Medical examination medical record information arrangement method for cardiac surgery department

Abstract

The invention relates to the technical field of medical record data mining, in particular to a cardiac surgery oriented medical examination medical record information arrangement method. The method comprises the following steps: firstly, obtaining a sentence vector under each information index in all medical record data; further analyzing to obtain a change deviation index of each medical record data, and screening out all reference medical record data; further analyzing and acquiring relativity of each information index in each medical record data, and further analyzing and acquiring an abnormal attention index of each information index in the medical record data; and finally, screening out all concerned information indexes in each medical record data. According to the invention, by analyzing and mining the change characteristics of each medical record data and the relativity among patients, the abnormal deviation condition of each information index in each medical record data is analyzed, and the abnormal attention index is determined, so that related medical personnel are assisted to analyze the information indexes in an emphasized manner; and therefore, the analysis efficiency of each electronic medical record table can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical case record data mining, and in particular to a method for collating medical examination case record information for cardiac surgery. Background Art

[0002] With the development of medical information technology, digital medical management tools such as electronic medical record systems are gradually being used in daily medical care. By recording the patient's basic information, physical signs, clinical test results and other information indicators, it is convenient for relevant doctors to quickly retrieve the patient's medical records for analysis and treatment.

[0003] However, the number of information indicators is huge, including name, home address, irrelevant medical history, etc. These information indicators have little impact on the current symptom analysis, making it difficult for doctors to quickly retrieve the information indicators that the patient should currently focus on. This is especially true for cardiac surgeons who often face many emergencies. The huge amount of information indicators will reduce the efficiency of analyzing medical record information. Summary of the invention

[0004] In order to solve the technical problem that the information indicators in the prior art are complex and thus the analysis efficiency of medical record information is low, the purpose of the present invention is to provide a method for collating medical examination medical record information for cardiac surgery. The technical solution adopted is as follows: The present invention proposes a method for collating medical examination case record information for cardiac surgery, the method comprising: Obtain the medical record data of all patients visiting the Department of Cardiac Surgery during each visit, wherein the medical record data includes the sentence vectors of the data recorded under each information indicator; take any visiting patient as the target patient, and take any medical record data of the target patient as the target medical record data; Among the different medical records of the target patient, according to the difference between the sentence vectors under each information indicator, obtain the change deviation index of the target medical record data relative to all other medical record data; according to the similarity between the change deviation index of the target medical record data and all the medical record data of all other patients, filter out all reference medical record data of the target medical record data from all the medical record data; Screen out all relatives of the target patient from all the patients who have visited the hospital; obtain the relative relevance of each information indicator in the target medical record data based on the difference between the sentence vectors under each information indicator between the target medical record data and all the medical record data of each relative of the target patient; Based on the difference between the sentence vectors of each information indicator between the target medical record data and each of the reference medical record data and combined with the relative association of each information indicator in the target medical record data, the abnormal attention index of each information indicator in the target medical record data is obtained; and all the information indicators of concern in the target medical record data are screened out according to the abnormal attention index.

[0005] Furthermore, the method for obtaining the change deviation index includes: Among all the medical records of the target patient, between the target medical record data and each of the remaining medical record data, according to the vector difference between the sentence vectors under each information indicator, obtaining a change parameter under each information indicator; the vector difference is positively correlated with the change parameter; Comprehensively calculate the change parameters under each information indicator between the target medical record data and all other medical record data to obtain the parameter deviation coefficient under each information indicator in the target medical record data; The parameter deviation coefficients under all information indicators in the comprehensive target medical record data are used to obtain the change deviation index of the target medical record data relative to all other medical record data.

[0006] Furthermore, the method for screening the reference medical record data includes: Among all the medical record data of all other patients except the target patient, if the difference between the change deviation index of the medical record data and the change deviation index of the target medical record data is less than a preset threshold, the medical record data will be used as reference medical record data for the target medical record data.

[0007] Furthermore, the method for obtaining the kinship association includes: Between the target medical record data and each of the medical record data of each relative of the target patient, according to the vector difference between the sentence vectors under each information indicator, obtaining a relative difference parameter under each information indicator; the vector difference is positively correlated with the relative difference parameter; The relative difference parameters under each information indicator are combined between the target medical record data and all the medical record data of each relative patient of the target patient to obtain the relative deviation coefficient under each information indicator in the target medical record data; the negative correlation normalization result of the relative deviation coefficient is used as the relative correlation under the corresponding information indicator in the target medical record data.

[0008] Furthermore, the method for obtaining the abnormal attention index includes: Between the target medical record data and each of the reference medical record data, according to the difference between the sentence vectors under each information indicator, combined with the relative association of each information indicator in the target medical record data, obtain a group difference parameter of each information indicator in the target medical record data; Between the target medical record data and each non-reference medical record data, according to the difference between the sentence vectors under each information indicator, combined with the relative association of each information indicator in the target medical record data, obtain a universal difference parameter for each information indicator in the target medical record data; According to the group difference parameter and the universal difference parameter, combined with the relative correlation of each information indicator in the target medical record data, and the difference in the sentence vector under each information indicator between the target medical record data and the reference medical record data, the abnormal attention index of each information indicator in the target medical record data is obtained.

