A method for analyzing electronic medical record data of gastroenterology patients

Abstract

The present invention relates to the technical field of electronic medical record analysis, and specifically to a method for analyzing electronic medical record data of patients in the Department of Gastroenterology, including preprocessing the outpatient electronic medical record data of new patients in the Department of Gastroenterology to obtain a medical consultation word array set; obtaining a symptom matching cluster based on the screening of the medical consultation word array set in the historical data cluster set; obtaining a matching cluster based on the names of historical patients in the symptom matching cluster, respectively matching each medical consultation data type cluster; and obtaining recommended word segments based on the number of word repetitions in each matching cluster and the number of occurrences of the names of historical patients in the symptom matching cluster. In the present invention, when the outpatient electronic medical record data of the new patient is connected to the medical management system, the symptom matching cluster is confirmed in the electronic medical record data of the same category based on the outpatient electronic medical record data, and then, based on the symptom matching cluster, the recommended keywords are confirmed to improve the diagnosis and treatment efficiency and treatment accuracy of the physician.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic medical record analysis, and in particular to a method for analyzing electronic medical record data of gastroenterology patients. Background Art

[0002] In today's medical system, electronic medical record systems have become an indispensable part of medical institutions at all levels, especially large and medium-sized hospitals. For gastroenterologists, electronic medical record systems provide them with great convenience. Physicians can use this system to record patients' medical history in detail, issue medical orders, and query patients' examination results quickly and easily. These functions greatly improve the efficiency of diagnosis and treatment, allowing physicians to focus more on the treatment and care of patients.

[0003] With the continuous advancement of medical technology and the increasing demand for medical care, more and more patients choose to go to hospitals for treatment. In particular, the number of patients in the Department of Gastroenterology is also increasing. In order to better manage the health information of these patients, a large amount of data has been accumulated in the electronic medical record system. This data includes outpatient medical records, inpatient medical records, test reports, medication records and other types. This data includes both structured parts, such as the patient's vital signs such as body temperature and blood pressure, and a large amount of unstructured text data, such as detailed text information such as medical history, physical examination, diagnosis, and treatment written by doctors.

[0004] However, although there is a lot of valuable medical knowledge and information in these rich unstructured text data, there is no effective way to analyze and utilize them. These unstructured data account for a very high proportion, and they often contain important information such as the physician's detailed description and analysis of the patient's condition, as well as the patient's personalized treatment plan. However, there is no method to deeply mine and utilize the unstructured text data in the above electronic medical records. Summary of the invention

[0005] In order to solve the above technical problems, the purpose of the present invention is to provide a method for analyzing electronic medical record data of patients in gastroenterology to solve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0006] The present application provides a method for analyzing electronic medical record data of patients in the Department of Gastroenterology, the method comprising: obtaining outpatient electronic medical record data of new patients visiting the Department of Gastroenterology; preprocessing the outpatient electronic medical record data to obtain a visit segmentation word array set; filtering and obtaining a symptom matching cluster according to the similarity of the visit segmentation word array set in a historical data cluster set, the historical data cluster set comprising a plurality of visit type clusters, one of the visit type clusters being an outpatient type cluster, the other visit type clusters being visit data type clusters, each of the visit type clusters comprising a plurality of symptom clusters, the symptom matching cluster being a symptom cluster in the outpatient type cluster; obtaining a matching cluster based on the repetition matching of the names of historical patients visiting the symptom matching cluster in each of the visit data type clusters; filtering and obtaining a recommended segmentation based on the number of repetitions of each segmentation word in each matching cluster and the number of occurrences of the names of historical patients visiting the symptom matching cluster in the symptom matching cluster.

[0007] In a possible implementation, a symptom matching cluster is obtained by screening according to the similarity of the consultation word array set in the historical data cluster set, including: performing a target word segmentation conversion process on the consultation word array set to obtain an English abbreviation word segmentation array corresponding to the outpatient electronic medical record data, the target word segmentation conversion process is to convert the target word in the consultation column into an English abbreviation word segmentation array based on a preset abbreviation dictionary; based on the English abbreviation word array corresponding to the outpatient electronic medical record data and the consultation word array set, the similarity between the outpatient electronic medical record data and each symptom cluster is calculated; and based on multiple similarities, a symptom matching cluster is obtained.

