Intensive care data processing method and device

Abstract

The invention discloses an intensive care data processing method and device, and relates to the field of intensive care data processing, and the method comprises the steps: S1, monitoring the real-time performance of each piece of intensive care equipment, and obtaining the performance difference correction value of each piece of intensive care equipment; s2, analyzing and processing to obtain a complexity evaluation value of each real-time intensive care data stream; and S3, based on the processing difficulty level of each real-time intensive care data stream, sending each real-time intensive care data stream to a data processing module under the corresponding processing difficulty level, performing corresponding processing, and sending the processed data stream to a management terminal. According to the intensive care data processing method provided by the invention, the data is corrected through the performance difference correction value, so that the accuracy and reliability of the data are ensured. Through the complexity evaluation value, a basis is provided for classification and processing of the data streams, a modular processing mode is realized, and the processing efficiency is improved. The pertinence and efficiency of data processing are improved, and the processing time and resource waste are reduced.

Description

Technical Field

[0001] The present invention relates to the field of intensive care data processing, and specifically to a method and device for intensive care data processing. Background Art

[0002] Monitoring devices in the ICU generate a large amount of data, including physiological parameters, laboratory test results, drug usage records, etc. Traditional manual processing methods are difficult to effectively manage this data, while advanced data processing methods can efficiently analyze and utilize this information. Through automated data processing methods, the workload of medical staff can be reduced, and through automated data monitoring and analysis, human errors can be reduced and medical safety can be improved. The research and application of intensive care data processing methods and devices are an inevitable requirement for the development of modern medicine, and are of great significance for improving the level of medical services, ensuring patient safety, and promoting the progress of medical technology.

[0003] The prior art, such as the invention patent with the publication number: CN117238507B, is an intensive care monitoring system based on the Internet of Things. The present invention obtains the physical sign stability evaluation of each dimension at the current moment by according to the change and distribution characteristics of the physiological monitoring data of each dimension in the system state vectors at all other moments within the historical range of the current moment; obtains the change weight value of the description element corresponding to each dimension at the current moment according to the difference of the description elements within the local data fluctuation range of each dimension at the current moment and the physical sign stability evaluation of each observation value sequence; and then adjusts the description elements to obtain the weighted sigma point set at the current moment; obtains the predicted state vector at the next moment, and monitors the abnormal state of the patient.

[0004] The prior art, such as the invention patent with the publication number: CN118098603B, is an intensive care clinical information system and method for eICU, which relates to the field of intelligent detection. It collects the vital sign data of patients in real-time monitoring, such as heart rate, blood pressure, and body temperature data, and introduces data processing and analysis algorithms at the backend to perform time-series collaborative analysis on these vital sign data, so as to automatically monitor whether there are abnormalities in the signs of critically ill patients.

[0005] Based on the above solutions, it can be seen that the prior art in the field of intensive care data processing usually focuses on directly processing and analyzing intensive care data itself. However, in practical applications, when obtaining intensive care data, the data quality will be affected by various factors, such as the performance of intensive care devices. At the same time, the transmission and processing of intensive care data are usually relatively complex, and the degree of demand for data processing often varies according to the actual situation. At present, only directly processing the data itself may reduce the accuracy of the data processing results due to the large data processing difficulty and low adaptability of the processing method. Summary of the Invention

[0006] In view of the deficiencies of the prior art, the present invention provides a method and device for processing intensive care data. To achieve the above objectives, the present invention is implemented through the following technical solutions: A method and device for processing intensive care data, including:

[0007] S1. Monitor the real-time performance of each intensive care device, obtain the real-time performance evaluation parameters of each intensive care device, analyze and process to obtain the real-time performance evaluation values of each intensive care device, and match to obtain the performance difference correction values of each intensive care device.

[0008] S2. Each intensive care device transmits each real-time intensive care data stream to the data processing center. After correction based on the performance difference correction values of each intensive care device, obtain the basic information of each real-time intensive care data stream, and analyze and process to obtain the complexity evaluation value of each real-time intensive care data stream.

[0009] S3. Based on the complexity evaluation values of each real-time intensive care data stream, match to obtain the processing difficulty levels of each real-time intensive care data stream. Based on the processing difficulty levels of each real-time intensive care data stream, send each real-time intensive care data stream to the data processing module under the corresponding processing difficulty level, and after corresponding processing, send it to the management terminal.

[0010] As a preferred technical solution, the process of monitoring the real-time performance of each intensive care device, obtaining the real-time performance evaluation parameters of each intensive care device, and analyzing and processing to obtain the real-time performance evaluation values of each intensive care device is as follows:

[0011] Obtain the real-time performance evaluation parameters of each intensive care device.

[0012] The real-time performance evaluation parameters of each intensive care device include the network real-time delay, device real-time load, CPU real-time usage rate, disk real-time read and write speed, and network real-time traffic of each intensive care device.

[0013] Extract the real-time performance evaluation verification parameters of each intensive care device from the database, including the real-time delay verification value, device real-time load verification value, CPU real-time usage rate verification value, disk real-time read and write speed verification value, and network real-time traffic verification value of each intensive care device.

[0014] Compare and analyze the real-time performance evaluation parameters of each intensive care device with the real-time performance evaluation verification parameters of each intensive care device to obtain the real-time performance evaluation values of each intensive care device.

[0015] The real-time performance evaluation values of each intensive care device are used to characterize the real-time performance of each intensive care device.

[0016] As a preferred technical solution, the process of obtaining the performance difference correction values for each intensive care device is as follows:

[0017] Based on the real-time performance evaluation values of each intensive care device, map and match them with the performance difference correction values corresponding to the pre-stored real-time performance evaluation value intervals in the database to obtain the performance difference correction values for each intensive care device.

