Nursing data analysis method and system based on artificial intelligence

By using an AI-based nursing data analysis method, clustering and similarity operations are performed on core point groups of real-time and preceding data segments. This solves the problems of delayed analysis results and large computational load in existing technologies, achieving efficient and accurate nursing data analysis and ensuring high real-time response in the medical and nursing field.

CN118626678BActive Publication Date: 2024-11-05THE THIRD AFFILIATED HOSPITAL OF ZHENGZHOU UNIVERSITY +1
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
CN202411089248.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-09
Publication Date
2024-11-05
Estimated Expiration
2044-08-09

AI Technical Summary

Technical Problem

Existing nursing data analysis methods cannot fully utilize the dynamism and real-time nature of real-time monitoring data, resulting in delayed analysis results, large computational load, and an inability to meet the medical and nursing field's demand for high real-time response, especially in emergency situations where the best treatment opportunity may be missed.

Method used

An AI-based nursing data analysis method is adopted. By acquiring the core point groups of real-time data segment combinations and preceding data segments, cluster analysis and similarity calculations are performed to update the core point groups of the target data segment, identify nursing data segments to be confirmed, and label them, so as to improve data processing efficiency and response speed.

Benefits of technology

It improves the efficiency of data processing, ensures the timeliness and effectiveness of patient care, enables the timely identification of potential health problems, and meets the medical care field's need for high real-time response.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118626678B_ABST
    Figure CN118626678B_ABST
Patent Text Reader

Abstract

The present application provides a nursing data analysis method and system based on artificial intelligence. When acquiring multiple real-time data segments in a real-time data segment combination, a current data segment core point group obtained according to bucketing of the multiple real-time data segments and a previous data segment core point group indicating past bucketing results are obtained. At the same time, the target data segment core point group is updated according to the previous data segment core point group and the current data segment core point group. The bucketing results obtained based on core point comparison and updating are prevented from being performed on the initial data segments and real-time data segments in the existing buckets, the initial data segments in the multiple data segment buckets, and the real-time data segments, so that the bucketing speed is improved, the efficiency of data processing is improved, and the high real-time response requirements during patient care are ensured.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of data processing, and specifically, to a nursing data analysis method and system based on artificial intelligence. Background Art

[0002] In the field of medical care, real-time monitoring data analysis of patients is an important means to ensure patient safety and improve the quality of care. Traditional nursing data analysis methods often rely on manual review, which is not only inefficient but also error-prone, especially when dealing with large amounts of real-time monitoring data. Its limitations are particularly obvious. With the continuous advancement of medical technology and the rapid development of technologies such as big data and artificial intelligence, how to achieve efficient and accurate nursing data analysis has become an urgent problem to be solved in the field of medical care. In the prior art, although some data analysis methods based on machine learning or statistics have been applied to nursing data analysis, these methods still have some shortcomings when processing real-time monitoring data.

[0003] First, existing data analysis methods often fail to fully utilize the dynamic and real-time characteristics of real-time monitoring data. The patient's physiological state changes over time, and real-time monitoring data can reflect the patient's current physiological state, but existing methods often cannot respond to these changes quickly, resulting in a lag in analysis results. Secondly, when processing a large amount of real-time monitoring data, existing data analysis methods often have large computational complexity and slow processing speed, and cannot meet the demand for high real-time response in the medical care field. Especially in some emergency situations, if the changes in the patient's physiological state cannot be analyzed in time, the best time for treatment may be missed.

[0004] Therefore, an efficient and accurate nursing data analysis method is needed that can make full use of the dynamic and real-time characteristics of real-time monitoring data, quickly respond to changes in the patient's physiological state, and meet the demand for high real-time response in the medical care field. At the same time, the method also needs to be able to simplify the amount of calculation and improve data processing efficiency to ensure the timeliness and effectiveness of patient care. Summary of the invention

[0005] The purpose of the present invention is to provide a nursing data analysis method and system based on artificial intelligence.

[0006] This application is implemented as follows:

[0007] In a first aspect, the present application provides a nursing data analysis method based on artificial intelligence, comprising: obtaining a real-time data segment combination obtained by intercepting a nursing monitoring data sequence, wherein the real-time data segment combination includes multiple real-time data segments intercepted from the nursing monitoring data sequence, and the real-time data segment includes multiple nursing data items arranged for the monitored objects according to monitoring timestamps; obtaining a current data segment core point group corresponding to the real-time data segment combination, and a preceding data segment core point group obtained based on the preceding data segment of the nursing monitoring data sequence before the real-time data segment combination, wherein the current data segment core point group indicates the bucketing results of all real-time data segments in the real-time data segment combination, and the preceding data segment core point group indicates the preceding data segment core point group. The bucketing result of each preceding data segment in the sequential data segment, the multiple current core points in the current data segment core point group and the multiple preceding data segment core points in the preceding data segment core point group respectively represent a data segment bucket; after the preceding data segment core point group is updated by the current data segment core point group to obtain the target data segment core point group, each of the real-time data segments in the real-time data segment combination is subjected to a similarity operation with the multiple target data segment core points in the target data segment core point group to obtain the target data segment vector interval corresponding to each real-time data segment; the real-time data segment whose target data segment vector interval is greater than the critical data segment vector interval is determined as a nursing data segment to be confirmed, and the nursing data segment to be confirmed is marked.

[0008] On the other hand, the present application provides a computer system, comprising: one or more processors; a memory; one or more computer programs; wherein the one or more computer programs are stored in the memory and configured to be executed by the one or more processors, and when the one or more computer programs are executed by the processors, the method as described above is implemented.

[0009] The beneficial effects of the present application are at least as follows: in the method provided by the present application, when obtaining multiple real-time data segments in a real-time data segment combination, the core point group of the current data segment obtained according to the bucketing of multiple real-time data segments and the core point group of the previous data segment indicating the past bucketing results are obtained, and at the same time, the core point group of the previous data segment and the core point group of the current data segment are updated to obtain the core point group of the target data segment, so as to prevent the initial data segments and the real-time data segments in the existing buckets from being used for similarity calculation and bucketing based on the core point comparison and updating of the bucketing results obtained, so that the bucketing speed is improved, the efficiency of data processing is improved, and the high real-time response requirements during patient care are ensured. In addition, the target data segment core point group based on the updated bucketing results is used for similarity calculation with the multiple real-time data segments in the real-time data segment combination to determine the nursing data segment to be confirmed, which can greatly reduce the order of magnitude of the similarity calculation, simplify the amount of calculation, and further improve the efficiency of the detection of the nursing data segment to be confirmed. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 This is a flowchart of an artificial intelligence-based nursing data analysis method provided in an embodiment of the present application.

[0011] Figure 2 It is a schematic diagram of the composition of a computer system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0012] The execution subject of the artificial intelligence-based nursing data analysis method in the embodiment of the present application is a computer system, including but not limited to a server, a personal computer, a laptop, a tablet computer, a smart phone, etc. Figure 1 As shown, the method includes:

[0013] Step S10: Obtain a real-time data segment combination intercepted from the nursing monitoring data sequence, wherein the real-time data segment combination includes multiple real-time data segments intercepted from the nursing monitoring data sequence, and the real-time data segment includes multiple nursing data items arranged according to monitoring timestamps for the monitored objects.

[0014] In an embodiment of the present application, step S10 obtains a real-time data segment combination from a continuous nursing monitoring data sequence. Specifically, when the computer system receives a nursing monitoring data sequence from a monitoring device, it identifies each data item in the data sequence. Each data item contains a series of nursing data about the monitored object, such as heart rate, blood pressure, blood oxygen saturation, etc., and these data items are arranged in the order of monitoring timestamps. For example, a data sequence containing continuous monitoring data can be represented as A1, A2, A3...An, where each Ai represents a data item, containing nursing data at a specific time point.

[0015] In order to conduct effective data analysis, the computer system divides these continuous data items into several real-time data segments. The length of the real-time data segment can be set according to the specific application scenario and requirements. Assuming that in this scenario, the length of the real-time data segment is set to 3 data items in advance, the computer system will intercept the real-time data segment combination from the data sequence according to this length.

[0016] Taking the data sequence A1, A2, A3...An as an example, according to the setting of the real-time data segment length of 3, the following real-time data segment combinations will be intercepted in sequence: the first real-time data segment: A1, A2, A3; the second real-time data segment: A2, A3, A4;...; the last real-time data segment: An-2, An-1, An. These data segments constitute a real-time data segment combination. Each real-time data segment contains monitoring data at three consecutive time points, which reflects the changes in the physiological state of the monitored object in a short period of time. It should be noted that the length setting of the real-time data segment is not fixed, and it can be adjusted according to specific business needs and data characteristics. For example, in scenarios where a more detailed analysis of the state changes of the monitored object is required, a shorter real-time data segment length can be set; while in scenarios where more attention is paid to long-term trends, a longer real-time data segment length can be set. In addition, the interception process of the real-time data segment can also adopt a sliding window method, that is, sliding one data item forward each time to obtain a continuous overlapping real-time data segment combination, so as to analyze the monitoring data more comprehensively.

[0017] Step S20: Obtain the current data segment core point group corresponding to the real-time data segment combination, and the preceding data segment core point group obtained based on the preceding data segment of the nursing monitoring data sequence before the real-time data segment combination, wherein the current data segment core point group indicates the bucketing results of all real-time data segments in the real-time data segment combination, the preceding data segment core point group indicates the bucketing results of each preceding data segment in the preceding data segment, and the multiple current core points in the current data segment core point group and the multiple preceding data segment core points in the preceding data segment core point group respectively represent a data segment bucket.

[0018] In an embodiment of the present application, step S20 clusters the real-time data segment combination and the preceding data segment to obtain the current data segment core point group and the preceding data segment core point group. The two core point groups represent the cluster centers of the real-time data segment and the preceding data segment, respectively. First, the computer system performs cluster analysis on the real-time data segment combination to obtain the current data segment core point group. Cluster analysis is an unsupervised learning method that can divide similar data points into the same cluster, while dissimilar data points are assigned to different clusters. In this scenario, cluster analysis is used to identify the cluster centers in the real-time data segment, i.e., the current core points. These current core points represent the main data patterns or trends in the real-time data segment.

[0019] Assume that the real-time data segment combination includes A1, A2, A3; A2, A3, A4; ...; An-2, An-1, An and other real-time data segments, each of which contains multiple nursing data items of the monitored object. The computer system can use clustering algorithms such as K-means and DBSCAN to perform clustering analysis on these real-time data segments. Taking the K-means algorithm as an example, the computer system randomly selects several points as the initial cluster centers, and then iteratively assigns each data segment to the nearest cluster center and recalculates the center of each cluster. After multiple iterations, a set of stable cluster centers will eventually be obtained, that is, the core point group of the current data segment.

[0020] At the same time, the computer system also obtains the core point group of the previous data segment. The previous data segment refers to the historical data segment before the real-time data segment is combined, which contains the monitoring data of the monitored object in the past period of time. Similar to the real-time data segment, the computer system will also perform cluster analysis on the previous data segment to obtain the core point group of the previous data segment. These core points of the previous data segment represent the main data patterns or trends in the historical data. The process of obtaining the core point group of the previous data segment is similar to that of the core point group of the current data segment, and is also achieved through cluster analysis. The computer system can use the same clustering algorithm (such as K-means) to cluster the previous data segments and obtain a set of stable cluster centers as the core point group of the previous data segment.

[0021] It should be noted that each core point in the current data segment core point group and the previous data segment core point group represents a data segment bucket. Data segment bucketing is the process of dividing data segments into several groups according to certain rules, and the data segments in each group have similar characteristics or trends. In this scenario, cluster centers are used as representatives of data segment buckets because they can reflect the main data patterns or trends in the data segments. By obtaining the current data segment core point group and the previous data segment core point group, the computer system can have a deeper understanding of the real-time and historical status of the monitored object. These two core point groups not only contain the main pattern or trend information of the data, but also provide similarity metrics between data segments. In subsequent data analysis and anomaly detection, the computer system can use these core point groups to compare the similarity between the real-time data segment and the previous data segment, so as to discover possible anomalies or trend changes.

[0022] Step S30: After the previous data segment core point group is updated by the current data segment core point group to obtain the target data segment core point group, each real-time data segment in the real-time data segment combination is subjected to similarity operation with multiple target data segment core points in the target data segment core point group to obtain the target data segment vector interval corresponding to each real-time data segment.

[0023] In an embodiment of the present application, step S30 involves updating the core point group of the previous data segment through the core point group of the current data segment to obtain the core point group of the target data segment. The purpose of this step is to combine historical data and real-time data to form a new core point group that takes into account past trends and reflects the current state. The update process can adopt a variety of strategies, such as weighted average, sliding window, etc. For example, a weight factor can be set so that the core point group of the current data segment occupies a larger weight in the update process, thereby emphasizing the impact of real-time data. Assume that the core point group of the previous data segment is C_prev ={Cp1, Cp2, ..., Cpn}, where each Cpi represents the core point of a previous data segment. The core point group of the current data segment is C_current ={Cc1, Cc2, ..., Ccm}, where each Cci represents the core point of a real-time data segment. The updating process can be performed according to a certain weight strategy, such as weighted averaging, to obtain the target data segment core point group C_target ={Ct1, Ct2, ..., Ctk}, where Ctj is the weighted average of the corresponding core points in C_prev and C_current.