[0008] In a possible implementation, the similarity between the outpatient electronic medical record data and each symptom cluster is calculated based on the English abbreviation segmentation array and the visit segmentation array set corresponding to the outpatient electronic medical record data, including: respectively calculating the similarity between the outpatient electronic medical record data and each electronic medical record data in the symptom cluster, each electronic medical record data in the symptom cluster corresponds to the outpatient electronic medical record data of a patient with a history of visiting the gastroenterology department, and the similarity is calculated from the English abbreviation segmentation array and the visit segmentation array set corresponding to new patients visiting the gastroenterology department and patients with a history of visiting the gastroenterology department; performing mean calculation based on all the similarities, and using the mean calculation result as the similarity.

[0009] In a possible implementation, recommended participles are obtained by screening based on the number of repetitions of each participle in each matching cluster and the number of occurrences of the names of patients who have visited the hospital in the disease matching cluster, including: respectively counting the total number of participles in the matching cluster and the total number of electronic medical record data in the matching cluster; obtaining corresponding electronic medical record data in the matching cluster based on the names of patients who have visited the hospital in the disease matching cluster, and counting the total number to obtain the total number of electronic medical records of matching patients; calculating the recommendation level of each participle based on the number of repetitions of each participle in the matching cluster, the total number of electronic medical records of matching patients, the total number of participles in the matching cluster, and the total number of electronic medical record data in the matching cluster; and obtaining recommended participles based on the recommendation level of each participle and a preset threshold.

[0010] In a possible implementation, the preset threshold is 0.6.

[0011] In one possible implementation, the method for acquiring historical data clusters includes: collecting electronic medical record data of all patients who have historically visited the Department of Gastroenterology; preprocessing each electronic medical record data to obtain a historical word segmentation array set corresponding to each electronic medical record data; and performing double clustering on all the historical word segmentation array sets to obtain historical data clusters.

[0012] In one possible implementation, all of the historical word segmentation array sets are double clustered to obtain historical data cluster sets, including: clustering the historical word segmentation array sets based on the titles of the electronic medical record data to obtain multiple consultation type clusters; and clustering each of the consultation type clusters based on the consultation column of the electronic medical record data to obtain multiple symptom clusters.

[0013] In a possible implementation, each of the visit type clusters is clustered based on the visit column of the electronic medical record data to obtain multiple symptom clusters, including: performing target segmentation conversion processing on each of the historical segmentation array sets one by one to obtain an English abbreviation segmentation array corresponding to each electronic medical record data, the target segmentation conversion processing is to convert the target segmentation in the visit column into an English abbreviation segmentation array based on a preset abbreviation dictionary; performing similarity calculation based on the English abbreviation segmentation array and the historical segmentation array set corresponding to every two electronic medical record data to obtain the similarity between every two electronic medical record data; clustering based on the similarity between the electronic medical record data to obtain multiple symptom clusters.

[0014] In a possible implementation, the clustering process is K-means clustering process.

[0015] In a possible implementation, the preprocessing includes word segmentation and stop word removal.

[0016] The present invention has the following beneficial effects:

[0017] The present invention incorporates the outpatient electronic medical record data of new patients visiting the gastroenterology department into the medical management system, calculates the similarity of the word segmentation phrases in the visit column of the electronic medical record data of the new patient and each disease cluster in the same category, so as to determine the disease matching cluster of the new patient, then performs a matching cluster analysis of the disease cluster matching under other visit type clusters among the patients in the disease matching cluster, and finally makes recommendations to physicians based on the electronic medical record text features of the matching cluster, so as to assist physicians in optimizing the diagnosis and treatment efficiency and treatment accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] 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.

[0019] Figure 1 A schematic diagram of a flow chart of a method for analyzing electronic medical record data of a gastroenterology patient provided in Embodiment 1 of the present invention;

[0020] Figure 2 A schematic diagram of the process of step 3 provided in embodiment 1 of the present invention;

[0021] Figure 3 A schematic diagram of a flow chart of a method for analyzing electronic medical record data of a gastroenterology patient provided in Embodiment 1 of the present invention;

[0022] Figure 4 A schematic flow chart of step S3 provided in Example 1 of the present invention;

[0023] Figure 5 This is a flowchart of step S5 provided in Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0024] 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 electronic medical record data analysis method for gastroenterology patients proposed by the present invention, its specific implementation method, structure, characteristics and effects, 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.

