Acute shock early recognition and typing system based on multi-parameter fusion

Abstract

The invention relates to the technical field of medical data processing, in particular to an acute shock early recognition and typing system based on multi-parameter fusion. Firstly, various related physical sign parameter time sequence data are acquired from the data acquisition module; the anomaly analysis module analyzes the deviation between the data value at each moment and a preset range to obtain an anomaly expression coefficient, reflect the degree of deviation from a normal range, compare the difference of the anomaly expression coefficients between the moments, determine the treatment influence expression degree and evaluate the treatment measure effect; the multi-feature fusion module analyzes the time sequence and similar features of the treatment influence expression degree, fuses the time sequence and similar features with abnormal expression coefficient difference features, calculates significance indexes, and comprehensively evaluates the importance of physical sign parameters; and finally, in a weight adjustment and neural network analysis module, adjusting a preset importance weight based on the significance index, so that the system adjusts different sign parameter weights in real time, automatically learns the feature importance degree, and improves the neural network identification and accurate typing capabilities.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical data processing, and in particular to an acute shock early identification and typing system based on multi-parameter fusion. BACKGROUND

[0002] In the medical field, acute shock is an extremely dangerous and life-threatening emergency condition. It can be caused by various causes such as severe infection, blood loss, anaphylaxis, etc., and the disease develops rapidly, so early identification and accurate typing of acute shock is crucial to improve the survival rate and quality of life of patients.

[0003] At present, when identifying and typing acute shock early, the time series data of multiple patient physical parameters such as blood pressure, blood oxygen saturation, etc. are usually collected, and then the multiple physical parameters are input into a neural network for classification. However, in the prior art, the importance weight of all physical parameters at each time is usually a uniform fixed preset value. In actual scenarios, different types of acute shock will cause different types of physical parameters to exhibit different characteristics due to different causes and pathological mechanisms. Therefore, if the importance weight of all types of physical parameters at each time is set to a uniform fixed preset value in the neural network, the information value of the data cannot be fully utilized, and the final recognition accuracy may be affected. SUMMARY

[0004] In order to solve the technical problem that in actual scenarios, different types of acute shock will cause different types of physical parameters to exhibit different characteristics due to different causes and pathological mechanisms, so if the importance weight of all types of physical parameters at each time is set to a uniform fixed preset value in the neural network, the information value of the data cannot be fully utilized, and the final recognition accuracy may be affected, the purpose of the present application is to provide an acute shock early identification and typing system based on multi-parameter fusion, and the technical solution adopted is as follows: The present application provides an acute shock early identification and typing system based on multi-parameter fusion, which comprises: A data acquisition module for acquiring time series data of multiple patient physical parameters; An abnormality analysis module for analyzing the deviation between the data value at each time and the preset range in each physical parameter time series data to obtain the abnormal performance coefficient of each physical parameter at each time, and comparing the difference characteristics of the abnormal performance coefficients between times to determine the treatment influence performance degree of each physical parameter at each time. a multi-feature fusion module, configured to analyze time sequence distribution characteristics of treatment influence performance degrees between time points and similar characteristics in the time sequence data of the vital sign parameters, and fuse the time sequence distribution characteristics of the treatment influence performance degrees between the time points and difference characteristics of abnormal performance coefficients, so as to obtain a significance index of each vital sign parameter at each time point; a weight adjustment and neural network analysis module, configured to adjust preset importance weights of the vital sign parameters based on the significance index corresponding to each time point in the time sequence data of the vital sign parameters, and then perform neural network analysis on all the time sequence data of the vital sign parameters.

[0005] Further, the method for obtaining the abnormal performance coefficient comprises: determining a preset range corresponding to each vital sign parameter, and obtaining an upper boundary value and a lower boundary value of the preset range, and taking a difference value between the upper boundary value and the lower boundary value as a range value; comparing, in the time sequence data of each vital sign parameter, a data value at each time point with the upper boundary value and the lower boundary value of the corresponding preset range; when the data value is greater than the upper boundary value, calculating a difference value between the data value and the upper boundary value as a deviation factor, and taking a ratio between the deviation factor and the corresponding range value as the abnormal performance coefficient of each vital sign parameter at each time point; when the data value is less than the lower boundary value, calculating a difference value between the lower boundary value and the data value as a deviation factor, and taking a ratio between the deviation factor and the corresponding range value as the abnormal performance coefficient of each vital sign parameter at each time point.

