A method for automatic extraction of preoperative anesthesia visit elements and risk item generation

Abstract

This invention discloses a method for automatically extracting elements and generating risk items during preoperative anesthesia visits, belonging to the field of preoperative risk generation technology. It addresses the problem of the inability to provide early warnings of potential anesthesia risks. The method involves analyzing visit update information of the target within a statistical timeframe, filtering update elements based on element labeling levels, constructing association links based on these update elements, retrieving element change records from a historical database, calculating element co-occurrence and occurrence counts, generating element association features and linkage weights, setting link activation coefficients to filter and mark association links, merging links based on the connection relationships of update elements to generate target association links, calling an element label database to count the number of elements of the same category and calculating an abnormal clustering index, generating element clustering status based on element abnormality frequency, and determining whether to generate an abnormality alert. This improves the accuracy and stability of preoperative anesthesia visit information.

Description

Technical Field

[0001] This invention relates to the field of preoperative risk generation technology, and more specifically, to a method for automatically extracting elements and generating risk items during preoperative anesthesia visits. Background Technology

[0002] In the existing preoperative anesthesia visit process, anesthesiologists usually need to conduct a comprehensive assessment of the patient's medical history, vital signs, laboratory test results, etc., in order to determine the anesthesia risk.

[0003] The existing technology has the following shortcomings:

[0004] Currently, existing technologies rely on manual or simple rule-based methods for processing preoperative visit information, which cannot detect potential cross-field correlations and linkage risks between visit elements. They also lack element association feature analysis based on historical change records and abnormal clustering quantification mechanisms, resulting in the inability to provide early warnings of potential anesthesia risks and insufficient accuracy and stability of visit information. Therefore, this paper proposes a method for automatic extraction of preoperative anesthesia visit elements and generation of risk items.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method for automatically extracting preoperative anesthesia visit elements and generating risk items. This method addresses the problems mentioned in the background art by employing automatic extraction of visit update elements, construction of association links based on historical element changes, calculation of linkage weights, and analysis of abnormal clustering indices.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for automatically extracting preoperative anesthesia visit elements and generating risk items, comprising the following steps:

[0008] Step S1: Set the statistical time, call the information database to obtain the visit update information of the visit target within the statistical time, retrieve the element labeling level of the visit target, and filter and extract the visit update elements in combination with the visit update information.

[0009] Step S2: Generate related links using visit update elements, access the historical database to obtain element change data of the visit target, evaluate element association characteristics based on element change data, and generate linkage weights for related links based on element association characteristics.

[0010] Step S3: Set the link activation coefficient, filter and mark related links in combination with linkage weights, perform link merging on the marked related links to generate target related links, and count the number of elements of the same category in each target related link.

[0011] Step S4: Calculate the abnormal clustering index based on the number of elements of the same category, retrieve the abnormal frequency of elements in the visited and updated elements, and combine the abnormal clustering index to assess the element clustering status of the target association link and determine whether to generate an abnormal prompt.

[0012] In a preferred embodiment, in step S1, a statistical time is preset, which is used to limit the range of updated records that participate in the analysis during the visit to the target.

[0013] Based on the time interval of the statistical period, a request is sent to the information database to retrieve and obtain all visit update information corresponding to the visit target within the statistical period.

[0014] The visit update information consists of visit content change entries recorded in the information database in chronological order.

[0015] In a preferred embodiment, in step S1, the feature labeling level of the visited target is invoked to indicate the sensitivity of different visited features in the anomaly identification process;

[0016] The visit update information is parsed, and the changes in the fields of each visit update information are read one by one. The changes before and after are mapped to the corresponding visit requirements.

[0017] All visit elements are filtered according to their element labeling level, and those elements with an element labeling level higher than the preset level threshold are selected as visit update elements.

[0018] In a preferred embodiment, in step S2, all visit update elements contained in the current visit update event are structured and organized to form an element set for link construction;

[0019] Combine any two visit update elements in the element set to form element association pairs;

[0020] After generating feature association pairs, access the historical database to obtain feature change data of the visit target during previous visits.