[0009] Furthermore, the method for obtaining the group difference parameter includes: Between the target medical record data and each of the reference medical record data, taking the vector difference between the sentence vectors under each information indicator as a reference difference parameter under each information indicator; The reference difference parameters under the corresponding information index are weighted by using the kinship association, and the weighted sum of all the reference difference parameters under the corresponding information index between the target medical record data and all the reference medical record data is normalized, and the normalized result is used as the group difference parameter of the target medical record data under the corresponding information indicator.

[0010] Furthermore, the method of obtaining the abnormal attention index of each information indicator in the target medical record data according to the group difference parameter and the universal difference parameter, combined with the relative association of each information indicator in the target medical record data, and the difference of the sentence vector under each information indicator between the target medical record data and the reference medical record data includes: Determining whether to perform relative screening on the target patient according to the magnitude relationship between the group difference parameter and the universal difference parameter; If relative screening is performed, then the abnormal concern index of each information indicator in the target medical record data is obtained based on the group difference parameter and the universal difference parameter, combined with the relative relevance of each information indicator in the target medical record data; If relative screening is not performed, the abnormal attention index of each information indicator in the target medical record data is obtained according to the difference in the sentence vector under each information indicator between the target medical record data and the reference medical record data.

[0011] Furthermore, the method of obtaining the abnormal attention index of each information indicator in the target medical record data according to the group difference parameter and the universal difference parameter in combination with the relative association of each information indicator in the target medical record data includes: Under each information indicator in the target medical record data, the relative association is used as the weight of the group difference parameter, the difference between the constant 1 and the relative association is used as the weight of the universal difference parameter, and the weighted sum result is used as the abnormal attention index of each information indicator in the target medical record data.

[0012] Furthermore, the method for obtaining the abnormal attention index of each information indicator in the target medical record data according to the difference of the sentence vector under each information indicator between the target medical record data and the reference medical record data includes: Between the target medical record data and each of the reference medical record data, the Hamming distance between the sentence vectors under each information indicator is used as a first abnormality parameter, and the cosine distance between the sentence vectors under each information indicator is used as a second abnormality parameter; The product of the first abnormal parameter and the second abnormal parameter is used as the abnormal parameter of each information indicator between the target medical record data and each of the reference medical record data; the sum of the abnormal parameters of each information indicator between the target medical record data and all the reference medical record data is normalized, and the normalized result is used as the abnormal attention index of each information indicator in the target medical record data.

[0013] Furthermore, the method for obtaining the attention information indicator includes: The information indicators in the target medical record data that are greater than the preset threshold are taken as the information indicators of interest.

[0014] The present invention has the following beneficial effects: The present invention obtains the medical record data of all patients visiting the cardiac surgery department during each visit, and the medical record data includes the sentence vectors of the data recorded under each information indicator; any visiting patient is taken as the target patient, and any medical record data of the target patient is taken as the target medical record data; between different medical record data of the target patient, according to the difference between the sentence vectors under each information indicator, the change deviation index of the target medical record data relative to all other medical record data is obtained, and the change deviation index roughly quantifies the deviation difference of the symptom characteristics of the target medical record data of the target patient during the visit relative to the symptom characteristics of the other visits, so as to facilitate the screening of all reference medical record data of the target medical record data from all medical record data, and the reference medical record data can reflect the similar symptom groups of the target patient from the side, so as to provide a comparison basis for subsequent analysis; all relatives of the target patient are screened out from all visiting patients; between the target medical record data and all medical record data of each relative patient of the target patient, according to each The differences between the sentence vectors under the information indicators are used to obtain the kinship relevance of each information indicator in the target medical record data. The kinship relevance analyzes the possibility of the target patient having a family genetic disease from the perspective of each information indicator. The greater the possibility of family inheritance, the greater the possibility of attention to the corresponding information indicator. Between the target medical record data and each reference medical record data, according to the differences between the sentence vectors of each information indicator, combined with the kinship relevance of each information indicator in the target medical record data, the abnormal attention index of each information indicator in the target medical record data is obtained. The abnormal attention index reflects the abnormal deviation of the sentence vector corresponding to the target medical record data and the target patient's similar disease characteristic group corresponding to all reference medical record data. It can be further used to assist in judging whether the target patient's disease is worse or better than other similar patients, and quantify the necessity of being analyzed and treated in detail. All information indicators of concern in the target medical record data are screened out according to the abnormal attention index. The present invention analyzes and mines the changing characteristics of each medical record data and the kinship relationships between the patients, thereby analyzing the abnormal deviation of each information indicator in each medical record data, determining its abnormal attention index, thereby assisting relevant medical personnel to analyze the information indicators in a focused manner, and further assisting in improving the analysis efficiency of each electronic medical record form. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] 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.

[0016] Figure 1A flowchart of a method for collating medical examination records information for cardiac surgery provided by one embodiment of the present invention; Figure 2 A flowchart of a method for obtaining an abnormal attention index provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a medical examination medical record information arrangement method for cardiac surgery proposed by the present invention in combination 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.

[0018] 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.

[0019] The following is a detailed description of a method for organizing medical examination case history information for cardiac surgery provided by the present invention in conjunction with the accompanying drawings.