[0024] Next, step S30 requires that each real-time data segment in the real-time data segment combination be subjected to a similarity operation with multiple target data segment core points in the target data segment core point group. The purpose of this step is to quantify the similarity or difference between the real-time data segment and the target data segment core points. The similarity operation can be performed in a variety of ways, such as Euclidean distance, cosine similarity, etc., and the specific selection depends on the characteristics of the data and business requirements. Taking Euclidean distance as an example, assuming that the real-time data segment combination contains real-time data segments R1, R2, ..., Rn, each real-time data segment Ri can be represented as a feature vector. At the same time, each core point Ctj in the target data segment core point group C_target can also be represented as a feature vector. For the real-time data segment Ri, the computer system will calculate the Euclidean distance between it and each core point Ctj in C_target, and these distances will constitute a distance vector Di = [d1, d2, ..., dk], where di represents the Euclidean distance between Ri and Ctj.

[0025] Finally, in order to obtain a comprehensive similarity measurement index, the computer system can further process each distance value in the distance vector Di, such as taking the minimum value, average value or weighted average value. Here, taking the minimum value as an example, for the real-time data segment Ri, its target data segment vector interval Si will be the minimum value in the distance vector Di, that is, Si=min(Di). This vector interval Si reflects the minimum distance between the real-time data segment Ri and the core point group of the target data segment, that is, the maximum similarity. Through the execution of step S30, the computer system can quantify the similarity or difference between the real-time data segment and the core point group of the target data segment, which provides an important basis for subsequent data analysis and anomaly detection. This similarity calculation method based on the core point group not only takes into account the overall trend of the data, but also takes into account the dynamic changes of the real-time data, making the data analysis results more accurate and reliable.

[0026] Step S40: determining the real-time data segments whose target data segment vector interval is greater than the critical data segment vector interval as nursing data segments to be confirmed, and marking the nursing data segments to be confirmed.

[0027] In the embodiment of the present application, step S40 is responsible for identifying and marking those real-time data segments that may be abnormal or require further attention. By comparing the target data segment vector interval and the critical data segment vector interval, the computer system can accurately identify those real-time data segments that deviate greatly from the target data segment core point group, thereby ensuring that medical staff can promptly discover and deal with potential health problems. The target data segment vector interval is obtained by performing a similarity operation on the real-time data segment and the target data segment core point group, which quantifies the similarity or difference between the real-time data segment and the target data segment core point group. The critical data segment vector interval is a preset threshold used to determine whether the real-time data segment is abnormal.

[0028] In the embodiment of the present application, it is assumed that the target data segment vector interval of each real-time data segment has been calculated by step S30. These vector intervals form a vector interval sequence, for example, Si = [s1, s2, ..., sn], where si represents the target data segment vector interval of the i-th real-time data segment. At the same time, a critical data segment vector interval s_threshold is also set as a threshold for abnormal judgment.

[0029] Next, the computer system will traverse the vector interval sequence Si and compare the size of each si with s_threshold one by one. If the target data segment vector interval si of a real-time data segment is greater than s_threshold, then the real-time data segment is regarded as a nursing data segment to be confirmed, which means that it may contain abnormal information or require further attention. For example, assuming that in the vector interval sequence Si, the target data segment vector interval sj of the j-th real-time data segment is greater than s_threshold, then the computer system will mark the real-time data segment Rj as a nursing data segment to be confirmed. This marking process can be implemented in a variety of ways, such as highlighting Rj on the data display interface, or generating an alarm message to send to medical staff. It will be understood that the present application does not involve specific disease diagnosis.

[0030] The labeling of the nursing data segments to be confirmed not only helps medical staff quickly locate data segments that may have problems, but also triggers further diagnosis or intervention measures. For example, after discovering that a patient's heart rate data segment is marked as a nursing data segment to be confirmed, medical staff can immediately view the patient's real-time monitoring data and make a comprehensive judgment based on other physiological indicators. If there is indeed an abnormal situation, medical staff can take timely treatment measures to ensure the patient's life safety.

[0031] In a possible implementation, step S30, before performing similarity operations on each real-time data segment in the real-time data segment combination and a plurality of target data segment core points in the target data segment core point group to obtain a target data segment vector interval corresponding to each real-time data segment, may further include:

[0032] Step S30A: respectively calculating the vector intervals between each current data segment core point in the current data segment core point group and each previous data segment core point in the previous data segment core point group.

[0033] In the embodiment of the present application, step S30A calculates the vector interval between each core point in the core point group of the current data segment and each core point in the core point group of the previous data segment. The purpose of this step is to evaluate the similarity between the current data segment and the previous data segment, and provide a basis for subsequent data integration and anomaly detection.

[0034] Specifically, assume that the core point group of the current data segment is represented by C_current = {Cc1, Cc2, ..., Ccn}, where each Cci is a multidimensional vector, representing a cluster center or core point in the current data segment. Similarly, the core point group of the previous data segment is represented by C_prev = {Cp1, Cp2, ..., Cpm}, where each Cpj is also a multidimensional vector, representing a cluster center or core point in the previous data segment.

[0035] In nursing data analysis, these multidimensional vectors usually contain multiple physiological index data of patients, such as heart rate, blood pressure, blood oxygen saturation, etc. Each index can be regarded as a dimension in the vector, and the value of the vector represents the specific value of the index. For example, a core point vector Cc1 may be expressed as [heart rate = 80, blood pressure = 120 / 80, blood oxygen saturation = 98%], while another core point vector Cp1 may be expressed as [heart rate = 75, blood pressure = 115 / 75, blood oxygen saturation = 97%]. During the execution of step S30A, the computer system will traverse each core point vector Cci in C_current and calculate the vector interval with each core point vector Cpj in C_prev. The calculation of vector interval usually adopts distance measurement method, such as Euclidean distance or cosine similarity. Here, Euclidean distance is taken as an example for explanation.

[0036] For each core point vector pair (Cci, Cpj), the computer system will calculate the Euclidean distance between them, and this distance value will quantitatively represent the similarity or difference between Cci and Cpj. By executing step S30A, the computer system will obtain a distance matrix, in which each element d(Cci, Cpj) represents the Euclidean distance between a core point in the core point group of the current data segment and a core point in the core point group of the previous data segment. This distance matrix will provide important reference information for subsequent data integration and anomaly detection. For example, if a core point of the current data segment is very close to a core point in the previous data segment, then the two core points may represent similar data patterns, and they can be considered to be merged; conversely, if the distance is far, it may indicate the emergence of a new data pattern or anomaly.

[0037] Step S30B: If there is a preceding data segment core point in the preceding data segment core point group whose vector interval is smaller than the reference core point vector interval, then multiple real-time data segments in the current data segment bucket corresponding to the current data segment core point are merged into the preceding data segment bucket corresponding to the preceding data segment core point to obtain an adjusted preceding data segment bucket, wherein each preceding data segment core point in the preceding data segment core point group corresponds to a reference core point vector interval.

[0038] In the embodiment of the present application, step S30B is a key data integration step, which involves evaluating the similarity between the core point group of the previous data segment and the core point group of the current data segment, and merging the current data segment according to the evaluation result. The purpose of this step is to ensure the continuity and consistency of the data, while reducing redundant information and improving the accuracy of subsequent data analysis. In step S30A, the computer system has calculated the vector interval between each core point in the core point group of the current data segment and each core point in the core point group of the previous data segment. These vector intervals quantitatively represent the similarity or difference between the two groups of core points. Next, in step S30B, the computer system will check whether these vector intervals are less than a preset reference core point vector interval. This reference core point vector interval is a threshold value used to determine whether the core point of the current data segment should be merged with the core point of the previous data segment.

[0039] Assume that in the nursing data analysis, the focus is on the patient's vital signs data, such as heart rate, blood pressure, etc. The current data segment core point group represents the patient's latest vital sign data pattern, while the previous data segment core point group represents the patient's vital sign data pattern in the past period of time. The reference core point vector interval is set to a smaller value to ensure that only very similar data patterns will be merged. Now, suppose a current data segment core point Cci, which has a vector interval with a core point Cpj in the previous data segment core point group that is smaller than the reference core point vector interval. This means that the data patterns represented by Cci and Cpj are very similar, and it is likely to reflect the continuous change trend of the patient's vital signs. In this case, according to step S30B, the computer system will perform a merging operation. Specifically, it will merge multiple real-time data segments in the current data segment bucket corresponding to the current data segment core point Cci into the previous data segment bucket corresponding to the previous data segment core point Cpj. These real-time data segments contain the patient's latest vital signs data. After being merged into the previous data segment buckets, they will form a continuous data sequence with the previous data, which is convenient for subsequent data analysis and trend prediction.

[0040] It should be noted that each of the preceding data segment core points in the preceding data segment core point group corresponds to a reference core point vector interval. This means that for different preceding data segment core points, different reference core point vector intervals may need to be set. This depends on the specific application scenario and data characteristics, and can be adjusted based on historical data and expert experience. By executing step S30B, the computer system can ensure the continuity and consistency of the data, while reducing redundant information and improving the accuracy of subsequent data analysis.

[0041] Step S30C: If the previous data segment core point group does not have a previous data segment core point whose vector interval is smaller than the reference core point vector interval, the current data segment bucket corresponding to the current data segment core point is used as a new previous data segment bucket in the previous data segment buckets.

[0042] Specifically, it is assumed that in the nursing data analysis, the focus is on the patient's real-time vital signs data, such as heart rate, blood pressure, etc. These data are collected in real time and divided into different data segments, each of which represents data within a time window. The current data segment core point group represents the main data pattern or trend in the latest collected data segment. In step S30B, the computer system attempts to match the current data segment core point with the previous data segment core point to determine whether there is a similar data pattern. However, if there is no core point in the previous data segment core point group whose vector interval with the current data segment core point is less than the reference core point vector interval, this means that there is no pattern in the previous data that is very similar to the current data segment.

[0043] In this case, according to step S30C, the computer system uses the current data segment bucket corresponding to the core point of the current data segment as a new preceding data segment bucket in the preceding data segment bucket. In other words, since the current data segment represents a new or rare data pattern, it will be added to the set of preceding data segment buckets as an independent data segment bucket. By executing step S30C, the computer system can ensure that all important data patterns, whether common or rare, can be accurately captured and recorded in the set of data segment buckets. This is crucial for subsequent anomaly detection, trend analysis, and a comprehensive assessment of the patient's health status.

[0044] Step S30D: using the data segment core points respectively corresponding to the preceding data segment buckets in the adjusted preceding data segment bucket combination as the target data segment core points in the target data segment core point group.

[0045] In the embodiment of the present application, step S30D is a key step to integrate and determine the target data segment core point group. After processing steps S30A to S30C, the computer system has adjusted the previous data segment core point group to form an adjusted previous data segment bucket combination. The role of step S30D is to use the data segment core point corresponding to each previous data segment bucket in this combination as a member of the target data segment core point group. In nursing data analysis, data segment bucketing is the process of dividing continuous data segments into several groups according to certain rules (such as clustering), and the data segments in each group have similar characteristics or trends. The data segment core point is the representative of each bucket, usually a multidimensional vector, which contains the main feature information of the data in the bucket.

[0046] In steps S30A to S30C, the computer system adjusts the core point group of the previous data segment. If the vector interval between the core point of the current data segment and a core point of a previous data segment is less than the reference core point vector interval, the current data segment will be merged into the corresponding previous data segment bucket (step S30B). If the vector interval between the core point of the current data segment and all the core points of the previous data segments is greater than or equal to the reference core point vector interval, the current data segment will be used as a new previous data segment bucket (step S30C). After these adjustments, the previous data segment bucket combination is updated.

[0047] The function of step S30D is to use the data segment core points corresponding to each bucket in the updated previous data segment bucket combination as members of the target data segment core point group. This process actually expresses the integration result of the previous data and the current data in the form of core points, providing a benchmark for subsequent similarity calculations and anomaly detection.

[0048] Through step S30D, the computer system can generate a target data segment core point group that contains both historical data patterns and reflects the current data status. This core point group will serve as a benchmark for subsequent similarity calculations and anomaly detection, helping medical staff to more accurately understand the changing trends of the patient's physiological status and discover potential health problems in a timely manner.

[0049] In a possible implementation, step S30D, using the data segment core points corresponding to the preceding data segment buckets in the adjusted preceding data segment bucket combination as the target data segment core points in the target data segment core point group, includes:

[0050] Step S30D1: If the target preceding data segment bucket in the adjusted preceding data segment bucket combination includes multiple preceding data segments and multiple real-time data segments in the current data segment bucket, then respectively obtain the data segment vectors corresponding to the multiple real-time data segments in the current data segment bucket, and multiply the vector modulus product with the inverse vector of the data segment vector, and use the vector modulus product corresponding to each of the multiple real-time data segments as the real-time feature scalar corresponding to the multiple real-time data segments.

[0051] In the embodiment of the present application, the main purpose of step S30D1 is to extract feature information from the real-time data segment and quantify it into a real-time feature scalar so that it can be used to determine the core point of the target data segment in the subsequent steps. Assume that in the nursing data analysis, the focus is on the vital signs data of a patient, such as heart rate, blood pressure, blood oxygen saturation, etc. These data are monitored in real time and divided into different data segments, each data segment represents the monitoring data within a time window. The current data segment bucket contains the patient's latest vital signs data, while the previous data segment represents the patient's vital signs data over a period of time in the past.

