An intelligent collection method, system and medium for clinical data of hepatobiliary surgery

Abstract

The present invention relates to the technical field of clinical data processing, and in particular to an intelligent collection method, system and medium for clinical data of hepatobiliary surgery, the method comprising: obtaining a target keyword set corresponding to clinical hepatobiliary surgery; screening out target clinical data from all clinical information corresponding to patients to be detected; the method for obtaining the target keyword set comprises: obtaining a clinical information set corresponding to all historical patients of hepatobiliary surgery, and performing word segmentation processing on the clinical information; determining the description possibility corresponding to each target segmentation; correcting the word frequency corresponding to each target segmentation; determining the topic possibility corresponding to each target segmentation; correcting the inverse document frequency corresponding to each target segmentation; and performing keyword extraction on all target segmentations according to the target word frequency and target inverse document frequency corresponding to all target segmentations. The present invention improves the rationality of keyword extraction, thereby improving the rationality of clinical data collection of hepatobiliary surgery.

Description

Technical Field

[0001] The present invention relates to the technical field of clinical data processing, and in particular to an intelligent collection method, system and medium for clinical data of hepatobiliary surgery. Background Art

[0002] In the process of collecting clinical data of hepatobiliary surgery, in order to facilitate doctors to quickly understand the patient's condition, key information of the patient is often extracted. For example, the required key information can be extracted based on the extracted keywords. Therefore, keyword extraction is very important. At present, the method commonly used to extract keywords is: keyword extraction through the inverse document frequency (TF-IDF) method.

[0003] However, when extracting keywords from hepatobiliary surgery clinical data using the inverse document frequency method, the following technical problems often occur:

[0004] When keywords are extracted from hepatobiliary surgery clinical data through the inverse document frequency method, the keyword extraction is often based on the uniqueness of the vocabulary. However, some words with relatively poor uniqueness are often also relatively key. For example, for hepatobiliary surgery clinical data, it often includes the rehabilitation treatment of patients in the clinic, and the recorded rehabilitation treatment topic names are often relatively large, which may lead to poor uniqueness, but they are often also relatively important and are key words. Therefore, when keywords are extracted from hepatobiliary surgery clinical data directly through the inverse document frequency method, the rationality of keyword extraction may be poor, thereby leading to poor rationality of rapid collection of hepatobiliary surgery clinical data. Summary of the invention

[0005] In order to solve the technical problem of poor rationality of keyword extraction, the present invention proposes an intelligent collection method, system and medium for clinical data of hepatobiliary surgery.

[0006] In a first aspect, the present invention provides an intelligent method for collecting clinical data of hepatobiliary surgery, the method comprising:

[0007] Obtain the target keyword set corresponding to hepatobiliary surgery clinical practice;

[0008] According to the target keyword set, target clinical data is screened out from all clinical information corresponding to the patient to be tested;

[0009] The method for obtaining the target keyword set comprises the following steps:

[0010] Obtain the clinical information set corresponding to all historical patients of hepatobiliary surgery, and perform word segmentation on the clinical information in each clinical information set to obtain the target word segmentation;

[0011] Determine the description possibility corresponding to each target participle according to the duration of each target participle appearing in the clinical information set to which it belongs and the number of times each target participle appears in the clinical information to which it belongs;

[0012] According to the description possibility corresponding to each target participle, the word frequency corresponding to each target participle is modified to obtain the target word frequency corresponding to each target participle;

[0013] According to the amount of clinical information to which each target participle belongs and the similarity between the clinical information to which each target participle belongs, the possibility of the topic corresponding to each target participle is determined;

[0014] According to the topic possibility corresponding to each target segmentation word, the inverse document frequency corresponding to each target segmentation word is corrected to obtain the target inverse document frequency;

[0015] According to the target word frequencies and target inverse document frequencies corresponding to all target segmented words, keywords are extracted for all target segmented words to obtain a target keyword set.

[0016] In combination with the first aspect above, in a possible implementation, determining the description possibility corresponding to each target participle according to the duration of appearance of each target participle in the clinical information set to which it belongs and the number of times each target participle appears in the clinical information to which it belongs includes:

[0017] Determine any target participle as a marked participle, and determine the number of times the marked participle appears in each clinical information to which it belongs as the target number of times the marked participle appears in each clinical information to which it belongs;

[0018] The maximum value of the target number of times of the marked word under all the clinical information to which it belongs is determined as the number peak value corresponding to the marked word;

[0019] Determine the duration between the earliest appearance time and the latest appearance time of the marked participle in each clinical information set to which it belongs as the target duration of the marked participle in each clinical information set to which it belongs, wherein the target duration of the marked participle in each clinical information set to which it belongs represents the duration of the appearance of the marked participle in each clinical information set to which it belongs;

[0020] The average of the target durations of the marked participle under all clinical information sets to which it belongs is determined as the overall representative duration corresponding to the marked participle;

[0021] The description possibility corresponding to the marked segmentation is determined according to the overall representative duration and the peak number of times corresponding to the marked segmentation.

