Postoperative intelligent monitoring and early warning system and method for vascular surgery

Abstract

The invention relates to the technical field of data processing, in particular to a postoperative intelligent monitoring and early warning system and method for vascular surgery. The method comprises the following steps: calculating a scoring range of patient states of patients of each age group, judging the patient state of each patient, then creating and training a postoperative monitoring model to obtain a trained postoperative monitoring model, and inputting real-time body data of the patients into the trained postoperative monitoring model to obtain real-time body states of the patients. The method comprises the following steps: acquiring a plurality of data items, calculating a deviation rate between the unqualified data items and a qualified interval, finally screening out a plurality of observation items according to a size sequence of the deviation rate, and resetting acquisition parameters of a patient based on the deviation rate of the observation items. And the acquisition frequency of the patient is recalculated in combination with the deviation rate, so that the acquisition parameters are automatically adjusted according to the body data of the patient, the comprehensiveness of the acquired data is ensured, and early warning is given out in time.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an intelligent monitoring and early warning system and method for post-operative vascular surgery. Background Art

[0002] Vascular surgery is a branch of surgery that focuses on the prevention, diagnosis and treatment of peripheral vascular diseases other than cerebrovascular and cardiac vessels. In addition to hair, nails, corneas, etc., blood vessels are spread throughout the human body, so vascular surgery covers a wide range.

[0003] A Chinese patent with announcement number CN116612899B discloses an Internet-based cardiovascular surgery data processing method and service platform, which generates a cardiovascular surgery data feature matrix projection diagram by visually projecting an interactive view of cardiovascular surgery data; generates a cardiovascular surgery convolution feature model by dilating the generated interactive view of cardiovascular surgery data; symmetrically encrypts the cardiovascular surgery convolution feature model; and uploads the cardiovascular surgery symmetric encryption model to the cardiovascular surgery data processing service platform, thereby achieving orderly and accurate management of cardiovascular surgery data. However, the prior art does not adjust the patient's real-time detection data in combination with the patient's physical data, and it is difficult to automatically adjust the body data collection parameters according to the patient's physical condition, which easily leads to incomplete body data. Summary of the invention

[0004] The purpose of the present invention is to address the problems existing in the background technology and to propose an intelligent monitoring and early warning system and method for postoperative vascular surgery.

[0005] The technical solution of the present invention: On the one hand, the present application provides a method for intelligent monitoring and early warning after vascular surgery, comprising: Setting collection parameters, collecting physical data of multiple patients, calculating the scoring range of the patient status of patients in each age group based on the physical data of the patients, and judging the patient status of each patient; Creating a postoperative monitoring model, inputting the body data and the patient status as a training set into the postoperative monitoring model to train the postoperative monitoring model, and obtaining a trained postoperative monitoring model; Collect the patient's real-time physical data, input the patient's real-time physical data into the trained postoperative monitoring model, obtain the patient's real-time physical status, screen out unqualified physical data based on the patient's real-time physical status, and calculate the deviation rate between the unqualified data items and the qualified interval; Multiple observation items are screened out in order of deviation rates, and the patient's acquisition parameters are reset based on the deviation rates of the observation items.

[0006] Preferably, set the acquisition parameters, acquire the physical data of multiple patients, calculate the scoring range of the patient status for each age group of patients based on the physical data of the patients, and determine the patient status of each patient, including: Create a physical data table; Set the acquisition parameters; the acquisition parameters include the acquisition period and the acquisition frequency; Based on the acquisition parameters, acquire the physical data of multiple different patients respectively, and put all the acquired physical data into the physical data table; the physical data includes patient age, blood pressure, blood oxygen saturation, heart rate, and body temperature.

[0007] Preferably, set the acquisition parameters, acquire the physical data of multiple patients, calculate the scoring range of the patient status for each age group of patients based on the physical data of the patients, and determine the patient status of each patient, further including: Divide the physical data of the patients into multiple data sets according to the patient age, and calculate the average value of the physical data included in each data set of each age group respectively; Set the deviation threshold, and calculate the scoring range corresponding to the patient status by combining the average value of each item of physical data with the deviation threshold; Randomly select the physical data of a patient from the physical data table; Judge the scoring range of the patient status where the physical data of the patient is located, so as to obtain the patient status of the patient, and put the patient status into the physical data table; Return to randomly select a piece of physical data from the physical data table until all the physical data in the physical data table have been selected, and obtain the patient status of each patient.