[0006] Further, the method for obtaining the treatment influence performance degree comprises: in the time sequence data of each vital sign parameter, taking a normalized value of an absolute value difference between an abnormal performance coefficient at each time point and an abnormal performance coefficient at a previous adjacent time point as a treatment influence performance factor corresponding to each time point; taking a mean value of the treatment influence performance factors corresponding to all time points in the time sequence data of each vital sign parameter as a mean characteristic value, and taking a normalized value of a ratio between the treatment influence performance factor corresponding to each time point and the mean characteristic value as the treatment influence performance degree of each vital sign parameter at each time point.

[0007] Further, the method for obtaining the significance index comprises: in the time sequence data of each vital sign parameter, taking a time point at which the treatment influence performance degree is greater than a preset performance threshold value as a significant change point, and taking a time point at which the treatment influence performance degree is less than or equal to the preset performance threshold value as a normal performance point; based on the performance of the treatment influence performance degree in the time sequence data of all kinds of vital sign parameters at each time point, determining a change performance sequence corresponding to each time point in the time sequence. In the time series data of each vital sign parameter, the similarity characteristics between time points are analyzed to determine the macroscopic difference significance value corresponding to each time point; In the time series data of each vital sign parameter, the temporal distribution characteristics between the time point and the point of significant change are compared and integrated with the difference characteristics of the abnormal performance coefficient to determine the significant value of the micro difference corresponding to each time point; In the time series data of each vital sign parameter, the normalized value of the product of the macroscopic difference significance value and the microscopic difference significance value at each time point is used as the significance index of each vital sign parameter at each time point.

[0008] Furthermore, the method for obtaining the change sequence includes: In terms of time series, for any given moment, when that moment is a significant change point in the time series data of a certain vital sign parameter, the performance value of that moment in the time series data of that vital sign parameter is recorded as the first preset value; when that moment is a normal performance point in the time series data of a certain vital sign parameter, the performance value of that moment in the time series data of that vital sign parameter is recorded as the second preset value. The performance values ​​of that moment in the time series data of all types of vital sign parameters are sorted according to the preset order of the types of vital sign parameters, thereby obtaining the change performance sequence corresponding to each moment.

[0009] Furthermore, the method for obtaining the macroscopic difference significance value includes: In the time series data of each vital sign parameter, the cosine similarity between the change sequence at each time point and the adjacent previous time point is negatively correlated and normalized, and the result is used as the macroscopic difference significance value corresponding to each time point.

[0010] Furthermore, the method for obtaining the significant value of the microscopic difference includes: In the time series data of each vital sign parameter, the normalized value of the absolute difference between the abnormal performance coefficients at each time point and the nearest significant change point in the time series is used as the significant value of the micro-difference corresponding to each time point.

[0011] Furthermore, in the time-series data of vital signs, after adjusting the preset importance weights of the vital signs parameters based on the significance index corresponding to the time, neural network analysis is performed on all time-series data of vital signs parameters, including: Based on the proportion of the significance index of each vital sign parameter in the overall significance index of all types of vital sign parameters at each time point, the adjustment factor of each vital sign parameter at each time point is determined. The sum of the adjustment factor and the preset importance weight for each vital sign parameter at each time point is normalized and used as the final importance weight. The final importance weights of all types of vital signs parameters at all times are combined into an importance weight matrix based on the time series and the preset order of the types of vital signs parameters. The importance weight matrix and time series data of all types of vital signs are used as inputs to the neural network for neural network analysis.

[0012] Furthermore, the method for obtaining the regulation factor includes: The sum of the significance indices of all time points across all types of vital signs is used as the significance and value characteristic values. In the time series data of each vital sign parameter, the ratio of the significance index corresponding to each time point to the significance and value characteristic value is used as the adjustment factor for each vital sign parameter at each time point.

[0013] Furthermore, the neural network is an RNN.

[0014] The present invention has the following beneficial effects: During the development of acute shock, the patient's condition may rapidly deteriorate or fluctuate. Time-series data of vital signs can dynamically capture these subtle changes. Therefore, the first step is to acquire time-series data of various vital signs closely related to acute shock in the data acquisition module. In the anomaly analysis module, the deviation between the data value at each time point and the preset range is analyzed to obtain the abnormality coefficient of each vital sign parameter at each time point. This quantitative method accurately reflects the degree to which the patient's vital signs parameters deviate from the normal range. Then, the differences in the abnormality coefficients between time points are compared to determine the treatment impact of each vital sign parameter at each time point. This indicator can assess the effect of current treatment measures on the patient's vital signs parameters. The multi-feature fusion module analyzes the temporal distribution and similarity characteristics of the treatment impact between time points and fuses them with the differences in the abnormality coefficients between time points to obtain the significance index of each vital sign parameter at each time point. This multi-feature fusion approach comprehensively considers information from multiple dimensions, including the degree of abnormality of vital signs parameters, treatment impact, and temporal changes, enabling a more comprehensive and accurate assessment of the importance of each vital sign parameter in the occurrence and development of acute shock. Finally, in the weight adjustment and neural network analysis module, the preset importance weights of the vital signs parameters are adjusted based on the significance index corresponding to the time. This enables the system to adjust the weights of different vital signs parameters in real time according to the dynamic changes in the patient's condition, automatically learn the importance of different features, highlight key information, and thus improve the neural network's ability to identify and accurately classify acute shock in the early stage. This can provide medical staff with more accurate suggestions. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, 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.