[0022] In combination with the first aspect above, in a possible implementation manner, determining the description possibility corresponding to the marked segmentation according to the overall representative duration and the peak number of times corresponding to the marked segmentation includes:

[0023] Determine the difference between the preset training phase duration and the overall representative duration corresponding to the marked segmentation as the description deviation factor corresponding to the marked segmentation;

[0024] The description possibility corresponding to the marked word segment is determined according to the description deviation factor and the peak number of times corresponding to the marked word segment, wherein the description deviation factor is negatively correlated with the description possibility, and the peak number of times is positively correlated with the description possibility.

[0025] In combination with the first aspect above, in a possible implementation manner, the word frequency corresponding to each target participle is modified according to the description possibility corresponding to each target participle to obtain the target word frequency corresponding to each target participle, including:

[0026] Determine the initial word frequency corresponding to each target word segmentation through the keyword extraction algorithm;

[0027] If the description possibility corresponding to the target word is greater than the preset description threshold, the target word is determined as a candidate training description word;

[0028] According to the description possibility and initial word frequency corresponding to each candidate training descriptor, the target word frequency corresponding to each candidate training descriptor is determined, wherein the description possibility and the initial word frequency are both positively correlated with the target word frequency;

[0029] Each target participle except the candidate training descriptor is determined as a reference participle, and the initial word frequency corresponding to each reference participle is determined as the target word frequency corresponding to each reference participle.

[0030] In combination with the first aspect above, in a possible implementation, determining the topic possibility corresponding to each target segmentation according to the amount of clinical information to which each target segmentation belongs and the similarity between the clinical information to which each target segmentation belongs includes:

[0031] Determine any target segmentation as a marked segmentation, and determine the total number of clinical information to which the marked segmentation belongs in all clinical information sets as the number of recurring information corresponding to the marked segmentation;

[0032] Determining an initial possible factor corresponding to the marked word segmentation according to the number of recurring information corresponding to the marked word segmentation;

[0033] If the initial possibility factor corresponding to the target word is greater than the preset suspected topic possibility threshold, the target word is determined as a suspected topic word;

[0034] According to the similarity between the clinical information to which each suspected topic participle belongs, the topic possibility corresponding to each suspected topic participle is determined;

[0035] Each target participle, except the suspected topic participle, is determined as a temporary participle, and a preset value is determined as the topic possibility corresponding to each temporary participle.

[0036] In combination with the first aspect above, in a possible implementation manner, determining the initial possible factor corresponding to the marked word segmentation according to the number of recurring information corresponding to the marked word segmentation includes:

[0037] The total number of all clinical information in all clinical information sets is determined as the number of historical reference information;

[0038] The ratio of the number of recurring information corresponding to the marked word segment to the number of historical reference information is determined as the initial possible factor corresponding to the marked word segment.

[0039] In combination with the first aspect above, in a possible implementation, the formula corresponding to the topic possibility corresponding to the suspected topic segmentation word is:

[0040] ;

[0041] ;in, It is The topic possibilities corresponding to the suspected topic participles; It is the type number of the suspected topic participle; is the normalization function; It is The number of clinical information sets to which the suspected subject participle belongs; It is The serial number of the clinical information set to which the suspected subject participle belongs; Characterization The suspected topic participle is Possible scenarios for the topics under each clinical information set; It is an exponential function with a natural constant as base; It is The suspected topic participle is The similarity between the earliest clinical information and the latest clinical information in a clinical information set; It is The suspected topic participle is The number of clinical information in a clinical information set; It is The suspected topic participle is The serial number of the clinical information to which it belongs in a clinical information set; It is The suspected topic participle is The clinical information set to which it belongs The information entropy of all target words in the clinical information; It is The suspected topic participle is In the clinical information set, it belongs to Clinical information and its The similarity between clinical information.

[0042] In combination with the first aspect above, in a possible implementation, the inverse document frequency corresponding to each target segmentation word is corrected according to the topic possibility corresponding to each target segmentation word to obtain the target inverse document frequency, including:

[0043] Determine the initial inverse document frequency corresponding to each target word segmentation through the keyword extraction algorithm;

[0044] If the topic probability corresponding to the target word is greater than the preset real topic possibility threshold, the target word is determined as a candidate topic word;

[0045] According to the topic probability and initial inverse document frequency corresponding to each candidate topic segmentation word, the target inverse document frequency corresponding to each candidate topic segmentation word is determined, wherein the topic probability and the initial inverse document frequency are both positively correlated with the target inverse document frequency;

[0046] Each target participle except the candidate topic participle is determined as a pending participle, and the initial inverse document frequency corresponding to each pending participle is determined as the target inverse document frequency corresponding to each pending participle.

[0047] In a second aspect, the present invention provides an intelligent collection system for clinical data of hepatobiliary surgery, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program implementing the method in the first aspect or any possible implementation of the first aspect when executed by the processor, specifically, the system comprising:

[0048] Keyword acquisition module, used to obtain the target keyword set corresponding to hepatobiliary surgery clinical practice;

[0049] The data screening module is used to screen out target clinical data from all clinical information corresponding to the patient to be tested according to the target keyword set.