[0052] In step S30D1, first, the computer system identifies a target preceding data segment bucket in the adjusted preceding data segment bucket combination, which not only contains multiple preceding data segments, but also contains multiple real-time data segments merged from the current data segment bucket. These real-time data segments contain the patient's latest vital sign data, which is crucial for understanding the patient's current state.

[0053] Next, for each real-time data segment in the target preceding data segment bucket, the computer system performs the following operations:

[0054] Obtaining a data segment vector: First, the computer system extracts the patient's vital signs data from the real-time data segment and represents it as a data segment vector. This data segment vector is a multidimensional vector, each dimension corresponds to a physiological indicator, and the value of the vector is the specific value of the indicator. For example, a data segment vector of a real-time data segment may be represented as [heart rate = 90, blood pressure = 120 / 80, blood oxygen saturation = 95%].

[0055] Calculate the vector modulus product (i.e., inner product) of the reverse vector: Since the data segment vector itself is a column vector, its reverse vector (i.e., transpose) is actually itself. Therefore, calculating the inner product of the data segment vector and its reverse vector is actually calculating the square of the modulus of the vector (for a unit vector, the square of the modulus is 1). However, what is more concerned here is the relative weight or importance between the dimensions of the vector, so the inner product can be simply regarded as the weighted sum of the values ​​of each dimension of the vector, and the weight defaults to 1 (i.e., no specific weighting is performed). In this way, the result of the inner product quantifies the comprehensive performance of each physiological indicator in the real-time data segment, forming a real-time feature scalar.

[0056] Generate real-time feature scalars: For each real-time data segment in the target pre-order data segment bucket, the computer system performs the above operations to generate a corresponding real-time feature scalar, i.e., feature value. These real-time feature scalars will be used to determine the core points of the target data segment in the subsequent steps.

[0057] Through step S30D1, key feature information can be extracted from the real-time data segment and quantified into real-time feature scalars. These real-time feature scalars not only reflect the patient's latest vital sign status, but also provide an important data basis for the subsequent determination of the core points of the target data segment.

[0058] Step S30D2: Determine the target data segment core point corresponding to the target pre-order data segment bucket according to the pre-order feature scalar corresponding to the pre-order data segment included in the target pre-order data segment bucket and the real-time feature scalar.

[0059] In the embodiment of the present application, step S30D2 determines the core point of the target data segment corresponding to the bucket according to the pre-order feature scalar corresponding to the pre-order data segment in the target pre-order data segment bucket and the real-time feature scalar corresponding to the real-time data segment. The pre-order feature scalar is obtained by calculating the inner product of the data segment vector corresponding to the pre-order data segment and its reverse vector, which quantifies the main features of the patient's vital signs data in the pre-order data segment. The real-time feature scalar is obtained by calculating the inner product of the data segment vector corresponding to the real-time data segment in the current data segment bucket and its reverse vector, reflecting the patient's latest vital sign state. In step S30D2, the computer system first obtains the pre-order feature scalars corresponding to all pre-order data segments in the target pre-order data segment bucket, and the real-time feature scalars corresponding to all real-time data segments in the current data segment bucket. These feature scalars provide rich information about the patient's vital signs data and are an important basis for determining the core point of the target data segment. Next, the computer system integrates these feature scalars to determine the core point of the target data segment. The specific integration method can be determined according to the specific application scenario and data characteristics. A common method is the weighted average method, that is, different weights are assigned to the previous data segment and the real-time data segment according to their importance or time sequence, and then the weighted average is calculated as the coordinate value of the core point of the target data segment. For example, if the real-time data segment is considered to better reflect the patient's current state, a higher weight can be assigned to the real-time feature scalar. In addition to the weighted average method, other more complex algorithms or models can also be used to determine the core point of the target data segment. For example, clustering algorithms in machine learning (such as K-means, DBSCAN, etc.) can be used, with the previous feature scalar and the real-time feature scalar as input, and the optimal clustering center is found through iterative calculation as the core point of the target data segment. This method can automatically discover hidden patterns in the data and generate more accurate core points.

[0060] Assume that in a specific nursing data analysis scenario, the target pre-order data segment bucket contains the patient's heart rate data. The pre-order feature scalar of the heart rate corresponding to the pre-order data segment represents the stable state of the patient's heart rate over the past period of time, while the real-time feature scalar of the heart rate corresponding to the real-time data segment reflects the patient's latest heart rate change. The computer system can use the weighted average method to assign a higher weight (such as 0.7) to the real-time feature scalar and a lower weight (such as 0.3) to the pre-order feature scalar, and then calculate the weighted average as the heart rate coordinate value of the core point of the target data segment. The core point of the target data segment obtained in this way will comprehensively consider the patient's historical heart rate data and the latest heart rate changes, and can more accurately reflect the patient's current physiological state.

[0061] Through step S30D2, the core point of the target data segment can be determined according to the previous feature scalar and the real-time feature scalar, which provides an important reference point for subsequent similarity calculation and anomaly detection.

[0062] Step S30D3: If the target preceding data segment bucket in the adjusted preceding data segment bucket combination only has multiple current data segments in the current data segment bucket, then obtain the data segment vectors corresponding to the multiple real-time data segments in the current data segment bucket respectively, and multiply them by the vector modulus product of the inverse vector of the data segment vector, and use the sum of the vector modulus products corresponding to the multiple real-time data segments as the real-time feature scalar corresponding to the multiple real-time data segments.

[0063] In the embodiment of the present application, step S30D3 occurs when a target preceding data segment bucket in the adjusted preceding data segment bucket combination contains only multiple real-time data segments in the current data segment bucket. In this case, since there is no preceding data segment as a reference, the computer system needs to rely only on the information of the current data segment to determine the real-time feature scalar of the bucket. Specifically, when the target preceding data segment bucket contains only the real-time data segment in the current data segment bucket, the computer system will perform the following operations: First, for each real-time data segment in the target preceding data segment bucket, the computer system will obtain its corresponding data segment vector. The data segment vector is a multidimensional vector, each dimension represents a physiological indicator, such as heart rate, blood pressure, blood oxygen saturation, etc., and the value of the vector is the specific monitoring value of these indicators. For example, the data segment vector of a real-time data segment may be expressed as [heart rate = 95, blood pressure = 110 / 70, blood oxygen saturation = 98%]. Then, the computer system will calculate the vector module product of each data segment vector and its reverse vector, that is, the inner product. Since the data segment vector is a column vector here, its reverse vector (i.e., transposition) is actually itself. Therefore, the inner product calculation is essentially the weighted sum of the values ​​of each dimension of the data segment vector, where the weight of each dimension defaults to 1 (i.e., no specific weighting processing is performed). This inner product result quantitatively represents the comprehensive characteristics of each physiological index in the real-time data segment, forming a real-time feature scalar. Then, for all real-time data segments in the target pre-order data segment bucket, the computer system will perform the above operations to obtain the real-time feature scalar corresponding to each real-time data segment. Since these real-time data segments all belong to the same bucket, their real-time feature scalars will jointly reflect the overall characteristics of the data in the bucket. Finally, the computer system can integrate these real-time feature scalars in some form to obtain a comprehensive real-time feature scalar representing the entire bucket. The integration method can be determined according to the specific application scenario and requirements, for example, the average or weighted average of all real-time feature scalars can be calculated. This comprehensive real-time feature scalar will serve as an important basis for determining the core point of the target data segment in the subsequent steps.

[0064] Step S30D4: Determine the target data segment core point corresponding to the target preceding data segment bucket according to the real-time feature scalar; wherein the sum of the vector intervals between the data segment vectors corresponding to the multiple data segments in the target preceding data segment bucket and the core point vector corresponding to the target data segment core point is not greater than the preset interval value.

[0065] In step S30D4 of the embodiment of the present application, the core point of the target data segment is determined based on the real-time feature scalar, and it is ensured that the vector interval between the data segment vectors corresponding to the multiple data segments in the target preceding data segment bucket and the core point vector corresponding to the core point of the target data segment satisfies certain constraints. In step S30D3, the computer system has calculated the inner product of the data segment vector corresponding to each real-time data segment in the target preceding data segment bucket and its reverse vector, and obtained the real-time feature scalar. These real-time feature scalars quantify the main features of the patient's vital sign data in the real-time data segment.

[0066] Next, in step S30D4, the computer system needs to determine the core point of the target data segment based on these real-time feature scalars. The specific method can be selected according to the characteristics of the data and business requirements. A common method is to calculate the average or weighted average of all real-time feature scalars and use the result as the coordinate value of the core point of the target data segment. For example, if it is believed that all real-time data segments are of equal importance, the arithmetic mean of the real-time feature scalars can be directly calculated; if different weights need to be assigned according to the time sequence of the data or other factors, the weighted average method can be used. In addition to determining the coordinate value of the core point of the target data segment, step S30D4 also introduces an important constraint: the sum of the vector intervals between the data segment vectors corresponding to the multiple data segments in the target preceding data segment bucket and the core point vector corresponding to the core point of the target data segment is not greater than the preset interval value. This constraint ensures that the core point of the target data segment can better represent all the data segments in the bucket and will not deviate too much from any data segment.

[0067] Specifically, for each data segment in the target pre-order data segment bucket, the computer system will calculate the vector interval (such as Euclidean distance) between the corresponding data segment vector and the core point vector corresponding to the core point of the target data segment. Then, the vector intervals of all data segments are added together to obtain a total vector interval. This total vector interval needs to meet the condition of not greater than the preset interval value. If the calculated total vector interval exceeds the preset interval value, it means that the currently determined core point of the target data segment may not be suitable, and the algorithm or parameters need to be readjusted to find a more suitable core point.

[0068] For example, suppose that in the nursing data analysis, the focus is on the heart rate data of a patient. The target pre-order data segment bucket contains three real-time data segments, and their corresponding data segment vectors are [heart rate = 90], [heart rate = 95] and [heart rate = 100] respectively. After calculation, the real-time feature scalars of these three real-time data segments are 90, 95 and 100 respectively. If the arithmetic mean method is used to determine the core point of the target data segment, the coordinate value of the core point will be (90+95+100) / 3=95. Next, it is necessary to calculate the vector interval between the three data segment vectors and the core point vector (i.e. [heart rate = 95]) (here it is actually the absolute value of the difference in heart rate values) and add them together. If the total vector interval obtained is less than the preset interval value (such as 5), it means that the core point [heart rate = 95] is appropriate; otherwise, it is necessary to adjust the algorithm or parameters to re-determine the core point. Through step S30D4, the computer system can determine the core point of the target data segment based on the real-time feature scalar, and ensure that the core point can better represent all data segments in the target pre-order data segment bucket. This will provide a benchmark for subsequent similarity calculations and anomaly detection, helping medical staff to more accurately understand the changing trends of patients' physiological states and detect potential health problems in a timely manner.

[0069] In a possible implementation, step S30D, when the data segment core points corresponding to the preceding data segment buckets in the adjusted preceding data segment bucket combination are used as the target data segment core points in the target data segment core point group, further includes:

[0070] Step S30D5: If the target preceding data segment bucket in the adjusted preceding data segment bucket combination includes multiple preceding data segments and multiple real-time data segments in the current data segment bucket, then obtain the vector interval addition result of the vector intervals of the core point vector corresponding to the core point of the current data segment, respectively, of the data segment vectors of the multiple real-time data segments in the current data segment bucket.

[0071] In the embodiment of the present application, the main purpose of step S30D5 is to quantify the degree of deviation between the real-time data segment in the current data segment and the core point of the current data segment, and provide a reference for the subsequent determination of the core point of the target data segment. When there is a target pre-order data segment bucket in the adjusted pre-order data segment bucket combination, which contains multiple pre-order data segments and multiple real-time data segments in the current data segment bucket, step S30D5 will be executed. The purpose of this step is to calculate the vector interval between the real-time data segment in the current data segment and the core point of the current data segment, and add these vector intervals to obtain a total deviation value.

[0072] Specifically, the computer system first obtains the data segment vector corresponding to each real-time data segment in the current data segment bucket. The data segment vector is a multidimensional vector, and each dimension represents the value of a physiological indicator. For example, the data segment vector of a real-time data segment may be [heart rate = 100, blood pressure = 120 / 80]. Next, the computer system obtains the core point vector corresponding to the core point of the current data segment. The core point vector is also a multidimensional vector, which represents the central trend or main feature of the data in the current data segment. For example, if the current data segment mainly reflects the trend of the patient's heart rate increasing, then the core point vector may have a higher value in the heart rate dimension. Then, for each real-time data segment in the current data segment bucket, the computer system calculates the vector interval between its data segment vector and the core point vector. This vector interval can be calculated by a variety of methods, such as Euclidean distance, cosine similarity, etc. Here, the Euclidean distance is used as an example for explanation. The Euclidean distance measures the straight-line distance between two vectors in a multidimensional space, and can reflect the similarity or deviation between the two vectors. Finally, the computer system adds the vector intervals between all real-time data segments and the core point vectors to obtain a total vector interval addition result. This addition result quantitatively represents the overall deviation between the real-time data segment in the current data segment and the core point of the current data segment. If the addition result is small, it means that the real-time data segment is closer to the core point and the data distribution is more concentrated; if the addition result is large, it means that the real-time data segment deviates far from the core point, and there may be anomalies or the data distribution is more dispersed. By executing step S30D5, the computer system can quantify the deviation between the real-time data segment in the current data segment and the core point of the current data segment, providing a reference for the subsequent determination of the core point of the target data segment. This will help to more accurately reflect the trend of changes in the patient's physiological state and discover potential health problems in a timely manner.

