Postoperative follow-up information recording system for patients with regional nerve block anesthesia

Abstract

This invention relates to the field of patient-related medical data processing technology, specifically to a postoperative follow-up information recording system for patients undergoing regional nerve block anesthesia. The invention first acquires all follow-up record tables and the word segments of all follow-up records within them; further, it acquires the record similarity coefficients between adjacent follow-up record tables and the column record similarity coefficients between the same columns; then, based on the frequency of occurrence of follow-up record word segments, it determines the recording method for each follow-up record word segment in each follow-up record table; finally, it performs follow-up recording based on the recording method of each follow-up record word segment in each follow-up record table. This invention determines the recording method based on the textual similarity and sentiment information of follow-up record word segments between adjacent follow-up processes, combined with the frequency of occurrence of each follow-up record word segment. Compared to existing methods, this reduces redundancy in the multiple postoperative follow-up information entry processes for patients undergoing regional nerve block anesthesia, thereby improving the recording effect of follow-up information.

Description

Technical Field

[0001] This invention relates to the field of patient-related medical data processing technology, specifically to a postoperative follow-up information recording system for patients undergoing local nerve block anesthesia. Background Technology

[0002] Local nerve block anesthesia is a procedure that reduces or eliminates pain in the surgical area by injecting local anesthetic drugs into the target nerve or interfascial space. In recent years, with the widespread use of bedside ultrasound in anesthesiology, ultrasound-guided nerve block anesthesia has become an important part of clinical anesthesia. However, nerve block anesthesia carries the risk of postoperative complications, such as incomplete block, nerve injury, infection, and bleeding. Therefore, postoperative follow-up is necessary to help physicians monitor patient recovery, promptly identify and manage potential complications, assess treatment effectiveness, and provide necessary rehabilitation guidance.

[0003] Compared to general anesthesia, postoperative follow-up after nerve block anesthesia often receives insufficient attention from medical staff. Current follow-up models typically involve multiple follow-ups for each patient, with the information collected at each visit sequentially entered into an electronic health record system. However, the follow-up information in these multiple visits to the same patient may be identical, leading to redundancy and inefficient data entry, which in turn affects subsequent retrieval. An accurate, efficient, and concise postoperative follow-up system would be more conducive to clinical research and help medical personnel explore better anesthesia strategies. Summary of the Invention

[0004] To address the problem of poor recording quality of follow-up information in existing technologies, the present invention aims to provide a postoperative follow-up information recording system for patients undergoing local nerve block anesthesia. The specific technical solution adopted is as follows:

[0005] A postoperative follow-up information recording system for patients undergoing regional nerve block anesthesia, the system comprising:

[0006] The follow-up information collection module is used to obtain the follow-up record table after each postoperative follow-up visit of the patient; it obtains word segments of all follow-up records in each column of each follow-up record table and constructs a set of word segments of records for each column;

[0007] The follow-up information analysis module is used to obtain the column record similarity coefficient between adjacent follow-up record tables based on the similarity between the record word segments of the same column and the sentiment information of each follow-up record word in the record word segments. It then combines the column record similarity coefficients between all columns to obtain the record similarity coefficient between adjacent follow-up record tables. Finally, based on the record similarity coefficients between adjacent follow-up record tables and the column record similarity coefficients between columns, and combined with the frequency of occurrence of each follow-up record word in the corresponding column, it determines the recording method for each follow-up record word in each follow-up record table.

[0008] The follow-up information processing module is used to record follow-up information according to the word segmentation method of each follow-up record in each follow-up record table.

[0009] Furthermore, the method for obtaining the segmented words of the follow-up records includes:

[0010] In each follow-up record table, the jieba word segmentation tool is used to segment the text information in each column to obtain all the segmented information in each column.

[0011] Based on the preset word segmentation set under each column, all follow-up record word segments are selected from all the information word segments in each column; the preset word segmentation set includes all template word segments in each column of the follow-up record form template.

[0012] Furthermore, the method for selecting all follow-up record word segments from all the information word segments in each column based on the preset word segmentation set under each column includes:

[0013] In each follow-up record table, any column is taken as the target column, and the preset word segmentation set under the target column is taken as the target word segmentation set; the information word segments in the target column are used as sequence elements in the order of appearance to construct the information word segmentation sequence of the target column.

[0014] Each information segment in the information segmentation sequence is treated as a one-dimensional segmentation set for iterative analysis. All concatenated information segments under each dimension are obtained sequentially to construct the corresponding segmentation set under the dimension until the preset dimension is met. Under each dimension, non-template segmented words are selected based on the Jaccard coefficient between the segmentation set corresponding to each concatenated information segmented word and the preset segmentation set. The concatenated information segmented words in each dimension are obtained by concatenating adjacent non-template segmented words in the information segmentation sequence according to the dimension requirements from all non-template segmented words in the previous dimension.

[0015] All non-template segmented words are used as follow-up record segmented words.

[0016] Furthermore, the method for filtering out non-template word segmentation includes:

[0017] In each dimension, the Jaccard coefficient between each word segmentation set and the preset word segmentation set is calculated. When the Jaccard coefficient is 0, the information word segmentation in the corresponding word segmentation set or all the constituent information word segmentation of the corresponding concatenated information word segmentation are used as template word segmentation.