[0008] Preferably, create a postoperative monitoring model, and input the physical data and the patient status as a training set into the postoperative monitoring model to train the postoperative monitoring model to obtain the trained postoperative monitoring model, including: Create a postoperative monitoring model; Divide the preprocessed physical data into multiple feature sets through cluster analysis, and divide the feature sets into a training set and a test set according to a random ratio; Input the training set into the postoperative monitoring model, so that the postoperative detection model learns the corresponding relationship between the physical data and the patient status, and obtain the trained postoperative monitoring model; Input the test set into the trained postoperative monitoring model, obtain the predicted patient status output by the trained postoperative monitoring model, and judge whether the trained postoperative monitoring model is trained completed.

[0009] Preferably, divide the preprocessed physical data into multiple feature sets through cluster analysis, and divide the feature sets into a training set and a test set according to a random ratio, including: Randomly select K pieces of data from all preprocessed body data as the initial cluster centers; Calculate the distances between other preprocessed body data and each cluster center, and assign the other preprocessed body data to the nearest initial cluster center to obtain K clusters; For each cluster, recalculate the mean of the distances of all sample points within each cluster, and use the position of the mean as the new cluster center; Set a distance threshold; Calculate whether the distance between the new cluster center and the initial clustering center is greater than the distance threshold; If the distance between the new cluster center and the initial clustering center is greater than the distance threshold, then return to calculate the distances between other preprocessed body data and each clustering center until the distance between the new cluster center and the initial clustering center is less than or equal to the distance threshold to obtain K clusters; Use the K clusters as K feature sets, and the sample points within each cluster as the feature samples in the feature sets.

[0010] Preferably, input the training set into the postoperative monitoring model so that the postoperative detection model learns the corresponding relationship between body data and patient status to obtain the trained postoperative monitoring model, including: Randomly select a feature sample from the training set; Based on this feature sample, establish a coupling relationship between body data - patient status, and input the body data, patient status, and the coupling relationship between body data - patient status as a training sample into the postoperative monitoring model so that the postoperative monitoring model learns the corresponding relationship between body data and patient status; Return to randomly select a feature sample from the training set until all feature samples in the training set have been selected to obtain the trained postoperative monitoring model.

[0011] Preferably, collect the patient's real-time body data, input the patient's real-time body data into the trained postoperative monitoring model to obtain the patient's real-time body status, and based on the patient's real-time body status, screen out unqualified body data and calculate the deviation rate of the unqualified data items from the qualified interval, including: Collect the patient's real-time body data, input the patient's real-time body data into the trained postoperative monitoring model to obtain the patient's real-time body status; Screen out unqualified data items based on the patient's real-time body status; Calculate the deviation rate of each unqualified data item from the qualified interval through Formula 1; Formula 1; Where, is the deviation rate of the i-th unqualified data item from the qualified interval, is the real-time data of the i-th data item, is the standard data of the i-th data.

[0012] Preferably, a plurality of observation items are screened according to the order of the deviation rates, and the acquisition parameters of the patient are reset based on the deviation rates of the observation items, including: Sort all unqualified data items according to the order of the deviation rates; Screen out the top N unqualified data items; record the top N unqualified data items as observation items; Obtain the original acquisition frequency of each observation item in turn according to the order of the deviation rates of the observation items; Calculate the new acquisition frequency of the observation item through formula 2 in combination with the deviation rate of the observation item; Formula 2; Wherein, is the new acquisition frequency of the observation item, is the original acquisition frequency of the observation item, is the deviation rate of the observation item; Recalculate the acquisition nodes of each observation item based on the new acquisition frequency, and collect real-time body data according to the new acquisition nodes.

[0013] On the other hand, the present application also provides an intelligent monitoring and warning system for after vascular surgery, including: An acquisition unit, which acquires the body data of the patient through the acquisition unit; A data unit, which stores and processes the body data of the patient through the data unit; A processing unit, which executes an intelligent monitoring and warning method for after vascular surgery as described above through the processing unit, and the acquisition unit, the data unit and the processing unit are communicatively connected.

[0014] Preferably, the data unit adopts a distributed storage system to realize real-time synchronization and access of the body data of the patient.

[0015] Preferably, the data unit adopts a distributed storage system to realize real-time synchronization and access of the body data of the patient.