[0050] In a third aspect, a server is provided, comprising a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, so that the device executes the method in the first aspect or any possible implementation of the first aspect.

[0051] In a fourth aspect, a computer program product is provided, comprising: a computer program code, which, when executed on a computer, enables the computer to execute the method in the first aspect or any possible implementation of the first aspect.

[0052] In a fifth aspect, an intelligent collection medium for clinical data of hepatobiliary surgery is provided, wherein the intelligent collection medium for clinical data of hepatobiliary surgery stores computer program code, and when the computer program code runs on a computer, the computer executes the method in the above-mentioned first aspect or any possible implementation of the first aspect.

[0053] The present invention has the following beneficial effects:

[0054] The intelligent collection method of clinical data of hepatobiliary surgery of the present invention solves the technical problem of poor rationality of keyword extraction, improves the rationality of keyword extraction, and thus improves the rationality of rapid collection of clinical data of hepatobiliary surgery. The present invention comprehensively considers multiple factors related to keywords of clinical data of hepatobiliary surgery, such as description possibility and subject possibility, thereby adaptively realizing the correction of word frequency and inverse document frequency in the inverse document frequency method, thereby improving the rationality of keyword extraction, thereby improving the rationality of rapid collection of clinical data of hepatobiliary surgery. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0056] Figure 1 A flowchart of an intelligent collection method for clinical data of hepatobiliary surgery according to the present invention;

[0057] Figure 2 A flowchart for obtaining a target keyword set of the present invention;

[0058] Figure 3 A schematic diagram of the structure of an intelligent collection system for clinical data of hepatobiliary surgery according to the present invention;

[0059] Figure 4 The figure is a schematic diagram of the structure of a computer device of the present invention. DETAILED DESCRIPTION

[0060] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the technical solutions proposed by the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0061] Unless defined otherwise, 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 belongs.

[0062] refer to Figure 1 , showing the process of some embodiments of a method for intelligently collecting clinical data of hepatobiliary surgery of the present invention. The method for intelligently collecting clinical data of hepatobiliary surgery comprises the following steps:

[0063] Step S1, obtaining a target keyword set corresponding to hepatobiliary surgery clinical practice.

[0064] Among them, hepatobiliary surgery clinic can represent the process of clinical treatment of patients by hepatobiliary surgery. The target keywords in the target keyword set can be keywords extracted to represent the process of clinical treatment of patients by hepatobiliary surgery.

[0065] In some embodiments, reference Figure 2 , a flow chart of a method for acquiring a target keyword set is shown. Specifically, the method for acquiring a target keyword set may include the following steps:

[0066] Step 201, obtain the clinical information set corresponding to all historical patients corresponding to hepatobiliary surgery, and perform word segmentation processing on the clinical information in each clinical information set to obtain target word segmentation.

[0067] Among them, the historical patients corresponding to the hepatobiliary surgery department can be the patients who have been treated in the hepatobiliary surgery department. In the hepatobiliary surgery department, in order to facilitate doctors to judge the condition of patients, the rehabilitation information of patients is often recorded regularly. Among them, the rehabilitation information can characterize the state of patients. The clinical information in the clinical information set can be represented by the recorded rehabilitation information of patients. For example, for patients who need to undergo hepatobiliary surgery, rehabilitation training is often arranged for patients during the treatment process before and after the surgery. If the recorded hepatobiliary surgery clinical data is rehabilitation training data, then the clinical information recorded at this time can characterize the situation of each rehabilitation training. The rehabilitation training before surgery can be, but is not limited to: abdominal breathing, pursed-lip breathing, and effective cough training. The rehabilitation training after surgery can be, but is not limited to: single leg lift in bed, marching in place, standing on one leg, and walking assisted by family members. The duration of each rehabilitation training can be preset. For example, the duration of rehabilitation training can be 15 minutes.

[0068] As an example, this step may include the following steps:

[0069] In the first step, through medical staff, record the completion situation of each rehabilitation training of historical patients as the initial rehabilitation information. Perform data preprocessing on the initial rehabilitation information, and record the initial rehabilitation information after data preprocessing as clinical information. And all the clinical information of this historical patient recorded constitutes the clinical information set corresponding to this historical patient.

[0070] Among them, data preprocessing may include, but is not limited to: text data cleaning and excluding common stop words. Text data cleaning can remove irrelevant characters such as special symbols and HTML (HyperText Markup Language) tags. The method of excluding common stop words can be: obtain stop words based on the existing stop word list, and remove the stop words contained in the data to reduce the noise impact during subsequent analysis. For example, "of", "in", etc. are all stop words.

[0071] In the second step, the jieba word segmentation algorithm can be used to perform word segmentation on the clinical information in each clinical information set, and each word obtained by word segmentation is recorded as the target word.

[0072] Among them, the same target words can be recorded as the same kind of target words.