[0016] Figure 1 This is a system block diagram of an early identification and classification system for acute shock based on multi-parameter fusion, provided in one embodiment of the present invention. Figure 2 This is a flowchart of a method for obtaining a significance index according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the system structure of an early identification and classification system for acute shock based on multi-parameter fusion, provided in one embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent monitoring system and method for building suspended platforms proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details a specific scheme for an early identification and classification system for acute shock based on multi-parameter fusion provided by this invention.

[0020] Please see Figure 1 The diagram shows a system block diagram of an early identification and classification system for acute shock based on multi-parameter fusion according to an embodiment of the present invention. The system includes: a data acquisition module 101, an anomaly analysis module 102, a multi-feature fusion module 103, and a weight adjustment and neural network analysis module 104.

[0021] The data acquisition module 101 is used to acquire time-series data of various vital signs parameters of patients.

[0022] The data acquisition module is the data source of the entire system. It is used to collect time-series data of various vital signs closely related to acute shock in patients. These time-series data of vital signs can dynamically reflect the changes in the physiological state of patients during the occurrence and development of acute shock, and provide key basic information for subsequent abnormal analysis, multi-feature fusion, weight adjustment and neural network analysis.

[0023] In this embodiment of the invention, the types of vital signs parameters include, but are not limited to, blood pressure, heart rate, and central venous pressure related to the cardiovascular system, blood oxygen saturation and arterial blood oxygen partial pressure related to the respiratory system, and blood glucose related to the metabolic system. Then, depending on the type of vital signs parameter to be acquired, appropriate professional medical monitoring equipment is selected. For example, a multi-functional monitor is used to acquire the patient's blood pressure time-series data, heart rate time-series data, and blood oxygen saturation time-series data; an arterial gas and blood pressure analyzer is used to acquire arterial blood oxygen partial pressure time-series data; and a blood glucose meter is used to measure blood glucose time-series data.

[0024] It should be noted that in this embodiment of the present invention, the acquisition frequency of various time series data is set to once every 5 seconds, the length of the time series data is set to 5 minutes, and various time series data need to be acquired synchronously to ensure time alignment; in other embodiments of the present invention, the acquisition frequency and length of the time series data can be adjusted according to the implementation scenario, and are not limited here.

[0025] In this embodiment of the invention, the collection and acquisition of patients' personal information data are all authorized by the relevant users, and the process does not violate relevant laws and regulations, nor does it violate public order and good morals.

[0026] The anomaly analysis module 102 is used to analyze the deviation between the data value at each time point and the preset range in the time series data of each vital sign parameter, to obtain the abnormal performance coefficient of each vital sign parameter at each time point; compare the difference characteristics of the abnormal performance coefficients between time points, and determine the degree of treatment impact of each vital sign parameter at each time point.

[0027] Because early identification of acute shock has a clear real-time requirement, and because it does not exhibit obvious categorization characteristics across multiple types of vital signs, its actual analysis is more characterized by abnormal data changes. Specifically, the development of the body's shock condition leads to significant anomalies in the time-series data of vital signs. Given that the essential differences between different types of shock lie in their different pathophysiological mechanisms and compensatory mechanisms, this results in significant differences in the time-series characteristics of vital signs in clinical manifestations. In other words, the overall abnormal structure of the time-series data of multiple vital signs can indicate the likelihood of early manifestations of a certain type of shock. The structural representation of the time-series data of vital signs depends on the extraction of abnormal features from the data values ​​over time. Therefore, in this embodiment of the invention, an anomaly analysis module is provided to analyze the abnormal data characteristics corresponding to the vital signs.

[0028] First, by analyzing the deviation between the data values ​​of each vital sign parameter under time series data and the preset range, the abnormal performance coefficient of each vital sign parameter at each time point is obtained. This can quickly and accurately identify vital sign parameters that deviate from the normal range, and is used to quantify the degree of abnormality of each vital sign parameter at each time point.