[0018] The performance difference correction values for each intensive care device are used to correct the impact of the performance differences of intensive care devices on the real-time intensive care data stream.

[0019] As a preferred technical solution, after correction based on the performance difference correction values of each intensive care device, the basic information of each real-time intensive care data stream is obtained, specifically including:

[0020] Based on the performance difference correction values of each intensive care device, and the real-time data stream quality correction parameters corresponding to the pre-stored performance difference correction values in the database, including filter cut-off frequency, filter passband ripple, and filter stopband attenuation.

[0021] The basic information of each real-time intensive care data stream includes the number of data types, the proportion of unstructured data, the total data volume, and the data noise ratio of each real-time intensive care data stream.

[0022] As a preferred technical solution, the complexity evaluation values of each real-time intensive care data stream are obtained through analysis and processing, specifically including:

[0023] Perform average value processing on the basic information of each real-time intensive care data stream to obtain the average value set of the basic information of the real-time intensive care data stream, including the average number of data types, the average proportion of unstructured data, the average total data volume, and the average data noise ratio.

[0024] Comprehensively analyze and process the basic information of each real-time intensive care data stream and the average value set of the basic information to obtain the complexity evaluation values of each real-time intensive care data stream.

[0025] The complexity evaluation values of each real-time intensive care data stream are used to characterize the complexity of each real-time intensive care data stream.

[0026] As a preferred technical solution, the process of obtaining the processing difficulty levels of each real-time intensive care data stream is as follows:

[0027] The processing difficulty levels include primary processing difficulty level, intermediate processing difficulty level, and advanced processing difficulty level.

[0028] Based on the complexity evaluation value of a certain real-time intensive care data stream, map and match it with the processing difficulty levels corresponding to the pre-stored complexity evaluation values in the database to obtain the processing difficulty level of the real-time intensive care data stream.

[0029] As an optimal technical solution, based on the processing difficulty levels of each real-time intensive care data stream, send each real-time intensive care data stream to the data processing module corresponding to the corresponding processing difficulty level, and after corresponding processing, send it to the management terminal. Specifically, it includes:

[0030] Based on the processing difficulty levels of each real-time intensive care data stream, the data processing center sends each real-time intensive care data stream to the primary data processing module, the intermediate data processing module, and the advanced data processing module respectively.

[0031] The primary data processing module is used to process each real-time intensive care data stream with a primary processing difficulty level, and this data processing module is used to process light tasks.

[0032] The intermediate data processing module is used to process each real-time intensive care data stream with an intermediate processing difficulty level, and this data processing module is used to process moderate tasks.

[0033] The advanced data processing module is used to process each real-time intensive care data stream with an advanced processing difficulty level, and this data processing module is used to process heavy tasks.

[0034] In each level of data processing module, feature extraction is performed on each real-time intensive care data stream respectively, and through analysis and processing, each data feature of each real-time intensive care data stream is obtained.

[0035] Integrate each data feature of each real-time intensive care data stream as each real-time intensive care data feature, send each real-time intensive care data feature back to the data processing center, process to obtain the variance explanation rate of each data feature, sort according to the size of the variance explanation rate of each real-time intensive care data feature, and send it to the management terminal for prompting according to the sorting.

[0036] As an optimal technical solution, it also includes monitoring the real-time processing performance of each level of data processing module, and analyzing and processing to obtain the real-time overload evaluation index of each level of data processing module. The specific processing conditions are:

[0037] Monitor the real-time processing performance of each level of data processing module, and obtain the real-time processing performance parameters of each level of data processing module, including the real-time memory usage rate, the real-time CPU usage rate, and the real-time processing queue length of each level of data processing module.

[0038] Extract the memory real-time utilization overload threshold, CPU real-time utilization overload threshold, and real-time processing queue length overload threshold of each level of data processing module from the database, and compare and process them with the real-time processing performance parameters of each level of data processing module to obtain the real-time overload evaluation index of each level of data processing module.

[0039] As a preferred technical solution, the matching to obtain the GPU real-time acceleration multiple of each level of data processing module specifically includes: based on the real-time overload evaluation index of each level of data processing module, perform mapping matching with the GPU real-time acceleration multiple corresponding to each real-time overload evaluation index in the database to obtain the GPU real-time acceleration multiple of each level of data processing module.

[0040] The GPU real-time acceleration multiple is used to improve the processing speed of each level of data processing module and prevent the data processing module from being overloaded.

[0041] A device is also provided, including: the device has one or more programs, and the one or more programs are executed by one or more processors.

[0042] Compared with the prior art, the embodiments of the present invention have at least the following beneficial effects:

[0043] (1) The present invention provides a method for processing intensive care data. By real-time monitoring the performance of intensive care equipment and obtaining evaluation parameters, the data is corrected with the obtained performance difference correction value to ensure the accuracy and reliability of the data. The quality of the data is improved, providing a solid foundation for subsequent analysis and processing.

[0044] (2) By analyzing and processing the basic information of the real-time intensive care data stream, the present invention obtains a complexity evaluation value, which can accurately reflect the characteristics of the data stream and the difficulty of processing. It provides a basis for the classification and processing of the data stream, helping to achieve reasonable resource allocation. Based on the complexity evaluation value, the processing difficulty level is matched, and the data stream is sent to the corresponding level of data processing module, realizing a modular processing method. Each module can be specifically optimized according to its processing difficulty level to improve the processing efficiency. The pertinence and efficiency of data processing are improved, reducing processing time and resource waste.

[0045] (3) By monitoring the real-time performance parameters of data processing modules at each level in real time, the real-time processing performance of each module can be accurately grasped. The real-time overload evaluation index calculated based on these performance parameters can reflect the load condition of the module. The real-time monitoring and dynamic adjustment of the performance of the data processing module are realized. By matching the real-time GPU acceleration multiple according to the real-time overload evaluation index, the GPU resources can be dynamically allocated according to the load conditions of different modules, realizing the optimal utilization of resources. Through GPU acceleration, the processing speed of the data processing module can be significantly improved, the efficiency of data processing is increased, and the data processing delay is reduced.