[0073] Step 202, determine the description possibility corresponding to each target word according to the duration of each target word appearing in its affiliated clinical information set and the number of times each target word appears in its affiliated clinical information.

[0074] Among them, if a certain clinical information set contains a certain target participle, then the clinical information set can be recorded as the clinical information set to which the target participle belongs. Similarly, if a certain clinical information contains a certain target participle, then the clinical information can be recorded as the clinical information to which the target participle belongs.

[0075] As an example, this step may include the following steps:

[0076] In the first step, any target participle is determined as a marked participle, and the number of times the marked participle appears in each clinical information to which it belongs is determined as the target number of times the marked participle appears in each clinical information to which it belongs.

[0077] In the second step, the maximum value of the target number of the above-mentioned marked word in all the clinical information to which it belongs is determined as the peak number of times corresponding to the above-mentioned marked word.

[0078] In the third step, the duration between the earliest appearance time and the latest appearance time of the above-mentioned marked word in each clinical information set to which it belongs is determined as the target duration of the above-mentioned marked word in each clinical information set to which it belongs.

[0079] The target duration of the above-mentioned marked word in each clinical information set to which it belongs may represent the duration of the above-mentioned marked word appearing in each clinical information set to which it belongs.

[0080] For example, the end time of the time period corresponding to the clinical information can be recorded as the reference time corresponding to the clinical information. The time period corresponding to the clinical information can be the time period of the rehabilitation training process represented by the clinical information. Any clinical information set to which the marked participle belongs is recorded as the marked information set, and the clinical information containing the marked participle is screened out from the marked information set as the standard information. The earliest reference time among the reference times corresponding to all the standard information can be determined as the earliest time when the marked participle appears in the marked information set, and the latest reference time among the reference times corresponding to all the standard information can be determined as the latest time when the marked participle appears in the marked information set; the target duration of the marked participle under the marked information set can be equal to the range between the reference times corresponding to all the standard information.

[0081] The fourth step is to determine the average of the target durations of the above-mentioned marked participles under all clinical information sets to which they belong as the overall representative duration corresponding to the above-mentioned marked participles.

[0082] In the fifth step, according to the overall representative duration and the peak number of times corresponding to the above-mentioned marked segmentation, determining the description possibility corresponding to the above-mentioned marked segmentation may include the following sub-steps:

[0083] In the first sub-step, the difference between the preset training phase duration and the overall representative duration corresponding to the above-mentioned marked segmentation is determined as the description deviation factor corresponding to the above-mentioned marked segmentation.

[0084] The preset training stage duration may represent the cycle of rehabilitation treatment, that is, a rehabilitation treatment plan for the next stage may be formulated for the patient every preset training stage duration. For example, the rehabilitation training cycle is generally set to 3 days.

[0085] In the second sub-step, the description possibility corresponding to the marked word segmentation is determined according to the description deviation factor and the peak value of the number of times corresponding to the marked word segmentation.

[0086] The description deviation factor may be negatively correlated with the description possibility, and the peak value of the number of times may be positively correlated with the description possibility.

[0087] For example, the formula for determining the description possibility corresponding to the target word segmentation can be:

[0088] ;in, It is The description possibilities corresponding to the target segmentation. It is the type number of the target word. is a normalization function. It is The peak number of target word segmentations. It is an exponential function with a natural constant as its base. It is the absolute value function. is the preset training phase duration. It is The overall representative duration corresponding to the target segmentation. It is Description deviation factor corresponding to the target word segmentation.

[0089] It should be noted that in actual situations, as time goes by, hepatobiliary surgery patients often enter different stages of recovery. During this period, the patient's condition may improve or worsen. Therefore, the recorded information is often different in the overall process, but is often similar in the local scope. Therefore, if a participle appears more frequently in a local area, it often means that the participle is more likely to represent the change in the condition, that is, it may be more critical. The larger the The more frequently the target word appears locally, the more likely it is that This target segmentation may be more critical in the clinical treatment of hepatobiliary surgery. Since the patient's physical recovery is often different in different stages of rehabilitation, the rehabilitation treatment plans they need are often different, which leads to the recorded clinical information often showing stages. Therefore, the segmentation representing the rehabilitation status will often show stages to a certain extent. The smaller the time, the more likely it is that The more likely the target participle is to show stages in the rehabilitation process of patients with different histories, the more likely it is that the first This target word segmentation may be more critical in the clinical treatment of hepatobiliary surgery. The larger the This target segmentation may be more critical in the clinical treatment of hepatobiliary surgery.

[0090] Step 203: According to the description possibility corresponding to each target segmentation, the word frequency corresponding to each target segmentation is modified to obtain the target word frequency corresponding to each target segmentation.

[0091] As an example, this step may include the following steps:

[0092] The first step is to determine the initial word frequency corresponding to each target word segmentation through the keyword extraction algorithm.

[0093] Among them, the keyword extraction algorithm can be a TF-IDF method.