[0029] Preferably, in one embodiment of the present invention, the method for obtaining the abnormal performance coefficient includes: Each vital sign parameter has its specific normal fluctuation range. Therefore, the first step is to determine the preset range corresponding to each vital sign parameter. The preset range provides a clear benchmark for judging whether the vital sign parameter is abnormal. At the same time, the upper and lower boundary values ​​of the preset range are obtained, and the difference between the upper and lower boundary values ​​is used as the range value. The range value reflects the normal fluctuation range of each vital sign parameter, providing a relative scale for the subsequent calculation of the abnormality coefficient, so that the degree of abnormality of different vital sign parameters can be compared within a unified framework.

[0030] Then, in the time-series data of each vital sign parameter, the data value at each time point is compared with the upper and lower boundary values ​​of the corresponding preset range. If the data value is between the upper and lower boundary values ​​of the corresponding preset range, the vital sign parameter is considered to be normal at that time point. If the data value exceeds the corresponding preset range, it indicates that the vital sign parameter is abnormal at that time point. When the data value is greater than the upper boundary value, it indicates that the data value is higher than the normal range. The difference between the data value and the upper boundary value is calculated as a deviation factor. The larger the deviation factor, the greater the degree to which the data value is higher than the normal range, and the more obvious the abnormality. Therefore, the deviation factor is compared with the corresponding range value. The ratio between the two values ​​is used as the abnormality coefficient for each vital sign parameter at each time point. The larger the abnormality coefficient, the higher the degree of abnormality of that vital sign parameter at that time point. When the data value is less than the lower boundary value, it means that the data value is below the normal range. The difference between the lower boundary value and the data value is calculated as the deviation factor. The larger the deviation factor, the greater the degree of abnormality of the data value below the normal range, and the more obvious the abnormality. Therefore, the ratio between the deviation factor and the corresponding range value is used as the abnormality coefficient for each vital sign parameter at each time point. Similarly, the larger the abnormality coefficient, the higher the degree of abnormality of that vital sign parameter at that time point.

[0031] It should be noted that the preset ranges for various vital signs and parameters can be obtained from medical data. For example, the preset range for blood glucose is 2.9 mmol / L to 6.1 mmol / L.

[0032] This yields the abnormality coefficient for each vital sign parameter at each time point. A larger abnormality coefficient at a given time point indicates a significant abnormality in that type of vital sign parameter. However, in actual early shock analysis, not all vital sign parameters exhibit abnormalities; the correspondence between conventional multi-type vital sign parameters is not obvious. Only some abnormalities can be effectively identified. Furthermore, in the actual early shock identification process, specific treatment measures will influence the outcome. For example, in hypovolemic shock, fluid resuscitation is highly effective, leading to a rapid rise in blood pressure. Therefore, from a temporal perspective, vital sign parameter types that change significantly due to treatment measures are more likely to reflect the characteristics of acute shock. The impact of treatment measures can be characterized by the dynamic changes in the abnormality coefficient of vital sign parameters between adjacent time points. Therefore, in this embodiment of the invention, the differences in abnormality coefficients between time points are compared in the temporal data of each vital sign parameter to determine the degree of treatment impact on each vital sign parameter at each time point, which is used to assess the impact of treatment on each vital sign parameter at each time point.

[0033] Preferably, in one embodiment of the present invention, the method for obtaining the therapeutic effect performance includes: In the time-series data of each vital sign parameter, the absolute value of the difference between the abnormal performance coefficient at each time point and the abnormal performance coefficient at the adjacent previous time point is calculated. This absolute value of the difference directly reflects the amount of change in the abnormality of the vital sign parameter at adjacent time points. The larger the value, the larger the variable, and the more obvious the influence of the treatment method. Therefore, the normalized value of this absolute value of the difference is used as the treatment influence performance factor corresponding to each time point. For the first time point, there is no adjacent previous time point. Therefore, in this embodiment of the invention, the treatment influence performance factor of the first time point can be set to be the same as that of the second time point.