[0018] In the information segmentation sequence, all information segments other than all template segments selected under each dimension are treated as non-template segments.

[0019] Furthermore, the method for obtaining the similarity coefficient of the column records includes:

[0020] Between adjacent follow-up record tables, the Jaccard coefficient between the word segmentation sets of the records in the same column is used as the text similarity coefficient;

[0021] In each of the record word segments, a set sentiment score is obtained based on the sentiment information of all follow-up record word segments; between adjacent follow-up record tables, the negative correlation mapping result of the difference in the set sentiment scores between the record word segments of the same column is used as the sentiment similarity coefficient;

[0022] The product of the emotional similarity coefficient and the text similarity coefficient is used as the column record similarity coefficient between the same columns in the corresponding adjacent follow-up record tables.

[0023] Furthermore, the method for obtaining the aggregate sentiment score includes:

[0024] The sentiment score of each follow-up record segment is obtained based on the relevant dictionary, and the sum of the sentiment scores of all follow-up record segments is used as the set sentiment score of the corresponding record segment set.

[0025] Furthermore, the method for obtaining the record similarity coefficient includes:

[0026] Between adjacent follow-up record tables, the average of the similarity coefficients of the records in the same column is used as the record similarity coefficient between adjacent follow-up record tables.

[0027] Furthermore, the method for determining the recording method of word segmentation for each follow-up record in each follow-up record table includes:

[0028] If the similarity coefficient between each follow-up record table and the previous adjacent follow-up record table is less than a preset threshold, each follow-up record in each follow-up record table will be segmented and recorded independently.

[0029] If the similarity coefficient between each follow-up record table and the previous adjacent follow-up record table is greater than or equal to a preset threshold, semantic matching is performed on the follow-up record word segments between the word segments of the same column; based on the similarity coefficient between the column records of the same column and the frequency of occurrence of the matched follow-up record word segments in the corresponding column, the recording method of each follow-up record word segment in each follow-up record table is determined.

[0030] Furthermore, the method for semantic matching of follow-up record segments among record segmentation sets of the same column includes:

[0031] The word vectors of each follow-up record in each column of each follow-up record table are obtained based on the Word2Vec algorithm; the word vectors of any follow-up record in each follow-up record table are used as the target word.

[0032] In the adjacent follow-up record table above each follow-up record table, calculate the cosine similarity between the word vector of the target word and the word vector of each follow-up record word in the same column, and take the follow-up record word with the maximum cosine similarity as the semantic matching word of the target word.

[0033] Furthermore, the method for determining the word segmentation record method for each follow-up record in each follow-up record table based on the similarity coefficient between the column records of the same column and the frequency of occurrence of the matching follow-up record word segmentation in the corresponding column includes:

[0034] The similarity coefficient between the same column of each follow-up record table and the previous adjacent follow-up record table is used as the first weight; the negative correlation normalization result of the first weight is used as the second weight; and the semantic matching word segment of the target word in the corresponding column of the previous adjacent follow-up record table is used as the reference word segment.

[0035] In each column of the previous adjacent follow-up record table, the frequency of occurrence of the reference word in the corresponding column is weighted using the first weight, and the weighted result is used as the first merging parameter of the target word; in each column of each follow-up record table, the frequency of occurrence of the target word in the corresponding column is weighted using the second weight, and the weighted result is used as the second merging parameter of the target word.

[0036] If the first merging parameter is less than or equal to the second merging parameter, the target word segment is merged and recorded in the previous adjacent follow-up record table; if the first merging parameter is greater than the second merging parameter, the target word segment is treated as a redundant record word and is not updated in the previous adjacent follow-up record table.

[0037] The present invention has the following beneficial effects:

[0038] This invention first obtains follow-up record forms after each postoperative follow-up visit and extracts word segments of all follow-up records in each column of each record form, constructing a word segment set for each column. This provides a foundation for subsequent analysis and evaluation of column record similarity coefficients and overall record similarity coefficients. Furthermore, between adjacent follow-up record forms, based on the similarity between word segment sets of records in the same column, and combined with the sentiment information of each follow-up record word segment in the word segment set, the column record similarity coefficient is obtained. The column record similarity coefficient combines not only text similarity but also the patient's sentiment information, accurately assessing the similarity of column records in the same column of adjacent follow-up record forms. This, in turn, comprehensively... The similarity coefficients of records in all identical columns are combined to obtain the similarity coefficients between adjacent follow-up record tables. These similarity coefficients reflect the similarity of follow-up records between adjacent tables, facilitating subsequent evaluation of the word segmentation processing method. Based on the similarity coefficients between adjacent follow-up record tables and between identical columns, and considering the frequency of each follow-up record word in its corresponding column (frequency reflecting importance or redundancy), the recording method for each follow-up record word in each table can be accurately determined, reducing redundancy in follow-up information. Finally, follow-up records are made according to the recording method of each follow-up record word in each table. This invention, based on the textual similarity and sentiment information of follow-up record words between adjacent follow-up processes, combined with the frequency of each follow-up record word to determine its recording method, can reduce redundancy and improve the recording effect of follow-up information in the process of multiple postoperative follow-up information entry for patients under local nerve block anesthesia. Attached Figure Description

[0039] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a system block diagram of a postoperative follow-up information recording system for patients undergoing local nerve block anesthesia, provided in one embodiment of the present invention.