[0046] Of course, it is not necessary for any product implementing the present invention to achieve all the above-mentioned advantages simultaneously. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0049] In the description of the present invention, it should be understood that the terms "opening", "upper", "lower", "thickness", "top", "middle", "length", "inner", "periphery", etc. indicating the orientation or positional relationship are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the components or elements referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

[0050] Please refer to Figure 1 As shown, an intensive care data processing method provided by an embodiment of the present invention includes:

[0051] S1. Monitor the real-time performance of each intensive care device, obtain the real-time performance evaluation parameters of each intensive care device, analyze and process to obtain the real-time performance evaluation values of each intensive care device, and match to obtain the performance difference correction values of each intensive care device.

[0052] The specific process of monitoring the real-time performance of each intensive care device, obtaining the real-time performance evaluation parameters of each intensive care device, and analyzing and processing to obtain the real-time performance evaluation values of each intensive care device is as follows:

[0053] Deploy real-time performance monitoring software tools at the integrated interfaces of each intensive care device to obtain real-time performance evaluation parameters of each intensive care device.

[0054] The real-time performance evaluation parameters of each intensive care device include the network real-time latency, device real-time load, CPU real-time usage rate, disk real-time read / write speed, and network real-time traffic of each intensive care device.

[0055] It should be noted that network real-time latency refers to the time required for data to be transmitted in the network from the sending end to the receiving end. It is used to reflect the network performance. The lower the latency, the faster the network response speed.

[0056] Device real-time load refers to the real-time workload or task volume borne by the device. It reflects the busyness of the device. The higher the load, the closer the device's task processing ability is to its limit.

[0057] CPU real-time usage rate refers to the percentage of CPU resources occupied by the tasks being processed by the CPU. It directly reflects the busyness of the CPU. The higher the usage rate, the more saturated the CPU's task processing ability.

[0058] Disk real-time read / write speed refers to the speed at which the disk performs data read / write operations, usually expressed in the amount of data read / written per second (such as MB / s). It reflects the performance of the disk. The faster the read / write speed, the higher the data access efficiency.

[0059] Network real-time traffic refers to the amount of data transmitted over the network, usually expressed in the number of bits transmitted per second (such as bps, Kbps, Mbps). It reflects the utilization of network bandwidth. The higher the traffic, the more fully the network bandwidth is occupied.

[0060] Extract the real-time performance evaluation verification parameters of each intensive care device from the database, including the real-time latency verification value, device real-time load verification value, CPU real-time usage rate verification value, disk real-time read / write speed verification value, and network real-time traffic verification value of each intensive care device.

[0061] Compare and analyze the real-time performance evaluation parameters of each intensive care device with the real-time performance evaluation verification parameters of each intensive care device to obtain the real-time performance evaluation value of each intensive care device, specifically including:

[0062]

[0063] Among them, Ma a is the real-time performance evaluation value of the a-th intensive care device, yc a is the network real-time latency of the a-th intensive care device, fz a is the device real-time load of the a-th intensive care device, is the CPU real-time usage rate of the a-th intensive care device, va is the real-time disk read / write speed of the ath intensive care device, l a is the real-time network traffic of the ath intensive care device, yc a0 is the real-time latency check value of the ath intensive care device, fz a0 is the device real-time load check value of the ath intensive care device, is the real-time CPU utilization check value of the ath intensive care device, v a0 is the real-time disk read / write speed check value of the ath intensive care device, l a0 is the real-time network traffic check value of the ath intensive care device, α1 is the network real-time latency weight factor, α2 is the device real-time load weight factor, α3 is the real-time CPU utilization weight factor, α4 is the real-time disk read / write speed weight factor, α5 is the real-time network traffic weight factor, a is the intensive care device number, a = 1, 2, 3,..., S, and S is the total number of intensive care devices.

[0064] It should be noted that the value ranges of the network real-time delay weight factor, device real-time load weight factor, CPU real-time utilization rate weight factor, disk real-time read / write speed weight factor, and network real-time traffic weight factor are all between 0 and 1, and satisfy α1 + α2 + α3 + α4 + α5 = 1. The network real-time delay weight factor is an influencing factor for the real-time performance evaluation value of each intensive care device preset in the database, indicating the influence degree of network real-time delay on the real-time performance evaluation value of each intensive care device. The device real-time load weight factor is an influencing factor for the real-time performance evaluation value of each intensive care device preset in the database, indicating the influence degree of device real-time load on the real-time performance evaluation value of each intensive care device. The CPU real-time utilization rate weight factor is an influencing factor for the real-time performance evaluation value of each intensive care device preset in the database, indicating the influence degree of CPU real-time utilization rate on the real-time performance evaluation value of each intensive care device. The disk real-time read / write speed weight factor is an influencing factor for the real-time performance evaluation value of each intensive care device preset in the database, indicating the influence degree of disk real-time read / write speed on the real-time performance evaluation value of each intensive care device. The disk real-time read / write speed weight factor is an influencing factor for the real-time performance evaluation value of each intensive care device preset in the database, indicating the influence degree of disk real-time read / write speed on the real-time performance evaluation value of each intensive care device. When in use, the network real-time delay weight factor, device real-time load weight factor, CPU real-time utilization rate weight factor, disk real-time read / write speed weight factor, and network real-time traffic weight factor can be directly obtained from the database, and the obtaining method is a preset mapping relationship. For example, the basic information of each intelligent vehicle involved in the embodiments of the present invention, including the network real-time delay, device real-time load, CPU real-time utilization rate, disk real-time read / write speed, and network real-time traffic of each intensive care device, is input into the preset mapping set in the database for mapping and matching to obtain the network real-time delay weight factor, device real-time load weight factor, CPU real-time utilization rate weight factor, disk real-time read / write speed weight factor, and network real-time traffic weight factor involved in the embodiments of the present invention, and the mapping relationship therein is one-to-one correspondence.