[0073] Step S30D6: Obtain the reference core point vector interval corresponding to the target preceding data segment bucket, and obtain the first importance factor corresponding to the core point of the current data segment, and the second importance factor corresponding to the core point of the preceding data segment.

[0074] In the embodiment of the present application, step S30D6 obtains multiple important parameters related to the core point of the data segment. These parameters not only help to quantify the degree of deviation of the data, but also can be weighted in subsequent calculations according to the real-time and historical importance of the data. The reference core point vector interval is a preset threshold used to measure whether the degree of deviation between the data segment and the core point is significant. The importance factor is a set of weight values ​​used to perform weighted processing in subsequent calculations according to the real-time and historical importance of the data.

[0075] In nursing data analysis, suppose we are interested in a patient's vital signs data, such as heart rate, blood pressure, etc. These data are monitored in real time and divided into different data segments, each of which contains monitoring data within a certain time window. The target previous data segment bucket is a collection of multiple previous data segments and real-time data segments in the current data segment bucket.

[0076] For the target preceding data segment bucket, the computer system first obtains the corresponding reference core point vector interval. This threshold is pre-set based on historical data and business requirements, and is used to determine whether the deviation between the data segment and the core point has reached a level that requires attention. For example, in heart rate data analysis, the reference core point vector interval may be set to 5 units, indicating that when the core point vector interval between the real-time heart rate data and the preceding heart rate data exceeds 5 units, attention should be paid. Next, the computer system obtains the first importance factor corresponding to the core point of the current data segment. This factor represents the importance of the current data segment in the overall data analysis. Since the current data segment contains the patient's latest vital sign data, it is usually given a higher weight. For example, the first importance factor may be set to 0.7, indicating that in subsequent calculations, the deviation of the current data segment will be given a greater weight. Finally, the computer system obtains the second importance factor corresponding to the core point of the preceding data segment. This factor represents the importance of the preceding data segment in the overall data analysis. Although the preceding data segment contains the patient's historical data, it is still of great significance for judging the patient's current status and future trends. However, due to its relatively low real-time nature, it is usually given a lower weight. For example, the second importance factor may be set to 0.3, indicating that in subsequent calculations, the deviation of the previous data segment will be given a smaller weight.

[0077] By executing step S30D6, the computer system obtains the reference core point vector interval and two importance factors, which provide important parameter basis for the subsequent calculation of the reference core point vector interval corresponding to the core point of the target data segment. These parameters not only help to quantify the degree of deviation of the data, but also perform weighted processing according to the real-time and historical importance of the data, thereby more accurately reflecting the trend of changes in the patient's physiological state.

[0078] Step S30D7: Determine the reference core point vector interval corresponding to the core point of the target data segment by a first weighted calculation result of the sum of the first importance factor and the vector interval, and a second weighted calculation result of the second importance factor and the reference core point vector interval.

[0079] In the embodiment of the present application, step S30D7 combines the importance of real-time data and historical data to determine the reference core point vector interval corresponding to the core point of the target data segment. This step comprehensively considers the degree of deviation between the real-time data segment and the core point of the current data segment (quantified by the vector interval addition result) and the importance of the previous data segment (represented by the reference core point vector interval) through weighted calculation, thereby obtaining a more accurate and comprehensive reference core point vector interval.

[0080] Suppose that in nursing data analysis, the focus is on the heart rate data of a patient. The real-time data segment contains the patient's latest heart rate data, while the previous data segment contains the patient's heart rate data over the past period of time. The core point of the current data segment represents the central trend of the heart rate in the real-time data segment, while the core point of the previous data segment represents the central trend of the heart rate in the historical data segment.

[0081] The computer system has calculated the vector interval addition result between the real-time data segment and the core point of the current data segment according to step S30D5. This addition result quantitatively represents the overall deviation between the real-time data segment and the core point of the current data segment. For example, if the heart rate of the real-time data segment is generally higher than the heart rate value represented by the core point of the current data segment, then the vector interval addition result will be larger. Then, the computer system obtains the first importance factor corresponding to the core point of the current data segment and the second importance factor corresponding to the core point of the previous data segment. These importance factors represent the importance of real-time data and historical data in the overall data analysis. Since real-time data can reflect the current state of the patient, it is usually given a higher weight (i.e., the first importance factor is larger); and although the previous data contains the patient's historical information, it is still of great significance for judging the patient's current state and future trends, but the weight is relatively low (i.e., the second importance factor is smaller). Then, the computer system determines the reference core point vector interval corresponding to the core point of the target data segment through weighted calculation. Specifically, it first multiplies the first importance factor by the sum of the vector intervals to obtain the weighted deviation degree of the real-time data segment; then multiplies the second importance factor by the reference core point vector interval to obtain the weighted deviation threshold of the previous data segment. Finally, the two weighted results are added to obtain the final reference core point vector interval.

[0082] Through the weighted calculation in step S30D7, the computer system obtains a reference core point vector interval that comprehensively considers the importance of real-time data and historical data. This interval value will serve as an important basis for subsequent judgment of whether the real-time data segment is abnormal or deviates from expectations, which will help medical staff to more accurately understand the trend of changes in the patient's physiological state and discover potential health problems in a timely manner.

[0083] Step S30D8: If the target preceding data segment bucket in the adjusted preceding data segment bucket combination only has multiple current data segments in the current data segment bucket, obtain the vector interval addition result of the vector intervals of the data segment vectors of each of the multiple real-time data segments in the current data segment bucket and the core point vector corresponding to the core point of the current data segment.

[0084] In an embodiment of the present application, when a target preceding data segment bucket in the adjusted preceding data segment bucket combination contains only real-time data segments in the current data segment bucket, this step will be executed. This situation usually occurs when there is not enough historical data for reference, or when the characteristics of the real-time data segment are significantly different from the historical data. Assume that in nursing data analysis, the focus is on the patient's vital signs data, such as heart rate, blood pressure, etc. These data are monitored in real time and divided into different data segments, each of which contains monitoring data within a certain time window. The current data segment bucket contains the patient's latest real-time data segment, while the preceding data segment bucket contains the patient's historical data segments over a period of time. However, in some cases, due to various reasons (such as the patient has just been admitted to the hospital, the monitoring equipment has just been activated, etc.), there may not be enough or suitable historical data for reference in the preceding data segment bucket. At this time, the target preceding data segment bucket will only contain the real-time data segments in the current data segment bucket. In this case, the computer system obtains the data segment vector corresponding to each real-time data segment in the current data segment bucket. These data segment vectors are multidimensional vectors, each dimension representing the value of a physiological indicator (such as heart rate, blood pressure, etc.). For example, the data segment vector of a real-time data segment may be [heart rate = 100, blood pressure = 120 / 80]. Next, the computer system obtains the core point vector corresponding to the core point of the current data segment. The core point vector is also a multidimensional vector, which represents the central trend or main feature of the data in the current data segment. For example, if the current data segment mainly reflects the trend of the patient's heart rate increasing, then the core point vector may have a higher value in the heart rate dimension. Then, for each real-time data segment in the current data segment bucket, the computer system will calculate the vector interval between its data segment vector and the core point vector. This vector interval can be calculated by a variety of methods, such as Euclidean distance, cosine similarity, etc. Here, the Euclidean distance is used as an example for explanation. The Euclidean distance measures the straight-line distance between two vectors in a multidimensional space and can reflect the similarity or deviation between the two vectors. Finally, the computer system adds the vector intervals between all real-time data segments and the core point vectors to obtain a total vector interval addition result. This addition result quantitatively represents the overall deviation between the real-time data segments in the current data segment and the core point of the current data segment. If the addition result is small, it means that the real-time data segment is close to the core point and the data distribution is relatively concentrated; if the addition result is large, it means that the real-time data segment deviates far from the core point, and there may be anomalies or the data distribution is relatively scattered. By executing step S30D8, the computer system can determine the degree of deviation between the real-time data segment and the core point of the current data segment only based on the real-time data segment without sufficient historical data reference. This helps medical staff to promptly discover potential health problems or data anomalies, so as to take corresponding nursing measures. At the same time, this analysis method based on real-time data also provides a flexible and effective method for nursing data analysis.

[0085] Step S30D9: The result of adding the vector intervals is used as the reference core point vector interval corresponding to the core point of the target data segment.

[0086] In an embodiment of the present application, step S30D9 occurs when a target preceding data segment bucket in the adjusted preceding data segment bucket combination contains only the real-time data segments in the current data segment bucket. In this case, since there is no preceding data segment to provide a historical reference, the computer system will directly rely on the real-time data segment to determine the reference core point vector interval corresponding to the core point of the target data segment. Specifically, when the target preceding data segment bucket contains only the real-time data segments of the current data segment bucket, it means that there is not enough historical data or preceding data segments to form a stable reference frame. Therefore, the computer system needs to find a method to quantify the degree of deviation between these real-time data segments and the core points for subsequent anomaly detection or trend analysis.

[0087] First, the computer system obtains the data segment vector corresponding to each real-time data segment in the current data segment bucket. These data segment vectors represent the patient's vital signs data monitored in real time, such as heart rate, blood pressure, etc. Each data segment vector is a multidimensional vector, which contains the specific values ​​of multiple physiological indicators. Then, the computer system calculates the vector interval between the data segment vector of each real-time data segment and the core point vector corresponding to the core point of the current data segment. This vector interval reflects the similarity or degree of deviation between the real-time data segment and the core point. In nursing data analysis, methods such as Euclidean distance or cosine similarity are usually used to calculate the vector interval. Then, the computer system adds the vector intervals between all real-time data segments and the core point vectors to obtain a total vector interval addition result. This addition result quantifies the overall deviation between all real-time data segments and the core point in the current data segment. If the addition result is small, it means that the real-time data segment is close to the core point and the data distribution is more concentrated; if the addition result is large, it means that the real-time data segment deviates far from the core point, and there may be anomalies or the data distribution is more dispersed.

[0088] In step S30D9, the vector interval addition result will be directly used as the reference core point vector interval corresponding to the core point of the target data segment. This means that in the absence of a previous data segment to provide a historical reference, the computer system will directly rely on the degree of deviation between the real-time data segment and the core point to determine whether the data is normal or abnormal.

[0089] Through step S30D9, the computer system can determine the reference core point vector interval corresponding to the core point of the target data segment by relying only on the real-time data segment without the reference of the previous data segment. This method provides a flexible and practical solution for nursing data analysis, especially for special cases such as newly admitted patients or monitoring equipment that has just been activated.

[0090] In a possible implementation, step S30 performs similarity calculations on each real-time data segment in the real-time data segment combination and multiple target data segment core points in the target data segment core point group to obtain a target data segment vector interval corresponding to each real-time data segment, including:

[0091] Step S31: Obtaining interval importance factors respectively corresponding to a plurality of target data segment core points in the target data segment core point group.

[0092] In the embodiment of the present application, step S31 assigns an interval importance factor to each core point of the target data segment. This importance factor will play an important role in the subsequent calculation of the vector interval between the real-time data segment and the core point of the target data segment, helping the computer system to perform weighted processing according to the different importance of each core point.

[0093] The target data segment core point group is obtained after a series of processing and integration. It contains multiple target data segment core points, each of which represents a central trend or main feature of the data segment. In nursing data analysis, these core points may represent different physiological states or data distribution patterns of patients.

[0094] In step S31, the computer system needs to assign an interval importance factor to each target data segment core point. This importance factor is actually a weight value used to quantify the importance or influence of the core point. The specific value of the importance factor can be determined based on a variety of factors, such as the historical performance of the core point, real-time performance, data quality, etc.

[0095] Assume that in the scenario of heart rate data analysis, the target data segment core point group contains two core points, representing the two main modes of heart rate data: stable mode and fluctuating mode. The first core point (stable mode) corresponds to a more concentrated range of heart rate values ​​and has a higher frequency of occurrence in historical data, so it may be regarded as a more important core point. Although the second core point (fluctuating mode) also represents a mode of heart rate data, it may be regarded as a relatively less important core point because of its larger range of variation and lower frequency of occurrence.

[0096] In this case, the computer system can assign a higher interval importance factor (such as 0.7) to the first core point and a lower interval importance factor (such as 0.3) to the second core point. This means that when the vector interval between the real-time data segment and the core point is subsequently calculated, the degree of deviation from the first core point (stable mode) will be given a greater weight, while the degree of deviation from the second core point (fluctuating mode) will be given a smaller weight. In this way, the computer system can more accurately reflect the importance of different core points in data analysis, providing a more meaningful reference for subsequent anomaly detection and trend analysis. This helps medical staff understand the changing trends of patients' physiological states more accurately and detect potential health problems in a timely manner.

[0097] Get a real-time data segment in the real-time data segment combination as the current real-time data segment, and execute the following steps in a loop until all the real-time data segments in the real-time data segment combination are executed:

[0098] Step S32: respectively calculating the reference data segment vector intervals between the current real-time data segment and the core points of multiple target data segments.

[0099] In the embodiment of the present application, step S32 provides a quantitative basis for subsequent anomaly detection or trend analysis by calculating the reference data segment vector interval between the current real-time data segment and the core points of multiple target data segments.