[0041] Figure 2 This is a flowchart illustrating a method for obtaining the similarity coefficient of column records according to an embodiment of the present invention. Detailed Implementation

[0042] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a postoperative follow-up information recording system for patients under local nerve block anesthesia proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0044] The following description, in conjunction with the accompanying drawings, details a specific scheme for a postoperative follow-up information recording system for patients undergoing local nerve block anesthesia provided by the present invention.

[0045] Please see Figure 1 The diagram illustrates a system block diagram of a postoperative follow-up information recording system for patients undergoing local nerve block anesthesia, according to an embodiment of the present invention. The system includes a follow-up information acquisition module 101, a follow-up information analysis module 102, and a follow-up information processing module 103.

[0046] The follow-up information collection module 101 is used to obtain the follow-up record table after each postoperative follow-up visit of the patient; to obtain the word segmentation of all follow-up records in each column of each follow-up record table, and to construct the word segmentation set of each column.

[0047] It should be noted that the embodiments of the present invention are aimed at the analysis and processing of postoperative follow-up records of patients with local nerve block anesthesia, hereinafter referred to as patients; in other embodiments, the implementer may also analyze and process postoperative follow-up records of patients with other types of diseases.

[0048] In one embodiment of the present invention, a follow-up record form, recorded by relevant medical personnel after each postoperative follow-up visit, is first collected. Since follow-ups are usually conducted via telephone or in-person interviews, the follow-up record form may be in different formats, such as electronic spreadsheets or paper forms. Paper forms are inconvenient for subsequent analysis and processing; therefore, OCR technology is used to convert the paper form into an editable electronic text format. The follow-up record form contains several columns, each recording postoperative follow-up information such as basic identity information, basic vital signs, postoperative anesthesia orders, and the patient's self-reported physical condition and recovery experience. It should be noted that OCR technology is existing technology and will not be elaborated upon here.

[0049] Considering that the follow-up record form is not completely blank before the patient is followed up, but contains the title of each column, the text information corresponding to the column title will have a certain impact when the follow-up record information is analyzed and processed in the future, so this embodiment of the invention will further obtain the word segments of all follow-up records in each column of each follow-up record form and construct a word segment set of records for each column; the word segments of follow-up records are the text segments corresponding to the follow-up information recorded during the follow-up process, and the word segment set of records includes all follow-up record word segments in the corresponding column.

[0050] Preferably, in one embodiment of the present invention, considering that the follow-up record form not filled out before the follow-up visit is usually a fixed template, its inner column title is also a fixed text segmentation, so the template segmentation corresponding to the column title can be determined first, and then it can be evaluated whether the information segmentation corresponding to all text information in the follow-up record form after the follow-up visit is the follow-up record segmentation; therefore, the method for obtaining the follow-up record segmentation includes:

[0051] In each follow-up record table, the jieba word segmentation tool is used to segment the text information in each column to obtain all the segmented information in each column.

[0052] Based on the preset word segmentation set under each column, all follow-up record word segments are selected from all information word segments of each column; the preset word segmentation set includes all template word segments of each column of the follow-up record form template.

[0053] It should be noted that the application of the jieba word segmentation tool is already existing technology, and will not be elaborated here; after using jieba word segmentation, it is also necessary to remove stop words to obtain all the useful information words in each column.

[0054] The template word segmentation in the preset word segmentation set is determined based on the follow-up record form that was not filled out before the follow-up. That is, the template word segmentation corresponds to the column title word segmentation. If the follow-up record form template remains unchanged, the preset word segmentation set under each column will also remain unchanged. Implementers can manually extract the column title word segmentation in each column to obtain the template word segmentation, which will not be elaborated further.

[0055] In a preferred embodiment of the present invention, considering that during the word segmentation process, the template word segmentation may also be divided into several words, such as the template word segmentation "anesthesia feeling" may be segmented into "anesthesia" and "feeling", which makes it difficult to accurately select the follow-up record word segmentation in the subsequent process; therefore, the method for selecting all follow-up record word segmentation from all information word segmentation of each column according to the preset word segmentation set includes:

[0056] In each follow-up record table, any column is taken as the target column, and the preset word segmentation set under the target column is taken as the target word segmentation set; the information word segments in the target column are used as sequence elements in the order of appearance to construct the information word segmentation sequence of the target column.

[0057] Each information segment in the information segmentation sequence is treated as a one-dimensional segmentation set for iterative analysis. All concatenated information segments under each dimension are obtained sequentially to construct the corresponding segmentation set under the dimension until the preset dimension is met. Under each dimension, non-template segmented words are selected based on the Jaccard coefficient between the segmentation set corresponding to each concatenated information segmented word and the preset segmentation set. The concatenated information segmented words in each dimension are obtained by concatenating adjacent non-template segmented words in the information segmentation sequence according to the dimension requirements from all non-template segmented words in the previous dimension. All non-template segmented words are used as follow-up record segmented words.