[0094] For example, the term frequency (TF) of each target word segmentation can be determined by the inverse document frequency method as the initial term frequency corresponding to each target word segmentation.

[0095] In the second step, if the description possibility corresponding to the target word segmentation is greater than the preset description threshold, the target word segmentation is determined as a candidate training description word.

[0096] The preset description threshold may be a preset threshold for screening candidate training description words. For example, the preset description threshold may be 0.68.

[0097] The third step is to determine the target word frequency corresponding to each candidate training descriptor according to the description possibility and initial word frequency corresponding to each candidate training descriptor.

[0098] Among them, description possibility and initial word frequency can both be positively correlated with target word frequency.

[0099] For example, the formula for determining the target word frequency corresponding to the candidate training description word can be:

[0100] ;in, It is The target word frequency corresponding to the candidate training description words. It is the type number of the candidate training description word. It is The description possibilities corresponding to the candidate training descriptors. It is The initial word frequencies corresponding to the candidate training description words.

[0101] It should be noted that when The larger the The candidate training descriptors may be more critical in the clinical treatment of hepatobiliary surgery. It can represent the word frequency before correction. The corrected word frequency can be represented.

[0102] In the fourth step, each target participle except the candidate training descriptor is determined as a reference participle, and the initial word frequency corresponding to each reference participle is determined as the target word frequency corresponding to each reference participle.

[0103] Step 204, determining the topic possibility corresponding to each target segmentation according to the amount of clinical information to which each target segmentation belongs and the similarity between the clinical information to which each target segmentation belongs.

[0104] As an example, this step may include the following steps:

[0105] In the first step, any target segmentation word is determined as a marked segmentation word, and the total number of clinical information to which the marked segmentation word belongs in all clinical information sets is determined as the number of recurring information corresponding to the marked segmentation word.

[0106] The second step, according to the number of recurring information corresponding to the above-mentioned marked word segmentation, determines the initial possible factor corresponding to the above-mentioned marked word segmentation, which may include the following sub-steps:

[0107] In the first sub-step, the total number of all clinical information in all clinical information sets is determined as the number of historical reference information.

[0108] In the second sub-step, the ratio of the number of recurring information corresponding to the above-mentioned marked word segmentation to the number of the above-mentioned historical reference information is determined as the initial possible factor corresponding to the above-mentioned marked word segmentation.

[0109] For example, the formula for determining the initial possible factors corresponding to the target word segmentation can be:

[0110] ;in, It is The initial possible factors corresponding to the target word segmentation. It is the type number of the target word. It is The number of recurring information corresponding to the target word. is the number of historical reference information.

[0111] It should be noted that, in actual situations, in the clinical data of hepatobiliary surgery, there are more rehabilitation training names because they can represent the theme of rehabilitation training and are often more critical. For example, if the theme of rehabilitation training is unclear, it is often difficult for doctors to judge the patient's condition and recovery. Among them, the name of rehabilitation training is also called the name of rehabilitation training program. Similarly, in the clinical data of hepatobiliary surgery, there are more rehabilitation treatment names because they can represent the theme of rehabilitation treatment and are often more critical. In general, the name of rehabilitation training can also be used as the name of rehabilitation treatment. When The larger the The more likely the target participle is to represent the rehabilitation training topic or rehabilitation treatment topic, the more likely it is that the This target segmentation may be more critical in the clinical treatment of hepatobiliary surgery.

[0112] In the third step, if the initial possibility factor corresponding to the target word segmentation is greater than the preset suspected topic possibility threshold, the target word segmentation is determined as a suspected topic word segmentation.

[0113] The preset suspected topic possible threshold may be a preset threshold for screening suspected topic segmentation words. For example, the suspected topic segmentation word may be 0.8.

[0114] The fourth step is to determine the topic possibility corresponding to each suspected topic segmentation based on the similarity between the clinical information to which each suspected topic segmentation belongs.

[0115] For example, the formula for determining the topic probability corresponding to the suspected topic segmentation word can be:

[0116] ;

[0117] ;in, It is The topic possibilities corresponding to the suspected topic participles. It is the type number of the suspected topic participle. is a normalization function. It is The number of clinical information sets to which the suspected subject participle belongs. It is The serial number of the clinical information set to which the suspected subject participle belongs. Characterization The suspected topic participle is Possible scenarios for topics under a clinical information collection. It is an exponential function with a natural constant as its base. It is The suspected topic participle is The similarity between the earliest clinical information and the latest clinical information in a clinical information set. The suspected topic participle is The earliest clinical information in a clinical information set can be: The suspected topic participle belongs to In the clinical information set, Among all the clinical information to which the suspected subject segment belongs, the clinical information with the earliest reference time. The suspected topic participle is The latest clinical information in a clinical information set can be: The suspected topic participle belongs to In the clinical information set, Among all clinical information to which the suspected topic segmentation belongs, the clinical information with the latest corresponding reference time. The method for obtaining the similarity between two clinical information can be: using the Jieba segmentation algorithm, obtain the segmentation vector of each target segmentation in the clinical information, record the two clinical information as the first clinical information and the second clinical information respectively, determine the cosine similarity between the segmentation vector of each target segmentation in the first clinical information and the segmentation vector of each target segmentation in the second clinical information as the similarity reference factor, and determine the mean of all similarity reference factors as the similarity between the first clinical information and the second clinical information. It is The suspected topic participle is The number of clinical information in a clinical information set. It is The suspected topic participle is The serial number of the clinical information to which it belongs in a clinical information set may be a serial number obtained by arranging in the time sequence of the reference time corresponding to the clinical information to which it belongs. It is The suspected topic participle is The clinical information set to which it belongs The information entropy of all target segmentations in clinical information. It is The suspected topic participle is In the clinical information set, it belongs to Clinical information and its The similarity between clinical information.