[0016] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects: By collecting the physical data of multiple patients, calculating the scoring range of the patient status for patients in each age group, and judging the patient status of each patient, then creating and training a postoperative monitoring model to obtain a trained postoperative monitoring model. Next, collecting the real-time physical data of the patient, inputting the real-time physical data of the patient into the trained postoperative monitoring model to obtain the real-time physical status of the patient, screening out unqualified physical data based on the real-time physical status of the patient, calculating the deviation rate of the unqualified data items from the qualified range, and finally screening out multiple observation items according to the magnitude order of the deviation rate, and resetting the collection parameters of the patient based on the deviation rate of the observation items. This application calculates the deviation rate of the unqualified data items from the qualified range, thereby screening out multiple observation items, and recombining the deviation rate to recalculate the collection frequency of the patient, so as to increase the collection frequency of the data items that the patient needs to focus on observing, realize the automatic adjustment of the collection parameters according to the physical data of the patient, ensure the comprehensiveness of the collected data, and thus issue an early warning in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 FIG. is a schematic flow chart of a method for intelligent monitoring and early warning after vascular surgery proposed by the present invention; Figure 2 FIG. is a schematic structural diagram of a system for intelligent monitoring and early warning after vascular surgery proposed by the present invention.

[0018] Reference numerals: 100, collection unit; 200, data unit; 300, processing unit. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] Embodiment 1, as Figure 1 shown, a method for intelligent monitoring and early warning after vascular surgery proposed by the present invention includes: S100, setting collection parameters, collecting the physical data of multiple patients, calculating the scoring range of the patient status for patients in each age group based on the physical data of the patients, and judging the patient status of each patient; S200, creating a postoperative monitoring model, inputting the physical data and the patient status as a training set into the postoperative monitoring model to train the postoperative monitoring model, and obtaining a trained postoperative monitoring model; S300, collecting the real-time physical data of the patient, inputting the real-time physical data of the patient into the trained postoperative monitoring model to obtain the real-time physical status of the patient, screening out unqualified physical data based on the real-time physical status of the patient, and calculating the deviation rate of the unqualified data items from the qualified range; S400, screening out multiple observation items according to the magnitude order of the deviation rate, and resetting the collection parameters of the patient based on the deviation rate of the observation items.

[0020] In the present invention, by collecting the physical data of multiple patients, calculating the scoring range of the patient status for patients in each age group, and determining the patient status of each patient, then creating and training a postoperative monitoring model to obtain a trained postoperative monitoring model. Next, collecting the real-time physical data of the patient, inputting the real-time physical data of the patient into the trained postoperative monitoring model to obtain the real-time physical status of the patient, screening out unqualified physical data based on the real-time physical status of the patient, calculating the deviation rate between the unqualified data items and the qualified interval, and finally screening out multiple observation items according to the magnitude order of the deviation rate, and resetting the collection parameters of the patient based on the deviation rate of the observation items. This application calculates the deviation rate between the unqualified data items and the qualified interval, thereby screening out multiple observation items, and recombining the deviation rate to recalculate the collection frequency of the patient, so as to improve the collection frequency of the data items that the patient needs to focus on observing, realize the automatic adjustment of the collection parameters according to the physical data of the patient, ensure the comprehensiveness of the collected data, and thus issue an early warning in a timely manner.

[0021] In an alternative embodiment, the S100 includes: S110, creating a physical data table; S120, setting collection parameters; the collection parameters include a collection period and a collection frequency; S130, respectively collecting the physical data of multiple different patients based on the collection parameters, and putting all the collected physical data into the physical data table; the physical data includes patient age, blood pressure, blood oxygen saturation, heart rate, and body temperature.

[0022] It should be noted that for different patients, different collection parameters can be set according to the patient's condition. Specifically, the more severe the patient's condition, the higher the collection frequency. After determining the collection parameters, it is necessary to collect the physical data of multiple physical items of the patient, so as to facilitate subsequent judgment of the patient's status based on the physical data.

[0023] After determining the collection period and the collection frequency, the collection nodes can be calculated through the collection period and the collection frequency. For example, the collection period of patient A is from 0:00 on December 1st to 24:00 on December 1st, and the collection frequency is 2 times. Then the collection nodes can be obtained as 00:00 on December 1st and 12:00 on December 1st.