[0100] The real-time data segment combination includes a plurality of real-time data segments arranged in chronological order, and each data segment represents the vital sign data of the patient within a specific time window. The target data segment core point group is a set composed of a plurality of target data segment core points, and each core point represents a central trend or main feature of the data segment. In step S32, the computer system selects a real-time data segment from the real-time data segment combination as the current real-time data segment. Then, for each core point in the target data segment core point group, the reference data segment vector interval between the current real-time data segment and the core point is calculated respectively.

[0101] The calculation of this vector interval is usually based on some similarity measurement method, such as Euclidean distance or cosine similarity. Taking Euclidean distance as an example, assume that the current real-time data segment is a multidimensional vector containing the values ​​of multiple physiological indicators (such as heart rate, blood pressure, blood oxygen saturation, etc.). Similarly, the core point of the target data segment is also a multidimensional vector, representing the main features of the data segment corresponding to the core point. The system calculates the straight-line distance (i.e., Euclidean distance) between the two vectors in multidimensional space to obtain the vector interval between the current real-time data segment and the core point. This vector interval quantitatively represents the similarity or deviation between the core point of the current real-time data segment and the core point of the target data segment. If the vector interval is small, it means that the current real-time data segment is closer to the core point of the target data segment, and the data distribution is more concentrated; conversely, if the vector interval is large, it means that the current real-time data segment deviates far from the core point of the target data segment, and there may be anomalies or the data distribution is more dispersed.

[0102] In order to understand the execution process of step S32 more specifically, an example can be given. Assume that in the nursing data analysis, the focus is on the heart rate data of a patient. The real-time data segment combination contains the patient's latest three real-time heart rate data segments, namely [heart rate 90], [heart rate 95], and [heart rate 100]. The target data segment core point group contains two core points, which correspond to the two main modes of heart rate data: stable mode (heart rate value is about 92) and slightly fluctuating mode (heart rate value is about 98). For the real-time data segment [heart rate 90], the system first calculates the Euclidean distance (i.e., the reference data segment vector interval) between it and the core point of the stable mode, and assumes that the result is 2. Then, the system calculates the Euclidean distance between [heart rate 90] and the core point of the slightly fluctuating mode, and assumes that the result is 8. These two vector intervals quantify the degree of deviation between the real-time data segment [heart rate 90] and the two target data segment core points. Then, step S32 is repeated for other real-time data segments in the real-time data segment combination until all real-time data segments have completed the vector interval calculation between the core points of the target data segments. These vector intervals will provide important quantitative basis for subsequent abnormality detection or trend analysis, helping medical staff to more accurately understand the changing trends of patients' physiological states.

[0103] Step S33: Determine the target data segment vector interval corresponding to the current real-time data segment through the interval importance factors corresponding to the core points of the multiple target data segments and the obtained multiple reference data segment vector intervals.

[0104] In the embodiment of the present application, step S33 comprehensively considers the degree of deviation and importance between the current real-time data segment and the core points of multiple target data segments by weighted calculation, thereby obtaining a more comprehensive and accurate vector interval value. The reference data segment vector interval is obtained by calculating the vector distance between the current real-time data segment and the core point of the target data segment, which quantitatively represents the degree of deviation between the two. The interval importance factor is a weight value assigned according to the importance and influence of the core point of the target data segment, which is used for weighted processing in subsequent calculations.

[0105] In step S33, the computer system first obtains the interval importance factors and reference data segment vector intervals corresponding to the core points of multiple target data segments. These values ​​are calculated through the previous steps and provide a quantitative basis for the similarity or deviation between the current real-time data segment and the core points of the target data segment.

[0106] Next, these values ​​are weighted. Specifically, each reference data segment vector interval is multiplied by the interval importance factor of the corresponding target data segment core point to obtain a weighted deviation value. This weighted calculation process actually weighs the importance of different core points, so that more important core points have a greater impact when determining the target data segment vector interval. Finally, all weighted deviation values ​​are summarized to obtain a comprehensive target data segment vector interval. This vector interval value comprehensively considers the degree of deviation and importance between the current real-time data segment and multiple target data segment core points, so it can more comprehensively and accurately reflect the overall deviation of the current real-time data segment. In step S33, a weighted calculation is performed: the weighted deviation value of the core point in the stable mode is 2*0.7=1.4, and the weighted deviation value of the core point in the slight fluctuation mode is 8*0.3=2.4. Then, the system adds these two weighted deviation values ​​to obtain a target data segment vector interval of 1.4+2.4=3.8. This vector interval value comprehensively considers the degree of deviation and importance between the core points of the current real-time data segment and the two target data segments, and can therefore more accurately reflect the overall deviation of the current real-time data segment.

[0107] In a possible implementation, step S32, respectively calculating the reference data segment vector intervals between the current real-time data segment and the core points of the plurality of target data segments, includes:

[0108] Step S321: sequentially obtain a target data segment core point from multiple target data segment core points, take it as a current data segment core point, and execute the following steps in a loop until all the target data segment core points are executed:

[0109] Step S322: Obtain a current data segment core point sub-combination corresponding to the current data segment core point, wherein the current data segment core point sub-combination includes the current data segment core point and multiple reference data segment core points, and the reference data segment core point is a target data segment core point that is before the current data segment core point.

[0110] In step S321, the computer system sequentially obtains a core point from multiple target data segment core points and regards it as the current data segment core point. This process is executed in a loop to ensure that each target data segment core point can be processed. This sequential processing method conforms to the characteristics of time series data, because each point in the time series data has a dependency relationship with its previous data point.

[0111] For example, suppose the target data segment core points are arranged in chronological order as C1, C2, C3, and C4. In the first cycle, C1 will be regarded as the current data segment core point; in the second cycle, C2 becomes the current data segment core point, and so on until C4.

[0112] In step S322, for each current data segment core point, the computer system constructs a current data segment core point sub-combination. This sub-combination includes not only the current data segment core point, but also multiple reference data segment core points. The reference data segment core point is the target data segment core point that is before the current data segment core point in the time series.

[0113] Continuing with the above example, when C2 is used as the core point of the current data segment, since C1 is before C2, C1 is the reference core point of C2. Therefore, the current core point sub-combination of C2 includes C2 and C1. Similarly, when processing C3, its reference core points are C1 and C2, so the current core point sub-combination of C3 includes C3, C2 and C1.

[0114] This way of constructing sub-combinations takes into account the dependencies and correlations in time series data. In nursing data analysis, the patient's physiological state often changes continuously, and the current state is often related to the previous one or several previous states. Therefore, combining the core points of the current data segment with the core points of the previous reference data segment for analysis can more accurately reflect the degree of deviation of the real-time data segment.

[0115] Step S323: respectively calculating the Chebyshev distances between the data segment vector corresponding to the current real-time data segment and the core point vectors corresponding to the multiple data segment core points in the current data segment core point sub-combination.

[0116] In the embodiment of the present application, the purpose of step S323 is to quantify the degree of deviation between the real-time data segment and different core points, so as to provide an important basis for subsequent data analysis and anomaly detection.

[0117] In step S323, the computer system first obtains the data segment vector corresponding to the current real-time data segment. Then, for each data segment core point in the current data segment core point sub-combination, the system calculates the Chebyshev distance between the core point vector corresponding to the core point and the data segment vector of the current real-time data segment. Chebyshev distance is a method for calculating the distance between two points in a multidimensional space, which considers the maximum difference in each dimension. In nursing data analysis, since the importance of different physiological indicators may be different, Chebyshev distance can capture the maximum deviation of data in each dimension, thereby more comprehensively reflecting the similarity or deviation between the real-time data segment and the core point.

[0118] Specifically, assuming that the data segment vector corresponding to the current real-time data segment is V_real = [heart rate 100, blood pressure 120 / 80, blood oxygen saturation 95%], the current data segment core point combination includes two core points C1 and C2, and their corresponding core point vectors are V_C1 = [heart rate 95, blood pressure 115 / 75, blood oxygen saturation 97%] and V_C2 = [heart rate 98, blood pressure 125 / 85, blood oxygen saturation 96%].

[0119] For the core point C1, the computer system calculates the Chebyshev distance between the data segment vector V_real and the core point vector V_C1. In this example, the Chebyshev distance will consider the maximum difference in the three dimensions of heart rate, blood pressure and blood oxygen saturation, that is, max(|100-95|, |120-115|, |95-97|) = 5 (here it is assumed that the difference in blood pressure values ​​is the difference between the average values ​​of systolic pressure and diastolic pressure).

[0120] Similarly, for core point C2, the system calculates the Chebyshev distance between the data segment vector V_real and the core point vector V_C2. This distance value will reflect the degree of deviation between the real-time data segment and the data segment corresponding to C2. By calculating the Chebyshev distance between the current real-time data segment and the core points of multiple data segments, the computer system can quantify the degree of deviation between the real-time data segment and different core points. These distance values ​​will provide an important basis for subsequent data analysis and anomaly detection, helping medical staff to more accurately understand the changing trends of the patient's physiological state and promptly discover potential health problems.

[0121] Step S324: taking the Chebyshev distance with the smallest value as the reference data segment vector interval between the current real-time data segment and the core point of the current data segment.

[0122] In the embodiment of the present application, step S324 selects the one with the smallest value as the reference data segment vector interval based on the multiple Chebyshev distances calculated in the previous step S323. This process aims to find which data segment core point the current real-time data segment is closest to, so as to quantify the degree of deviation of the real-time data segment. Specifically, in step S323, the computer system has calculated the Chebyshev distance between the data segment vector corresponding to the current real-time data segment and the core point vector corresponding to each data segment core point in the current data segment core point sub-combination. These distance values ​​reflect the degree of deviation between the real-time data segment and different core points. Then, in step S324, the computer system selects the one with the smallest value from multiple Chebyshev distances. This minimum distance represents the degree of deviation between the current real-time data segment and the most similar data segment core point in the current data segment core point sub-combination. Since the Chebyshev distance considers the maximum difference in each dimension, the minimum Chebyshev distance can capture the maximum degree of deviation between the real-time data segment and the most similar core point in each physiological index. By executing step S324, the computer system can determine the reference data segment vector interval between the current real-time data segment and the core point of the current data segment. This interval value will serve as an important basis for subsequent data analysis, helping medical staff to more accurately understand the trend of changes in the patient's physiological state and timely discover potential health problems. At the same time, since multiple core points and their previous reference data segment core points are taken into account, this interval value can more comprehensively reflect the deviation degree and context information of the real-time data segment.

[0123] In a possible implementation, step S33, determining the target data segment vector interval corresponding to the current real-time data segment by using the interval importance factors corresponding to the core points of the multiple target data segments and the obtained multiple reference data segment vector intervals, includes:

[0124] Step S331: Based on the interval importance factors corresponding to the core points of multiple target data segments, the importance integration results of multiple reference data segment vector intervals are calculated to obtain the first data segment vector interval corresponding to the current real-time data segment.

[0125] In an embodiment of the present application, step S331 combines the interval importance factors corresponding to multiple target data segment core points with the reference data segment vector interval to obtain a comprehensive, weighted result, namely, the first data segment vector interval corresponding to the current real-time data segment. The target data segment core points are the central trends or main feature points of the data segments obtained through a series of data processing steps, which represent different modes or states in the nursing data. The interval importance factor is the weight value associated with each target data segment core point, which reflects the importance of the core point in data analysis. The reference data segment vector interval is obtained by calculating the vector distance between the real-time data segment and the core point, which quantitatively represents the degree of deviation between the two.

[0126] In step S331, the computer system first obtains the interval importance factors and reference data segment vector intervals corresponding to multiple target data segment core points. Then, the computer system uses the interval importance factors to perform weighted summation on the reference data segment vector intervals. Specifically, for each target data segment core point, the system multiplies its corresponding interval importance factor with the reference data segment vector interval to obtain a weighted deviation value. This process actually weighs the influence of different core points, so that more important core points have a greater contribution in determining the first data segment vector interval.

[0127] Finally, all weighted deviation values ​​are summed up to obtain a comprehensive importance integration result, that is, the first data segment vector interval corresponding to the current real-time data segment. This vector interval value comprehensively considers the deviation degree and importance between the real-time data segment and the core points of multiple target data segments, so it can more comprehensively reflect the overall deviation of the real-time data segment.

[0128] In order to understand the execution process of step S331 more specifically, an example can be given. Assume that in the nursing data analysis, the focus is on the heart rate data of a patient. The heart rate of the current real-time data segment is 90 beats / minute, and the target data segment core point group contains two core points: a stable mode core point (heart rate is about 92 beats / minute) and a slightly fluctuating mode core point (heart rate is about 98 beats / minute). Through calculation, it is obtained that the reference data segment vector interval between the current real-time data segment and the stable mode core point is 2, and the reference data segment vector interval with the slightly fluctuating mode core point is 8. At the same time, a higher interval importance factor of 0.7 is assigned to the stable mode core point, and a lower interval importance factor of 0.3 is assigned to the slightly fluctuating mode core point.

[0129] In step S331, the system will perform weighted calculation: the weighted deviation value of the core point in the stable mode is 2*0.7=1.4, and the weighted deviation value of the core point in the slightly fluctuating mode is 8*0.3=2.4. Then, the system adds these two weighted deviation values ​​to obtain the first data segment vector interval of 1.4 + 2.4 = 3.8. This vector interval value comprehensively considers the deviation degree and importance between the current real-time data segment and the core points of the two target data segments, so it can more accurately reflect the overall deviation of the current real-time data segment.