[0025] 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.

[0026] Example 1:

[0027] The following specifically describes the specific solution of an electronic medical record data analysis method for patients in the department of gastroenterology provided by the present invention in conjunction with the accompanying drawings.

[0028] Please refer to Figure 1 , which shows a schematic flowchart of an electronic medical record data analysis method for patients in the department of gastroenterology provided by an embodiment of the present invention. Specifically, in this embodiment, a total of steps Step1 - Step3 are included.

[0029] Step1: Collect the electronic medical record data of all historical patients in the department of gastroenterology.

[0030] First of all, it should be noted that in this embodiment, the electronic medical record data of all historical patients in the department of gastroenterology can be obtained from the hospital's background management system. Specifically, these electronic medical record data can include various unstructured data categories such as outpatient medical records, inpatient medical records, and medication records. These unstructured data categories are not restricted in this embodiment.

[0031] Among them, these electronic medical record data cover the detailed information of patients at different stages of medical treatment. For example, outpatient medical records record the chief complaint, current medical history, physical examination results, diagnosis conclusion, and treatment measures of patients during outpatient visits; inpatient medical records record the detailed diagnosis and treatment process of patients during hospitalization, including admission records, progress records, surgical records, discharge records, etc.; medication records detail the medication conditions of patients during outpatient and inpatient treatments, including drug names, dosages, usage methods, and treatment courses.

[0032] At the same time, it should be noted that each patient may have multiple historical electronic medical record data. This means that the same patient may have multiple medical treatment records at different time periods, including outpatient medical records, inpatient medical records, and medication records. These historical electronic medical record data can all help physicians assist in judging the conditions of new patients in the department of gastroenterology in the future.

[0033] Step2: Preprocess each of the electronic medical record data to obtain a corresponding historical word segmentation array set for each of the electronic medical record data.

[0034] It should be noted that the preprocessing mentioned in this embodiment is to use the Jieba word segmentation tool to perform word segmentation on the text and remove stop words. The Jieba word segmentation tool is a commonly used Chinese word segmentation tool that can split continuous Chinese text into independent words, facilitating subsequent text analysis and processing. Removing stop words means that after word segmentation, some commonly used words that have no practical meaning for text analysis (such as "de", "le", "zai", etc.) are removed from the text to improve the accuracy and efficiency of analysis.

[0035] Furthermore, in this embodiment, the object that needs to be preprocessed is the consultation column in the electronic medical record data. Specifically, the consultation column includes but is not limited to the following contents: chief complaint, description of current medical history, past medical history record, physical examination results, outpatient diagnosis results, and outpatient treatment measures. The medical record title and patient name do not need to be preprocessed.

[0036] In order to facilitate the subsequent electronic medical record data analysis, the consultation columns that must be processed include the main complaint, current medical history, outpatient diagnosis results, and outpatient treatment measures. These columns contain the patient's main symptoms, disease progression, physician's diagnosis, and treatment measures taken, and are the key parts of electronic medical record data analysis.

[0037] As for other description columns, such as past medical history records and physical examination results, technicians in this field can selectively perform pre-processing according to actual conditions and specific needs, and this embodiment does not make specific restrictions on this.

[0038] Step 3: Perform double clustering on all the historical word segmentation array sets to obtain historical data clusters.

[0039] To clarify the specific process of dual clustering, see Figure 2 , the figure shows that this step includes steps Step31 and Step32.

[0040] Step 31, clustering the historical word segmentation array set based on the title of the electronic medical record data to obtain multiple medical consultation type clusters.