[0020] See also Figure 1 , which shows a flow chart of a method for collating medical examination medical record information for cardiac surgery provided by an embodiment of the present invention, specifically comprising: Step S1, obtain the medical record data of all patients visiting the Department of Cardiac Surgery during each visit, the medical record data including the sentence vectors recording the data under each information indicator; any visiting patient is taken as the target patient, and any medical record data of the target patient is taken as the target medical record data.

[0021] It should be noted that the embodiment of the present invention is aimed at specially marking the electronic medical record information of cardiac surgery patients who have visited the hospital multiple times, so as to help improve the efficiency of relevant medical personnel in analyzing and judging symptoms, that is, it is assumed that each patient corresponds to multiple medical record data.

[0022] In one embodiment of the present invention, the electronic medical record forms of all patients visiting the cardiac surgery department during each visit are first retrieved from the electronic medical record system, wherein the electronic medical record forms generally include various information indicators such as the patient's basic information, physical sign information, and clinical test results, as well as the corresponding record results under each information indicator.

[0023] As an example, the information indicators mainly include: the information indicators under the basic information at least include the name, age, gender, description of the work environment, height, weight, etc. of the patient; the information indicators under the physical sign information at least include heart rate, blood pressure, respiratory rate, blood oxygen saturation, chief complaint, past history, family history, drug allergy history, medication history, etc.; the information indicators under the clinical test results at least include myocardial enzyme spectrum, blood lipids, blood sugar, electrolytes, troponin, electrocardiogram diagnosis opinion, echocardiogram diagnosis opinion, coronary angiography diagnosis opinion, etc.; by combining the above three categories of situations, all information indicators can be obtained.

[0024] It should be noted that the types and quantities of information indicators are determined based on the design structure of the electronic medical record form, and implementers can also determine the types and quantities of information indicators based on actual conditions.

[0025] Considering that in different electronic medical record forms, the recorded data under the same information indicators usually involve complex text content, which may contain rich semantic information and contextual relationships, etc., direct comparison of the recorded data may not be able to fully and accurately evaluate the differences; and based on the semantic processing model to obtain the sentence vector of each record data, and then by comparing the differences between the sentence vectors, compared with directly comparing the text content differences of the record data, it can better capture the deep semantic relationship between the texts, thereby improving the accuracy of the difference analysis of the record data; considering that the BERT model is a pre-trained language model based on deep learning, which can be trained on large-scale text data, so as to learn rich context-related semantic information; therefore, in one embodiment of the present invention, the sentence vector of the record data under each information indicator in each medical record data is obtained based on the BERT model.

[0026] It should be noted that obtaining sentence vectors for text data based on the BERT model is an existing technology well known to those skilled in the art and will not be elaborated here.

[0027] In order to facilitate subsequent analysis to obtain the information indicators that should be focused on in each electronic medical record form of each patient and perform special marking processing, the embodiment of the present invention will combine the changes and differences in the recorded data under the same information indicators in different electronic medical record forms between each patient and other patients, to quantitatively evaluate the importance of each information indicator in each electronic medical record form; since the method for determining the importance of information indicators in each electronic medical record form, that is, medical record data, is consistent, only any medical record data of any patient is used as an example for description, and the analysis and processing methods of other medical record data are consistent, and they will not be repeated here one by one; therefore, the embodiment of the present invention will take any patient as the target patient, and any medical record data of the target patient as the target medical record data.

[0028] Step S2, among different medical records of the target patient, based on the difference between the sentence vectors under each information indicator, obtain the change deviation index of the target medical record data relative to all other medical record data; based on the similarity between the change deviation index of the target medical record data and all other medical record data of all the patients who visited the hospital, filter out all reference medical record data of the target medical record data from all medical record data.

[0029] Taking into account the different diagnosis and treatment stages, the recorded data under the same information indicator may have certain changes in the different medical records of each patient. For example, under the information indicator corresponding to the electrocardiogram diagnosis opinion, the recorded data corresponding to different treatment processes may be different due to factors such as changes in the patient's condition, that is, the sentence vector corresponding to the recorded data under the same information indicator may change; and considering that a medical record data can roughly reflect the patient's condition during the treatment process, by analyzing the differences between the medical record data corresponding to different treatment processes of the same patient, it is possible to evaluate the changes in the patient's condition during each treatment process, that is, to judge the approximate stage of change of the patient's symptom characteristics, and then facilitate the subsequent screening of similar symptom groups of the patient according to the similarity of the symptom characteristics, so as to comprehensively analyze the medical record data of similar groups and determine important information indicators.

[0030] Therefore, the embodiment of the present invention first obtains the change deviation index of the target medical record data relative to all other medical record data according to the difference between the sentence vectors under each information indicator between different medical record data of the target patient; the change deviation index roughly quantifies the deviation difference of the symptom characteristics of the target medical record data corresponding to the treatment process of the target patient relative to the symptom characteristics of the other treatment processes; further based on the change deviation index of each medical record data, all reference medical record data with similar symptom characteristics corresponding to the target medical record data are screened out from all medical record data of all other patients except the target patient. The reference medical record data can reflect the similar symptom groups of the target patients from the side, so as to provide a comparison basis for subsequent analysis.