[0034] Then, the mean value of the treatment impact factor corresponding to all time points in the time series data of each vital sign parameter is used as the mean characteristic value. The mean characteristic value reflects the average level of the treatment impact factor of each vital sign parameter over the entire time period, providing an overall reference standard. Finally, the ratio of the treatment impact factor to the mean characteristic value at each time point is calculated. The ratio reflects the relative prominence of the local impact of treatment on the vital sign parameter at each time point during the entire monitoring process. The larger the value, the greater the prominence and the higher the importance. The normalized value of this ratio is used as the treatment impact performance degree of each vital sign parameter at each time point. Based on the aforementioned logic, the larger the treatment impact performance degree, the more obvious the response of that vital sign parameter to acute shock treatment is at that time point, and the higher the degree of influence. Normalization is a technique well-known to those skilled in the art. The choice of normalization function can be linear normalization or standard normalization, such as maximum-minimum normalization. The specific normalization method is not limited here.

[0035] The multi-feature fusion module 103 is used to analyze the temporal distribution characteristics and similarity characteristics of the treatment impact performance between time points in the time series data of vital signs parameters, and to fuse them with the difference characteristics of the abnormal performance coefficients between time points, so as to obtain the significance index of each vital sign parameter at each time point.

[0036] Since the treatment impact performance obtained in the anomaly analysis module reflects the abnormal performance of vital signs parameters at a single time point, and in reality, as the patient's physical condition gradually deteriorates, the time series data of vital signs parameters show more obvious abnormal characteristics, the abnormalities of vital signs parameters corresponding to different types of shock will show obvious evolution. Based on this, the significance of each vital sign parameter at each time point can be judged, which can be used to adjust the preset importance weight in the subsequent process, thereby enabling more accurate identification and classification of early acute shock.

[0037] In clinical practice, diseases often cause multiple vital signs to change simultaneously. These changes may be related, and the temporal distribution characteristics of the data can reveal the changes in the data. Through multi-feature fusion, these potential correlations and patterns can be discovered. Therefore, in this module, we analyze the temporal distribution characteristics and similarity characteristics of the treatment impact performance between time points in all vital sign data, and fuse them with the difference characteristics of the abnormal performance coefficients between time points to calculate the significance index of each vital sign parameter at each time point, which is used to characterize its importance in subsequent identification and classification.

[0038] Preferably, in one embodiment of the present invention, the method for obtaining the significance index includes: Please see Figure 2 The diagram illustrates a method flowchart for obtaining a significance index according to an embodiment of the present invention. The method includes the following steps: Step S301: In the time series data of each vital sign parameter, the time points are divided into points of significant change and points of normal performance based on the degree of treatment impact.

[0039] Treatment impact performance reflects the degree to which vital signs are affected by treatment at each time point, that is, the degree of response to treatment. Therefore, in the time series data of each vital sign, the time when the treatment impact performance is greater than the preset performance threshold is taken as the point of significant change, and the time when the treatment impact performance is less than or equal to the preset performance threshold is taken as the point of normal performance.

[0040] It should be noted that in this embodiment of the present invention, the preset performance threshold is set to 0.85. The specific value can be adjusted according to the implementation scenario and is not limited here.

[0041] Step S302: In terms of time series, based on the performance of the therapeutic effect in all types of time series data of vital signs parameters at each time point, determine the sequence of changes corresponding to each time point.

[0042] In terms of time sequence, for any given moment, when that moment is a significant change point in the time sequence data of a certain vital sign parameter, the performance value of that moment in the time sequence data of that vital sign parameter is recorded as a first preset value. In this embodiment of the invention, the first preset value is 1. When that moment is a normal performance point in the time sequence data of a certain vital sign parameter, the performance value of that moment in the time sequence data of that vital sign parameter is recorded as a second preset value. In this embodiment of the invention, the second preset value is 0. The performance values ​​of that moment in the time sequence data of all types of vital sign parameters are sorted according to the preset order of the types of vital sign parameters, thereby obtaining the change performance sequence corresponding to each moment. If the types of vital sign parameters are blood pressure, heart rate, central venous pressure, blood oxygen saturation, arterial blood oxygen partial pressure, and blood glucose, then the preset order in this embodiment of the invention can be arranged according to [blood pressure, heart rate, central venous pressure, blood oxygen saturation, arterial blood oxygen partial pressure, and blood glucose]. At this time, for each moment in the time sequence, a change performance sequence composed of the numbers 1 and 0 can be obtained, and the length of the change performance sequence is equal to the number of types of vital sign parameters.

[0043] Step S303: In the time series data of each vital sign parameter, analyze the similarity characteristics between time intervals to determine the macroscopic difference significance value corresponding to each time interval.