[0041] It should be noted that, since the title of each electronic medical record data usually indicates the category of the electronic medical record data. Therefore, it is first necessary to classify the electronic medical record data of the same category together, which is more conducive to subsequent accurate analysis. Regarding the classification method, technicians in this field can directly use the title of each electronic medical record data for classification. For example, the electronic medical record data can be divided into categories such as outpatient medical records, inpatient medical records, and medication records. In addition, technicians in this field can also use the Jieba word segmentation tool to segment the title of the electronic medical record data to obtain a word segmentation array. Next, the Jaccard method can be used to calculate the similarity of different title word segmentation arrays. After obtaining the similarity measure, the K-means clustering algorithm can be used to cluster the electronic medical record data of all patients with a history of gastroenterology, where the number of clusters can be determined by the silhouette coefficient method. The silhouette coefficient is an indicator for evaluating the clustering effect. The quality of clustering is measured by calculating the average distance between each data point and other data points in its cluster and the average distance between each data point and the nearest cluster. Through the above process, the distance between electronic medical record data can be quantified by the similarity between the title word segmentation arrays. Assume that there are N clusters after clustering, and the electronic medical records in each cluster are the same type of electronic medical record data. In this way, the same type of electronic medical record data can be clustered together, so as to better perform subsequent analysis and processing. Among them, the Jaccard method is a commonly used similarity measurement method, which measures the similarity of two sets by calculating the ratio of the intersection and the union of two sets. K-means is a commonly used clustering algorithm, which maximizes the similarity between data points in the same cluster by dividing data points into K clusters. The above implementation methods are all prior art and will not be repeated in this embodiment.

[0042] Step 32: Based on the consultation column of the electronic medical record data, each consultation type cluster is clustered to obtain multiple disease clusters.

[0043] In the current step of this embodiment, the same calculation method as step Step 31 can be used to perform clustering processing. However, in the actual operation process, it must be taken into account that when physicians write electronic medical records for patients, due to the physician's writing habits and preferences, they often use some abbreviations or abbreviations to represent certain drug names or diseases. For example, Proton Pump Inhibitors may be abbreviated as PPI, and Gastroesophageal Reflux Disease may be abbreviated as GERD. Such abbreviations are very common in electronic medical record data, so when performing calculations, these abbreviations need to be properly processed and converted to ensure the accuracy and reliability of the calculation results. Only in this way can we ensure that the results of the clustering processing can truly reflect the similarities between electronic medical record data, thereby providing strong support for subsequent decision-making.

[0044] Therefore, in this embodiment, step Step32 also includes steps Step321 to Step323.

[0045] Step 321, perform target segmentation conversion processing on each of the historical segmentation array sets one by one to obtain an English abbreviation segmentation array, wherein the target segmentation conversion processing is to convert the target segmentation in the consultation column into an English abbreviation segmentation array based on a preset abbreviation dictionary.

[0046] In this embodiment, in order to effectively deal with various abbreviations, an abbreviation dictionary can be pre-constructed. The purpose of the abbreviation dictionary is to correspond and link common English abbreviations with their respective complete forms. Through this method, it can be ensured that before the cluster analysis of the consultation type is performed, the English abbreviation segmentation extraction can be performed on the segmentation array set of each patient with a history of visiting the gastroenterology department. If the corresponding English abbreviation is found in the abbreviation dictionary, it is extracted. Conversely, if the corresponding English abbreviation is not found in the abbreviation dictionary, the segmentation will be ignored and no further processing will be performed. This processing method ensures that various abbreviations can be accurately processed, thereby improving the accuracy of the entire cluster analysis.

[0047] Step 322: perform similarity calculation based on the English abbreviation word segmentation array and the historical word segmentation array set corresponding to each two electronic medical record data to obtain the similarity between each two electronic medical record data.

[0048] Specifically, in this step, the similarity calculation formula between each two electronic medical record data is as follows:

[0049]

[0050] In the formula, Indicates Patients with a history of visiting the Department of Gastroenterology and The similarity between the electronic medical record data of patients with a history of gastroenterology visits; Indicates The word segmentation array of the consultation column of the patients with a history of consultation in the Department of Gastroenterology; Indicates The word segmentation array of the consultation column of the patients with a history of consultation in the Department of Gastroenterology; Indicates The English abbreviations of all the words in the word array of the consultation column of the patient with a history of visiting the Department of Gastroenterology are in the first The number of times a patient with a history of visiting the Department of Gastroenterology appears in the word segmentation array of the consultation column; Represents the number of word segmentation arrays in the target intersection, the target intersection is The word segmentation array of the consultation column of the patient with a history of visiting the Department of Gastroenterology and the The intersection of the word segmentation arrays of the consultation columns of the patients who have visited the Department of Gastroenterology; Represents the number of word segmentation arrays in the target union, where the target union is the The word segmentation array of the consultation column of the patient with a history of visiting the Department of Gastroenterology and the The union of the word segmentation arrays of the consultation columns of the patients with a history of visiting the Department of Gastroenterology.