[0065] It should also be noted that there is a certain correlation between the network real-time delay, device real-time load, CPU real-time usage, disk real-time read and write speed, and network real-time traffic of each critical care device. When the network real-time traffic increases, if the network bandwidth is limited, it may cause network congestion, thereby increasing the network real-time delay. An increase in the device real-time load usually leads to an increase in CPU usage because more tasks require the CPU to process. A high load may mean more data read and write operations, thereby increasing the disk read and write speed. A high CPU usage rate may mean that a large amount of data that needs to be read and written to the disk is being processed, which affects the disk read and write speed. Conversely, a slow disk read and write speed may also cause the CPU to wait for data, thereby reducing the CPU usage rate. A large network traffic may mean that there is a large amount of data that needs to be stored or retrieved, thereby increasing the disk read and write operations. An abnormality in any one parameter may cause changes in other parameters. For example, an increase in network delay may cause an increase in device load, which in turn increases CPU usage, which may ultimately affect the disk read and write speed.

[0066] The real-time performance evaluation value of each intensive care device is used to characterize the real-time performance of each intensive care device.

[0067] The matching obtains the performance difference correction value of each intensive care device, and the specific process is:

[0068] Based on the real-time performance evaluation value of each critical care device, a mapping match is performed with the performance difference correction value corresponding to each real-time performance evaluation value interval pre-stored in the database to obtain the performance difference correction value of each critical care device. The performance difference correction value of each critical care device is used to correct the impact of the performance difference of the critical care device on the real-time critical care data stream.

[0069] S2. Each intensive care device transmits each real-time intensive care data stream to the data processing center. After correction based on the performance difference correction value of each intensive care device, the basic information of each real-time intensive care data stream is obtained, and the complexity evaluation value of each real-time intensive care data stream is obtained through analysis and processing.

[0070] After the correction is performed based on the performance difference correction value of each intensive care device, basic information of each real-time intensive care data stream is obtained, specifically including:

[0071] Based on the performance difference correction value of each intensive care device, mapping and matching are performed with the real-time data stream quality correction parameters corresponding to each performance difference correction value pre-stored in the database to obtain the real-time data stream quality correction parameters of each intensive care device, including the filter cutoff frequency, filter passband ripple and filter stopband attenuation of each intensive care device.

[0072] It should be noted that the filter cut-off frequency, filter passband ripple, and filter stopband attenuation are three key setting parameters of the filter, which jointly determine the performance and characteristics of the filter.

[0073] The filter cut-off frequency refers to the frequency point at which the filter transitions from the passband to the stopband. At this point, the output power of the filter drops to a certain proportion of the maximum output power (usually -3dB, that is, the power drops to half). The cut-off frequency is used to distinguish the signal components to be retained (within the passband) and the interference components to be filtered out (within the stopband). By setting the cut-off frequency, the bandwidth of the filter can be controlled, thereby affecting the signal passing range.

[0074] The filter passband ripple refers to the degree of fluctuation of the output signal amplitude relative to the ideal amplitude within the filter passband. It is usually expressed in decibels (dB). The passband ripple reflects the distortion degree of the signal amplitude within the passband by the filter. The smaller the ripple, the smaller the signal distortion and the better the filter performance.

[0075] The filter stopband attenuation refers to the suppression ability of the filter for unwanted frequency components within the stopband, usually expressed in decibels (dB). The greater the attenuation, the stronger the suppression ability of the filter for interference signals. The stopband attenuation is used to determine the suppression effect of the filter on interference signals. The greater the attenuation, the more thoroughly the interference signals are filtered out.

[0076] The basic information of each real-time intensive care data stream includes the number of data types, the proportion of unstructured data, the total data volume, and the data noise ratio of each real-time intensive care data stream, which are obtained through the data acquisition identifier.

[0077] It should be noted that the number of data types refers to the number of different data types included in the real-time intensive care data stream. For example, electrocardiogram data, blood pressure data, respiratory rate data, etc. It reflects the complexity and diversity of the data stream and helps to understand the difficulty of data integration and analysis.

[0078] The proportion of unstructured data refers to the proportion of unstructured data (such as text, images, videos, etc.) in the real-time intensive care data stream. Unstructured data is usually more difficult to analyze and process, and a high proportion means that more complex analysis tools and methods are required.

[0079] The total data volume refers to the total data volume of the real-time intensive care data stream, usually expressed in bytes, megabytes or larger units. It reflects the scale of the data stream and helps to evaluate the resources required for data processing and analysis.

[0080] The data noise ratio refers to the proportion of noise data (i.e., incorrect, incomplete or irrelevant data) in the real-time intensive care data stream. It reflects the overall quality of the data stream, and a high noise ratio means a large amount of work for data cleaning and preprocessing.

[0081] The above analysis and processing obtain the complexity evaluation values of each real-time intensive care data stream, specifically including:

[0082] The basic information of each real-time intensive care data stream is averaged to obtain a set of average values of the basic information of the real-time intensive care data stream, including the average number of data types, the average proportion of unstructured data, the average total amount of data volume, and the average proportion of data noise.

[0083] The basic information of each real-time intensive care data stream and the set of average values of the basic information are comprehensively analyzed and processed to obtain the complexity evaluation value of each real-time intensive care data stream.