[0044] In the time series data of each vital sign parameter, the cosine similarity between the change sequence at each time point and the adjacent time point is obtained. A higher cosine similarity indicates a more consistent change sequence between the two time points, resulting in a more consistent overall performance across all vital sign parameters and lower discriminative power. Therefore, the cosine similarity is negatively correlated and normalized to correct the logical relationship, obtaining the macroscopic difference significance value for each time point. A higher macroscopic difference significance value indicates a greater discriminative power between that time point and the adjacent time point across all vital sign parameters, thus increasing its reference value for subsequent importance weight adjustment and the identification and classification of acute shock. Since the cosine similarity ranges from -1 to 1, the negative correlation mapping and normalization can be performed using the formula... ,in, Let x represent the normalization function, and let x represent the independent variable. In this embodiment of the invention, the normalization function can be normalized by the maximum and minimum values. The specific method is a well-known technique and will not be described in detail here.

[0045] Step S304: In the time series data of each vital sign parameter, compare the time series distribution characteristics between the time point and the point of significant change, and integrate them with the difference characteristics of the abnormal performance coefficient to determine the significant value of the micro difference corresponding to each time point.

[0046] In the time series data of each vital sign parameter, the absolute value of the difference between the abnormal performance coefficient and the most recent significant change point in the time series is calculated. The significant change point is the moment when the treatment has a significant impact. Therefore, the larger the absolute value of the difference in the abnormal performance coefficient, the more obvious the difference in the degree of data abnormality between that moment and the moment when the treatment has a significant impact. So the value after normalizing the absolute value of the difference is used as the micro-difference significance value corresponding to each moment. The micro-difference significance value focuses on the time series data of a single vital sign parameter and is used to characterize the degree of prominence of each moment. The larger the value, the higher the reference value of that moment for subsequent importance weight adjustment and the identification and classification of acute shock.

[0047] Step S305: In the time series data of each vital sign parameter, combine the macroscopic difference significance value and the microscopic difference significance value corresponding to each time point to obtain the significance index of each vital sign parameter at each time point.

[0048] In the time series data of each vital sign parameter, the product of the macroscopic and microscopic significant differences at each time point is normalized and used as the significance index for each vital sign parameter at each time point. Based on the logical analysis in steps S303 and S304, it can be seen that the larger the significance index, the higher its priority as the primary shock type analysis and the greater its reference value. Normalization is a technique well-known to those skilled in the art. The normalization function can be linear normalization or standard normalization, such as maximum-minimum normalization. The specific normalization method is not limited here.

[0049] The weight adjustment and neural network analysis module 104 is used to perform neural network analysis on all time series data of vital signs parameters after adjusting the preset importance weights of the vital signs parameters based on the significance index corresponding to the time.

[0050] In medical monitoring scenarios, the importance of different vital signs parameters for disease analysis is not fixed, but rather varies. Therefore, in this module, the preset importance weights of vital signs parameters are adaptively adjusted based on the significance indicators corresponding to each time point in the time series data of vital signs parameters. This allows the final importance weights to better adapt to the data characteristics exhibited by the patient's vital signs parameters, thereby helping to build a more accurate neural network for subsequent analysis and judgment.

[0051] Preferably, in one embodiment of the present invention, after adjusting the preset importance weights of the vital signs parameters based on the significance index corresponding to the time in the time-series data of vital signs parameters, neural network analysis is performed on all time-series data of vital signs parameters, including: Based on the proportion of the significance index of each vital sign parameter in the overall significance index of all types of vital sign parameters at each time point, the adjustment factor of each vital sign parameter at each time point is determined: the sum of the significance indices of all types of vital sign parameters at all time points is used as the significance and value characteristic values. In the time series data of each vital sign parameter, the ratio of the significance index corresponding to each time point to the significance and value characteristic value is used as the adjustment factor of each vital sign parameter at each time point. By calculating the proportion, the importance of each vital sign parameter relative to other vital sign parameters at each time point can be clearly identified. The larger the adjustment factor, the greater the importance, that is, the more critical that vital sign parameter is for the final identification and classification at that time point.

[0052] Different adjustment factors contribute differently to the final identification and classification. Therefore, the sum of the adjustment factor and the preset importance weight for each vital sign parameter at each time step is normalized and used as the final importance weight for each vital sign parameter at each time step. Based on the aforementioned analysis, a larger final importance weight is considered to represent more critical data information. Normalization is a technique well-known to those skilled in the art. The normalization function can be linear normalization, standard normalization, or, for example, maximum-minimum normalization. The specific normalization method is not limited here.

[0053] Next, the final importance weights of all types of vital signs parameters at all times are used to form an importance weight matrix based on the time series and the preset order of the types of vital signs parameters. The rows and columns of the importance weight matrix can represent the time series and the types of vital signs parameters, respectively. Specifically, if there are 3 types of vital signs parameters and the length of the time series data is also 3, then the size of this importance weight matrix is ​​3*3.