[0051] In the above calculation formula, The larger the value, the Gastroenterology patients and The higher the similarity between the word arrays of the consultation column of the gastroenterology patients, the higher the similarity. The larger the value of , the higher the similarity between the arrays formed by the two patients' medical records after word segmentation. This increase in similarity can reflect the similarity between their symptom descriptions, medical history records, and other related medical information. Therefore, by calculating The value of can effectively evaluate and compare the similarity of medical records between patients in different gastroenterology departments, thus providing more valuable reference for clinical diagnosis and treatment.

[0052] Step 323, clustering is performed based on the similarities between the electronic medical record data to obtain multiple disease clusters.

[0053] It should be noted that the clustering mentioned in this step is based on the K-means algorithm using The distance between two electronic medical record data is clustered. The number of clusters in this step is determined by the silhouette coefficient method, which is a prior art and will not be described in detail in this embodiment. Through the above steps, multiple visit type clusters are obtained by double clustering in this embodiment; and multiple disease clusters under each visit type cluster.

[0054] In this embodiment, the electronic medical record data of historical gastroenterology patients are first divided into categories by calculating the titles. Subsequently, the electronic medical records in the same category are clustered by calculating the similarity of the word segmentation arrays in the visit column between different types of electronic medical record data. Through this dual clustering method, the data can be effectively classified and a strong data foundation support can be provided for the subsequent medical treatment of new patients. It should be noted that the first heavy clustering is to cluster the historical word segmentation array set based on the title of the electronic medical record data to obtain multiple visit type clusters, and the second heavy clustering is to cluster the visit column of the electronic medical record data in each visit type cluster to obtain multiple symptom clusters.

[0055] Embodiment 2:

[0056] The specific scheme of the method for analyzing electronic medical record data of gastroenterology patients provided by the present invention is described in detail below with reference to the accompanying drawings.

[0057] See also Figure 3 , which shows a flow chart of a method for analyzing electronic medical record data of gastroenterology patients provided by an embodiment of the present invention. Specifically, in this embodiment, a total of steps S1 to S5 are included.

[0058] S1. Obtain the outpatient electronic medical record data of new patients visiting the Department of Gastroenterology.

[0059] In this embodiment, the method for obtaining the outpatient electronic medical record data of new patients visiting the Department of Gastroenterology can be directly extracted from the hospital's backend management system. Specifically, these outpatient electronic medical record data cover multiple aspects such as medical record title, patient name, chief complaint content, current medical history description, past medical history record, physical examination results, outpatient diagnosis results, and outpatient treatment measures. The above-mentioned outpatient data is the data generated by new patients visiting the Department of Gastroenterology who register and then conduct related diagnosis and treatment activities, such as consultation, examination, and testing. It should be noted that in the implementation of this embodiment in different hospitals, due to the differences in outpatient medical record templates, the items included may be different. Therefore, in this embodiment, this embodiment does not limit the content included in the outpatient electronic medical record data.

[0060] In this embodiment, the acquisition method of the outpatient electronic medical record data of new patients in the gastroenterology department can be directly extracted from the hospital's backend management system. Specifically, these outpatient electronic medical record data cover multiple aspects of information, including medical record title, patient name, chief complaint, current medical history description, past medical history record, physical examination results, outpatient diagnosis results, and outpatient treatment measures.

[0061] These outpatient data are generated by a series of medical activities after new patients in the Department of Gastroenterology register. These medical activities include, but are not limited to, medical history taking, examinations, tests, etc. It should be noted that during the implementation of this embodiment in different hospitals, due to differences in outpatient medical record templates, the included items may vary. Therefore, in this embodiment, no specific restrictions are imposed on the specific content of outpatient electronic medical record data. Those skilled in the art can flexibly adjust and expand the content of outpatient electronic medical record data according to their actual situations and needs.

