referring to Figure 2 The pre-trained critical recovery condition analysis model can include a static feature processing branch, a dynamic feature processing branch, and a fusion prediction branch. The static feature processing branch is configured to extract features from the personal information and the medical record information to obtain a static feature vector. The dynamic feature processing branch is configured to extract features from the physiological parameter information detected in the plurality of continuous time periods to obtain a dynamic feature vector. The fusion prediction branch is configured to fuse the static feature vector and the dynamic feature vector to obtain the current risk assessment information of the critical patient.

[0030] The static feature processing branch is configured to extract features from the personal information and the medical record information. Referring to Figure 2The static feature processing branch can include an embedding layer and a normalization layer. The embedding layer is configured to map the categorical data in the personal information and the medical record information into a low-dimensional vector. The normalization layer is configured to normalize the numerical data in the personal information and the medical record information. The low-dimensional vector and the numerical data after normalization can be spliced together to obtain the static feature vector.

[0031] For example, the embedding layer can be used to convert the gender, disease type, and surgery type in the personal information and the medical record information into a vector containing 16 numbers. The normalization layer can be used to normalize the age, height, and weight in the personal information and the medical record information to a value in the range of 0 to 1.

[0032] The dynamic feature processing branch is configured to extract features from the physiological parameter information detected in the recent continuous time periods to obtain a dynamic feature vector. Please refer to Figure 2 The dynamic feature processing branch can include an LSTM encoder and an attention mechanism module. The LSTM encoder is configured to extract the time sequence dependency between the physiological parameter information in the recent continuous time periods and generate a feature representation sequence containing global context. The attention mechanism module is configured to learn the weight of each feature in the feature representation sequence and integrate the feature representation sequence into the dynamic feature vector through weighted summation.

[0033] The attention mechanism module can automatically evaluate the importance of each time period. The attention mechanism module can calculate an attention score for each time period. The higher the score, the more critical the time period is for the final prediction. After calculating the attention score of each time period, the features in the feature representation sequence can be weighted based on the attention score of each time period to obtain the dynamic feature vector.

[0034] The fusion prediction branch is configured to fuse the static feature vector and the dynamic feature vector to obtain the current risk assessment information of the critical patient. Please refer to Figure 2 The fusion prediction branch includes a feature splicing layer, a fully connected layer, and an output layer. The feature splicing layer is configured to splice the static feature vector and the dynamic feature vector to obtain a comprehensive feature vector. The fully connected layer is configured to perform nonlinear transformation processing on the comprehensive feature vector. The output layer is configured to activate the comprehensive feature vector after nonlinear transformation processing through a softmax activation function to output the current risk assessment information of the critical patient.

[0035] The current risk assessment information of the critical patient can be various risks that the critical patient can have and probabilities of the various risks. The various risks can be risks directly induced by the disease of the critical patient (such as hypotension, hypoxia, etc.), and can also be complication risks induced by the disease of the critical patient, which are not specifically limited in the embodiments of the present application.

[0036] In one or more embodiments, after obtaining the current risk assessment information of the critical patient, a risk assessment report of the critical patient can also be generated based on the current risk assessment information of the critical patient. In this way, the risk that can occur can be warned in advance, and the optimization strategy can be automatically adjusted, so as to improve the success rate of treatment of the critical patient.

[0037] The pre-trained critical recovery analysis model can be used to predict the current risk assessment information of the critical patient. In one or more embodiments, the training process of the critical recovery analysis model can include, but is not limited to, the following steps S201-S202.

[0038] Step S201. Obtain a sample data set.

[0039] The sample data in the sample data set includes patient basic information of a sample critical patient and risk assessment information of the sample critical patient evaluated by a doctor according to the patient basic information of the sample critical patient. The patient basic information of the sample critical patient includes personal information, medical record information of the sample critical patient, and physiological parameter information of the sample critical patient detected in a plurality of continuous time periods before being evaluated.

[0040] Step S202. Take the patient basic information of the sample critical patient in the sample data as a sample input of the critical recovery analysis model, and take the risk assessment information of the sample critical patient in the sample data as a sample output of the critical recovery analysis model to train, so as to obtain the pre-trained critical recovery analysis model.