[0021] In a preferred embodiment, in step S2, the feature change data includes the number of times the feature co-occurs and the number of times the feature appears;

[0022] Count the number of times any two visit elements co-occur in the historical record, i.e., the element co-occurrence count;

[0023] Record the number of times each visit element appears independently throughout the entire historical process, i.e., the number of times the element appears;

[0024] Add the occurrence counts of two co-occurring visit elements to get the total occurrence count of the element. Calculate the ratio of the co-occurrence count to the total occurrence count of the element to obtain the element association feature.

[0025] A standardized formula is used, which divides the element association feature by the maximum value of the element association feature to obtain the linkage weight.

[0026] In a preferred embodiment, in step S3, a preset link activation coefficient is established, and the preset link activation coefficient is compared with the linkage weight of each associated link to filter and mark the associated links.

[0027] If the preset link activation coefficient is less than the linkage weight, the associated link is marked.

[0028] Conversely, no associated links are marked;

[0029] Based on the link merging process of the marked association links to generate the target association link, the visit update elements in the marked association links are compared one by one. If there are the same visit update elements in the marked association links, the same visit update elements are used as the link connection nodes. The marked association links are connected in sequence at the connection nodes to form the target association link.

[0030] In a preferred embodiment, in step S3, the feature classification tags of the visit update elements are retrieved from the feature tag database, and the visit update elements with the same feature classification tags in the target association link are grouped into the same category group.

[0031] The number of visit update elements in each category group is counted in the target association link as the number of elements in the same category.

[0032] In a preferred embodiment, in step S4, the ratio of the number of elements of the same category to the total number of visited and updated elements in the same target association link is used as the abnormal clustering index.

[0033] The frequency of element anomalies under the visit update elements is retrieved from the element parameter library. The frequency of element anomalies refers to the cumulative number of times that the visit update elements have triggered anomaly prompt events in historical visits.

[0034] The average frequency of anomalies in visit update elements of the same category is used to obtain the anomaly frequency of elements of the same category.

[0035] The abnormal frequency coefficient is obtained by standardizing the abnormal frequency of elements of the same category.

[0036] In a preferred embodiment, in step S4, the product of the abnormal frequency coefficient and the abnormal clustering index is used as the clustering coefficient of elements of the same category.

[0037] The maximum value of the feature clustering coefficient is taken as the feature clustering state of the target association link;

[0038] If the clustering state of elements is greater than the preset clustering threshold, then an abnormal prompt is generated for the target association link.

[0039] Conversely, if the condition is not met, then no abnormal prompt will be generated for the target-related link.

[0040] The technical effects and advantages of this invention are as follows:

[0041] This invention analyzes visit update information of the target within a statistical period, obtains visit update elements by combining element labeling hierarchy, constructs related links based on visit update elements, obtains element change records from historical databases, calculates element co-occurrence and occurrence frequency, generates element association features and linkage weights, sets link activation coefficients to filter and mark related links, merges links according to the connection relationship of visit update elements to generate target related links, calls element label database to count the number of elements of the same category and calculates abnormal clustering index, generates element clustering status based on element abnormal frequency and determines whether to generate abnormal prompts, realizes early warning of potential risks caused by cross-field linkage, improves the accuracy and stability of preoperative anesthesia visit information, and provides more reliable data support for anesthesia risk assessment. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating the implementation of a method for automatically extracting preoperative anesthesia visit elements and generating risk items according to the present invention.

[0043] Figure 2 This is a schematic diagram illustrating the steps of a method for automatically extracting elements and generating risk items during preoperative anesthesia visits according to the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] This invention analyzes visit update information of the target within a statistical period, obtains visit update elements by combining element labeling hierarchy, constructs related links based on visit update elements, obtains element change records from historical databases, calculates element co-occurrence frequency and occurrence frequency, generates element association features and linkage weights, sets link activation coefficients to filter and mark related links, merges links according to the connection relationship of visit update elements to generate target related links, calls element label database to count the number of elements of the same category and calculates abnormal clustering index, generates element clustering status based on element abnormal frequency and determines whether to generate abnormal prompts, realizes early warning of potential risks caused by cross-field linkage, and improves the accuracy and stability of preoperative anesthesia visit information.

[0046] Example 1, as Figures 1 to 2 As shown, a method for automatically extracting elements and generating risk items during preoperative anesthesia visits includes the following steps:

[0047] Step S1: Set the statistical time, call the information database to obtain the visit update information of the visit target within the statistical time, retrieve the element labeling level of the visit target, and filter and extract the visit update elements in combination with the visit update information.