[0058] The specific iterative analysis process is as follows:

[0059] Each information segment in the information segmentation sequence is used as a set element to construct a corresponding one-dimensional segmentation set; based on the Jaccard coefficient between each one-dimensional segmentation set and the preset segmentation set, all first non-template segmentations are selected from the information segments corresponding to all one-dimensional segmentation sets.

[0060] In the information segmentation sequence, two first non-template segmented words with adjacent sequence element numbers are concatenated and combined to obtain two-dimensional concatenated information segmented words. Each two-dimensional concatenated information segmented word is used as a set element to construct a corresponding two-dimensional segmented word set. Based on the Jaccard coefficient between each two-dimensional segmented word set and the preset segmented word set, all second non-template segmented words are selected from the constituent information segmented words of the two-dimensional concatenated information segmented words corresponding to all two-dimensional segmented word sets.

[0061] Iteratively obtain concatenated information word segmentation under different dimensions and construct word segmentation sets under corresponding dimensions until the preset dimensions are met; based on the Jaccard coefficient between each word segmentation set under the preset dimensions and the preset word segmentation set, select all non-template word segments from the component information word segments of the concatenated information word segments corresponding to the word segments under all preset dimensions; use all non-template word segments as follow-up record word segments.

[0062] The method for selecting all non-template segmented words from the segmented word sets corresponding to all preset dimensions, considering that the Jaccard coefficient can be used to evaluate the similarity between two sets (a Jaccard coefficient of 0 indicates that the two sets are completely dissimilar, while a coefficient of 0 indicates that they have overlap), involves calculating the Jaccard coefficient between each segmented word set under the preset dimensions and the preset segmented word set. When the Jaccard coefficient is 0, all the constituent information segmented words corresponding to the concatenated information segmented words under the preset dimensions are taken as template segmented words. In the information segmented word sequence, all information segmented words other than all template segmented words selected under all dimensions are taken as non-template segmented words. It should be noted that the Jaccard coefficient is a well-known technique and will not be elaborated upon further.

[0063] As an example, for ease of understanding, we will use any follow-up record table as an example for explanation. We will take any column in the follow-up record table as the target column, and the preset word segmentation set under the target column as the target word segmentation set. The steps for obtaining the word segments of all follow-up records in the target column of the follow-up record table are described as follows:

[0064] First, the information in the target column is segmented into a word segmentation sequence according to the order of appearance. Assuming that the target column mainly records the patient's self-reported recovery experience, the corresponding text information is "Recovery experience: There is pain after the anesthesia wears off, especially unbearable during activity, and I am not satisfied with the recovery progress."; where the template word segmentation corresponding to the column title is "recovery experience", that is, the target word segmentation set is {recovery experience}, and the corresponding information word segmentation sequence after word segmentation is: {recovery, experience, anesthesia, wear off, after, have, pain, especially, is, in, activity, time, difficult, bearable, I, on, recovery, progress, unsatisfied};

[0065] Then, each information segmentation word is treated as a set element to construct all one-dimensional segmentation sets {recovery}, {feeling}, {anesthesia}, {subsidence}, {after}, {have}, {pain}...{recovery}, {progress}, {dissatisfaction}; calculate the Jaccard coefficient between each one-dimensional segmentation set and the target segmentation set {recovery feeling}; after calculation, it can be found that the Jaccard coefficient between each one-dimensional segmentation set and the target segmentation set is 0, so all information segmentation words corresponding to the one-dimensional segmentation sets are used as the first non-template segmentation words;

[0066] By concatenating two adjacent first non-template word segments in the sequence element index, all two-dimensional word sets are obtained, such as {recovery feeling}, {feeling anesthesia}, {anesthesia dissipation}, {after dissipation}, {afterwards}, {feeling pain}...{recovery progress}, {unsatisfactory progress}. The Jaccard coefficient between each two-dimensional word set and the target word set is calculated. It is found that only the Jaccard coefficient between the two-dimensional word set {recovery feeling} and the target word set {recovery feeling} is not 0. Therefore, the constituent information words "recovery" and "feeling" of the two-dimensional concatenated information words corresponding to the two-dimensional word set {recovery feeling} are used as template words, and the remaining information words in the information word sequence are used as second non-template words.

[0067] The non-template word segments selected from the previous dimension are continuously iterated and concatenated, such as continuing to concatenate to obtain three-dimensional concatenated word segments, four-dimensional concatenated word segments, etc., until the preset dimension is met. In this example, the preset dimension is three-dimensional. Then all three-dimensional word segments are {after anesthesia wears off}, {after wears off}, {after pain}...{unsatisfactory recovery progress}. The Jaccard coefficient between each three-dimensional word segment set and the target word segment set is calculated. Since the Jaccard coefficient is 0, the constituent information words of the concatenated information words corresponding to all word segments under the preset dimension are non-template words. Non-template words are the follow-up record words.

[0068] In other embodiments, the implementer may also set other preset dimensions according to the template word segmentation length of each column in the follow-up record table.