[0054] Finally, the importance weight matrix and time series data of all types of vital signs parameters are used as input to the neural network for neural network analysis to obtain the possible types of acute shock in patients, thereby assisting doctors in further identification and classification.

[0055] In this embodiment of the invention, the neural network may be an RNN.

[0056] It should be noted that the preset importance weight of the vital signs parameter at each time point in this embodiment of the present invention is 1, and the specific value can also be obtained based on the neural network-related information in the implementation scenario.

[0057] In summary, during the development of acute shock, the patient's condition may rapidly deteriorate or fluctuate. Time-series data of vital signs can dynamically capture these subtle changes. Therefore, the first step is to acquire time-series data of various vital signs closely related to acute shock in the data acquisition module. In the anomaly analysis module, by analyzing the deviation between the data values ​​at each time point and the preset range, the abnormality coefficient of each vital sign parameter at each time point is obtained. This quantitative method can accurately reflect the degree to which the patient's vital signs deviate from the normal range. Then, by comparing the differences in the abnormality coefficients between time points, the treatment impact of each vital sign parameter at each time point is determined. This indicator can assess the effect of current treatment measures on the patient's vital signs. The multi-feature fusion module analyzes the temporal distribution and similarity characteristics of the treatment impact between time points and fuses them with the differences in the abnormality coefficients between time points to obtain the significance index of each vital sign parameter at each time point. This multi-feature fusion approach comprehensively considers information from multiple dimensions, including the degree of abnormality of vital signs, treatment impact, and temporal changes, enabling a more comprehensive and accurate assessment of the importance of each vital sign parameter in the occurrence and development of acute shock. Finally, in the weight adjustment and neural network analysis module, the preset importance weights of the vital signs parameters are adjusted based on the significance index corresponding to the time. This enables the system to adjust the weights of different vital signs parameters in real time according to the dynamic changes in the patient's condition, automatically learn the importance of different features, highlight key information, and thus improve the neural network's ability to identify and accurately classify acute shock in the early stage. This can provide medical staff with more accurate suggestions.

[0058] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above.

[0059] Please see Figure 3This illustration shows a schematic diagram of the system structure of an early acute shock identification and classification system based on multi-parameter fusion according to an embodiment of the present invention. It includes a processor 300, a memory 301, a bus 302, and a communication interface 303. The processor 300, communication interface 303, and memory 301 are connected via the bus 302. The memory 301 may include a high-speed random access memory, and the bus 302 may be an ISA bus, PCI bus, or EISA bus, etc. The processor 300 may be an integrated circuit chip with signal processing capabilities. The memory 301 stores at least one instruction, at least one program, code set, or instruction set. When the processor loads and executes the at least one instruction, at least one program, code set, or instruction set, it implements the steps of each module in the early acute shock identification and classification system based on multi-parameter fusion.

[0060] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0061] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0062] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A system for early identification and classification of acute shock based on multi-parameter fusion, characterized in that, The system includes: The data acquisition module is used to acquire time-series data of various vital signs parameters of patients; The anomaly analysis module is used to analyze the deviation between the data value at each time point and the preset range in the time series data of each vital sign parameter, to obtain the abnormal performance coefficient of each vital sign parameter at each time point; compare the differences in the abnormal performance coefficients between time points to determine the degree of treatment impact of each vital sign parameter at each time point; The multi-feature fusion module is used to analyze the temporal distribution characteristics and similarity characteristics of the treatment impact performance between time points in the time series data of vital signs parameters, and to fuse them with the difference characteristics of the abnormal performance coefficients between time points, so as to obtain the significance index of each vital sign parameter at each time point. The weight adjustment and neural network analysis module is used to adjust the preset importance weights of the vital signs parameters based on the significance index corresponding to the time in the time series data of vital signs parameters, and then perform neural network analysis on all time series data of vital signs parameters. The neural network is an RNN.

2. The acute shock early identification and classification system based on multi-parameter fusion according to claim 1, characterized in that, The method for obtaining the abnormal behavior coefficient includes: Determine the preset range corresponding to each vital sign parameter and obtain the upper and lower boundary values ​​of the preset range, and use the difference between the upper and lower boundary values ​​as the range value; In the time-series data of each vital sign parameter, the data value at each time point is compared with the corresponding upper and lower boundary values ​​of a preset range: When the data value is greater than the upper boundary value, the difference between the data value and the upper boundary value is calculated as the deviation factor. The ratio between the deviation factor and the corresponding range value is used as the abnormal performance coefficient of each vital sign parameter at each time. When the data value is less than the lower boundary value, the difference between the lower boundary value and the data value is calculated as a deviation factor. The ratio between the deviation factor and the corresponding range value is used as the abnormal performance coefficient of each vital sign parameter at each time point.