[0048] Step S2: Generate related links using visit update elements, access the historical database to obtain element change data of the visit target, evaluate element association characteristics based on element change data, and generate linkage weights for related links based on element association characteristics.

[0049] Step S3: Set the link activation coefficient, filter and mark related links in combination with linkage weights, perform link merging on the marked related links to generate target related links, and count the number of elements of the same category in each target related link.

[0050] Step S4: Calculate the abnormal clustering index based on the number of elements of the same category, retrieve the abnormal frequency of elements in the visited and updated elements, and combine the abnormal clustering index to assess the element clustering status of the target association link and determine whether to generate an abnormal prompt.

[0051] The specific implementation is as follows:

[0052] In step S1, during the preoperative anesthesia visit, multiple entries and dynamic updates are required for the patient's medical history, allergies, medication records, and physiological indicators. In actual visit records, anomalies may occur, such as fields being repeatedly modified, multiple fields forming abnormal combinations in the same update, and multiple elements unexpectedly interacting during continuous updates. These anomalies are triggered by multiple visit elements linked together over time, leading to inconsistencies and structural disorder in the visit information. Therefore, this embodiment automatically extracts visit elements, constructs element relationships, identifies element clustering patterns, and generates risk warnings during the visit update process. It intelligently analyzes the visit content from the perspective of data structure evolution to solve the problem of existing technologies failing to effectively identify abnormal evolution of visit information.

[0053] First, a statistical timeframe is preset. This timeframe is used to limit the scope of updated records that are included in the analysis during the visit to the target audience.

[0054] Based on the time interval of the statistical time, a call request is initiated to the information database to retrieve and obtain all visit update information corresponding to the visit target within the statistical time. The visit update information consists of visit content change entries recorded in the information database in chronological order. Each visit update information includes the values ​​of the field content before and after the change, the field category identifier, and the source information that triggered the content change, which is used to characterize the content changes of the visit target in one or more visits.

[0055] It should be noted that the information database is an information management carrier used to store data related to preoperative anesthesia visits. It records various types of visit data associated with the visit target in a structured or semi-structured manner, including the original record of the visit content, the updated entries of the visit content at different time points, field category identifiers, field values, change source information, and record generation time, etc.

[0056] The feature annotation level of the visited target is retrieved. The feature annotation level is stored in the attribute record of the visited target and is used to indicate the sensitivity of different visited elements in the anomaly identification process. By statistically analyzing the frequency of occurrence of each element of the visited target in past visit records, the co-occurrence intensity with other elements, and the participation ratio in historical anomaly events, a comprehensive score value of the element is formed according to a preset weighted summation formula. Then, the element is divided into different annotation levels according to the preset score range, thereby obtaining the feature annotation level of the visited target.

[0057] The visit update information is parsed, and the changes in the field content of each visit update information are read one by one. The content before and after the change is mapped to the corresponding visit elements. Then, all visit elements are filtered according to the element labeling level. Visit elements with an element labeling level higher than the preset level threshold are selected as visit update elements.

[0058] It should be noted that the preset level thresholds are quantitatively set based on the statistical distribution results of historical visit data. First, the numerical ranges of all element labeling levels are statistically analyzed to form a level distribution sequence, and the trigger rate, false trigger rate, and correlation strength with the anomaly link of each element labeling level in historical anomaly identification tasks are calculated. Subsequently, a level performance curve is constructed based on the above quantitative indicators. By analyzing the effective contribution of different element labeling levels in actual anomaly identification, the level critical point that can improve the identification accuracy and maintain a reasonable recall rate is determined, and this is used as the preset level threshold.

[0059] In step S2, after obtaining the visit update elements, association links for subsequent analysis are generated based on these elements. These association links describe the combination relationships between multiple visit elements triggered simultaneously in the same visit update. All visit update elements included in the current visit update event are structured and organized to form an element set for link construction. Subsequently, any two visit update elements in the element set are paired to form element association pairs as association links. Each element association pair represents the co-occurrence relationship of two elements in the same visit update behavior.

[0060] After generating feature association pairs, access the historical database to obtain feature change data of the visit target during previous visits.

[0061] It should be noted that the historical database is used to store the complete trajectory of changes in the elements of the target being visited throughout all past visit events, including the set of visit elements that appeared in each visit event and their corresponding time records.