[0062] S2. Preprocess the outpatient electronic medical record data to obtain a set of visit word segmentation arrays.

[0063] Specifically, in this embodiment, the preprocessing step includes using the Jieba word segmentation tool to perform word segmentation on the text and removing stop words. The Jieba word segmentation tool is a commonly used Chinese word segmentation tool that can split continuous Chinese text into independent words, facilitating subsequent text analysis and processing. For example, a description of the present illness: Stomachache at midnight yesterday. Ate more hawthorns yesterday afternoon. Persistent pain, took 250 ml of Coke orally, and now has paroxysmal pain. Can eat normally. Gastroscopy last year showed superficial gastritis. CT showed no abnormalities. Blood test was basically normal. The word segmentation array after processing with the Jieba word segmentation tool is [yesterday, midnight, stomachache, the day before yesterday, afternoon, ate, more, hawthorns, persistent, pain, took, Coke, 250 ml, now, paroxysmal, pain, normal, diet, last year, gastroscopy, showed, superficial, gastritis, CT, showed, no, abnormalities, blood test, basically, normal]. Removing stop words means removing some commonly used words (such as "de", "le", "zai", etc.) that have no practical significance for text analysis from the text after word segmentation to improve the accuracy and efficiency of analysis.

[0064] Further, in this embodiment, the object to be preprocessed is the visit column in the electronic medical record. Specifically, the visit column includes, but is not limited to, the following contents: chief complaint content, description of the present illness, record of past medical history, physical examination results, outpatient diagnosis results, and outpatient treatment measures. Among these columns, the medical record title and patient name do not need to be preprocessed.

[0065] For the convenience of subsequent electronic medical record data analysis, the visit columns that must be processed include the chief complaint content, description of the present illness, outpatient diagnosis results, and outpatient treatment measures. These columns contain the patient's main symptoms, the development of the disease, the doctor's diagnosis conclusion, and the treatment measures taken, which are the key parts of electronic medical record analysis.

[0066] As for other description columns, such as the record of past medical history and physical examination results, etc., those skilled in the art can selectively perform preprocessing according to the actual situation and specific needs, and no specific restrictions are made in this embodiment.

[0067] S3. A symptom matching cluster is obtained by screening according to the similarity of the medical consultation word set in the historical data cluster set, wherein the historical data cluster set includes multiple medical consultation type clusters, one of the medical consultation type clusters is an outpatient type cluster, and the other medical consultation type clusters are medical consultation data type clusters, each of the medical consultation type clusters includes multiple symptom clusters, and the symptom matching cluster is a symptom cluster within the outpatient type cluster.

[0068] For those skilled in the art, how to select the disease matching clusters according to the similarity can be obtained by referring to calculating the distance between the medical consultation word set and each disease cluster in the outpatient type cluster. At the same time, a step similar to Step 32 in this embodiment 1 can also be used. Specifically, in this embodiment, see Figure 4 , the figure shows that this step also includes steps S31 to S33.

[0069] S31, performing a target segmentation conversion process on the consultation segmentation array set to obtain an English abbreviation segmentation array corresponding to the outpatient electronic medical record data, wherein the target segmentation conversion process is to convert the target segmentation in the consultation column into an English abbreviation segmentation based on a preset abbreviation dictionary.

[0070] It should be noted that this step has the same logic as step 321 in Example 1, so it will not be described in detail.

[0071] S32. Calculate the similarity between the outpatient electronic medical record data and each symptom cluster based on the English abbreviation word segmentation array and the consultation word segmentation array set corresponding to the outpatient electronic medical record data.

[0072] In order to clarify the similarity between the outpatient electronic medical record data and each disease cluster, the specific calculation process can be referred to step S321 and step S322.

[0073] S321. Calculate the similarity between the outpatient electronic medical record data and each electronic medical record data in the disease cluster respectively. Each electronic medical record data in the disease cluster corresponds to the outpatient electronic medical record data of a patient with a history of visiting the gastroenterology department. The similarity is calculated from the English abbreviation segmentation array and the medical consultation segmentation array set corresponding to the new patients visiting the gastroenterology department and the patients with a history of visiting the gastroenterology department.