[0031] Preferably, in one embodiment of the present invention, considering that each medical record data contains many information indicators, when analyzing the change deviation index of the target medical record data relative to the remaining medical record data, the difference between the sentence vectors under the same information indicator in different medical record data can be analyzed first, and then all information indicators are evaluated comprehensively; the method for obtaining the change deviation index includes: In all the medical records of the target patient, the change parameter under each information indicator is obtained based on the vector difference between the sentence vectors under each information indicator between the target medical record data and each of the remaining medical record data; the vector difference is positively correlated with the change parameter; Comprehensively calculate the change parameters under each information indicator between the target medical record data and all other medical record data, and obtain the parameter deviation coefficient under each information indicator in the target medical record data; The parameter deviation coefficients of all information indicators in the comprehensive target medical record data are used to obtain the change deviation index of the target medical record data relative to all other medical record data.

[0032] As an example, the calculation formula for the change deviation indicator is: ; in, The serial number of the target medical record data of the target patient; Among all the medical records of the target patient, the change deviation index of the target medical record data relative to all other medical record data; is the serial number of the information indicator; is the total number of information indicators in each medical record data; is an S-type function; is the serial number of all the medical records of the target patient except the target medical record; The total number of all medical records of the target patient; The target medical record data Sentence vectors under information indicators; For the target patient The medical records Sentence vectors under information indicators; is the vector difference metric function; The target medical record data of the target patient and the The medical records are in The change parameters under the information index; The target patient's target medical record data The parameter deviation coefficient under the information index.

[0033] In the calculation formula of the change deviation index, since sentence vectors are usually high-dimensional vectors, it is difficult to use traditional Euclidean distance to measure vector differences. Since cosine similarity can be used to measure the semantic similarity between texts, the vector difference measurement function is specifically cosine distance, thereby obtaining a change parameter. The larger the cosine distance, the larger the change parameter, indicating that the change in the data recorded by the information indicator is large. Then all the change parameters under the information indicator are accumulated. The larger the accumulated value, the greater the change in the target medical record data under the information indicator relative to all other medical record data, and the larger the parameter deviation coefficient. Then the parameter deviation coefficients under all information indicators are accumulated to obtain the change deviation index.

[0034] In other examples, implementers may also use other distance measurement methods such as Manhattan distance to obtain vector differences, which and cosine distance are well-known technologies and will not be described in detail here.

[0035] After obtaining the change deviation index reflecting the change characteristics of the symptoms corresponding to the target medical record data, all reference medical record data of the target medical record data can be further obtained.

[0036] Preferably, in one embodiment of the present invention, among all the medical records of all the patients except the target patient, if the absolute value of the difference between the change deviation index of the medical record data and the change deviation index of the target medical record data is less than a preset threshold, the medical record data is used as the reference medical record data of the target medical record data. The preset threshold is 0.1, which can also be set by the implementer, but should not be too large to avoid the reference value of the reference medical record data screened out being low.

[0037] At this point, all the reference medical record data of the target medical record data are screened out from all the medical record data of all other patients in the Department of Cardiac Surgery except the target patient.

[0038] Step S3, screen out all relatives of the target patient from all the patients who have visited the hospital; obtain the relative correlation of each information indicator in the target medical record data based on the difference between the sentence vectors under each information indicator between the target medical record data and all the medical record data of each relative of the target patient.

[0039] Considering that cardiac surgical diseases usually have a certain degree of familial inheritance, special attention should be paid to the differences between the medical records of the target patient and his close relatives; therefore, the embodiment of the present invention first screens out all relatives of the target patient from all the patients who have visited the hospital.

[0040] It should be noted that close relatives mainly include direct relatives and collateral relatives; when retrieving all patients visiting the Department of Cardiac Surgery from the electronic medical record system, the consent of each patient will be sought first, and all patients corresponding to their direct relatives and collateral relatives will be regarded as their relative patients. Then, they and their relative patients will be labeled with the same label, and then all relative patients of the target patient will be screened out according to the label.

[0041] Since some information indicators in the medical record data may be affected by genetic factors, which may lead to similar recorded data, for patients with familial hereditary cardiovascular diseases, the greater the possibility that such information indicators should be the information indicators that should be focused on, so that doctors can discover potential risks as early as possible and make intervention treatments; therefore, the embodiment of the present invention will obtain the relative relevance of each information indicator in the target medical record data based on the difference between the sentence vectors under each information indicator between the target medical record data and all the medical record data of each relative patient of the target patient. Relative relevance analyzes the possibility of the target patient having a familial hereditary disease from the perspective of each information indicator; the greater the relative relevance, the greater the possibility of familial inheritance, and the greater the possibility of paying attention to the corresponding information indicator.

[0042] Preferably, in one embodiment of the present invention, considering that under the same information index, if the sentence vector difference between the target medical record data of the target patient and the data recorded in all the medical record data of the other relative patients is smaller, it means that the disease characteristics caused by family genetic factors are more similar, and the possibility of inheritance is greater; therefore, the method for obtaining relative association includes: Between the target medical record data and each medical record data of each relative of the target patient, the relative difference parameter under each information indicator is obtained according to the vector difference between the sentence vectors under each information indicator; the vector difference is positively correlated with the relative difference parameter; The relative difference parameters under each information indicator are combined between the target medical record data and all the medical record data of each relative of the target patient to obtain the relative deviation coefficient under each information indicator in the target medical record data; the negative correlation normalization result of the relative deviation coefficient is used as the relative correlation under the corresponding information indicator in the target medical record data.