[0062] The system retrieves the combination of elements that appeared in each visit event from the historical database, counts the number of times any two visit elements co-occur in the historical records (i.e., element co-occurrence count), and records the number of times each visit element appeared independently throughout the entire historical process (i.e., element occurrence count). The element co-occurrence count is used to characterize the actual linkage strength between two visit elements in historical visits, while the element occurrence count is used to characterize the frequency of use of each visit element.

[0063] Add the occurrence counts of two co-occurring visit elements to obtain the total occurrence count of the element. Calculate the ratio of the co-occurrence count to the total occurrence count of the element to obtain the element association feature, which represents the relative co-occurrence ratio of the two visit elements in historical visit data. The calculation formula is as follows:

[0064] ;

[0065] in, as elements and elements The characteristics of element association, as elements and elements The number of times the elements co-occur. as elements The number of times the element appears, as elements The number of times the element appears.

[0066] After obtaining the element association features, they are further converted into linkage weights for link filtering. The linkage weights are calculated using a standardized formula to eliminate the magnitude differences in the association feature values ​​of different elements, enabling them to participate in comparisons on a uniform scale. The specific calculation method is as follows:

[0067] ;

[0068] in, as elements and elements The linkage weight, This is the maximum value of the feature association features calculated across all feature pairs, used for global normalization of the association features.

[0069] The above process enables the quantification and structural representation of the potential linkage relationships between visit update elements. It generates element association features and calculates linkage weights through historical element change data, quantifies and identifies the actual association strength of different elements in the historical visit process, and ensures that abnormal clustering analysis and risk warnings are only performed on links with actual statistical associations, thereby improving the accuracy and timeliness of anomaly identification.

[0070] In step S3, a preset link activation coefficient is established, and the preset link activation coefficient is compared with the linkage weight of each associated link to filter and mark the associated links.

[0071] If the preset link activation coefficient is less than the linkage weight, the associated link is marked.

[0072] Conversely, no associated links are marked;

[0073] The preset link activation coefficient is compared with the linkage weight to ensure that the marked associated links have a high structural correlation. When the linkage weight is greater than the preset link activation coefficient, it indicates that the associated links show a high intensity of coordinated change. By using the combined labeling of associated links as a screening method, potential structural shift trends can be identified from scattered field changes, providing input for generating clustering anomaly prompts and making the judgment of abnormal clustering more accurate. When the linkage weight is less than or equal to the preset link activation coefficient, it indicates that the combination of visit update elements of the associated links is a low-frequency combination and is not labeled.

[0074] It should be explained that the preset link activation coefficient is used to quantify the minimum linkage strength required for the association link to participate in the analysis. It can be set according to the co-occurrence frequency of visit update elements in historical update records, the stability of element changes, and the distribution of historical linkage features.

[0075] Based on the link merging process of the marked association links to generate the target association link, the visit update elements in the marked association links are compared one by one. If there are the same visit update elements in the marked association links, it indicates that the link units have inherent continuity. The marked association links are then spliced ​​together in sequence according to the order of the visit update elements. The same visit update elements are used as the link connection nodes to realize the sequential connection of the marked association links at the connection nodes, so that the originally scattered association relationship is transformed into a continuous relationship chain that can extend from the starting point to the end point.

[0076] For example, the associated links are marked as follows: , Both of the tag-associated links contain visit update elements. Visit update elements These are common elements belonging to two tag-associated links, therefore the visit update element will be included. Treating each element as a connecting node, and based on the order of elements, the two links are sequentially spliced ​​together at the connecting node; after splicing, the two originally separate tag-associated links are merged into a continuous link structure. This forms a target association link that can fully express the path of changes in visit update elements.

[0077] By using target-related links as the overall structural basis, potential structural deviations in visit records can be identified, enabling early warning of potential anomalies in changes to visit update elements. Anomalies can be determined from the perspective of structural correlation, improving the accuracy and interpretability of anomaly identification, ensuring the quality and completeness of preoperative anesthesia visit information, and providing more reliable basic data for subsequent anesthesia risk assessment.

[0078] The feature classification tags of the visit update elements are retrieved from the feature tag database. The visit update elements with the same feature classification tags in the target association link are divided into the same category group. The number of visit update elements in each category group in the target association link is counted as the number of elements of the same category.