[0130] Step S332: scaling the first data segment vector interval according to the importance integration result corresponding to each real-time data segment to obtain the target data segment vector interval corresponding to the current real-time data segment.

[0131] In the embodiment of the present application, the main purpose of step S332 is to perform data scaling on the first data segment vector interval calculated previously, so as to obtain the target data segment vector interval corresponding to the current real-time data segment. This process helps to standardize the vector interval values ​​between different real-time data segments so that they can be compared and analyzed on the same scale. In step S331, the computer system calculates the first data segment vector interval corresponding to the current real-time data segment based on the interval importance factors and reference data segment vector intervals corresponding to the core points of multiple target data segments. This vector interval value comprehensively considers the degree of deviation and importance between the real-time data segment and the core points of multiple target data segments, but it is still an original value and may vary depending on the real-time data segment. In order to enable these vector interval values ​​to be compared on the same scale, step S332 introduces the concept of data scaling. Data scaling is a common data preprocessing method, which transforms the original data so that the transformed data has specific attributes or ranges. Here, the main purpose of data scaling is to map the first data segment vector interval to a fixed numerical range (such as normalized to between 0 and 1) for comparison and analysis across data segments.

[0132] Specifically, in step S332, the computer system first obtains the first data segment vector intervals corresponding to all real-time data segments. Then, it determines a scaling range or target range (such as normalized to between 0 and 1). Next, the system calculates the maximum and minimum values ​​of all first data segment vector intervals to determine the range of the original data. Finally, the system uses a scaling formula to map each first data segment vector interval value into the target range to obtain the corresponding target data segment vector interval.

[0133] The specific form of the scaling formula depends on the selected scaling method. In the case of normalization, the scaling formula is usually as follows: target data segment vector interval = (first data segment vector interval - minimum value) / (maximum value - minimum value). In this way, the vector interval values ​​between different real-time data segments can be compared on the same scale.

[0134] In a possible implementation, after determining the target data segment vector interval corresponding to the current real-time data segment by using the interval importance factors corresponding to the core points of the multiple target data segments and the obtained multiple reference data segment vector intervals, step S33 further includes:

[0135] Sequentially obtain a target data segment core point in the target data segment core point group as the current data segment core point, and execute the following steps in a loop until each target data segment core point in the target data segment core point group is executed;

[0136] Step S34: Obtain the centrality density coefficient of the core point vector corresponding to the core point of the current data segment.

[0137] In an embodiment of the present application, step S34 evaluates the compactness of the distribution of data points around the core point of the target data segment, which is achieved by calculating the centrality density coefficient of the core point vector corresponding to the core point of the current data segment. The centrality density coefficient is a quantitative indicator used to reflect the representativeness or stability of the core point in its neighborhood. In nursing data analysis, each core point of the target data segment corresponds to a core point vector, which represents the main characteristics or central trend of the data segment corresponding to the core point. The core point vector is usually a multidimensional vector that contains the values ​​of multiple physiological indicators, such as heart rate, blood pressure, blood oxygen saturation, etc.

[0138] In step S34, the computer system first selects a core point in the core point group of the target data segment as the core point of the current data segment. Then, it calculates the centrality density coefficient of the core point vector corresponding to the core point. This coefficient measures the centrality of the core point in its neighborhood, that is, the degree of closeness of the surrounding data points around the core point. One method of calculating the centrality density coefficient is to calculate the square value of the modulus of the core point vector. The square value of the modulus is actually the sum of the squares of the values ​​of each dimension of the core point vector, which reflects the "size" or "influence" of the core point vector in multidimensional space. In nursing data analysis, if the square value of the modulus of the core point vector is large, it means that the data segment corresponding to the core point has a large range of numerical changes in multiple physiological indicators, so it may represent a relatively stable or representative data pattern. On the contrary, if the square value of the modulus is small, it means that the data segment corresponding to the core point has a small range of changes in physiological indicators, which may not be stable enough or representative. After calculating the centrality density coefficient, the computer system will use this coefficient to evaluate the importance of the core point of the current data segment, and in subsequent steps, the reference data segment vector interval will be weighted according to this importance factor. In this way, the influence of different core points in data analysis can be reflected more accurately, improving the accuracy and reliability of the analysis.

[0139] Step S35: respectively calculating the interval summation result of the vector intervals between the core point vector corresponding to the core point of the current data segment and the core point vectors corresponding to the remaining target data segment core points in the target data segment core point group.

[0140] In the embodiment of the present application, step S35 involves calculating the vector interval between the core point vector corresponding to the core point of the current data segment and the core point vector corresponding to the remaining core points in the core point group of the target data segment, and summing these interval values ​​to obtain the interval summation result. The purpose of this step is to quantify the overall deviation between the current core point and other core points, and to provide a basis for the subsequent adjustment of the interval importance factor. The target data segment core point group is obtained after a series of data processing steps, and it contains multiple target data segment core points, each core point corresponds to a core point vector, and this vector represents the main characteristics or central trend of the data segment corresponding to the core point. In step S35, the computer system first selects a core point in the target data segment core point group as the current data segment core point. Then, the vector interval between the core point vector corresponding to the core point of the current data segment and the core point vector corresponding to the remaining core points in the core point group of the target data segment is calculated. This vector interval can be calculated by a variety of methods, such as Euclidean distance, cosine similarity, etc., depending on the characteristics of the data and analysis requirements. Here, the Euclidean distance is used as an example for explanation. Euclidean distance can quantify the straight-line distance between two vectors in multidimensional space, thereby reflecting the degree of deviation between two core points. Specifically, assuming that the core point vector corresponding to the core point of the current data segment is V_current, the core point vectors corresponding to the remaining core points in the core point group of the target data segment are V_1, V_2, ..., V_n. The system first calculates the Euclidean distance d_1 between V_current and V_1, then calculates the Euclidean distance d_2 between V_current and V_2, and so on, until the Euclidean distance d_n between V_current and V_n is calculated.

[0141] Finally, add these Euclidean distance values ​​to get the interval summation result Sum_d = d_1 + d_2 + ... + d_n. This summation result quantitatively represents the overall deviation between the core point of the current data segment and other core points. If the interval summation result is small, it means that the current core point is close to other core points and the data distribution is more concentrated; conversely, if the interval summation result is large, it means that the current core point deviates far from other core points, there may be anomalies or the data distribution is more dispersed.

[0142] Step S36: Determine the adjusted interval importance factor of the core point of the current data segment by the ratio of the centrality density coefficient to the interval summation result.

[0143] In the embodiment of the present application, step S36 determines the adjusted interval importance factor of the core point of the current data segment based on the previously calculated central density coefficient and interval summation result. This adjusted interval importance factor will more accurately reflect the importance of the core point in data analysis, thereby helping medical staff to more accurately understand the trend of changes in the patient's physiological state.

[0144] In step S35, the computer system calculates the vector interval between the core point vector corresponding to the core point of the current data segment and the core point vector corresponding to the remaining core points in the core point group of the target data segment, and sums these interval values ​​to obtain the interval summation result. This interval summation result quantitatively represents the overall deviation between the current core point and other core points. At the same time, in step S34, the system calculates the centrality density coefficient of the core point vector corresponding to the current core point, which evaluates the compactness of the distribution of data points around the core point. In step S36, the computer system combines the centrality density coefficient with the interval summation result, and determines the adjusted interval importance factor by calculating their ratio. This ratio is actually a re-evaluation of the importance of the core point. Specifically, if the centrality density coefficient is large (indicating that the data points around the core point are compactly distributed and the core point is representative), and the interval summation result is small (indicating that the current core point does not deviate much from other core points), then the ratio will be large, indicating that the current core point is more important in data analysis. On the contrary, if the centrality density coefficient is small or the interval summation result is large, the ratio will be small, indicating that the importance of the current core point is relatively low. By calculating this ratio and adjusting the interval importance factor accordingly, the computer system can more finely quantify the importance of different core points in data analysis. This will help medical staff more accurately identify key data patterns in subsequent data analysis and anomaly detection, and discover potential health problems in a timely manner.

[0145] Through the processing of step S36, the computer system can more accurately evaluate the importance of different core points in data analysis, thereby improving the accuracy and reliability of data analysis. This will help medical staff better understand the changing trends of patients' physiological states, discover potential health problems in a timely manner, and take corresponding nursing measures.

[0146] In a possible implementation, step S36, determining the adjusted interval importance factor of the core point of the current data segment by the ratio of the centrality density coefficient to the interval summation result, includes:

[0147] Step SI361: If the data segment bucket corresponding to the core point of the current data segment includes the real-time data segment in the real-time data segment combination, the ratio of the centrality density coefficient and the interval summation result is used as the reference importance factor.

[0148] In the embodiment of the present application, step SI361 specifically evaluates the importance of the core point of the current data segment, especially when the data segment bucket corresponding to the core point contains a real-time data segment. This step provides a reference importance factor for the core point associated with the real-time data segment by calculating the ratio of the centrality density coefficient to the interval summation result, reflecting its relative importance in the current data analysis.

[0149] The centrality density coefficient is a quantitative indicator used to evaluate the compactness of the distribution of data points around the core point. It reflects the representativeness or stability of the core point in its neighborhood. The interval summation result is obtained by calculating and summing the vector intervals between the current core point and other core points. It quantitatively represents the overall deviation between the current core point and other core points.

[0150] In step SI361, the computer system first checks whether the data segment bucket corresponding to the core point of the current data segment contains the real-time data segment in the real-time data segment combination. This step is to identify the core points that are directly related to the latest real-time monitoring data, because these core points may better reflect the patient's current physiological state. If the data segment bucket corresponding to the core point of the current data segment does contain the real-time data segment, the system will perform the following operations: calculate the ratio of the centrality density coefficient of the core point vector corresponding to the core point to the interval summation result, and use this ratio as a reference importance factor. This ratio is actually a weighted coefficient that comprehensively considers the centrality of the core point and the degree of deviation from other core points. Specifically, if the centrality density coefficient is large (indicating that the data points around the core point are compactly distributed), and the interval summation result is small (indicating that the current core point does not deviate much from other core points), then the ratio will be large, indicating that the current core point is more important in data analysis.

[0151] For example, suppose that in the heart rate data analysis, there are two target data segment core points C1 and C2. The data segment bucket corresponding to C1 contains the latest real-time heart rate data segment, while the data segment bucket corresponding to C2 does not contain the real-time data segment. Through calculation, the centrality density coefficient of C1 is 0.8, and the interval sum is 3; the centrality density coefficient of C2 is 0.6, and the interval sum is 5. Since the data segment bucket corresponding to C1 contains the real-time data segment, according to step SI361, the reference importance factor of C1 is calculated to be 0.8 / 3 = 0.267.

[0152] Through the processing of step SI361, the computer system can more accurately evaluate the importance of core points directly related to real-time data in data analysis.

[0153] Step SI362: Obtain a first tuning parameter V1, wherein the value of the tuning parameter is not less than 0 and not greater than 1.

[0154] In the embodiment of the present application, the first tuning parameter V1 controls the weight distribution of new and old data when updating the interval importance factor. By adjusting the value of V1, the computer system can flexibly balance the importance of newly acquired data (such as real-time data segments) and historical data (such as previous data segments) in data analysis. V1 is a numerical value whose value range is not less than 0 and not greater than 1. This range ensures that V1 has a certain flexibility in adjusting the weight of new and old data without exceeding reasonable limits.

[0155] In nursing data analysis, the specific value of the tuning parameter V1 can be determined according to the actual application scenario and needs. When the value of V1 is close to 1, it means that the new data (such as the core point corresponding to the real-time data segment) will occupy a greater weight when updating the interval importance factor. This helps the computer system to quickly respond to changes in the patient's physiological state and adjust the analysis strategy in time. On the contrary, when the value of V1 is close to 0, the old data (such as the core point corresponding to the previous data segment) will have a greater influence, which helps to maintain the stability and continuity of data analysis. In practical applications, the setting of the tuning parameter V1 often needs to be adjusted according to the dynamics of the data and business needs. For example, in a scenario where the patient's condition changes dramatically, V1 may need to be set higher so that the computer system can capture these changes faster and respond. In the case of a relatively stable condition, the value of V1 can be appropriately reduced to make full use of historical data to improve the accuracy of the analysis. In addition, it is worth noting that the tuning parameter V1 is associated with the second tuning parameter V2, and their sum is always 1. This setting ensures that the sum of the weights of the new and old data when updating the interval importance factor remains unchanged, thereby avoiding the deviation of the analysis results caused by improper weight distribution. In practical applications, the values ​​of V1 and V2 can be adjusted according to specific needs to achieve the best analysis effect.

[0156] Step SI363: The weighted calculation result of the first tuning parameter and the reference importance factor, and the weighted calculation result of the second tuning parameter V2 and the interval importance factor corresponding to the core point of the current data segment are determined as the adjusted interval importance factor of the core point of the current data segment, where V2=1-V1.

[0157] In the embodiment of the present application, step SI363 is a key step, which combines new data (such as the core point corresponding to the real-time data segment) and old data (such as the original interval importance factor of the core point of the current data segment), and balances the weights of the two in updating the interval importance factor through tuning parameters. This step ensures the flexibility and accuracy of data analysis, enabling the computer system to more accurately reflect the changing trend of the patient's physiological state. In step SI361, if the data segment bucket corresponding to the core point of the current data segment contains a real-time data segment, the computer system uses the ratio of the centrality density coefficient of the core point to the interval summation result as a reference importance factor. This factor reflects the relative importance of the current core point in the new data. In step SI362, the computer obtains the first tuning parameter V1, which is used to control the weight distribution of new and old data when updating the interval importance factor.