[0024] In an alternative embodiment, the S100 further includes: S140, dividing the physical data of the patient into multiple data sets according to the patient age, and respectively calculating the average value of the physical data included in each age group's data set; Optionally, in addition to using age groups as the classification criterion, other criteria such as gender can also be used as the criterion for dividing different data sets. Regardless of which index is selected as the criterion for dividing the data set, the aim is to enable different data sets to reflect the patient status of different patients; S150, set a deviation threshold, and calculate the score range corresponding to the patient status by combining the average value of each piece of physical data with the deviation threshold; Specifically, the deviation threshold refers to the maximum deviation amplitude of the average value relative to the average value. If the average value of a certain piece of physical data is A and the deviation threshold is C, then the score range of this piece of physical data is [(1 - C)*A, (1 + C)*A]. When the patient's physical data is within this score range, it means that the patient's status is qualified. When the patient's physical data is outside this score range, it means that the patient's status is unqualified; S160, randomly select the physical data of a patient from the physical data table; S170, determine the score range of the patient status in which the patient's physical data is located, so as to obtain the patient's status, and put the patient status into the physical data table; Specifically, when determining the score range of the patient status in which the patient's physical data is located, in fact, each piece of physical data included in the patient's physical data is judged. As long as there is at least one piece of physical data that is unqualified, the patient's status is unqualified; S180, return to randomly select a piece of physical data from the physical data table until all the physical data in the physical data table have been selected, and obtain the patient status of each patient.

[0025] It should be noted that since there are certain differences in the physical data of patients in different age groups. For example, the normal systolic blood pressure of men aged 20 - 40 is between 115 - 120, while the normal systolic blood pressure of men aged 40 - 60 is between 124 - 128. It can be seen that there are obvious differences between the two. Therefore, when processing the patient's physical data, it is necessary to divide the patient's physical data into multiple data sets according to age groups, so as to make the prediction of the patient's status by the postoperative monitoring model after training more accurate.

[0026] Since the physical data standards of patients in different age groups are different, the score ranges of patients in different age groups are also different. Taking systolic blood pressure as an example, when the deviation threshold is set to 10%, the score range of the systolic blood pressure of men aged 20 - 40 is [104, 132], and the score range of the systolic blood pressure of men aged 40 - 60 is [112, 141]. Then, the patient status of men aged 20 - 40 with systolic blood pressure within the range of [104, 132] is qualified, and the patient status of men aged 40 - 60 with systolic blood pressure within the range of [112, 141] is qualified.

[0027] In an alternative embodiment, the 200 includes: S210, creating a postoperative monitoring model; S220, dividing the preprocessed body data into multiple feature sets through cluster analysis, and dividing the feature sets into a training set and a test set according to a random ratio; S230, inputting the training set into the postoperative monitoring model, enabling the postoperative detection model to learn the correspondence between the body data and the patient status, and obtaining the trained postoperative monitoring model; S240, inputting the test set into the trained postoperative monitoring model, obtaining the predicted patient status output by the trained postoperative monitoring model, and determining whether the trained postoperative monitoring model is trained.

[0028] It should be noted that by creating and training the postoperative monitoring model, the patient status of the patient can be directly monitored through the postoperative monitoring model, and an early warning can be issued in a timely manner when the patient status is abnormal, so as to realize the real-time monitoring of the patient.

[0029] In an alternative embodiment, the S220 includes: S221, randomly selecting K data from all the preprocessed body data as the initial cluster centers; S222, calculating the distances between the other preprocessed body data and each cluster center, and assigning the other preprocessed body data to the nearest initial cluster center to obtain K clusters; S223, for each cluster, recalculating the mean of the distances of all the sample points within each cluster, and taking the position of the mean as the new cluster center; S224, setting a distance threshold; S225, calculating whether the distance between the new cluster center and the initial cluster center is greater than the distance threshold; S226, if the distance between the new cluster center and the initial cluster center is greater than the distance threshold, then return to calculate the distances between the other preprocessed body data and each cluster center until the distance between the new cluster center and the initial cluster center is less than or equal to the distance threshold to obtain K clusters; taking the K clusters as K feature sets, and the sample points within each cluster as the feature samples in the feature sets.