[0079] When the number of elements of a certain category in the target association link increases, it indicates that the information corresponding to the category has undergone concentrated changes. The number of elements of the same category serves as a clue to judge structural offset and captures potential abnormal changes in the overall link structure.

[0080] It should be noted that the feature label database is a pre-built tagged data resource library used to store the category label information of each visit feature.

[0081] In step S4, within the same target association link, the ratio of the number of elements of the same category to the total number of visited and updated elements is used as the abnormal clustering index.

[0082] The abnormal clustering index reflects the degree of concentration of this type of element in the target association link. When the abnormal clustering index is large, it indicates that the proportion of this type of element has increased significantly, which may indicate the risk of structural deviation. When the abnormal clustering index is small, it indicates that the distribution of this type of element in the link is relatively balanced, and the link structure is still within the normal range.

[0083] The frequency of element anomalies under the visit update elements is retrieved from the element parameter library. The frequency of element anomalies refers to the cumulative number of times that the visit update elements have triggered anomaly prompt events in historical visits.

[0084] The higher the frequency of element anomalies, the more likely the visit update elements are to participate in forming abnormal combinations or continuous changes during the historical visit process, and the greater the impact on the overall structural stability of preoperative anesthesia visit information; the lower the frequency of element anomalies, the smaller the disturbance effect of visit update elements on the evolution of visit record structure.

[0085] The average frequency of anomalies in visit update elements of the same category is used to obtain the anomaly frequency of elements of the same category.

[0086] The abnormal frequency coefficient is obtained by standardizing the abnormal frequency of elements of the same category.

[0087] The product of the abnormal frequency coefficient and the abnormal clustering index is used as the clustering coefficient of elements of the same category.

[0088] The maximum value of the feature clustering coefficient is taken as the feature clustering state of the target association link;

[0089] The clustering status of elements is compared with a preset clustering threshold to determine whether to generate an anomaly alert for the target association link:

[0090] If the clustering state of elements is greater than the preset clustering threshold, then an abnormal prompt is generated for the target association link.

[0091] Conversely, if the condition is not met, then no abnormal prompt will be generated for the target-related link.

[0092] When the clustering of elements is greater than the preset clustering threshold, it indicates that there is a concentrated shift of category elements in the target association link, and the link structure has an abnormal evolution trend. An abnormal prompt is generated for the target association link. Conversely, when the clustering of elements is less than or equal to the preset clustering threshold, it indicates that the link structure maintains a normal change pattern and no abnormal prompt is needed.

[0093] This step standardizes the frequency of anomalies in the same category and calculates the combined effect with the anomaly clustering index to identify the structural anomaly trend caused by the superposition of category concentration shift and historical anomaly sensitivity. This enables early warning of the abnormal evolution of visit information, captures potential risks brought about by cross-field linkage, and improves the accuracy of preoperative anesthesia visit information.

[0094] It should be noted that the standardization methods include, but are not limited to, standard linear transformation based on interval scaling, statistical Z-Score standardization method, or normalization method based on nonlinear mapping function. The application methods of standardization will not be elaborated here. The element parameter library is a database of element statistics and risk characteristics pre-built based on historical visit records. The preset element clustering threshold can be set according to the distribution of historical abnormal events and the visit target's sensitivity to abnormalities.

[0095] Finally, it should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0096] Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0097] In this document, the singular forms “a,” “an,” and “the” may also include the plural forms unless the context clearly indicates otherwise. It should also be understood that terms such as “comprising / including” or “having” specify the presence of the stated features, integrals, steps, operations, components, parts, or combinations thereof, but do not preclude the possibility of the presence or addition of one or more other features, integrals, steps, operations, components, parts, or combinations thereof. Meanwhile, the term “and / or” as used in this specification includes any and all combinations of the associated listed items.