[0043] As an example, the formula for calculating kinship relatedness is: ; in, The serial number of the target medical record data of the target patient; is the serial number of the information indicator; The target patient's target medical record data Relative relatedness under information indicators; The serial numbers of all medical records of all relatives of the target patient; The total number of all medical records of all relatives of the target patient; is an exponential function with the natural constant e as base; is the vector difference metric function; The target medical record data Sentence vectors under information indicators; The first The medical records Sentence vectors under information indicators; The target medical record data and the relative patient's The medical records are in The relative difference parameters under the information index; The target patient's target medical record data The relative deviation coefficient under the information index.

[0044] In the calculation formula of kinship association, the vector difference measurement function is the cosine distance. The smaller the cosine distance, the smaller the kinship difference parameter, which indirectly indicates that the more similar the recorded data under the information indicator, the greater the possibility of being affected by family heredity; then all kinship difference parameters under the information indicator are accumulated. The smaller the accumulated value, the smaller the kinship deviation coefficient, which further indicates that the more similar the recorded data of the target medical record data relative to all medical record data of the relative patients under the information indicator, the greater the possibility of being affected by family heredity; then the negative correlation mapping logic of the kinship deviation coefficient under the information indicator is adjusted so that the smaller the kinship deviation coefficient, the greater the kinship association.

[0045] In other examples, implementers may also use other distance measurement methods such as Manhattan distance to obtain vector differences, or use other negative correlation mapping methods such as performing inverse operations on the kinship deviation coefficient, which are well-known technologies like cosine distance and will not be described in detail here.

[0046] At this point, the relative correlation under each information indicator in the target medical record data can be obtained to prepare for the subsequent evaluation of the differences between the target medical record data and the reference medical record data.

[0047] Step S4, between the target medical record data and each reference medical record data, based on the difference between the sentence vectors of each information indicator and combined with the relative association of each information indicator in the target medical record data, obtain the abnormal attention index of each information indicator in the target medical record data; and filter out all the information indicators of concern in the target medical record data according to the abnormal attention index.

[0048] Considering that the symptom characteristics of the corresponding patients should be relatively similar between the target medical record data and the reference medical record data, the difference between the sentence vectors under each information indicator should also be smaller; but if the difference between the sentence vectors under the corresponding information indicator is larger, it means that the recorded data under the information indicator is more abnormal, and it should be focused on and analyzed more, so as to assist doctors in discovering disease risks as early as possible; considering that the greater the kinship relevance of the information indicator, the greater the necessity of focusing on the information indicator; therefore, the embodiment of the present invention obtains the abnormal attention index of each information indicator in the target medical record data based on the difference between the sentence vectors of each information indicator between the target medical record data and each reference medical record data, combined with the kinship relevance of each information indicator in the target medical record data.

[0049] The abnormal attention index reflects the abnormal deviation of the sentence vector corresponding to the target medical record data and the target patient's similar disease characteristic group corresponding to all reference medical record data. The more abnormal the deviation, the more this information indicator can reflect the target patient's disease characteristics in the diagnosis and treatment process corresponding to the target medical record data. The more it deviates from other similar patient groups, and can be further used to assist in judging whether the target patient's disease is worse or better than other similar patients, and the more it needs to be analyzed and treated by the doctor.

[0050] Preferably, in one embodiment of the present invention, the method for obtaining the abnormal attention index includes: See also Figure 2 , which shows a flow chart of a method for obtaining an abnormal attention index provided by an embodiment of the present invention, specifically comprising: Step S201, between the target medical record data and each reference medical record data, based on the difference between the sentence vectors under each information indicator, combined with the relative association of each information indicator in the target medical record data, obtain the group difference parameter of each information indicator in the target medical record data.

[0051] Considering that the greater the difference in sentence vectors under each information indicator between the target medical record data and the reference medical record data and the greater the kinship correlation of the information indicator, it means that under this information indicator, the greater the difference between the recorded data in the target medical record data and the recorded data in the remaining reference medical record data, that is, the more the symptom characteristics of the target patient in the corresponding medical treatment process of the target medical record data deviate from the similar symptom group, the greater the group difference parameter of each information indicator in the target medical record data.

[0052] In a preferred embodiment of the present invention, the method for obtaining group difference parameters includes: Between the target medical record data and each reference medical record data, the vector difference between the sentence vectors under each information indicator is used as the reference difference parameter under each information indicator; The reference difference parameters under the corresponding information indicators are weighted by using relative association, and the weighted sum of all reference difference parameters under the corresponding information indicators between the target medical record data and all reference medical record data are normalized, and the normalized result is used as the group difference parameter of the target medical record data under the corresponding information indicator.