[0030] It should be noted that by using the K-means clustering algorithm, the body data of multiple patients is divided into K feature sets, and the body data within each feature set is similar, thereby reducing the computational complexity in training the postoperative monitoring model. The K-means clustering algorithm is a classic clustering method and belongs to the unsupervised learning algorithm. Its core idea is to divide the data set into K categories through iteration, so that the data points within each category have a high similarity, while the data points between different categories have a low similarity.

[0031] In an optional embodiment, the S230 includes: S231, randomly select a feature sample from the training set; S232, establish a coupling relationship between body data and patient status based on the feature sample, and input the body data, patient status, and the coupling relationship between body data and patient status as a training sample into the postoperative monitoring model, so that the postoperative monitoring model learns the corresponding relationship between body data and patient status; S233, return to randomly select a feature sample from the training set until all the feature samples in the training set are selected, and obtain the trained postoperative monitoring model.

[0032] It should be noted that through the foregoing steps, K feature sets are obtained, and then the K feature sets are divided into a training set and a test set according to a random ratio. Generally speaking, the ratio of the training set is higher than that of the test set. By inputting multiple training samples in the training set into the postoperative monitoring model, the postoperative monitoring model continuously learns the coupling relationship between body data and patient status until the postoperative monitoring model can automatically output the patient status according to the input body data, thereby obtaining the trained postoperative monitoring model.

[0033] The existence of the postoperative monitoring model enables the user to only collect the real-time body data of the patient and input it into the trained postoperative monitoring model to obtain the current patient status of the patient.

[0034] In an optional embodiment, the S300 includes: S310, collect the real-time body data of the patient, input the real-time body data of the patient into the trained postoperative monitoring model, and obtain the real-time body status of the patient; S320, screen out unqualified data items based on the real-time body status of the patient; S330, calculate the deviation rate of each unqualified data item from the qualified interval through Formula 1; Formula 1; Wherein, is the deviation rate of the i-th unqualified data item from the qualified interval, is the real-time data of the i-th data item, is the standard data of the i-th data.

[0035] It should be noted that the "standard data" is the standard requirement for the physical data of healthy people according to clinical standards. For example, for the physical data item of blood pressure, the standard data for people aged 0-3 is 90 / 60, for people aged 3-15 is 117 / 75, and for people aged 20-40 is 125 / 80. Since the standard data is relatively healthy data, the greater the deviation rate between the real-time physical data and the standard data, the greater the possibility that the patient's body has problems. Therefore, for patients with a greater deviation rate, the acquisition frequency should be set higher to ensure that any abnormalities in the patient's body can be detected in a timely manner.

[0036] In an optional embodiment, the S400 includes: S410, sorting all unqualified data items according to the magnitude order of the deviation rate; S420, screening out the top N unqualified data items; denoting the top N unqualified data items as observation items; S430, sequentially obtaining the original acquisition frequency of each observation item according to the magnitude order of the deviation rate of the observation items; S440, calculating the new acquisition frequency of the observation items through Formula 2 in combination with the deviation rate of the observation items; Formula 2; Wherein, is the new acquisition frequency of the observation item, is the original acquisition frequency of the observation item, is the deviation rate of the observation item; Optionally, the new acquisition frequency of the observation items can also be calculated through Formula 3; Formula 3; Wherein, is the new acquisition frequency of the i-th observation item, is the original acquisition frequency of the i-th observation item, is the deviation rate of the i-th observation item, is the age parameter of the patient; Specifically, , wherein, is the age of the patient, is the age parameter of the patient; As the patient's age increases, the patient's physical fitness continuously declines. For older patients, the collection frequency for elderly patients can be appropriately increased. Therefore, the patient's age parameter is added to Formula 3, and the patient's age is also used as a reference condition for setting the collection frequency.

[0037] S450, recalculate the collection nodes for each observation item based on the new collection frequency, and collect real-time body data according to the new collection nodes; Optionally, in addition to increasing the collection frequency of observation items, the collection frequency of qualified items can also be reduced, so as to invest more monitoring resources in observation items. The collection frequency of qualified items can be calculated by Formula 4; Formula 4; Wherein, is the new collection frequency of the i-th qualified item, is the original collection frequency of the i-th qualified item, is the deviation rate of the i-th qualified item.

[0038] It should be noted that the deviation rates of multiple unqualified data items are obtained in the foregoing embodiments. The reason for screening out N unqualified data items according to the order of the deviation rates is that these N unqualified data items have the most problems and can best reflect the patient's condition. Therefore, these N unqualified data items are marked as observation items for key observation.