[0158] Next, step SI363 combines this information to calculate the adjusted interval importance factor. Specifically, the computer first multiplies the first tuning parameter V1 by the reference importance factor to obtain the importance weight of the new data in the update. At the same time, the computer calculates the second tuning parameter V2 (V2 = 1 - V1) multiplied by the original interval importance factor of the core point of the current data segment to obtain the importance weight of the old data in the update. Finally, the computer adds the two weighted calculation results to obtain the adjusted interval importance factor. This process is similar to the weight update strategy in machine learning. Here, the first tuning parameter V1 and the second tuning parameter V2 play the role of learning rate and new and old weights respectively. By adjusting the value of V1, the computer system can flexibly control the importance weight of new and old data in the update. When the value of V1 is large, the new data is more important in the update, which helps the computer system to quickly respond to changes in the patient's physiological state; when the value of V1 is small, the old data occupies a larger weight in the update, which helps to maintain the stability and continuity of data analysis.

[0159] For example, suppose that in the heart rate data analysis, there are two target data segment core points C1 and C2. The data segment bucket corresponding to C1 contains the real-time heart rate data segment, and the reference importance factor calculated by step SI361 is 0.3. Assuming that the first tuning parameter V1 is set to 0.7, then the second tuning parameter V2 is 0.3. The original interval importance factor of C1 is 0.5. According to step SI363, the adjusted interval importance factor of C1 is calculated as: 0.7*0.3+0.3*0.5=0.21+0.15=0.36. This new interval importance factor will be used in subsequent data analysis and anomaly detection.

[0160] In a possible implementation, step S36, determining the adjusted interval importance factor of the core point of the current data segment by the ratio of the centrality density coefficient to the interval summation result, includes:

[0161] Step SII361: If the data segment bucket corresponding to the core point of the current data segment does not have a real-time data segment in the real-time data segment combination, the ratio of the centrality density coefficient to the interval summation result is used as a reference importance factor.

[0162] In an embodiment of the present application, step SII361 focuses on the relationship between the core point of the current data segment and the real-time data segment, and quantifies the importance of the core point based on this relationship. When the data segment bucket corresponding to the core point of the current data segment does not contain the real-time data segment, this step will calculate a reference importance factor to reflect the relative importance of the core point in the current data analysis. In nursing data analysis, the data is divided into multiple data segments, each of which represents monitoring data within a time window. The core point is obtained by clustering the data segments, representing the main features or central trends of the data segments. The core point of the current data segment refers to the core point in the data segment currently being analyzed, and the data segment bucket is a set of data segments containing the core point.

[0163] The execution logic of step SII361 is as follows: the computer system checks whether the data segment bucket corresponding to the core point of the current data segment contains the real-time data segment in the real-time data segment combination. The real-time data segment combination is composed of the latest acquired real-time monitoring data, which reflects the patient's current physiological state. If the data segment bucket corresponding to the core point of the current data segment does not contain the real-time data segment, it means that the core point is mainly related to historical data and has little to do with the current real-time monitoring data.

[0164] In this case, the importance of the core point in the current data analysis is evaluated. To this end, the system calculates the ratio of the centrality density coefficient of the core point vector corresponding to the current core point to the interval summation result, and uses this ratio as a reference importance factor. The centrality density coefficient measures the compactness of the distribution of data points around the core point and reflects the stability or representativeness of the core point. The interval summation result is obtained by calculating and summing the vector intervals between the current core point and other core points, which quantitatively represents the overall degree of deviation between the current core point and other core points. This ratio is actually a weighted coefficient that comprehensively considers the centrality of the core point and the degree of deviation from other core points. Specifically, if the centrality density coefficient is large (indicating that the data points around the core point are compactly distributed) and the interval summation result is small (indicating that the current core point does not deviate much from other core points), then the ratio will be large, indicating that the current core point is more important in data analysis. Conversely, if the ratio is small, it means that the importance of the current core point is relatively low.

[0165] For example, suppose that in the heart rate data analysis, there are two target data segment core points C1 and C2. The data segment bucket corresponding to C1 does not contain the real-time heart rate data segment, while the data segment bucket corresponding to C2 contains the real-time data segment. Through calculation, the centrality density coefficient of C1 is 0.7, and the interval sum is 4. Since the data segment bucket corresponding to C1 does not contain the real-time data segment, according to step SII361, the reference importance factor of C1 is calculated as 0.7 / 4 = 0.175. This value will be used as a reference for the subsequent evaluation of the importance of C1 in the current data analysis.

[0166] Through the processing of step SII361, the computer system can more accurately evaluate the importance of core points that are not directly related to real-time data in data analysis, thereby providing strong support for subsequent data analysis and anomaly detection. This will help medical staff better understand the changing trends of patients' physiological states and discover potential health problems in a timely manner.

[0167] Step SII362: Obtain the weakening parameter corresponding to the core point of the current data segment, wherein the weakening parameter is equal to 1 divided by the time point difference, the time point difference is the result of subtracting the second time point from the first time point, the first time point refers to the moment corresponding to the real-time data segment combination, and the second time point refers to the moment corresponding to the last addition of the preceding data segment to the preceding data segment bucket corresponding to the core point of the current data segment.

[0168] In an embodiment of the present application, step SII362 introduces a weakening parameter to quantify the importance of the previous data (i.e., old data) corresponding to the core point of the current data segment in data analysis. This weakening parameter is calculated based on the time difference between the core point of the current data segment and the real-time data segment, and is intended to help the computer system more accurately evaluate the relative importance of new and old data in data analysis. The first time point refers to the moment corresponding to the real-time data segment combination, that is, the moment of the latest real-time monitoring data acquired. The second time point refers to the moment corresponding to the last addition of the previous data segment to the bucket of the previous data segment corresponding to the core point of the current data segment, which represents the time relationship between the core point and the latest data point in the previous data. The time point difference is the result of subtracting the second time point from the first time point, which quantitatively represents the time interval between the current core point and the previous data.

[0169] In step SII362, the computer system calculates a weakening parameter as a quantitative indicator for evaluating the importance of old data. The weakening parameter is equal to 1 divided by the time point difference. This calculation method makes the size of the weakening parameter inversely proportional to the time interval: when the time interval is large, the value of the weakening parameter is small, which means that the importance of old data in data analysis is low; conversely, when the time interval is small, the value of the weakening parameter is large, and the importance of old data is relatively high.

[0170] This design helps computer systems process time series data more flexibly. In nursing data analysis, the patient's physiological state changes over time, so newly acquired real-time monitoring data often reflects the patient's current state better than old data. By introducing a weakening parameter, the computer system can give new data a greater weight when updating the interval importance factor, thereby more accurately reflecting the patient's real-time physiological state. For example, suppose that in the heart rate data analysis, there is a core point C in the current data segment, and the time corresponding to the last data point added in the previous data segment bucket corresponding to it is 1 hour ago, while the time corresponding to the real-time data segment combination is the current time. Then the time point difference is 1 hour, and the weakening parameter is 1 / 1 hour = 1. If the time corresponding to the last data point added in the previous data segment bucket corresponding to another core point D is 6 hours ago, and the time point difference is 6 hours, then the weakening parameter is 1 / 6 hours ≈ 0.17. At this time, since the time interval of C is short, its corresponding weakening parameter is large, which means that when updating the interval importance factor, the old data related to C will have a larger weight; while the weakening parameter of D is small, and the weight of the old data related to D will be relatively low.

[0171] Through the calculation of step SII362, the computer system can dynamically adjust the weight of new and old data in data analysis according to the time interval, so that the algorithm is more adaptable to the dynamic change characteristics of time series data. This will help improve the accuracy and reliability of data analysis, help medical staff better understand the changing trends of patients' physiological status, and discover potential health problems in a timely manner.

[0172] Step SII363: Obtain a first tuning parameter V1, wherein the value of the tuning parameter is not less than 0 and not greater than 1.

[0173] Step SII364: The weighted calculation result of the first tuning parameter V1, the weakening parameter and the reference importance factor, and the weighted calculation result of the second tuning parameter V2 and the interval importance factor corresponding to the core point of the current data segment are taken as the adjusted interval importance factor of the core point of the current data segment, where V2=1-V1.

[0174] In the embodiment of the present application, steps SII363 and SII364 comprehensively consider the importance of new and old data in data analysis, as well as the impact of time factors on old data, by introducing tuning parameters and weakening parameters, thereby ensuring the accuracy and adaptability of data analysis.

[0175] In step SII363, the computer system obtains a first tuning parameter V1. This tuning parameter is used to control the weight distribution of new and old data when updating the interval importance factor. The value of the tuning parameter V1 is not less than 0 and not greater than 1, and its specific value can be determined according to the actual application scenario and data characteristics.

[0176] In nursing data analysis, the setting of tuning parameter V1 reflects the importance of new and old data. When the value of V1 is large, new data (such as the core points corresponding to the real-time data segment) will have a greater weight when updating the interval importance factor. This helps the computer system to quickly respond to changes in the patient's physiological state and adjust the analysis strategy in time. On the contrary, when the value of V1 is small, old data (such as the core points corresponding to the previous data segment) will have a greater influence, which helps to maintain the stability and continuity of data analysis.

[0177] In step SII364, the adjusted interval importance factor is calculated according to the first tuning parameter V1, the weakening parameter, the reference importance factor and the second tuning parameter V2 (V2=1-V1). This calculation process comprehensively considers the importance of new and old data and the impact of time factors on old data.

[0178] Specifically, first add the first tuning parameter V1 to the product of the weakening parameter and the reference importance factor. This product is actually the importance weight of the new data in the update. Since the weakening parameter is calculated based on the time difference, it reflects the degree of decay of the old data over time. Therefore, this product not only takes into account the importance of the new data, but also takes into account the impact of time factors on the old data. Then, add the above weighted calculation result to the product of the second tuning parameter V2 and the original interval importance factor of the core point of the current data segment. This product represents the importance weight of the old data in the update. Since V2=1-V1, the sum of the weights of the new and old data always remains at 1, ensuring the rationality of the weight distribution.

[0179] Finally, the obtained result is used as the interval importance factor of the core point after adjustment in the current data segment. This new importance factor will be used in subsequent data analysis and anomaly detection to more accurately reflect the importance of different core points in data analysis.

[0180] Assume that in the heart rate data analysis, there is a core point C of the current data segment, and the corresponding data segment bucket does not contain the real-time data segment. Through calculation, the centrality density coefficient of C is 0.6, the interval sum result is 5, and the time point difference is 3 hours (assuming that the time corresponding to the real-time data segment combination is the current time, and the time corresponding to the last addition of the previous data segment to the previous data segment bucket corresponding to C is 3 hours ago). Therefore, the weakening parameter is 1 / 3 = 0.33. Assuming that the first tuning parameter V1 is set to 0.7, then the second tuning parameter V2 is 0.3. The original interval importance factor of C is 0.5. According to the calculation method of step SII364, the adjusted interval importance factor of C is: 0.7*0.33*(0.6 / 5)+0.3*0.5=0.0277+0.15=0.1777. This new interval importance factor comprehensively considers the importance of new and old data and the influence of time factors on old data, and will be used in subsequent heart rate data analysis.

[0181] In a possible implementation, step S40, after determining the real-time data segment whose target data segment vector interval is greater than the critical data segment vector interval as the nursing data segment to be confirmed and marking the nursing data segment to be confirmed, further includes:

[0182] Step S50: Acquire a real-time data segment combination corresponding to the target time obtained by intercepting the nursing monitoring data sequence, wherein the real-time data segment combination includes a plurality of real-time data segments whose monitoring time is after the target time.

[0183] In the embodiment of the present application, step S50 intercepts a new real-time data segment combination from the nursing monitoring data sequence based on the target time. Specifically, in step S40, the computer system may have identified some real-time data segments whose target data segment vector interval is greater than the critical data segment vector interval, and marked them as nursing data segments to be confirmed. These data segments may reflect the physiological abnormalities of the patient at a certain point in time. However, it is not enough to make a final diagnosis or decision based solely on these data, because the patient's physiological state changes over time.

[0184] Therefore, in step S50, the computer system sets a target time, which is usually a specific moment after the current time. The setting of the target time depends on the specific business needs and data characteristics. It can be a time point after a few minutes, hours or days. Then, the system intercepts a combination of real-time data segments corresponding to the target time from the nursing monitoring data sequence. This combination includes multiple real-time data segments whose monitoring time is after the target time, which reflect the patient's physiological state after the target time.

[0185] The process of intercepting the real-time data segment combination is similar to the process described in step S10, but the time range and target are different. In step S10, the system intercepts the real-time data segment combination before the current time for analyzing the patient's historical physiological state; while in step S50, the system intercepts the real-time data segment combination after the target time for analyzing the patient's future physiological state change trend.

[0186] Step S60: In the real-time data segment combination corresponding to the target time, the real-time data segment whose target data segment vector interval is greater than the critical data segment vector interval is taken as the nursing data segment to be confirmed.