[0053] As an example, the population difference parameter is calculated as: ; in, The serial number of the target medical record data of the target patient; is the serial number of the information indicator; The target patient's target medical record data Group difference parameters under information indicators; is an S-type function; The serial number of the reference medical record data of the target medical record data; The total number of reference medical record data for the target medical record data; The target patient's target medical record data Relative relatedness under information indicators; is the vector difference metric function; The target medical record data Sentence vectors under information indicators; The target medical record data Reference medical records Sentence vectors under information indicators; The target medical record data and Reference medical records data in The reference difference parameter under the information index.

[0054] In the calculation formula of the group difference parameter, the weighted summation result is mapped to the S-type function for normalization, and the vector difference measurement function is the cosine distance. The implementer can also use other normalization methods and vector difference measurement methods, which are not repeated here; the kinship weight is used to weight the sentence vector difference under the corresponding information indicator between the target medical record data and each reference medical record data, and the attention paid to the sentence vector difference under each information indicator is adjusted. The greater the kinship correlation, the greater the reference value of the sentence vector difference under the corresponding information indicator. Therefore, from the perspective of relative screening, the deviation between the target medical record data and the reference medical record data under each information indicator can be accurately evaluated, that is, in the corresponding diagnosis and treatment process of the target medical record data, its symptom characteristics deviate more from other similar patient groups.

[0055] Step S202, between the target medical record data and each non-reference medical record data, according to the difference between the sentence vectors under each information indicator, combined with the relative association of each information indicator in the target medical record data, obtain the universal difference parameter of each information indicator in the target medical record data.

[0056] It should be noted that the non-reference medical record data refers to all medical record data of all other patients except the target patient, except the reference medical record data.

[0057] Based on the same idea and method as that for obtaining group difference parameters, non-reference medical record data are used as comparison objects, and the cosine distance between the target medical record data and each non-reference medical record data is measured. The sentence vectors under each information indicator are used as the non-reference difference parameter under each information indicator, and then the weighted summation of kinship association is used to obtain the universal difference parameter. The specific steps are consistent with the method described in S201 and will not be repeated here.

[0058] Step S203, according to the group difference parameter and the universal difference parameter, combined with the relative correlation of each information indicator in the target medical record data, and the difference in sentence vectors under each information indicator between the target medical record data and the reference medical record data, obtain the abnormal attention index of each information indicator in the target medical record data.

[0059] Considering that the group difference parameter and the universal difference parameter respectively reflect the deviation difference between the target medical record data and the general population with dissimilar disease characteristics under each information indicator; if the group difference parameter is larger and the universal difference parameter is smaller, it means that the judgment of the disease characteristics of the corresponding medical treatment process recorded under this information indicator in the target medical record data is more affected by family inheritance, and relative screening should be considered, focusing on information indicators with greater relative correlation; conversely, if the group difference parameter is smaller and the universal difference parameter is larger, relative screening can be omitted, and the differences in the corresponding sentence vectors under each information indicator between different medical record data can be directly analyzed to evaluate the abnormal deviation of the sentence vector under each information indicator in the target medical record data, and then the abnormal attention index under each information indicator in the target medical record data can be obtained.

[0060] In a preferred embodiment of the present invention, whether relative screening is performed for the target patient is determined based on the size relationship between the group difference parameter and the universal difference parameter; if relative screening is performed, the abnormal attention index of each information indicator in the target medical record data is obtained based on the group difference parameter and the universal difference parameter, combined with the relative relevance of each information indicator in the target medical record data; if relative screening is not performed, the abnormal attention index of each information indicator in the target medical record data is obtained based on the difference in sentence vectors under each information indicator between the target medical record data and the reference medical record data.

[0061] As an example, when the group difference parameter is greater than the universal difference parameter, relative screening is performed; when the group difference parameter is less than or equal to the universal difference parameter, relative screening is not performed.

[0062] Among them, when conducting relative screening, the method for obtaining the abnormal concern index includes: Under each information indicator in the target medical record data, relative association is used as the weight of the group difference parameter, the difference between the constant 1 and the relative association is used as the weight of the universal difference parameter, and the weighted sum result is used as the abnormal attention index of each information indicator in the target medical record data.

[0063] The calculation formula of abnormal attention index is: ; in, The serial number of the target medical record data of the target patient; is the serial number of the information indicator; In the case of relative screening, the target patient's target medical record data Abnormal attention index under information indicators; The target patient's target medical record data Group difference parameters under information indicators; The target patient's target medical record data The universal difference parameter under the information index; The target patient's target medical record data Relative relatedness under information indicators.

[0064] In the above formula, the greater the relative correlation is, and the group difference parameter is greater than the universal difference parameter, more attention will be paid to the information indicators affected by genetics from the perspective of relative screening; on the contrary, if the relative correlation is smaller, it means that the corresponding information indicator is less affected by genetics, then the abnormal attention index will be more affected by the universal difference parameter.

[0065] Among them, when relative screening is not performed, the methods for obtaining the abnormal concern index include: Between the target medical record data and each reference medical record data, the Hamming distance between the sentence vectors of each information indicator is used as the first abnormal parameter, and the cosine distance between the sentence vectors of each information indicator is used as the second abnormal parameter; the product of the first abnormal parameter and the second abnormal parameter is used as the abnormal parameter of each information indicator between the target medical record data and each reference medical record data; the sum of the abnormal parameters of each information indicator between the target medical record data and all reference medical record data is normalized, and the normalized result is used as the abnormal attention index of each information indicator in the target medical record data.