[0039] After obtaining the observation items, it is necessary to calculate the new collection frequency of each observation item in combination with the deviation rate. The greater the deviation rate, the higher the new collection frequency.

[0040] This application calculates the deviation rate between the patient's body data and the standard data, thereby obtaining multiple observation items that the patient needs to focus on monitoring, and calculates the new collection frequency of each observation item in combination with the deviation rate of each observation item, so as to ensure that the real-time body data of the patient is collected more comprehensively and an alarm can be issued more timely.

[0041] As Figure 2 shown, this application also provides an intelligent monitoring and warning system for post-vascular surgery, including a collection unit 100, a data unit 200, and a processing unit 300.

[0042] The collection unit 100 collects the patient's body data, the data unit 200 stores and processes the patient's body data, the processing unit 300 executes a method for intelligent monitoring and warning after vascular surgery as described in any one of the first embodiment, and the collection unit 100, the data unit 200, and the processing unit 300 are communicatively connected.

[0043] It should be noted that the acquisition unit 100 includes multiple different types of sensors for acquiring the physical data of different items of the patient. The physical data acquired by the acquisition unit 100 is transmitted to the data unit 200 for storage, and the processing unit 300 can call and process the physical data of the patient in the data unit 200.

[0044] In an alternative embodiment, the data unit 200 adopts a distributed storage system to achieve real-time synchronization and access of the patient's physical data.

[0045] It should be noted that distributed storage is a data storage technology that uses the disk space on each machine in an enterprise through a network and constructs these scattered storage resources into a virtual storage device. The data is scattered and stored in every corner of the enterprise. By adopting distributed storage, the storage efficiency of the data unit 200 is improved.

[0046] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made without departing from the spirit of the present invention within the scope of knowledge possessed by those skilled in the art to which the present invention pertains.

Claims

1. A method for intelligent monitoring and early warning after vascular surgery, characterized in that: include: Setting collection parameters, collecting physical data of multiple patients, calculating the scoring range of the patient status of patients in each age group based on the physical data of the patients, and judging the patient status of each patient; Creating a postoperative monitoring model, inputting the body data and the patient status as a training set into the postoperative monitoring model to train the postoperative monitoring model, and obtaining a trained postoperative monitoring model; Collect the patient's real-time physical data, input the patient's real-time physical data into the trained postoperative monitoring model, obtain the patient's real-time physical status, screen out unqualified physical data based on the patient's real-time physical status, and calculate the deviation rate between the unqualified data items and the qualified interval; Multiple observation items are screened out in order of deviation rates, and the patient's acquisition parameters are reset based on the deviation rates of the observation items.

2. The intelligent monitoring and early warning method for postoperative vascular surgery according to claim 1, characterized in that: Set collection parameters, collect physical data of multiple patients, calculate the score range of the patient status of each age group of patients based on the physical data of the patients, and judge the patient status of each patient, including: Create a body data table; Set acquisition parameters; the acquisition parameters include acquisition period and acquisition frequency; Based on the acquisition parameters, the physical data of multiple different patients are respectively collected, and all the collected physical data are put into a physical data table; the physical data includes the patient's age, blood pressure, blood oxygen saturation, heart rate and body temperature.

3. The intelligent monitoring and early warning method for postoperative vascular surgery according to claim 2, characterized in that: Setting acquisition parameters, acquiring physical data of multiple patients, calculating the scoring range of the patient status of each age group of patients based on the physical data of the patients, and judging the patient status of each patient, further comprising: Dividing the patient's physical data into multiple data sets according to the patient's age, and calculating the average value of the physical data contained in the data set of each age group; Set a deviation threshold, and calculate the score range corresponding to the patient's status by combining the average value of each physical data with the deviation threshold; Randomly select a patient's physical data from the physical data table; Determine the patient status score range of the patient's physical data, thereby obtaining the patient status of the patient, and put the patient status into the physical data table; Return to randomly select a body data from the body data table until all the body data in the body data table are selected, and obtain the patient status of each patient.