[0084] The above analysis and processing obtain the complexity evaluation values of each real-time intensive care data stream, and the specific processing conditions are:

[0085]

[0086] Among them, SJ i is the complexity evaluation value of the i-th real-time intensive care data stream, k i is the evaluation value of the number of data types of the i-th real-time intensive care data stream, g i is the proportion of unstructured data in the number of data types of the i-th real-time intensive care data stream, q i is the total amount evaluation value of the data volume of the number of data types of the i-th real-time intensive care data stream, z i is the data noise ratio evaluation value of the number of data types of the i-th real-time intensive care data stream, i is the number of the real-time intensive care data stream, i = 1, 2, 3,..., n, n is the total number of real-time intensive care data streams, kc i is the number of data types of the i-th real-time intensive care data stream, gc i is the proportion of unstructured data of the i-th real-time intensive care data stream, qc i is the total amount of data volume of the i-th real-time intensive care data stream, zc i is the data noise ratio of the i-th real-time intensive care data stream, is the average number of data types, is the average proportion of unstructured data, is the average total amount of data volume, is the average proportion of data noise, χ1 is the complexity evaluation value corresponding to the primary processing difficulty level, χ2 is the complexity evaluation value corresponding to the intermediate processing difficulty level, χ3 is the complexity evaluation value corresponding to the advanced processing difficulty level, and ψ is the data stream complexity threshold.

[0087] It should be noted that there is a certain correlation among the basic information of each real-time intensive care data stream, including the number of data types, the proportion of unstructured data, the total data volume, and the data noise ratio of each real-time intensive care data stream. The larger the number of data types, usually the more types of information need to be collected and processed, which may lead to an increase in the total data volume. Different types of data may be generated at different rates, so the increase in the number of data types may bring about a non-linear growth in the total data volume. Unstructured data (such as text, images, videos) usually occupies more storage space than structured data (such as numerical values, time series data). If the proportion of unstructured data is high, then the total overall data volume may increase significantly. Different types of data may have different noise characteristics. For example, physiological signal data may be subject to electromagnetic interference, while text data may contain spelling mistakes. The increase in the number of data types may lead to a change in the overall data noise ratio, depending on the noise level of each data type. Unstructured data is often more difficult to standardize and process, so it may contain a higher noise ratio. The increase in the proportion of unstructured data may increase the overall data noise ratio.

[0088] It should also be noted that the data stream complexity threshold is preset by the data processing center in the embodiments of the present invention. There are various methods for setting the data stream complexity threshold. For example, it can be obtained through statistical analysis or experts can determine the boundary between high complexity and low complexity through corresponding historical data to set and adjust the threshold.

[0089] The complexity evaluation value of each real-time intensive care data stream is used to characterize the complexity of each real-time intensive care data stream.

[0090] S3. Based on the complexity evaluation values of each real-time intensive care data stream, match and obtain the processing difficulty levels of each real-time intensive care data stream. Based on the processing difficulty levels of each real-time intensive care data stream, send each real-time intensive care data stream to the data processing module corresponding to the processing difficulty level, and after corresponding processing, send it to the management terminal.

[0091] Obtaining the processing difficulty levels of each real-time intensive care data stream specifically includes:

[0092] Based on the complexity evaluation value of a certain real-time intensive care data stream, perform mapping and matching with the processing difficulty levels corresponding to each complexity evaluation value pre-stored in the database to obtain the processing difficulty level of this real-time intensive care data stream.

[0093] The processing difficulty levels include the primary processing difficulty level, the intermediate processing difficulty level, and the advanced processing difficulty level.

[0094] When the complexity evaluation value of a certain real-time intensive care data stream is χ1, the corresponding processing difficulty level matched is the primary processing difficulty level, which is the lowest level among the processing difficulty levels. This indicates that the data of this real-time intensive care data stream is relatively simple and has a high degree of structurality. Processing such data streams usually does not require complex algorithms or a large amount of computing resources. For example, simple physiological parameter monitoring data, such as body temperature, blood pressure, etc., which are directly obtained from sensors in a structured form.

[0095] When the complexity evaluation value of a certain real-time intensive care data stream is χ2, the corresponding processing difficulty level matched is the intermediate processing difficulty level, which is the middle level among the processing difficulty levels. This indicates that the data of this real-time intensive care data stream contains some unstructured components or requires certain preprocessing, and some intermediate algorithms and an appropriate amount of computing resources are needed. For example, patient history records containing text descriptions or partial image data, which need to undergo certain preprocessing before being used for analysis.

[0096] When the complexity evaluation value of a certain real-time intensive care data stream is χ3, the corresponding processing difficulty level matched is the advanced processing difficulty level. The advanced processing difficulty level is the highest level among the processing difficulty levels. This indicates that the data of this real-time intensive care data stream is highly complex and has a high degree of unstructurality. Processing such data streams requires the use of advanced algorithms and a large amount of computing resources. The data may contain a high proportion of noise, outliers, or missing values, and complex data cleaning and preprocessing steps are required. For example, high-resolution medical image data, continuous multi-parameter physiological signal data streams, etc., which require complex analysis methods and a large amount of computing resources to extract useful information.

[0097] Sending each real-time intensive care data stream to the data processing module corresponding to the processing difficulty level based on the processing difficulty level of each real-time intensive care data stream, and after corresponding processing, sending it to the management terminal, specifically including:

[0098] Based on the processing difficulty level of each real-time intensive care data stream, the data processing center sends each real-time intensive care data stream to the primary data processing module, the intermediate data processing module, and the advanced data processing module respectively.

[0099] The primary data processing module is used to process each real-time intensive care data stream with the primary processing difficulty level, and this data processing module is used to process light tasks.

[0100] The intermediate data processing module is used to process each real-time intensive care data stream with the intermediate processing difficulty level, and this data processing module is used to process medium tasks.

[0101] The advanced data processing module is used to process each real-time intensive care data stream with an advanced processing difficulty level, and this data processing module is used to handle heavy tasks.