[0069] By iteratively concatenating the data, we can accurately filter out the word segments of all follow-up records. Then, by changing the target category, we iteratively obtain the word segments of all follow-up information in each category of each follow-up record table. Finally, we use the word segments of the follow-up information as set elements to construct a set of word segments for each category.

[0070] The follow-up information analysis module 102 is used to obtain the column record similarity coefficient between adjacent follow-up record tables based on the similarity between the record segmentation sets of the same column and the sentiment information of each follow-up record segmentation word in the record segmentation set; to obtain the record similarity coefficient between adjacent follow-up record tables by combining the column record similarity coefficients between all column records of the same column; and to determine the recording method of each follow-up record segmentation word in each follow-up record table based on the record similarity coefficient between adjacent follow-up record tables and the column record similarity coefficient between the same column, combined with the frequency of occurrence of each follow-up record segmentation word in the corresponding column.

[0071] Considering that the more similar the word segmentation of follow-up records under the same column are between adjacent follow-up record tables, and the more similar the emotional information they express, it indicates that the column records between adjacent columns of the corresponding adjacent follow-up record tables are more similar; and considering that the column records between all the same columns are more similar, it further indicates that the follow-up records between adjacent follow-up record tables are more similar.

[0072] Therefore, in this embodiment of the invention, the similarity coefficient between adjacent follow-up record tables is obtained based on the similarity between record word segments of the same column and the sentiment information of each follow-up record word segment in the record word segment set; further, the similarity coefficient between all column records of the same column is combined to obtain the record similarity coefficient between adjacent follow-up record tables.

[0073] Among them, the column record similarity coefficient reflects the similarity between column records in the same column of adjacent follow-up record tables, and the record similarity coefficient reflects the similarity between follow-up records in adjacent follow-up record tables, so as to evaluate the processing method of word segmentation of follow-up records in the future, thereby improving the data entry effect of follow-up record tables.

[0074] Preferably, in one embodiment of the present invention, the method for obtaining the similarity coefficient of column records includes:

[0075] Please see Figure 2 The flowchart illustrates a method for obtaining the similarity coefficient of column records according to an embodiment of the present invention, specifically including:

[0076] Step S201: Between adjacent follow-up record tables, the Jaccard coefficient between the word segmentation sets of records in the same column is used as the text similarity coefficient.

[0077] Since the Jaccard coefficient can be used to evaluate the similarity between sets and is widely used in information retrieval and data mining, in one embodiment of the present invention, the Jaccard coefficient between the word segmentation sets of records in the same category is directly used as the text similarity coefficient; the larger the Jaccard coefficient, the larger the text similarity coefficient, indicating that the category records may be more similar.

[0078] In step S202, in each record segmentation set, obtain the set sentiment score based on the sentiment information of all follow-up record segments; between adjacent follow-up record tables, use the negative correlation mapping result of the difference in set sentiment scores between record segmentation sets of the same column as the sentiment similarity coefficient.

[0079] Considering that each column may also contain the patient's feelings during the recovery process, and the patient's feelings also reflect the similarity of the column records in that column; for example, "good recovery feeling" and "extremely poor recovery feeling" have high text similarity, but express completely opposite emotions, so it is necessary to further evaluate the emotional information in each column; the more similar the emotional information, the greater the possibility that the column records in the corresponding columns of the two follow-up record tables are similar; therefore, one embodiment of the present invention first obtains the set of sentiment scores of each record word segmentation set, and then evaluates the column sentiment differences between the same columns of adjacent follow-up record tables to obtain the sentiment similarity coefficient.

[0080] In a preferred embodiment of the present invention, considering that a relevant dictionary is a very practical Chinese sentiment analysis tool that can be used to analyze the sentiment tendency of text, the sentiment score of each follow-up record word segment can be obtained based on the relevant dictionary, and then the aggregate sentiment score can be comprehensively evaluated; therefore, the method for obtaining the sentiment score of the sentiment processing aggregate includes:

[0081] The sentiment score of each follow-up record word segment is obtained based on the relevant dictionary, and the sum of the sentiment scores of all follow-up record word segments is used as the set sentiment score of the corresponding record word segment set.

[0082] It should be noted that obtaining sentiment scores from word segmentation based on relevant dictionaries is a well-known technique, and will not be elaborated upon here.

[0083] As an example, after obtaining the aggregate sentiment score of each record segmentation set, the absolute value of the difference between the aggregate sentiment scores of the same column of records in adjacent follow-up record tables is first calculated. The absolute value of the difference is used as x in the exponential function exp(-x) with the natural constant e as the base, so as to perform negative correlation mapping normalization. The logic is then adjusted so that the negative correlation mapping result is used as the sentiment similarity coefficient, so that the smaller the absolute value of the difference between the aggregate sentiment scores, the larger the sentiment similarity coefficient.

[0084] Step S203: The product of the emotional similarity coefficient and the text similarity coefficient is used as the column record similarity coefficient between the same columns of the corresponding adjacent follow-up record tables.

[0085] Since both the sentiment similarity coefficient and the text similarity coefficient reflect the similarity of the same column records between adjacent follow-up record tables to a certain extent, the sentiment similarity coefficient and the text similarity coefficient are multiplied together to obtain the column record similarity coefficient.