4. The intelligent monitoring and early warning method for postoperative vascular surgery according to claim 3, characterized in that: A postoperative monitoring model is created, and the body data and the patient status are input into the postoperative monitoring model as a training set to train the postoperative monitoring model, and the trained postoperative monitoring model is obtained, including: Create a post-operative monitoring model; The preprocessed body data is divided into multiple feature sets through cluster analysis, and the feature sets are divided into training sets and test sets according to random proportions; Inputting the training set into the postoperative monitoring model, so that the postoperative monitoring model learns the corresponding relationship between the body data and the patient's status, and obtains a trained postoperative monitoring model; The test set is input into the trained postoperative monitoring model, the patient prediction status output by the trained postoperative monitoring model is obtained, and it is determined whether the trained postoperative monitoring model is trained.

5. The intelligent monitoring and early warning method for postoperative vascular surgery according to claim 4, characterized in that: The preprocessed body data is divided into multiple feature sets through cluster analysis, and the feature sets are divided into training sets and test sets according to random proportions, including: Randomly select K data from all preprocessed body data as initial cluster centers; Calculate the distance between other preprocessed body data and each cluster center, assign other preprocessed body data to the initial cluster center closest to it, and obtain K clusters; For each cluster, recalculate the mean distance of all sample points in each cluster, and use the location of the mean as the new cluster center; Set distance threshold; Calculate whether the distance between the new cluster center and the initial cluster center is greater than the distance threshold; If the distance between the new cluster center and the initial cluster center is greater than the distance threshold, return to calculate the distance between other preprocessed body data and each cluster center until the distance between the new cluster center and the initial cluster center is less than or equal to the distance threshold, and K clusters are obtained; the K clusters are used as K feature sets, and the sample points in each cluster are used as feature samples in the feature set.

6. The intelligent monitoring and early warning method for postoperative vascular surgery according to claim 5, characterized in that: The training set is input into the postoperative monitoring model, so that the postoperative detection model learns the corresponding relationship between the body data and the patient's status, and the trained postoperative monitoring model is obtained, including: Randomly select a feature sample from the training set; A coupling relationship between body data and patient status is established based on the feature sample, and the body data, patient status, and the coupling relationship between body data and patient status are input into a postoperative monitoring model as a training sample, so that the postoperative monitoring model learns the corresponding relationship between body data and patient status; Return to randomly select a feature sample from the training set until all feature samples in the training set have been selected to obtain the trained postoperative monitoring model.

7. The intelligent monitoring and early warning method for postoperative vascular surgery according to claim 6, characterized in that: Collect the patient's real-time physical data, input the patient's real-time physical data into the trained postoperative monitoring model, obtain the patient's real-time physical status, screen out unqualified physical data based on the patient's real-time physical status, and calculate the deviation rate between the unqualified data items and the qualified interval, including: Collect the patient's real-time physical data, input the patient's real-time physical data into the trained postoperative monitoring model, and obtain the patient's real-time physical status; Screen out unqualified data items based on the patient's real-time physical status; The deviation rate between each unqualified data item and the qualified interval is calculated by formula 1; Formula 1; in, is the deviation rate between the i-th unqualified data item and the qualified interval, is the real-time data of the i-th data item, is the standard data of the i-th data.

8. The intelligent monitoring and early warning method for postoperative vascular surgery according to claim 7, characterized in that: Multiple observation items are screened out in order of deviation rate, and the patient's collection parameters are reset based on the deviation rate of the observation items, including: Sort all unqualified data items in order of deviation rate; Screen out the first N unqualified data items; record the first N unqualified data items as observation items; Obtain the original collection frequency of each observation item in order according to the deviation rate of the observation item; Combined with the deviation rate of the observation items, the new collection frequency of the observation items is calculated by formula 2; Formula 2; in, is the frequency of new acquisitions of observation items, is the original acquisition frequency of the observation item, is the deviation rate of the observed items; The collection nodes of each observation item are recalculated based on the new collection frequency, and real-time body data is collected according to the new collection nodes.

9. An intelligent monitoring and early warning system for postoperative vascular surgery, characterized in that: include: A collection unit, through which the patient's physical data is collected; a data unit, through which the patient's physical data is stored and processed; A processing unit, through which a method for intelligent monitoring and early warning after vascular surgery as described in any one of claims 1 to 8 is executed, and the acquisition unit, the data unit and the processing unit are communicatively connected.

10. The intelligent monitoring and early warning system for postoperative vascular surgery according to claim 9, characterized in that: The data unit adopts a distributed storage system to achieve real-time synchronization and access to the patient's physical data.

Images