[0102] Each data processing module is equipped with a basic filtering algorithm, a signal processing algorithm, and a deep learning algorithm.

[0103] The primary data processing module preferentially uses the basic filtering algorithm, which is the moving average filtering algorithm in the embodiment of the present invention. The moving average filtering algorithm (Moving Average Filter) is a simple data processing technique used to smooth data sequences, reduce random fluctuations, and highlight data trends. The principle is to take the average of N consecutive data points in the data sequence as the new data point. Move on the data sequence, taking one step each time, and calculate the new average value. In the analysis of intensive care data streams, the moving average filtering is used to smooth the real-time data of physiological parameters such as heart rate and blood pressure to reduce the impact of instantaneous noise and make the data easier to observe and analyze.

[0104] The intermediate data processing module preferentially uses the signal processing algorithm, which is the Fourier transform algorithm in the embodiment of the present invention. The Fourier transform algorithm (Fourier Transform) is a mathematical transform that converts a signal from the time domain to the frequency domain. The principle is that any continuous signal can be represented as a combination of sine waves and cosine waves with different frequencies, amplitudes, and phases. The Fourier transform decomposes the complex time-domain signal into these basic frequency components. In the analysis of intensive care data streams, the Fourier transform can be used to analyze the frequency components of physiological signals (such as electrocardiograms and electroencephalograms).

[0105] The advanced data processing module preferentially uses the deep learning algorithm, which is the convolutional neural network algorithm in the embodiment of the present invention. The convolutional neural network algorithm (Convolutional Neural Network, CNN) is a deep learning model suitable for processing data with a grid-like topology (such as images). The principle is that the CNN extracts features in the data through a series of convolutional layers, pooling layers, and fully connected layers. The convolutional layer automatically learns and extracts local features through convolutional operations. The pooling layer is used to reduce the dimension of the data and increase the robustness of the model to position. The fully connected layer is used for the final classification or regression task. In the analysis of intensive care data streams, the CNN is used for complex tasks such as medical image analysis (such as X-ray films and CT scans) and signal recognition (such as arrhythmia detection).

[0106] In each level of data processing module, feature extraction is performed on each real-time intensive care data stream, and each data feature of each real-time intensive care data stream is obtained through analysis and processing.

[0107] In the embodiment of the present invention, the data features of each real-time critical care data stream include, but are not limited to: electrocardiogram data features, blood pressure data features, mechanical ventilation parameter data features, blood sugar data features, and intracranial pressure data features, etc. For example, the main parameter types in the mechanical ventilation parameter data features are tidal volume, respiratory rate, inspiration-expiration ratio, and airway pressure, etc.

[0108] The data features of each real-time intensive care data stream are integrated and recorded as each real-time intensive care data feature, and each real-time intensive care data feature is sent back to the data processing center for processing to obtain the variance explanation rate of each data feature, and the data are sorted according to the size of the variance explanation rate of each real-time intensive care data feature, and sent to the management terminal for prompting according to the sorting.

[0109] The principal component analysis (PCA) is a statistical method used to reduce a large number of possibly related variables to a few unrelated variables while retaining the variation information of the original data as much as possible. These unrelated variables are called "principal components".

[0110] The principle of principal component analysis is to transform a set of variables that may be correlated into a set of linearly uncorrelated variables through orthogonal transformation. The transformed set of variables is the principal component. The principal component is a linear combination of the original variables and is uncorrelated with each other. The first principal component contains the most data variation information, the second one is second, and so on.

[0111] The functions of principal component analysis include converting high-dimensional data into low-dimensional data, simplifying the data structure, and reducing computational complexity. It eliminates the correlation between the original variables and makes the new variables independent of each other. It identifies the main features in the data. In practical applications, the eigenvalues of each feature are usually calculated through a subspace iteration algorithm.

[0112] The variance explanation rate refers to the percentage of the total variance of the original data explained by each principal component. In principal component analysis, each principal component is a set of linear combinations of the original variables, and these principal components are sorted according to the variance (i.e. variance) they explain. The first principal component explains the largest variance, the second principal component explains the second largest variance, and so on.

[0113] The formula is:

[0114] In this embodiment of the present invention, FCrate x is the variance explanation rate of the x-th real-time critical care data feature, λ xis the eigenvalue of the x-th real-time intensive care data feature, where x is the real-time intensive care data feature number, x = 1, 2, 3,..., M, and M is the total number of real-time intensive care data features.

[0115] Monitor the real-time processing performance of each level of data processing module, analyze and obtain the real-time overload evaluation index of each level of data processing module, and match the GPU real-time acceleration multiple of each level of data processing module. The specific processing conditions are as follows:

[0116] Monitor the real-time processing performance of each level of data processing module through the network data Netdata tool, and obtain the real-time processing performance parameters of each level of data processing module, including the real-time memory usage rate, CPU real-time usage rate, and real-time processing queue length of each level of data processing module.

[0117] It should be noted that the real-time memory usage rate refers to the ratio between the memory capacity occupied by the current data processing module and the total system memory capacity. It is used to understand the memory usage situation in real time and avoid system crashes caused by memory leaks or excessive occupation.

[0118] The CPU real-time usage rate refers to the ratio between the CPU time occupied by the current data processing module and the total CPU time. It is used to understand the CPU usage situation in real time and avoid processing delays or system crashes caused by CPU overload.

[0119] The real-time processing queue length refers to the number of data tasks waiting to be processed currently. It is used to understand the load situation of the data processing module in real time and avoid processing delays caused by task backlogs.