[0118] It should be noted that in actual situations, different rehabilitation stages often correspond to different rehabilitation training topics, and the rehabilitation training situations in the same rehabilitation stage are often the same. Generally speaking, in the hepatobiliary surgery clinical data of the same patient, the clinical information at the two ends of the same rehabilitation training topic often represents the beginning and end of a rehabilitation stage. The similarity between these two clinical information is often low, and the similarity between the clinical information of rehabilitation training topics that are adjacent in time is often high. The smaller the time, the more likely it is that The suspected topic participle is The smaller the similarity between the earliest clinical information and the latest clinical information in a clinical information set, the smaller the similarity between the earliest clinical information and the latest clinical information. The more the suspected topic segmentation matches the similarity characteristics of the two ends of the rehabilitation training topic, the more likely it is that the The more likely the suspected topic participle is, the more likely it is to be a key topic name. Can be used as When The larger the The suspected topic participle is The clinical information set to which it belongs The more chaotic the distribution of target words in clinical information is, the more likely it is that the first The suspected topic participle is The clinical information set to which it belongs The richer the information contained in the first clinical information, the more likely it is that the The suspected topic participle is The clinical information set to which it belongs The more important the clinical information is, the greater its weight should be. The larger the The suspected topic participle is In the clinical information set, it belongs to Clinical information and its The greater the similarity between the clinical information, the The more the suspected topic segmentation matches the similarity characteristics of the rehabilitation training topics when they appear adjacent to each other, the more likely it is that the The more likely the suspected topic participle is, the more likely it is a key topic name. The larger the The more likely the suspected topic participle is, the more likely it is to be a key topic name.

[0119] In the fifth step, each target segmentation word except the suspected topic segmentation word is determined as a temporary segmentation word, and a preset value is determined as the topic possibility corresponding to each temporary segmentation word.

[0120] The preset value may be a preset value representing the topic possibility that the temporary segmentation word is not a topic name. For example, the preset value may be 0.

[0121] Step 205: According to the topic probability corresponding to each target segmentation word, the inverse document frequency corresponding to each target segmentation word is corrected to obtain a target inverse document frequency.

[0122] As an example, this step may include the following steps:

[0123] The first step is to determine the initial inverse document frequency corresponding to each target word through the keyword extraction algorithm.

[0124] For example, the inverse document frequency method can be used to determine the inverse document frequency (IDF) of each target word segmentation as the initial inverse document frequency corresponding to each target word segmentation.

[0125] In the second step, if the topic possibility corresponding to the target segmentation word is greater than the preset true topic possibility threshold, the target segmentation word is determined as a candidate topic segmentation word.

[0126] The preset real topic possible threshold may be a preset threshold for screening candidate topic segmentation words. For example, the preset real topic possible threshold may be 0.88.

[0127] The third step is to determine the target inverse document frequency corresponding to each candidate topic segmentation word based on the topic possibility and initial inverse document frequency corresponding to each candidate topic segmentation word.

[0128] Among them, both the topic likelihood and the initial inverse document frequency can be positively correlated with the target inverse document frequency.

[0129] For example, the formula for determining the target inverse document frequency corresponding to the candidate topic segmentation word can be:

[0130] ;in, It is The target inverse document frequency corresponding to the candidate topic segmentation words. It is the type number of the candidate topic segmentation word. It is The topic possibilities corresponding to the candidate topic segmentations. It is The initial inverse document frequency corresponding to the candidate topic segmentation words.

[0131] It should be noted that when The larger the The more likely the candidate topic segmentation words are, the more likely they are key topic names. The inverse document frequency before correction can be characterized. The corrected inverse document frequency can be characterized.

[0132] In the fourth step, each target participle except the candidate topic participle is determined as a pending participle, and the initial inverse document frequency corresponding to each pending participle is determined as the target inverse document frequency corresponding to each pending participle.

[0133] Step 206: perform keyword extraction on all target segmented words according to the target word frequencies and target inverse document frequencies corresponding to all target segmented words to obtain a target keyword set.

[0134] As an example, the word frequency and inverse document frequency of the target word segmentation are updated to their corresponding target word frequency and target inverse document frequency respectively. The inverse document frequency method is used to extract keywords from all target word segmentations. The keywords extracted at this time are the target keywords, and all target keywords constitute a target keyword set.