[0187] In the embodiment of the present application, step S60 is intended to further confirm and analyze the patient's physiological state. In step S50, the computer system has intercepted a new real-time data segment combination from the nursing monitoring data sequence based on the target time, and these data segments reflect the patient's physiological state after the target time. Step S60 is to perform specific analysis and judgment on these new real-time data segments to determine which data segments need to be marked as nursing data segments to be confirmed. Specifically, in step S60, the computer system first calculates the target data segment vector interval of each real-time data segment in the real-time data segment combination corresponding to the target time. This vector interval is obtained by performing a similarity operation on the real-time data segment and the integrated and adjusted target data segment core point group, which quantitatively represents the degree of deviation between the real-time data segment and the target data segment.

[0188] Next, the computer system compares the target data segment vector intervals of these real-time data segments with the preset critical data segment vector intervals. The critical data segment vector interval is a threshold used to determine whether the real-time data segment is abnormal or deviates from expectations. If the target data segment vector interval of a real-time data segment is greater than the critical data segment vector interval, it means that there is a large deviation between the data segment and the target data segment, which may reflect that the patient's physiological state has become abnormal or changed.

[0189] In this case, the computer system marks the real-time data segment as a nursing data segment to be confirmed. These nursing data segments to be confirmed will be the focus of attention for subsequent data analysis and abnormality detection. By continuously monitoring and analyzing these data segments, medical staff can promptly discover changes in the patient's physiological state and take corresponding nursing measures. Through the processing of step S60, the computer system can further confirm and analyze changes in the patient's physiological state based on the real-time monitoring data after the target time. This will help medical staff to promptly discover potential health problems and provide patients with more accurate and timely nursing services.

[0190] The present application also provides a computer system, such as Figure 2As shown, the computer system 100 includes: a processor 101 and a memory 103. The processor 101 and the memory 103 are connected, such as through a bus 102. Optionally, the computer system 100 may also include a transceiver 104. It should be noted that in actual applications, the transceiver 104 is not limited to one, and the structure of the computer system 100 does not constitute a limitation on the embodiments of the present application.

[0191] An embodiment of the present application provides a computer system. The computer system in the embodiment of the present application includes: one or more processors; a memory; and one or more computer programs, wherein the one or more computer programs are stored in the memory and configured to be executed by the one or more processors, and when the one or more programs are executed by the processor, the method of the present application is implemented.

Claims

1. A nursing data analysis method based on artificial intelligence, characterized in that: include: Acquire a real-time data segment combination obtained by intercepting a nursing monitoring data sequence, wherein the real-time data segment combination includes a plurality of real-time data segments intercepted from the nursing monitoring data sequence, and the real-time data segment includes a plurality of nursing data items arranged for a monitored object according to a monitoring timestamp; Obtain a current data segment core point group corresponding to the real-time data segment combination, and a preceding data segment core point group obtained according to a preceding data segment of the nursing monitoring data sequence before the real-time data segment combination, wherein the current data segment core point group indicates a bucketing result for all real-time data segments in the real-time data segment combination, the preceding data segment core point group indicates a bucketing result for each preceding data segment in the preceding data segment, and a plurality of current core points in the current data segment core point group and a plurality of preceding data segment core points in the preceding data segment core point group respectively represent a data segment bucket; After the previous data segment core point group is updated by the current data segment core point group to obtain the target data segment core point group, each of the real-time data segments in the real-time data segment combination is subjected to a similarity operation with a plurality of target data segment core points in the target data segment core point group to obtain a target data segment vector interval corresponding to each of the real-time data segments; Determine the real-time data segment whose target data segment vector interval is greater than the critical data segment vector interval as a nursing data segment to be confirmed, and mark the nursing data segment to be confirmed; The step of performing a similarity operation on each of the real-time data segments in the real-time data segment combination and a plurality of target data segment core points in the target data segment core point group to obtain a target data segment vector interval corresponding to each of the real-time data segments comprises: Obtaining interval importance factors respectively corresponding to a plurality of the target data segment core points in the target data segment core point group; Obtain one of the real-time data segments in the real-time data segment combination as the current real-time data segment, and execute the following steps in a loop until all the real-time data segments in the real-time data segment combination are executed: respectively calculating reference data segment vector intervals between the current real-time data segment and a plurality of target data segment core points; Determine the target data segment vector interval corresponding to the current real-time data segment through the interval importance factors respectively corresponding to the core points of the multiple target data segments and the obtained multiple reference data segment vector intervals; After determining the target data segment vector interval corresponding to the current real-time data segment by using the interval importance factors respectively corresponding to the core points of the multiple target data segments and the obtained multiple reference data segment vector intervals, the method further includes: Sequentially obtain one of the target data segment core points in the target data segment core point group as the current data segment core point, and execute the following steps in a loop until each of the target data segment core points in the target data segment core point group is executed; Obtaining the centrality density coefficient of the core point vector corresponding to the core point of the current data segment; Respectively calculating the interval summation result of the vector intervals between the core point vector corresponding to the core point of the current data segment and the core point vectors corresponding to the remaining target data segment core points in the target data segment core point group; Determining the interval importance factor of the core point of the current data segment after adjustment by the ratio of the centrality density coefficient to the interval summation result; The determining the interval importance factor of the core point of the current data segment after adjustment by the ratio of the centrality density coefficient to the interval summation result includes: If the data segment bucket corresponding to the core point of the current data segment includes the real-time data segment in the real-time data segment combination, the ratio of the centrality density coefficient to the interval summation result is used as a reference importance factor; Obtaining a first tuning parameter V1, wherein the value of the tuning parameter is not less than 0 and not greater than 1; Determine the sum of the weighted calculation result of the first tuning parameter and the reference importance factor, and the weighted calculation result of the second tuning parameter V2 and the interval importance factor corresponding to the core point of the current data segment as the interval importance factor after adjustment of the core point of the current data segment, wherein V2=1-V1; Or: determining the interval importance factor after the core point of the current data segment is adjusted by the ratio of the centrality density coefficient to the interval summation result, including: If the data segment bucket corresponding to the core point of the current data segment does not have the real-time data segment in the real-time data segment combination, the ratio of the centrality density coefficient to the interval summation result is used as a reference importance factor; Obtain a weakening parameter corresponding to the core point of the current data segment, wherein the weakening parameter is equal to 1 divided by a time point difference, the time point difference is a result of subtracting a second time point from a first time point, the first time point refers to the moment corresponding to the real-time data segment combination, and the second time point refers to the moment corresponding to the previous data segment bucket corresponding to the core point of the current data segment is last added to the previous data segment; Obtaining a first tuning parameter V1, wherein the value of the tuning parameter is not less than 0 and not greater than 1; The weighted calculation result of the first tuning parameter V1, the weakening parameter and the reference importance factor, and the weighted calculation result of the second tuning parameter V2 and the interval importance factor corresponding to the core point of the current data segment are taken as the interval importance factor after the core point of the current data segment is adjusted, where V2=1-V1.

2. The method according to claim 1, characterized in that Before performing similarity operations on each of the real-time data segments in the real-time data segment combination and a plurality of target data segment core points in the target data segment core point group to obtain a target data segment vector interval corresponding to each of the real-time data segments, the method further includes: Respectively calculating the vector intervals between each of the current data segment core points in the current data segment core point group and each of the previous data segment core points in the previous data segment core point group; If the preceding data segment core point group includes a preceding data segment core point whose vector interval is smaller than the reference core point vector interval, then a plurality of the real-time data segments in the current data segment bucket corresponding to the current data segment core point are merged into the preceding data segment bucket corresponding to the preceding data segment core point to obtain the adjusted preceding data segment bucket, wherein each preceding data segment core point in the preceding data segment core point group corresponds to one of the reference core point vector intervals; If the core point group of the previous data segment does not have a core point of the previous data segment whose vector interval is smaller than the vector interval of the reference core point, the current data segment bucket corresponding to the core point of the current data segment is used as a new previous data segment bucket in the previous data segment bucket; The data segment core points respectively corresponding to the preceding data segment buckets in the adjusted preceding data segment bucket combination are used as the target data segment core points in the target data segment core point group.

3. The method according to claim 2, characterized in that The step of using the data segment core points respectively corresponding to the preceding data segment buckets in the adjusted preceding data segment bucket combination as the target data segment core points in the target data segment core point group includes: If the target preceding data segment bucket in the adjusted preceding data segment bucket combination includes a plurality of preceding data segments and a plurality of real-time data segments in the current data segment bucket, then respectively obtaining data segment vectors corresponding to the plurality of real-time data segments in the current data segment bucket, multiplying the vector module of the reverse vector of the data segment vector by the vector module of the reverse vector of the data segment vector, and taking the vector module product corresponding to each of the plurality of real-time data segments as the real-time feature scalar corresponding to the plurality of real-time data segments; Determine the target data segment core point corresponding to the target pre-order data segment bucket according to the pre-order feature scalar corresponding to the pre-order data segment included in the target pre-order data segment bucket and the real-time feature scalar; If the target preceding data segment bucket in the adjusted preceding data segment bucket combination only has a plurality of the current data segments in the current data segment bucket, then respectively obtaining data segment vectors corresponding to the plurality of real-time data segments in the current data segment bucket, multiplying the vector module products of the inverse vectors of the data segment vectors by the vector module products, and taking the sum of the vector module products corresponding to the plurality of real-time data segments as the real-time feature scalar corresponding to the plurality of real-time data segments; The target data segment core point corresponding to the target preceding data segment bucket is determined according to the real-time feature scalar; wherein the sum of the vector intervals between the data segment vectors corresponding to multiple data segments in the target preceding data segment bucket and the core point vector corresponding to the target data segment core point is not greater than a preset interval value.

4. The method according to claim 2, characterized in that: When the data segment core points respectively corresponding to the preceding data segment buckets in the adjusted preceding data segment bucket combination are used as the target data segment core points in the target data segment core point group, the method further includes: If the target preceding data segment bucket in the adjusted preceding data segment bucket combination includes a plurality of preceding data segments and a plurality of real-time data segments in the current data segment bucket, then obtaining a vector interval addition result of the vector intervals of the respective data segment vectors of the plurality of real-time data segments in the current data segment bucket and the core point vector corresponding to the core point of the current data segment; Obtain the reference core point vector interval corresponding to the target preceding data segment bucket, and obtain the first importance factor corresponding to the core point of the current data segment, and the second importance factor corresponding to the core point of the preceding data segment; Determine the reference core point vector interval corresponding to the core point of the target data segment by a first weighted calculation result of the sum of the first importance factor and the vector interval, and a second weighted calculation result of the second importance factor and the reference core point vector interval; If the target preceding data segment bucket in the adjusted preceding data segment bucket combination only has a plurality of the current data segments in the current data segment bucket, obtaining a vector interval addition result of vector intervals between respective data segment vectors of the plurality of real-time data segments in the current data segment bucket and a core point vector corresponding to a core point of the current data segment; The result of adding the vector intervals is used as the reference core point vector interval corresponding to the core point of the target data segment.

5. The method according to claim 1, characterized in that The respectively calculating the reference data segment vector intervals between the current real-time data segment and the core points of the plurality of target data segments comprises: Sequentially obtain a target data segment core point from the multiple target data segment core points, use it as a current data segment core point, and loop through the following steps until all the multiple target data segment core points are executed: Acquire a current data segment core point sub-combination corresponding to the current data segment core point, wherein the current data segment core point sub-combination includes the current data segment core point and a plurality of reference data segment core points, and the reference data segment core point is a target data segment core point that is before the current data segment core point; Respectively calculating the Chebyshev distance between the data segment vector corresponding to the current real-time data segment and the core point vectors corresponding to the multiple data segment core points in the current data segment core point sub-combination; The Chebyshev distance with the smallest value is used as the reference data segment vector interval between the current real-time data segment and the core point of the current data segment; The determining the target data segment vector interval corresponding to the current real-time data segment by using the interval importance factors respectively corresponding to the core points of the multiple target data segments and the obtained multiple reference data segment vector intervals includes: Based on the interval importance factors respectively corresponding to the core points of the plurality of target data segments, the importance integration results of the plurality of reference data segment vector intervals are calculated to obtain a first data segment vector interval corresponding to the current real-time data segment; The first data segment vector interval is data scaled according to the importance integration result corresponding to each of the real-time data segments to obtain the target data segment vector interval corresponding to the current real-time data segment.

6. The method according to any one of claims 1 to 5, characterized in that After determining the real-time data segment whose target data segment vector interval is greater than the critical data segment vector interval as the nursing data segment to be confirmed and marking the nursing data segment to be confirmed, the method further includes: Acquire a real-time data segment combination corresponding to a target time obtained by intercepting a nursing monitoring data sequence, wherein the real-time data segment combination includes a plurality of real-time data segments whose monitoring time is after the target time; In the real-time data segment combination corresponding to the target time, the real-time data segment whose target data segment vector interval is greater than the critical data segment vector interval is used as the nursing data segment to be confirmed.

7. A computer system, characterized in that: include: one or more processors; Memory; one or more computer programs; The one or more computer programs are stored in the memory and configured to be executed by the one or more processors, and when the one or more computer programs are executed by the processors, the method according to any one of claims 1 to 6 is implemented.

Citation Information

Patent Citations

  • Clustering method and device, storage medium and electronic equipment

    CN110807474A

  • Data stream clustering method and device based on density peak value

    CN113269238A

  • Severe care monitoring system based on Internet of Things

    CN117238507A

  • Intelligent management method for monitoring data of critical patient

    CN117609813A

  • Time sequence extraction method based on data bucket sampling interval and DTW optimization

    CN117668565A