[0120] Extract the real-time memory usage rate overload threshold, CPU real-time usage rate overload threshold, and real-time processing queue length overload threshold of each level of data processing module from the database, and compare and analyze them with the real-time processing performance parameters of each level of data processing module to obtain the real-time overload evaluation index of each level of data processing module, specifically including:

[0121]

[0122] Among them, G w is the real-time overload evaluation index of the w-th level data processing module, nc w is the real-time memory usage rate of the w-th level data processing module, CPU w is the CPU real-time usage rate of the w-th level data processing module, dl w is the real-time processing queue length of the w-th level data processing module, nc w0 is the real-time memory usage rate overload threshold of the w-th level data processing module, CPU w0 is the CPU real-time usage rate overload threshold of the w-th level data processing module, dlw0 is the overload threshold of the real-time processing queue length of the w-th level data processing module, ξ1 is the weight factor of the real-time memory usage rate, ξ2 is the weight factor of the real-time CPU usage rate, ξ3 is the weight factor of the real-time processing queue length, w is the number of the level data processing module, and w = 1, 2, 3.

[0123] It should be noted that the value ranges of the weight factor of the real-time memory usage rate, the weight factor of the real-time CPU usage rate, and the weight factor of the real-time processing queue length are all between 0 and 1, and satisfy ξ1 + ξ2 + ξ3 = 1. The weight factor of the real-time memory usage rate is an influencing factor of the real-time overload evaluation index of each level data processing module preset in the database, indicating the influence degree of the real-time memory usage rate on the real-time overload evaluation index of each level data processing module; the weight factor of the real-time CPU usage rate is an influencing factor of the real-time overload evaluation index of each level data processing module preset in the database, indicating the influence degree of the real-time CPU usage rate on the real-time overload evaluation index of each level data processing module; the weight factor of the real-time processing queue length is an influencing factor of the real-time overload evaluation index of each level data processing module preset in the database, indicating the influence degree of the real-time processing queue length on the real-time overload evaluation index of each level data processing module; when in use, the weight factor of the real-time memory usage rate, the weight factor of the real-time CPU usage rate, and the weight factor of the real-time processing queue length can be directly obtained from the database, and the acquisition method is a preset mapping relationship. For example: the real-time memory usage rate, the real-time CPU usage rate, and the real-time processing queue length of each level data processing module involved in the embodiments of the present invention are input into the preset mapping set in the database for mapping and matching to obtain the weight factor of the real-time memory usage rate, the weight factor of the real-time CPU usage rate, and the weight factor of the real-time processing queue length involved in the embodiments of the present invention, and the mapping relationship therein is one-to-one.

[0124] It should also be noted that there is a certain correlation among the parameters such as the real-time memory utilization rate, CPU real-time utilization rate, and real-time processing queue length of each level of data processing module. When the memory utilization rate is relatively high, more memory page swapping may be required, which will increase the burden on the CPU and thus lead to an increase in the CPU utilization rate. An overly high memory utilization rate may cause new tasks to be unable to be allocated memory space in a timely manner, thereby increasing the real-time processing queue length. The CPU utilization rate reflects the processing ability of the data processing module. When the CPU utilization rate is low, it indicates that there is surplus processing capacity and the processing queue length can be shortened. The utilization rates of memory and CPU directly affect the processing ability of the data processing module, and thus affect the real-time processing queue length. The change in the queue length will in turn feedback to the utilization rates of memory and CPU, forming a performance feedback loop. These three parameters restrict each other, and any abnormality in one parameter may cause changes in other parameters. For example, insufficient memory may lead to an increase in CPU utilization rate and an increase in queue length; a CPU bottleneck may lead to an increase in queue length and thus an increase in memory occupancy. Therefore, in practical applications, it is necessary to comprehensively monitor these three parameters in order to fully evaluate the performance status of the data processing module.

[0125] Based on the real-time overload evaluation index of each level of data processing module, map and match with the GPU real-time acceleration multiple corresponding to each real-time overload evaluation index in the database to obtain the GPU real-time acceleration multiple of each level of data processing module. It should be added that if the GPU real-time acceleration multiple is 1, it means that the CPU can independently complete data processing and there is no need to send data to the GPU for acceleration.

[0126] When the real-time overload evaluation index of a certain level of data processing module is greater than 1, the specific acceleration process includes: transferring the data of this level of data processing module from CPU memory to GPU memory, using the GPU scheduler to adjust the parallel quantity based on the GPU real-time acceleration multiple and then perform parallel execution. After the parallel calculation is completed, transfer the data processing result from GPU memory back to CPU memory.

[0127] For example, if the GPU real-time acceleration multiple matched by the real-time overload evaluation index of a certain level of data processing module is 1.2 times, then perform 1.2 times acceleration processing on the GPU processor of this level of data processing module.

[0128] The GPU real-time acceleration multiple is used to improve the processing speed of each level of data processing module and avoid overload situations in the data processing module.

[0129] In this embodiment, the present invention also provides a device, including: the device has one or more programs, and the one or more programs are executed by one or more processors.

[0130] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0131] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. As long as it does not deviate from the structure of the present invention or exceed the scope defined by the present invention, it should fall within the protection scope of the present invention.

Claims

1. A method for processing intensive care data, characterized in that, Including: S1. Monitor the real-time performance of each intensive care device, obtain the real-time performance evaluation parameters of each intensive care device, analyze and process to obtain the real-time performance evaluation values of each intensive care device, and match to obtain the performance difference correction values of each intensive care device; S2. Each intensive care device transmits each real-time intensive care data stream to the data processing center. After correction based on the performance difference correction values of each intensive care device, obtain the basic information of each real-time intensive care data stream, and analyze and process to obtain the complexity evaluation value of each real-time intensive care data stream; S3. Based on the complexity evaluation values of each real-time intensive care data stream, match to obtain the processing difficulty levels of each real-time intensive care data stream, and send each real-time intensive care data stream to the data processing module under the corresponding processing difficulty level, and after corresponding processing, send it to the management terminal.