[0086] It should be noted that the range of values ​​for both the sentiment similarity coefficient and the text similarity coefficient is 0 to 1, and the range of values ​​for the column record similarity coefficient is also 0 to 1.

[0087] At this point, the similarity coefficient of column records between the same columns of adjacent follow-up record tables has been obtained, and the record similarity coefficient between adjacent follow-up record tables can be further obtained.

[0088] Preferably, in one embodiment of the present invention, the method for obtaining the similarity coefficient includes:

[0089] Between adjacent follow-up record tables, the average of the similarity coefficients of all columns with the same content is used as the record similarity coefficient between adjacent follow-up record tables; it should be noted that the record similarity coefficient ranges from 0 to 1.

[0090] Considering that the higher the similarity coefficient between adjacent follow-up record tables, the more likely there is information redundancy between two adjacent follow-up records; also considering that the similarity coefficient between each column of adjacent follow-up record tables is different, the higher the similarity coefficient, the greater the possibility of word segmentation redundancy between follow-up records of the same column; redundant information does not need to be recorded separately in the follow-up information system to improve the effectiveness of follow-up information recording; furthermore, considering that the frequency of word segmentation in each follow-up record in a column may also be different, the higher the frequency of occurrence and the lower the similarity coefficient between column records of the same column, the more likely it is to correspond to important follow-up information, and therefore it should not be treated as redundant information when recording adjacent follow-up record information;

[0091] Therefore, in this embodiment of the invention, the recording method of each follow-up record word in each follow-up record table will be determined based on the similarity coefficient between adjacent follow-up record tables and the similarity coefficient between column records in the same column, combined with the frequency of occurrence of each follow-up record word in the corresponding column.

[0092] Preferably, in one embodiment of the present invention, considering that the follow-up records in the same column during different follow-up processes may not be completely consistent, it is first necessary to perform semantic matching on the follow-up record word segments between the record word segments of the same column, and then evaluate the importance of the matching follow-up record word segments in the corresponding column and determine their recording method.

[0093] Based on this, the methods for determining the word segmentation recording method for each follow-up record in each follow-up record table include:

[0094] If the similarity coefficient between each follow-up record table and the previous adjacent follow-up record table is less than a preset threshold, each follow-up record in each follow-up record table will be segmented and recorded independently.

[0095] If the similarity coefficient between each follow-up record table and the previous adjacent follow-up record table is greater than or equal to a preset threshold, semantic matching is performed on the follow-up record word segments between the word segments of the same column; based on the similarity coefficient between column records of the same column and the frequency of occurrence of the matched follow-up record word segments in the corresponding column, the word segmentation recording method of each follow-up record in each follow-up record table is determined.

[0096] Since the similarity coefficient ranges from 0 to 1, the preset threshold is set to 0.7; in other embodiments, the implementer may also set it according to the actual situation.

[0097] In a preferred embodiment of the present invention, considering that the Word2Vec algorithm can map semantically similar word vectors to regions that are close in distance in the vector space, since word vectors with similar semantics are also similar; and considering that cosine similarity can be used to measure the similarity between word vectors; based on this, the method for semantic matching of follow-up record word segments between record word segments of the same category includes:

[0098] The word vectors of each follow-up record in each column of each follow-up record table are obtained based on the Word2Vec algorithm; the word vectors of any follow-up record in each follow-up record table are used as the target word.

[0099] In the adjacent follow-up record table above each follow-up record table, calculate the cosine similarity between the word vector of the target word and the word vector of each follow-up record word in the same column. The follow-up record word with the highest cosine similarity is used as the semantic matching word of the target word.

[0100] It should be noted that the word vectors and cosine similarity obtained from word segmentation based on the Word2Vec algorithm are well-known techniques and will not be elaborated here.

[0101] By changing the target word segmentation, semantic matching can be performed on the word segmentation of each follow-up record in each column of each follow-up record table to obtain the semantically matching word segmentation in the corresponding column of the previous adjacent follow-up record table. However, due to the different sizes of the record word segmentation sets and the different word vectors, the matching relationship between each follow-up record word segmentation in each column of each follow-up record table and the follow-up record word segmentation in the corresponding column of the previous adjacent follow-up record table is not one-to-one.

[0102] Preferably, in one embodiment of the present invention, in order to minimize redundancy during the processing of two adjacent follow-up record tables and improve the entry effect of follow-up information, it is necessary to determine whether each follow-up record word in each column of each follow-up record table is an important follow-up word; considering that the higher the frequency of the follow-up record word in the corresponding column and the smaller the similarity coefficient between the column records of the same column, the more likely it is to correspond to important follow-up information, and thus determine its recording method;

[0103] Based on this, the method for determining the word segmentation recording method for each follow-up record in each follow-up record table, according to the similarity coefficient between records in the same column and the frequency of occurrence of the matching follow-up record word segmentation in the corresponding column, includes:

[0104] The similarity coefficient between the same column of each follow-up record table and the previous adjacent follow-up record table is used as the first weight; the negative correlation normalization result of the first weight is used as the second weight; and the semantic matching word segment of the target word in the corresponding column of the previous adjacent follow-up record table is used as the reference word segment.