[0041] The ECMO intelligent monitoring method based on multi-modal data fusion provided by the application, by acquiring patient basic information of a critical patient, the patient basic information including personal information, medical record information and physiological parameter information detected in a plurality of continuous time periods, then taking the patient basic information as an input of a pre-trained critical recovery condition analysis model to analyze the critical recovery condition, and obtaining current risk assessment information of the critical patient. The pre-trained critical recovery condition analysis model includes a static feature processing branch, a dynamic feature processing branch and a fusion prediction branch. The static feature processing branch is used for feature extraction on the personal information and the medical record information to obtain a static feature vector. The dynamic feature processing branch is used for feature extraction on the physiological parameter information detected in the plurality of continuous time periods to obtain a dynamic feature vector. The fusion prediction branch is used for feature fusion on the static feature vector and the dynamic feature vector to obtain the current risk assessment information of the critical patient. In this way, by means of the double-branch design and the feature fusion mechanism, the static features in the heterogeneous data can be effectively captured, and the key dynamic patterns in the time sequence can also be captured, so that the model can better understand and analyze the data, improve the accuracy and efficiency of the critical patient risk condition analysis, so as to early warn the possible risks, automatically adjust and optimize the strategy, significantly reduce the complication risk, improve the success rate of critical patient treatment, and at the same time, the analysis can be performed without relying on professional medical staff, the medical resource consumption is reduced, and the actual application and promotion are facilitated.

[0042] Please refer to Figure 3 The second aspect of the embodiments of the application provides an ECMO intelligent monitoring system based on multi-modal data fusion, which comprises: An acquisition unit is configured to acquire patient basic information of a critical patient, the patient basic information including personal information, medical record information and physiological parameter information detected in a plurality of continuous time periods; An analysis unit is configured to analyze the patient basic information as an input of a pre-trained critical recovery condition analysis model to analyze the critical recovery condition, and obtain current risk assessment information of the critical patient; The pre-trained critical recovery condition analysis model includes a static feature processing branch, a dynamic feature processing branch and a fusion prediction branch. The static feature processing branch is configured to extract features from the personal information and the medical record information to obtain a static feature vector. The dynamic feature processing branch is configured to extract features from the physiological parameter information detected in the plurality of continuous time periods to obtain a dynamic feature vector. The fusion prediction branch is configured to fuse the static feature vector and the dynamic feature vector to obtain the current risk assessment information of the critical patient.

[0043] The working process, working details and technical effects of the ECMO intelligent monitoring system based on multi-modal data fusion provided by the second aspect of the embodiment can be referred to the first aspect of the embodiment, and will not be described here.

[0044] Please refer to Figure 4 The third aspect of the embodiment of the present application provides an electronic device, which comprises a memory, a processor and a transceiver connected in sequence and in communication. The memory is configured to store a computer program. The transceiver is configured to receive and send messages. The processor is configured to read the computer program and execute the ECMO intelligent monitoring method based on multi-modal data fusion as described in the first aspect of the embodiment.

[0045] For example, the memory can include, but is not limited to, random access memory (RAM), read-only memory (ROM), flash memory, first-in-first-out memory (FIFO) and / or first-in-last-out memory (FILO), etc. The processor can be, but is not limited to, a microprocessor of STM32F105 series, an ARM (Advanced RISC Machines) processor, an X86 architecture processor or a processor integrated with NPU (neural-network processing units). The transceiver can be, but is not limited to, a WiFi (Wireless Fidelity) wireless transceiver, a Bluetooth wireless transceiver, a General Packet Radio Service (GPRS) wireless transceiver, a ZigBee wireless transceiver, a 3G transceiver, a 4G transceiver and / or a 5G transceiver, etc.

[0046] The fourth aspect of the embodiment provides a computer readable storage medium storing instructions of the ECMO intelligent monitoring method based on multi-modal data fusion as described in the first aspect of the embodiment, i.e. the computer readable storage medium stores instructions, when the instructions are run on a computer, the ECMO intelligent monitoring method based on multi-modal data fusion as described in the first aspect is executed. The computer readable storage medium refers to a carrier storing data, which can include, but is not limited to, floppy disks, optical disks, hard disks, flash memories, USB flash drives and / or Memory Sticks, etc. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices.

[0047] The fifth aspect of the embodiment provides a computer program product containing instructions, when the instructions are run on a computer, the computer executes the ECMO intelligent monitoring method based on multi-modal data fusion as described in the first aspect of the embodiment, wherein the computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0048] Finally, it should be noted that the above only describes the preferred embodiments of the present application and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An ECMO intelligent monitoring method based on multi-modal data fusion, characterized in that, The method comprises: obtaining patient basic information of a critical patient, the patient basic information comprising personal information, medical record information, and physiological parameter information detected in a plurality of continuous time periods in the past; performing critical recovery condition analysis on the patient basic information as input of a pre-trained critical recovery condition analysis model to obtain current risk assessment information of the critical patient; wherein the pre-trained critical recovery condition analysis model comprises a static feature processing branch, a dynamic feature processing branch, and a fusion prediction branch, the static feature processing branch is configured to extract features from the personal information and the medical record information to obtain a static feature vector; the dynamic feature processing branch is configured to extract features from the physiological parameter information detected in the plurality of continuous time periods in the past to obtain a dynamic feature vector; the fusion prediction branch is configured to fuse the static feature vector and the dynamic feature vector to obtain the current risk assessment information of the critical patient.