[0135] Step S2: Filter out target clinical data from all clinical information corresponding to the patient to be tested according to the target keyword set.

[0136] The patient to be tested may be a patient to be tested for a disease condition.

[0137] In some embodiments, data containing target keywords can be screened out from all clinical information corresponding to the patient to be tested as target clinical data, thereby forming an electronic medical record. The electronic medical record is transmitted to the medical record system, which can realize the intelligent collection of hepatobiliary surgery clinical data.

[0138] refer to Figure 3 Based on the same inventive concept as the above method embodiment, the present invention provides an intelligent collection system for clinical data of hepatobiliary surgery, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the above computer program is executed by the processor, the steps of implementing an intelligent collection method for clinical data of hepatobiliary surgery may specifically include:

[0139] Keyword acquisition module 301, used to acquire a target keyword set corresponding to hepatobiliary surgery clinical practice;

[0140] The data screening module 302 is used to screen out target clinical data from all clinical information corresponding to the patient to be tested according to the above target keyword set.

[0141] Figure 4 is a schematic diagram of the structure of a computer device provided by an embodiment of the present invention. Figure 4As shown, the computer device 400 includes: a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402, wherein when the processor 402 executes the computer program 403, the computer device can execute any one of the intelligent collection methods for hepatobiliary surgery clinical data introduced above.

[0142] Based on the same inventive concept as the above method embodiment, the present invention provides a server, including a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, so that the device executes any of the above-mentioned intelligent collection methods for clinical data of hepatobiliary surgery.

[0143] Based on the same inventive concept as the above-mentioned method embodiment, the present invention provides a computer program product, which includes: a computer program code, when the computer program code runs on a computer, enables the computer to execute any one of the above-mentioned intelligent collection methods for hepatobiliary surgery clinical data.

[0144] Based on the same inventive concept as the above-mentioned method embodiments, the present invention provides an intelligent collection medium for hepatobiliary surgery clinical data, which stores computer program code. When the computer program code runs on a computer, the computer executes any one of the above-mentioned intelligent collection methods for hepatobiliary surgery clinical data.

[0145] In summary, the present invention comprehensively considers multiple factors related to the keywords of hepatobiliary surgery clinical data, such as description possibility and topic possibility, so as to adaptively realize the correction of word frequency and inverse document frequency in the inverse document frequency method, thereby improving the rationality of keyword extraction, thereby improving the rationality of rapid collection of hepatobiliary surgery clinical data.