[0066] The calculation formula of abnormal attention index is: ; in, The serial number of the target medical record data of the target patient; is the serial number of the information indicator; In the case of no relative screening, the target patient's target medical record data Abnormal attention index under information indicators; is an S-type function; The serial number of the reference medical record data of the target medical record data; The total number of reference medical record data for the target medical record data; The target medical record data Sentence vectors under information indicators; The target medical record data Reference medical records Sentence vectors under information indicators; The target medical record data The first abnormal parameter under the information index; The target medical record data The second abnormal parameter under the information index; The target medical record data Abnormal parameters under information indicators.

[0067] In the above calculation formula, the larger the cosine distance and the Hamming distance, the larger the product accumulation sum, which means that under this information indicator, the greater the difference between the record content of the target medical record data and the record content of all reference medical record data, and the more abnormal it is, the more attention it needs, and the larger the corresponding abnormal attention index; then the product accumulation sum is mapped to the S-type function and normalized to obtain the abnormal attention index. The larger the abnormal attention index, the more attention this information indicator should receive.

[0068] In other examples, the implementer may also combine the first anomaly parameter and the second anomaly parameter through basic mathematical operations such as addition or weighted summation, and may also use other vector difference measurement methods to obtain the first anomaly parameter or the second anomaly parameter. For example, the edit distance between the character strings corresponding to the sentence vectors under each information indicator is used as the first anomaly parameter or the second anomaly parameter. This is a well-known technology along with the cosine distance and the Hamming distance and will not be elaborated here.

[0069] After obtaining the abnormal attention index of each information indicator in the target medical record data, all the information indicators of interest in the target medical record data can be screened out from all the information indicators in the target medical record data according to the abnormal attention index.

[0070] Preferably, in one embodiment of the present invention, considering that the larger the abnormal attention index is, the more attention should be paid; therefore, the method for obtaining the attention information index includes: taking the information index greater than the preset threshold in the target medical record data as the attention information index. The preset threshold is set to 0.8, and the implementer can also set it according to the actual situation.

[0071] In one embodiment of the present invention, after screening out all the information indicators of interest in the target medical record data, the corresponding information indicators and their recorded data in the electronic medical record record table corresponding to the target medical record data can be highlighted and bolded, etc., to prompt relevant medical personnel to focus on the recorded data under the information indicator in the current electronic medical record table; by changing the target medical record data, all specially marked electronic medical record tables of all cardiac surgery patients in the electronic medical record system can be obtained, thereby helping to improve the analysis efficiency of relevant medical personnel on each electronic medical record table, thereby improving medical efficiency and patient satisfaction.

[0072] In summary, the present invention first obtains the sentence vector of the recorded data under each information indicator in each medical record data of all the patients who visited the hospital; further, in all the medical record data of each patient who visited the hospital, the change deviation index of each medical record data relative to all other medical record data is analyzed and obtained, so as to screen out all the reference medical record data of each medical record data; further, between each medical record data of each patient who visited the hospital and all the medical record data of the corresponding relative patients, the relative relevance of each information indicator in each medical record data is analyzed and obtained, and then, between the target medical record data and each reference medical record data, the difference between the sentence vectors of each information indicator is combined to obtain the abnormal attention index of each information indicator in the target medical record data; finally, all the concerned information indicators in each medical record data are screened out according to the abnormal attention index. The present invention analyzes and mines the change characteristics of each medical record data and the relative relationship between the patients who visited the hospital, so as to analyze the abnormal deviation of each information indicator in each medical record data, determine its abnormal attention index, thereby assisting the relevant medical personnel to analyze the information indicators in a focused manner, and then assisting in improving the analysis efficiency of each electronic medical record form.

[0073] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0074] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for organizing medical examination record information for cardiac surgery, characterized in that: The method comprises: Obtain the medical record data of all patients visiting the Department of Cardiac Surgery during each visit, wherein the medical record data includes the sentence vectors of the data recorded under each information indicator; take any visiting patient as the target patient, and take any medical record data of the target patient as the target medical record data; Among the different medical records of the target patient, according to the difference between the sentence vectors under each information indicator, obtain the change deviation index of the target medical record data relative to all other medical record data; according to the similarity between the change deviation index of the target medical record data and all the medical record data of all other patients, filter out all reference medical record data of the target medical record data from all the medical record data; Screen out all relatives of the target patient from all the patients who have visited the hospital; obtain the relative relevance of each information indicator in the target medical record data based on the difference between the sentence vectors under each information indicator between the target medical record data and all the medical record data of each relative of the target patient; Based on the difference between the sentence vectors of each information indicator between the target medical record data and each of the reference medical record data and combined with the relative association of each information indicator in the target medical record data, the abnormal attention index of each information indicator in the target medical record data is obtained; and all the information indicators of concern in the target medical record data are screened out according to the abnormal attention index.