[0074] The similarity calculation method proposed in this step is the same as that in step 322 of embodiment 1. Therefore, it will not be described in detail.

[0075] S322: Calculate the mean value of all the similarities, and use the mean value calculation result as the similarity.

[0076] S33, obtaining a disease matching cluster based on the multiple similarities.

[0077] Specifically, in this step, the one with the greatest similarity to the symptom cluster within the clinic type cluster is selected as the symptom matching cluster.

[0078] Assuming that through the above steps, in this embodiment, it is assumed that there are S electronic medical record data in the disease matching cluster. Since the first outpatient medical record is generated when the internal medicine patient visits the hospital, followed by the medication record, hospitalization medical record, etc. Therefore, the subsequent medication records, hospitalization medical records and other data of the patients corresponding to the S electronic medical record data in the disease matching cluster have important reference significance for various data in the subsequent treatment process of new patients visiting the gastroenterology department. The recommended medical record data of new patients can be generated by analyzing the subsequent medication records, hospitalization medical records and other data of the patients corresponding to the S electronic medical record data in these disease matching clusters, thereby providing guidance for the doctor's subsequent treatment and medical record generation, optimizing the doctor's diagnosis and treatment efficiency and improving the accuracy of treatment. For details, please refer to step S4 and step S5.

[0079] S4. Based on the names of patients with historical medical consultations in the symptom matching cluster, the names are matched in each of the medical consultation data type clusters to obtain a matching cluster.

[0080] That is, in this step, the names of the patients who have visited the hospital in the symptom matching cluster are matched in each of the medical data type clusters to obtain a matching cluster with the most names of the patients who have visited the hospital in the symptom matching cluster. Specifically, the patient matching degree calculation formula can be used for evaluation:

[0081]

[0082] In the formula, Indicates The first The degree of patient fit between the disease cluster and the disease matching cluster; It indicates the number of patients with a history of visiting the Department of Gastroenterology in the disease matching cluster, which can also be called the total number of electronic medical record data in the disease matching cluster; It indicates the number of patients with a history of visiting the Department of Gastroenterology in the jth symptom cluster in the mth visit data type cluster and the number of patients with a history of visiting the Department of Gastroenterology in the Sth symptom matching cluster who have the same name.

[0083] In the above calculation formula, The larger the value, the greater the In the cluster of visit data types, The patient fit between the first disease cluster and the disease matching cluster is higher. Specifically, this means that in this disease cluster, the patient's symptoms, medical history, and other relevant data are more consistent with the characteristics of the disease matching cluster, reflecting a higher fit. The clusters of medical visit data types are further analyzed, and the disease cluster with the highest degree of fit with the disease matching cluster patients is selected and marked as a matching cluster. Based on this series of steps, this embodiment first searches for the disease matching cluster with the highest similarity to the outpatient data in the historical data cluster set. Through this disease matching cluster, those matching patients who show relevant disease characteristics in the subsequent electronic medical record data are further located. This method not only improves the accuracy of diagnosis, but also helps doctors better understand the patient's condition, thereby formulating a more personalized treatment plan.

[0084] S5. Based on the number of repetitions of each segmentation in each matching cluster and the number of occurrences of the names of historical patients in the disease matching cluster, the recommended segmentations are screened.

[0085] For the steps of screening recommended word segmentation, please refer to Figure 5 , the figure shows that step S5 includes steps S51 to S54 in total.

[0086] S51. Count the total number of word segments in the matching cluster and the total number of electronic medical record data in the matching cluster.

[0087] S52, based on the names of the patients who have visited the hospital in the disease matching cluster, the corresponding electronic medical record data is matched in the matching cluster, and the total number of electronic medical records of the matching patients is obtained by counting the total number.

[0088] It should be noted that the matching patients mentioned in this step are patients with a history of visiting the gastroenterology department who appear in both the disease matching cluster and the matching cluster. Correspondingly, the non-matching patients in the matching cluster are patients with a history of visiting the gastroenterology department who only appear in the matching cluster but not in the disease matching cluster.