3. The acute shock early identification and classification system based on multi-parameter fusion according to claim 1, characterized in that, The methods for obtaining the treatment effect performance include: In the time series data of each vital sign parameter, the normalized value of the absolute difference between the abnormal performance coefficient at each time point and the abnormal performance coefficient at the adjacent previous time point is used as the treatment impact performance factor corresponding to each time point. The mean value of the treatment impact performance factor corresponding to all time points in the time series data of each vital sign parameter is used as the mean feature value. The normalized value of the ratio of the treatment impact performance factor corresponding to each time point to the mean feature value is used as the treatment impact performance degree of each vital sign parameter at each time point.

4. The acute shock early identification and classification system based on multi-parameter fusion according to claim 1, characterized in that, The methods for obtaining the significance index include: In the time series data of each vital sign parameter, the moment when the treatment effect performance is greater than the preset performance threshold is taken as the significant change point, and the moment when the treatment effect performance is less than or equal to the preset performance threshold is taken as the normal performance point. In terms of time series, based on the performance of the therapeutic impact on all types of vital signs parameters in time series data at each time point, the corresponding change sequence at each time point is determined; In the time series data of each vital sign parameter, the similarity characteristics between time points are analyzed to determine the macroscopic difference significance value corresponding to each time point; In the time series data of each vital sign parameter, the temporal distribution characteristics between the time point and the point of significant change are compared and integrated with the difference characteristics of the abnormal performance coefficient to determine the significant value of the micro difference corresponding to each time point; In the time series data of each vital sign parameter, the normalized value of the product of the macroscopic difference significance value and the microscopic difference significance value at each time point is used as the significance index of each vital sign parameter at each time point.

5. The acute shock early identification and classification system based on multi-parameter fusion according to claim 4, characterized in that, The method for obtaining the change sequence includes: In terms of time series, for any given moment, when that moment is a significant change point in the time series data of a certain vital sign parameter, the performance value of that moment in the time series data of that vital sign parameter is recorded as the first preset value; when that moment is a normal performance point in the time series data of a certain vital sign parameter, the performance value of that moment in the time series data of that vital sign parameter is recorded as the second preset value. The performance values ​​of that moment in the time series data of all types of vital sign parameters are sorted according to the preset order of the types of vital sign parameters, thereby obtaining the change performance sequence corresponding to each moment.

6. The acute shock early identification and classification system based on multi-parameter fusion according to claim 4, characterized in that, The methods for obtaining the macroscopic difference significance values ​​include: In the time series data of each vital sign parameter, the cosine similarity between the change sequence at each time point and the adjacent previous time point is negatively correlated and normalized, and the result is used as the macroscopic difference significance value corresponding to each time point.

7. The acute shock early identification and classification system based on multi-parameter fusion according to claim 4, characterized in that, The method for obtaining the significance value of the microscopic differences includes: In the time series data of each vital sign parameter, the normalized value of the absolute difference between the abnormal performance coefficients at each time point and the nearest significant change point in the time series is used as the significant value of the micro-difference corresponding to each time point.

8. The acute shock early identification and classification system based on multi-parameter fusion according to claim 1, characterized in that, In the time-series data of vital signs, after adjusting the preset importance weights of the vital signs parameters based on the significance index corresponding to the time, neural network analysis is performed on all time-series data of vital signs parameters, including: Based on the proportion of the significance index of each vital sign parameter in the overall significance index of all types of vital sign parameters at each time point, the adjustment factor of each vital sign parameter at each time point is determined. The normalized sum of the adjustment factor and the preset importance weight for each vital sign parameter at each time point is used as the final importance weight. The final importance weights of all types of vital signs parameters at all times are combined into an importance weight matrix based on the time series and the preset order of the types of vital signs parameters. The importance weight matrix and time series data of all types of vital signs are used as inputs to the neural network for neural network analysis.

9. The acute shock early identification and classification system based on multi-parameter fusion according to claim 8, characterized in that, The method for obtaining the regulation factor includes: The sum of the significance indices of all time points across all types of vital signs is used as the significance and value characteristic values. In the time series data of each vital sign parameter, the ratio of the significance index corresponding to each time point to the significance and value characteristic value is used as the adjustment factor for each vital sign parameter at each time point.

Images