[0098] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0099] The above description of the disclosed embodiments will enable those skilled in the art to make or use various modifications to these embodiments. It will be readily apparent to those skilled in the art that the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for automatically extracting elements and generating risk items during preoperative anesthesia visits, characterized in that: Includes the following steps: Step S1: Set the statistical time, call the information database to obtain the visit update information of the visit target within the statistical time, retrieve the element labeling level of the visit target, and filter and extract the visit update elements in combination with the visit update information. In step S1, the feature annotation level of the visited target is invoked to indicate the sensitivity of different visited features in the anomaly identification process; The visit update information is parsed, and the changes in the fields of each visit update information are read one by one. The changes before and after are mapped to the corresponding visit elements. All visit elements are filtered according to the element labeling level, and visit elements with an element labeling level higher than the preset level threshold are selected as visit update elements. Step S2: Generate related links using visit update elements, access the historical database to obtain element change data of the visit target, evaluate element association characteristics based on element change data, and generate linkage weights for related links based on element association characteristics. Step S3: Set the link activation coefficient, filter and mark related links in combination with linkage weights, perform link merging on the marked related links to generate target related links, and count the number of elements of the same category in each target related link. In step S3, a preset link activation coefficient is established, and the preset link activation coefficient is compared with the linkage weight of each associated link to filter and mark the associated links. If the preset link activation coefficient is less than the linkage weight, the associated link is marked. Conversely, no associated links are marked; Based on the link merging process of the marked association links to generate the target association link, the visit update elements in the marked association links are compared one by one. If there are the same visit update elements in the marked association links, the same visit update elements are used as the link connection nodes. The marked association links are connected in the order of the connection nodes to form the target association link. Step S4: Calculate the abnormal clustering index based on the number of elements of the same category, retrieve the abnormal frequency of elements in the visited and updated elements, and combine the abnormal clustering index to assess the element clustering status of the target association link and determine whether to generate an abnormal prompt.

2. The method for automatically extracting preoperative anesthesia visit elements and generating risk items according to claim 1, characterized in that: In step S1, a preset statistical time is used to limit the scope of updated records that participate in the analysis during the visit to the target. Based on the time interval of the statistical period, a request is sent to the information database to retrieve and obtain all visit update information corresponding to the visit target within the statistical period. The visit update information consists of visit content change entries recorded in the information database in chronological order.

3. The method for automatically extracting preoperative anesthesia visit elements and generating risk items according to claim 1, characterized in that: In step S2, all visit update elements contained in the current visit update event are structured and organized to form an element set for link construction; Combine any two visit update elements in the element set to form element association pairs; After generating feature association pairs, access the historical database to obtain feature change data of the visit target during previous visits.

4. The method for automatically extracting preoperative anesthesia visit elements and generating risk items according to claim 3, characterized in that: In step S2, the feature change data includes the number of times features co-occur and the number of times features appear; Count the number of times any two visit update elements co-occur in the historical record, i.e., the number of element co-occurrences; Record the number of times each visit update element appears independently throughout the entire historical process, i.e., the number of times the element appears; Add the occurrence counts of the two co-occurring visit update elements to get the total occurrence count of the elements. Calculate the ratio of the co-occurrence count to the total occurrence count of the elements to obtain the element association feature. The linkage weight is obtained by using a standardized formula, which is the maximum value of the element association feature divided by the element association feature.

5. The method for automatically extracting preoperative anesthesia visit elements and generating risk items according to claim 1, characterized in that: In step S3, the feature classification tags of the visit update features are retrieved from the feature tag database, and the visit update features with the same feature classification tags in the target association link are grouped into the same category group; The number of visit update elements in each category group is counted in the target association link as the number of elements in the same category.

6. The method for automatically extracting preoperative anesthesia visit elements and generating risk items according to claim 5, characterized in that: In step S4, within the same target association link, the ratio of the number of elements of the same category to the total number of visited and updated elements is used as the abnormal clustering index. The frequency of element anomalies under the visit update elements is retrieved from the element parameter library. The frequency of element anomalies refers to the cumulative number of times that the visit update elements have triggered anomaly prompt events in historical visits. The average frequency of anomalies in visit update elements of the same category is used to obtain the anomaly frequency of elements of the same category. The abnormal frequency coefficient is obtained by standardizing the abnormal frequency of elements of the same category.

7. The method for automatically extracting preoperative anesthesia visit elements and generating risk items according to claim 6, characterized in that: In step S4, the product of the abnormal frequency coefficient and the abnormal clustering index is used as the clustering coefficient of elements of the same category. The maximum value of the feature clustering coefficient is taken as the feature clustering state of the target association link; If the clustering state of elements is greater than the preset clustering threshold, then an abnormal prompt is generated for the target association link. Conversely, if the condition is not met, then no abnormal prompt will be generated for the target-related link.

Images