[0105] In each column of the adjacent follow-up record table, the frequency of occurrence of the reference word in the corresponding column is weighted using the first weight, and the weighted result is used as the first merging parameter of the target word; in each column of each follow-up record table, the frequency of occurrence of the target word in the corresponding column is weighted using the second weight, and the weighted result is used as the second merging parameter of the target word.

[0106] If the first merging parameter is less than or equal to the second merging parameter, the target word segmentation is merged and recorded in the previous adjacent follow-up record table; if the first merging parameter is greater than the second merging parameter, the target word segmentation is treated as a redundant record word segmentation and is not updated in the previous adjacent follow-up record table.

[0107] As an example, to facilitate understanding of the method for determining the word segmentation recording method for each follow-up record in each follow-up record table, we will use any follow-up record table obtained other than the first follow-up as the current follow-up record table for explanation. The word segmentation recording method for each follow-up record in the current follow-up record table is as follows:

[0108] Between the current follow-up record table and the previous adjacent follow-up record table, the similarity coefficient between the records of the same column is used as the first weight. The larger the first weight, the more similar the column records are. The first weight is used as x in 1-x for negative correlation normalization to obtain the second weight. The larger the first weight and the smaller the second weight, the lower the similarity of the column records. The second weight is multiplied by the frequency of occurrence of the target word in the corresponding column and combined to obtain the second merging parameter. The first weight is multiplied by the frequency of occurrence of the reference word of the target word in the corresponding column and combined to obtain the first merging parameter.

[0109] If the first merging parameter is greater than the second merging parameter, it means that the target word segment is more likely to be a redundant information segment, and the record will not be updated in the previous adjacent follow-up record table; if the first merging parameter is less than or equal to the second merging parameter, it means that the target word segment may be important follow-up information, and it will be merged and recorded in the previous adjacent follow-up record table.

[0110] Change the target word segmentation in the current follow-up record table and determine the recording method for each follow-up record word segmentation.

[0111] The follow-up information processing module 103 is used to record follow-up information according to the word segmentation method of each follow-up record in each follow-up record table.

[0112] In one embodiment of the present invention, after determining the recording method of word segmentation for each follow-up record in each follow-up record table, the follow-up record table after the first follow-up is taken as the initial follow-up record. Then, based on the recording method of word segmentation for each follow-up record in the second follow-up record table, the initial follow-up record table is updated to obtain the second follow-up record. Based on the recording method of word segmentation for each follow-up record in the third follow-up record table, the second follow-up record is updated to obtain the third follow-up record. The follow-up records are iteratively updated until the patient's follow-up is completed, resulting in a complete patient follow-up information record table. The complete patient follow-up information record representation includes important follow-up words added at each follow-up, reducing the redundancy of follow-up information and thus improving the recording effect of follow-up information.

[0113] In summary, this invention first obtains follow-up record tables after each postoperative follow-up visit, and then obtains word segments of all follow-up records in each column of each follow-up record table to construct a set of word segments for each column. Further, based on the textual similarity and sentiment information of the word segments in adjacent follow-up record tables, it obtains the column record similarity coefficient between the same columns, thereby obtaining the record similarity coefficient between adjacent follow-up record tables. Furthermore, it determines the recording method for each word segment in each follow-up record table by combining the frequency of occurrence of each word segment in the corresponding column. Finally, follow-up records are made according to the recording method of each word segment in each follow-up record table. This invention, based on the textual similarity and sentiment information of word segments in follow-up records between adjacent follow-up processes, and combined with the frequency of occurrence of each word segment, determines its recording method. Compared with existing methods, this can reduce redundancy in the multiple postoperative follow-up information entry process for patients under local nerve block anesthesia, thereby improving the recording effect of follow-up information.