2. The method for processing intensive care data according to claim 1, characterized in that: The specific process of monitoring the real-time performance of each intensive care device, obtaining the real-time performance evaluation parameters of each intensive care device, and analyzing and processing to obtain the real-time performance evaluation values of each intensive care device is as follows: Obtain the real-time performance evaluation parameters of each intensive care device; The real-time performance evaluation parameters of each intensive care device include the network real-time delay, device real-time load, CPU real-time usage rate, disk real-time read / write speed, and network real-time traffic of each intensive care device; Extract the real-time performance evaluation verification parameters of each intensive care device from the database, including the real-time delay verification value, device real-time load verification value, CPU real-time usage rate verification value, disk real-time read / write speed verification value, and network real-time traffic verification value of each intensive care device; Compare and analyze the real-time performance evaluation parameters of each intensive care device with the real-time performance evaluation verification parameters of each intensive care device to obtain the real-time performance evaluation values of each intensive care device; The real-time performance evaluation values of each intensive care device are used to characterize the real-time performance of each intensive care device.

3. The method for processing intensive care data according to claim 1, wherein: The specific process of matching to obtain the performance difference correction values of each intensive care device is as follows: Based on the real-time performance evaluation values of each intensive care device, perform mapping and matching with the performance difference correction values corresponding to each pre-stored real-time performance evaluation value interval in the database to obtain the performance difference correction values of each intensive care device; The performance difference correction values of each intensive care device are used to correct the influence of the performance difference of the intensive care device on the real-time intensive care data stream.

4. The method for processing intensive care data according to claim 1, wherein: After correction based on the performance difference correction values of each intensive care device, the basic information of each real-time intensive care data stream is obtained, specifically including: Based on the performance difference correction values of each intensive care device, the real-time data stream quality correction parameters corresponding to each performance difference correction value pre-stored in the database, including the filter cut-off frequency, filter passband ripple, and filter stopband attenuation; The basic information of each real-time intensive care data stream includes the number of data types, unstructured data ratio, total data volume, and data noise ratio of each real-time intensive care data stream.

5. The method for processing intensive care data according to claim 1, wherein: The specific process of analyzing and processing to obtain the complexity evaluation value of each real-time intensive care data stream includes: Average the basic information of each real-time intensive care data stream to obtain a set of average values of the basic information of the real-time intensive care data stream, including the average number of data types, the average proportion of unstructured data, the average total amount of data volume, and the average proportion of data noise; Perform comprehensive analysis and processing on the basic information of each real-time intensive care data stream and the set of average values of the basic information to obtain the complexity evaluation value of each real-time intensive care data stream; The complexity evaluation value of each real-time intensive care data stream is used to characterize the complexity of each real-time intensive care data stream.

6. The method for processing intensive care data according to claim 1, wherein: Obtaining the processing difficulty level of each real-time intensive care data stream specifically includes: The processing difficulty level includes the primary processing difficulty level, the intermediate processing difficulty level, and the advanced processing difficulty level; Based on the complexity evaluation value of a certain real-time intensive care data stream, perform mapping and matching with the processing difficulty levels corresponding to the pre-stored complexity evaluation values in the database to obtain the processing difficulty level of this real-time intensive care data stream.

7. The method for processing intensive care data according to claim 1, wherein: Sending each real-time intensive care data stream to the data processing module under the corresponding processing difficulty level based on the processing difficulty level of each real-time intensive care data stream, and sending it to the management terminal after corresponding processing, specifically including: Based on the processing difficulty level of each real-time intensive care data stream, the data processing center sends each real-time intensive care data stream to the primary data processing module, the intermediate data processing module, and the advanced data processing module respectively; The primary data processing module is used to process each real-time intensive care data stream at the primary processing difficulty level, and this data processing module is used to process light tasks; The intermediate data processing module is used to process each real-time intensive care data stream at the intermediate processing difficulty level, and this data processing module is used to process medium tasks; The advanced data processing module is used to process each real-time intensive care data stream at the advanced processing difficulty level, and this data processing module is used to process heavy tasks; In each level of data processing module, perform feature extraction on each real-time intensive care data stream respectively, and analyze and process to obtain each data feature of each real-time intensive care data stream; Integrate each data feature of each real-time intensive care data stream and record it as each real-time intensive care data feature, send each real-time intensive care data feature back to the data processing center, process to obtain the variance interpretation rate of each data feature, sort according to the size of the variance interpretation rate of each real-time intensive care data feature, and send it to the management terminal for prompting according to the sorting.

8. The method for processing intensive care data according to claim 1, wherein: It also includes monitoring the real-time processing performance of each level of data processing module, and analyzing and processing to obtain the real-time overload evaluation index of each level of data processing module. The specific processing conditions are: Monitor the real-time processing performance of each level of data processing module, and obtain the real-time processing performance parameters of each level of data processing module, including the real-time memory usage rate, the real-time CPU usage rate, and the real-time processing queue length of each level of data processing module; Extract the memory real-time usage overload threshold, CPU real-time usage overload threshold, and real-time processing queue length overload threshold of each level of data processing module from the database, and compare and analyze them with the real-time processing performance parameters of each level of data processing module to obtain the real-time overload evaluation index of each level of data processing module.

9. The method for processing intensive care data according to claim 1, characterized in that: Perform corresponding processing, specifically including: Based on the real-time overload evaluation index of each level of data processing module, map and match it with the GPU real-time acceleration multiple corresponding to each real-time overload evaluation index in the database to obtain the GPU real-time acceleration multiple of each level of data processing module; The GPU real-time acceleration multiple is used to improve the processing speed of each level of data processing module and avoid overload of the data processing module.

10. A device, characterized in that, Include: The device has one or more programs, and the one or more programs are executed by one or more processors to implement the method according to any one of claims 1-9.

Images