2. A method for arranging medical examination records for cardiac surgery according to claim 1, characterized in that: The method for obtaining the change deviation index includes: Among all the medical records of the target patient, between the target medical record data and each of the remaining medical record data, according to the vector difference between the sentence vectors under each information indicator, obtaining a change parameter under each information indicator; the vector difference is positively correlated with the change parameter; Comprehensively calculate the change parameters under each information indicator between the target medical record data and all other medical record data to obtain the parameter deviation coefficient under each information indicator in the target medical record data; The parameter deviation coefficients under all information indicators in the comprehensive target medical record data are used to obtain the change deviation index of the target medical record data relative to all other medical record data.

3. A method for arranging medical examination records for cardiac surgery according to claim 1, characterized in that: The screening method of the reference medical record data comprises: Among all the medical record data of all other patients except the target patient, if the difference between the change deviation index of the medical record data and the change deviation index of the target medical record data is less than a preset threshold, the medical record data will be used as reference medical record data for the target medical record data.

4. A method for arranging medical examination records for cardiac surgery according to claim 1, characterized in that: The method for obtaining the relative association includes: Between the target medical record data and each of the medical record data of each relative of the target patient, according to the vector difference between the sentence vectors under each information indicator, obtaining a relative difference parameter under each information indicator; the vector difference is positively correlated with the relative difference parameter; The relative difference parameters under each information indicator are combined between the target medical record data and all the medical record data of each relative patient of the target patient to obtain the relative deviation coefficient under each information indicator in the target medical record data; the negative correlation normalization result of the relative deviation coefficient is used as the relative correlation under the corresponding information indicator in the target medical record data.

5. The method for organizing medical examination records for cardiac surgery according to claim 1, characterized in that: The method for obtaining the abnormal attention index includes: Between the target medical record data and each of the reference medical record data, according to the difference between the sentence vectors under each information indicator, combined with the relative association of each information indicator in the target medical record data, obtain a group difference parameter of each information indicator in the target medical record data; Between the target medical record data and each non-reference medical record data, according to the difference between the sentence vectors under each information indicator, combined with the relative association of each information indicator in the target medical record data, obtain a universal difference parameter for each information indicator in the target medical record data; According to the group difference parameter and the universal difference parameter, combined with the relative correlation of each information indicator in the target medical record data, and the difference in the sentence vector under each information indicator between the target medical record data and the reference medical record data, the abnormal attention index of each information indicator in the target medical record data is obtained.

6. A method for organizing medical examination records for cardiac surgery according to claim 5, characterized in that: The method for obtaining the population difference parameter includes: Between the target medical record data and each of the reference medical record data, taking the vector difference between the sentence vectors under each information indicator as a reference difference parameter under each information indicator; The reference difference parameters under the corresponding information index are weighted by using the kinship association, and the weighted sum of all the reference difference parameters under the corresponding information index between the target medical record data and all the reference medical record data is normalized, and the normalized result is used as the group difference parameter of the target medical record data under the corresponding information indicator.

7. A method for organizing medical examination records for cardiac surgery according to claim 5, characterized in that: The method of obtaining the abnormal attention index of each information indicator in the target medical record data according to the group difference parameter and the universal difference parameter, combined with the relative association of each information indicator in the target medical record data, and the difference of the sentence vector under each information indicator between the target medical record data and the reference medical record data includes: Determining whether to perform relative screening on the target patient according to the magnitude relationship between the group difference parameter and the universal difference parameter; If relative screening is performed, then the abnormal concern index of each information indicator in the target medical record data is obtained based on the group difference parameter and the universal difference parameter, combined with the relative relevance of each information indicator in the target medical record data; If relative screening is not performed, the abnormal attention index of each information indicator in the target medical record data is obtained according to the difference in the sentence vector under each information indicator between the target medical record data and the reference medical record data.

8. A method for arranging medical examination records for cardiac surgery according to claim 7, characterized in that: The method for obtaining the abnormal attention index of each information indicator in the target medical record data according to the group difference parameter and the universal difference parameter in combination with the relative association of each information indicator in the target medical record data includes: Under each information indicator in the target medical record data, the relative association is used as the weight of the group difference parameter, the difference between the constant 1 and the relative association is used as the weight of the universal difference parameter, and the weighted sum result is used as the abnormal attention index of each information indicator in the target medical record data.

9. A method for arranging medical examination records for cardiac surgery according to claim 7, characterized in that: The method for obtaining the abnormal attention index of each information indicator in the target medical record data according to the difference of the sentence vector under each information indicator between the target medical record data and the reference medical record data includes: Between the target medical record data and each of the reference medical record data, the Hamming distance between the sentence vectors under each information indicator is used as a first abnormality parameter, and the cosine distance between the sentence vectors under each information indicator is used as a second abnormality parameter; The product of the first abnormal parameter and the second abnormal parameter is used as the abnormal parameter of each information indicator between the target medical record data and each of the reference medical record data; the sum of the abnormal parameters of each information indicator between the target medical record data and all the reference medical record data is normalized, and the normalized result is used as the abnormal attention index of each information indicator in the target medical record data.

10. The method for organizing medical examination records for cardiac surgery according to claim 1, characterized in that: The method for obtaining the attention information indicator includes: The information indicators in the target medical record data that are greater than the preset threshold are taken as the information indicators of interest.

Images