[0146] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features can be replaced by equivalents. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. An intelligent method for collecting clinical data of hepatobiliary surgery, characterized in that: The following steps are involved: Obtain the target keyword set corresponding to hepatobiliary surgery clinical practice; According to the target keyword set, target clinical data is screened out from all clinical information corresponding to the patient to be tested; The method for obtaining the target keyword set comprises the following steps: Obtain the clinical information set corresponding to all historical patients of hepatobiliary surgery, and perform word segmentation on the clinical information in each clinical information set to obtain the target word segmentation; Determine the description possibility corresponding to each target participle according to the duration of each target participle appearing in the clinical information set to which it belongs and the number of times each target participle appears in the clinical information to which it belongs; According to the description possibility corresponding to each target participle, the word frequency corresponding to each target participle is modified to obtain the target word frequency corresponding to each target participle; According to the amount of clinical information to which each target segmentation belongs and the similarity between the clinical information to which each target segmentation belongs, the possibility of the topic corresponding to each target segmentation is determined; According to the topic possibility corresponding to each target segmentation word, the inverse document frequency corresponding to each target segmentation word is corrected to obtain the target inverse document frequency; According to the target word frequency and target inverse document frequency corresponding to all target segmented words, keyword extraction is performed on all target segmented words to obtain a target keyword set; Determining the description possibility corresponding to each target participle according to the duration of appearance of each target participle in the clinical information set to which it belongs and the number of times each target participle appears in the clinical information to which it belongs includes: Determine any target participle as a marked participle, and determine the number of times the marked participle appears in each clinical information to which it belongs as the target number of times the marked participle appears in each clinical information to which it belongs; The maximum value of the target times of the marked word under all the clinical information to which it belongs is determined as the peak number of times corresponding to the marked word; Determine the duration between the earliest appearance time and the latest appearance time of the marked participle in each clinical information set to which it belongs as the target duration of the marked participle in each clinical information set to which it belongs, wherein the target duration of the marked participle in each clinical information set to which it belongs represents the duration of the appearance of the marked participle in each clinical information set to which it belongs; The average of the target durations of the marked participle under all clinical information sets to which it belongs is determined as the overall representative duration corresponding to the marked participle; Determining the description possibility corresponding to the marked segmentation according to the overall representative duration and the peak number of times corresponding to the marked segmentation; Determining the description possibility corresponding to the marked segmentation according to the overall representative duration and the peak number of times corresponding to the marked segmentation includes: Determine the difference between the preset training phase duration and the overall representative duration corresponding to the marked segmentation as the description deviation factor corresponding to the marked segmentation; Determine the description possibility corresponding to the marked word according to the description deviation factor and the number peak value corresponding to the marked word, wherein the description deviation factor is negatively correlated with the description possibility, and the number peak value is positively correlated with the description possibility; Determining the topic possibility corresponding to each target segmentation according to the amount of clinical information to which each target segmentation belongs and the similarity between the clinical information to which each target segmentation belongs includes: Determine any target segmentation as a marked segmentation, and determine the total number of clinical information to which the marked segmentation belongs in all clinical information sets as the number of recurring information corresponding to the marked segmentation; Determining an initial possible factor corresponding to the marked word segmentation according to the number of recurring information corresponding to the marked word segmentation; If the initial possibility factor corresponding to the target word is greater than the preset possible threshold of the suspected topic, the target word is determined as the suspected topic word; According to the similarity between the clinical information to which each suspected topic participle belongs, the topic possibility corresponding to each suspected topic participle is determined; Determine each target participle except the suspected topic participle as a temporary participle, and determine a preset value as the topic possibility corresponding to each temporary participle; The determining, according to the number of recurring information corresponding to the marked word segmentation, the initial possible factor corresponding to the marked word segmentation includes: The total number of all clinical information in all clinical information sets is determined as the number of historical reference information; Determine the ratio of the number of recurring information corresponding to the marked word segment to the number of historical reference information as the initial possible factor corresponding to the marked word segment; The formula corresponding to the topic possibility corresponding to the suspected topic segmentation word is: Among them, Y a is the topic probability corresponding to the a-th suspected topic segmentation word; a is the type number of the suspected topic segmentation word; sigmod() is the normalization function; N a is the number of clinical information sets to which the a-th suspected subject segmentation belongs; b is the serial number of the clinical information set to which the a-th suspected subject segmentation belongs; Y ab represents the possible situation of the topic of the a-th suspected topic participle under the b-th clinical information set to which it belongs; exp( ) is an exponential function with a natural constant as the base; c ab is the similarity between the earliest clinical information and the latest clinical information of the a-th suspected topic participle in the b-th clinical information set; n ab is the number of clinical information to which the a-th suspected topic participle belongs in the b-th clinical information set to which it belongs; f is the sequence number of the clinical information to which the a-th suspected topic participle belongs in the b-th clinical information set to which it belongs; σ abf is the information entropy of all target segmentations in the a-th suspected topic segmentation in the b-th clinical information set to which it belongs and in the f-th clinical information to which it belongs; c ab,f,f+1 It is the similarity between the a-th suspected topic participle and the f-th clinical information to which it belongs and the f+1-th clinical information to which it belongs in the b-th clinical information set.

2. The intelligent collection method of clinical data of hepatobiliary surgery according to claim 1 is characterized in that: The word frequency corresponding to each target participle is modified according to the description possibility corresponding to each target participle to obtain the target word frequency corresponding to each target participle, including: Determine the initial word frequency corresponding to each target word segmentation through the keyword extraction algorithm; If the description possibility corresponding to the target word is greater than the preset description threshold, the target word is determined as a candidate training description word; According to the description possibility and initial word frequency corresponding to each candidate training descriptor, the target word frequency corresponding to each candidate training descriptor is determined, wherein the description possibility and the initial word frequency are both positively correlated with the target word frequency; Each target participle except the candidate training description word is determined as a reference participle, and the initial word frequency corresponding to each reference participle is determined as the target word frequency corresponding to each reference participle.

3. The intelligent collection method of clinical data of hepatobiliary surgery according to claim 1 is characterized in that: The method of correcting the inverse document frequency corresponding to each target segmentation word according to the topic possibility corresponding to each target segmentation word to obtain the target inverse document frequency includes: Determine the initial inverse document frequency corresponding to each target word segmentation through the keyword extraction algorithm; If the topic probability corresponding to the target word is greater than the preset real topic possibility threshold, the target word is determined as a candidate topic word; According to the topic possibility and initial inverse document frequency corresponding to each candidate topic segmentation word, the target inverse document frequency corresponding to each candidate topic segmentation word is determined, wherein the topic possibility and the initial inverse document frequency are both positively correlated with the target inverse document frequency; Each target participle except the candidate topic participle is determined as a pending participle, and the initial inverse document frequency corresponding to each pending participle is determined as the target inverse document frequency corresponding to each pending participle.

4. An intelligent collection system for clinical data of hepatobiliary surgery, characterized in that: It comprises a processor and a memory, wherein the processor is used to process instructions stored in the memory to implement an intelligent collection method for clinical data of hepatobiliary surgery according to any one of claims 1-3.

5. An intelligent collection medium for clinical data of hepatobiliary surgery, characterized in that: Computer program codes are stored, and when the computer program codes are run on a computer, the computer is enabled to execute an intelligent method for collecting clinical data of hepatobiliary surgery as described in any one of claims 1-3.

Images