[0114] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0115] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A postoperative follow-up information recording system for patients undergoing regional nerve block anesthesia, characterized in that, The system includes: The follow-up information collection module is used to obtain the follow-up record table after each postoperative follow-up visit of the patient; it obtains word segments of all follow-up records in each column of each follow-up record table and constructs a set of word segments of records for each column; The follow-up information analysis module is used to obtain the column record similarity coefficient between adjacent follow-up record tables based on the similarity between the record word segments of the same column and the sentiment information of each follow-up record word in the record word segments. It then combines the column record similarity coefficients between all columns to obtain the record similarity coefficient between adjacent follow-up record tables. Finally, based on the record similarity coefficients between adjacent follow-up record tables and the column record similarity coefficients between columns, and combined with the frequency of occurrence of each follow-up record word in the corresponding column, it determines the recording method for each follow-up record word in each follow-up record table. The follow-up information processing module is used to record follow-up information according to the word segmentation recording method of each follow-up record in each follow-up record table. The methods for obtaining the similarity coefficient of the column records include: Between adjacent follow-up record tables, the Jaccard coefficient between the word segmentation sets of the records in the same column is used as the text similarity coefficient; In each of the record word segments, a set sentiment score is obtained based on the sentiment information of all follow-up record word segments; between adjacent follow-up record tables, the negative correlation mapping result of the difference in the set sentiment scores between the record word segments of the same column is used as the sentiment similarity coefficient; The product of the emotional similarity coefficient and the text similarity coefficient is used as the column record similarity coefficient between the same columns in the corresponding adjacent follow-up record tables; The method for determining the word segmentation recording method for each follow-up record in each follow-up record table includes: If the similarity coefficient between each follow-up record table and the previous adjacent follow-up record table is less than a preset threshold, each follow-up record in each follow-up record table will be segmented and recorded independently. If the similarity coefficient between each follow-up record table and the previous adjacent follow-up record table is greater than or equal to a preset threshold, semantic matching is performed on the follow-up record word segments of the same column. Based on the similarity coefficient between the column records of the same column and the frequency of occurrence of the matched follow-up record word segments in the corresponding column, the recording method for each follow-up record word segment in each follow-up record table is determined. Specifically, this includes: using the similarity coefficient between the column records of the same column of each follow-up record table and the previous adjacent follow-up record table as a first weight; using the negative correlation normalization result of the first weight as a second weight; and using the semantic similarity coefficient of the target word segment in the corresponding column of the previous adjacent follow-up record table. The matched word segment is used as a reference word segment; in each column of the previous adjacent follow-up record table, the frequency of occurrence of the reference word in the corresponding column is weighted using the first weight, and the weighted result is used as the first merging parameter of the target word; in each column of each follow-up record table, the frequency of occurrence of the target word in the corresponding column is weighted using the second weight, and the weighted result is used as the second merging parameter of the target word; if the first merging parameter is less than or equal to the second merging parameter, the target word is merged and recorded in the previous adjacent follow-up record table; if the first merging parameter is greater than the second merging parameter, the target word is treated as a redundant record word and is not updated in the previous adjacent follow-up record table.

2. The postoperative follow-up information recording system for patients undergoing local nerve block anesthesia according to claim 1, characterized in that, The method for obtaining the segmented words of the follow-up records includes: In each follow-up record table, the jieba word segmentation tool is used to segment the text information in each column to obtain all the segmented information in each column. Based on the preset word segmentation set under each column, all follow-up record word segments are selected from all the information word segments in each column; the preset word segmentation set includes all template word segments in each column of the follow-up record form template.

3. The postoperative follow-up information recording system for patients undergoing local nerve block anesthesia according to claim 2, characterized in that, The method for selecting all follow-up record word segments from all the information word segments in each column based on the preset word segmentation set under each column includes: In each follow-up record table, any column is taken as the target column, and the preset word segmentation set under the target column is taken as the target word segmentation set; the information word segments in the target column are used as sequence elements in the order of appearance to construct the information word segmentation sequence of the target column. Each information segment in the information segmentation sequence is treated as a one-dimensional segmentation set for iterative analysis. All concatenated information segments under each dimension are obtained sequentially to construct the corresponding segmentation set under the dimension until the preset dimension is met. Under each dimension, non-template segmented words are selected based on the Jaccard coefficient between the segmentation set corresponding to each concatenated information segmented word and the preset segmentation set. The concatenated information segmented words in each dimension are obtained by concatenating adjacent non-template segmented words in the information segmentation sequence according to the dimension requirements from all non-template segmented words in the previous dimension. All non-template segmented words are used as follow-up record segmented words.

4. A postoperative follow-up information recording system for patients undergoing local nerve block anesthesia according to claim 3, characterized in that, The method for filtering out non-template word segmentation includes: In each dimension, the Jaccard coefficient between each word segmentation set and the preset word segmentation set is calculated. When the Jaccard coefficient is 0, the information word segmentation in the corresponding word segmentation set or all the constituent information word segmentation of the corresponding concatenated information word segmentation are used as template word segmentation. In the information segmentation sequence, all information segments other than all template segments selected under each dimension are treated as non-template segments.

5. A postoperative follow-up information recording system for patients undergoing local nerve block anesthesia according to claim 1, characterized in that, The methods for obtaining the sentiment score of the set include: The sentiment score of each follow-up record segment is obtained based on the relevant dictionary, and the sum of the sentiment scores of all follow-up record segments is used as the set sentiment score of the corresponding record segment set.

6. A postoperative follow-up information recording system for patients undergoing local nerve block anesthesia as described in claim 1, characterized in that, The method for obtaining the record similarity coefficient includes: Between adjacent follow-up record tables, the average of the similarity coefficients of the records in the same column is used as the record similarity coefficient between adjacent follow-up record tables.

7. A postoperative follow-up information recording system for patients undergoing local nerve block anesthesia according to claim 1, characterized in that, The method for semantic matching of follow-up record word segments among record word segments of the same column includes: The word vectors of each follow-up record in each column of each follow-up record table are obtained based on the Word2Vec algorithm; the word vectors of any follow-up record in each follow-up record table are used as the target word. In the adjacent follow-up record table above each follow-up record table, calculate the cosine similarity between the word vector of the target word and the word vector of each follow-up record word in the same column, and take the follow-up record word with the maximum cosine similarity as the semantic matching word of the target word.

Images