2. The ECMO intelligent monitoring method based on multi-modal data fusion according to claim 1, characterized in that, The static feature processing branch comprises an embedding layer and a normalization layer, the embedding layer is configured to map category data in the personal information and the medical record information into a low-dimensional vector, and the normalization layer is configured to normalize numerical data in the personal information and the medical record information, so as to splice the low-dimensional vector and the numerical data after normalization to obtain the static feature vector; the dynamic feature processing branch comprises an LSTM encoder and an attention mechanism module, the LSTM encoder is configured to extract time sequence dependency between the physiological parameter information in the plurality of continuous time periods in the past and generate a feature representation sequence containing global context, and the attention mechanism module is configured to learn weights of each feature in the feature representation sequence and integrate the feature representation sequence into the dynamic feature vector through weighted summation; the fusion prediction branch comprises a feature splicing layer, a full connection layer, and an output layer, the feature splicing layer is configured to splice the static feature vector and the dynamic feature vector to obtain a comprehensive feature vector, the full connection layer is configured to perform nonlinear transformation processing on the comprehensive feature vector, and the output layer is configured to activate the comprehensive feature vector after nonlinear transformation processing through a softmax activation function to obtain the current risk assessment information of the critical patient.

3. The ECMO intelligent monitoring method based on multi-modal data fusion according to claim 1, characterized in that, The personal information comprises gender, age, height, and / or weight, the medical record information comprises disease type and / or surgery type, and the physiological parameter information comprises blood flow, average anticoagulation index, average pre- and post-membrane pressure difference, average oxygenation index, average heart rate, average respiratory rate, average blood pressure, average blood glucose, average blood oxygen saturation, and / or average body temperature.

4. The ECMO intelligent monitoring method based on multi-modal data fusion according to claim 1, characterized in that, The method further comprises: obtaining a sample data set, sample data in the sample data set comprising patient basic information of a sample critical patient and risk assessment information of the sample critical patient evaluated by a doctor according to the patient basic information of the sample critical patient; The patient basic information of the sample severe patient in sample data is taken as sample input of a severe recovery condition analysis model, and the risk assessment information of the sample severe patient in sample data is taken as sample output of the severe recovery condition analysis model for training, to obtain a pre-trained severe recovery condition analysis model.

5. The ECMO intelligent monitoring method based on multi-modal data fusion according to claim 1, characterized in that, After obtaining the patient basic information of the severe patient, the method further includes: The patient basic information is subjected to outlier rejection and missing value padding processing.

6. The ECMO intelligent monitoring method based on multi-modal data fusion according to claim 1, characterized in that, After obtaining the current risk assessment information of the severe patient, the method further includes: Generating a risk assessment report of the severe patient based on the current risk assessment information of the severe patient.

7. An ECMO intelligent monitoring system based on multi-modal data fusion, characterized in that, Comprise: An acquisition unit is configured to acquire patient basic information of a severe patient, the patient basic information comprising personal information, medical record information, and physiological parameter information detected in a plurality of consecutive time periods; An analysis unit is configured to analyze the patient basic information as input of a pre-trained severe recovery condition analysis model to obtain current risk assessment information of the severe patient; The pre-trained severe recovery condition analysis model comprises a static feature processing branch, a dynamic feature processing branch, and a fusion prediction branch, the static feature processing branch is configured to extract features from the personal information and the medical record information to obtain a static feature vector; The dynamic feature processing branch is configured to extract features from the physiological parameter information detected in the plurality of consecutive time periods to obtain a dynamic feature vector; The fusion prediction branch is configured to fuse the static feature vector and the dynamic feature vector to obtain the current risk assessment information of the severe patient.

8. An electronic device, comprising: The memory, the processor, and the transceiver are connected in sequence and communicate with each other, the memory is configured to store a computer program, the transceiver is configured to receive and send messages, and the processor is configured to read the computer program and execute the ECMO intelligent monitoring method based on multi-modal data fusion according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores instructions, and when the instructions are executed on the computer, the ECMO intelligent monitoring method based on multi-modal data fusion according to any one of claims 1-6 is executed.

10. A computer program product comprising computer programs or instructions, characterized in that, The computer program or the instructions realize the ECMO intelligent monitoring method based on multi-modal data fusion according to any one of claims 1-6 when executed on the computer.

Images