[0089] S53. Calculate the recommendation level of each segmentation based on the number of repetitions of each segmentation in the matching cluster, the total number of electronic medical records of the matching patient, the total number of segmentations in the matching cluster, and the total number of electronic medical record data in the matching cluster.

[0090] The calculation formula for the recommendation level is as follows:

[0091]

[0092] In the formula, Indicates The first cluster of the medical visit data type clusters The recommendation level of each participle; Indicates The total number of segmentations in the clusters that match the visit data type clusters; Indicates that the kth word is among all The number of repetitions in a participle; Indicates The total number of electronic medical records of patients in the clusters that match the visit data type cluster; Indicates The number of times a word appears in the electronic medical record data corresponding to the patient.

[0093] In the above calculation formula, considering that the electronic medical record data of the matching cluster contains both the electronic medical record data of matching patients and the electronic medical record data of non-matching patients, and the electronic medical record data of matching patients has greater reference value, we use As the weight value, The larger the value of , the greater the recommendation degree of the k-th word.

[0094] S54: Obtain recommended word segments based on the recommendation level of each word segment and a preset threshold.

[0095] It should be noted that the preset threshold in this embodiment is 0.6. Through the above embodiment, the word segmentation greater than the preset threshold can be found in each matching cluster to recommend the physician, thereby providing experience support when the physician formulates a corresponding medical plan for a new patient.

[0096] In this embodiment, when the outpatient electronic medical record data of new patients in the Department of Gastroenterology is incorporated into the medical management system, the similarity of the word segmentation phrases in the medical record data of the new patient and each disease cluster in the same category is calculated to determine the disease matching cluster of the new patient. Subsequently, by performing a matching cluster analysis of the disease cluster matching under other medical type clusters among the patients in the disease matching cluster, finally, recommendations are made to physicians based on the electronic medical record text features of the matching clusters to assist physicians in optimizing diagnosis and treatment efficiency and treatment accuracy.

[0097] 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.

[0098] 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 analyzing electronic medical record data of gastroenterology patients, characterized in that: The method comprises: Obtain outpatient electronic medical record data of new patients visiting the gastroenterology department; Preprocessing the outpatient electronic medical record data to obtain a medical consultation word segmentation array set; A symptom matching cluster is obtained by screening according to the similarity of the medical consultation word set in the historical data cluster set, wherein the historical data cluster set includes a plurality of medical consultation type clusters, one of which is an outpatient type cluster, and the other medical consultation type clusters are medical consultation data type clusters, each of which includes a plurality of symptom clusters, and the symptom matching cluster is a symptom cluster in the outpatient type cluster; Based on the names of patients with historical medical consultations in the symptom matching cluster, the names are matched in each of the medical consultation data type clusters to obtain a matching cluster; The recommended segmentation is obtained based on the number of repetitions of each segmentation in each matching cluster and the number of occurrences of the names of historical patients in the disease matching cluster; The symptom matching clusters are obtained by screening according to the similarity of the medical consultation word set in the historical data cluster set, including: Performing a target segmentation conversion process on the consultation segmentation array set to obtain an English abbreviation segmentation array corresponding to the outpatient electronic medical record data, wherein the target segmentation conversion process is to convert the target segmentation in the consultation column into an English abbreviation segmentation based on a preset abbreviation dictionary; Based on the English abbreviation segmentation array and the consultation segmentation array set corresponding to the outpatient electronic medical record data, the similarity between the outpatient electronic medical record data and each disease cluster is calculated; A disease matching cluster is obtained based on a plurality of similarity judgments.

2. The method for analyzing electronic medical records of gastroenterology patients according to claim 1, characterized in that: Based on the English abbreviation segmentation array corresponding to the outpatient electronic medical record data, the similarity between the outpatient electronic medical record data and each disease cluster is calculated, including: The similarity between the outpatient electronic medical record data and each electronic medical record data in the disease cluster is calculated respectively. Each electronic medical record data in the disease cluster corresponds to the outpatient electronic medical record data of a patient with a history of visiting the gastroenterology department. The similarity is calculated from the English abbreviation segmentation array and the visit segmentation array set corresponding to the new patients visiting the gastroenterology department and the patients with a history of visiting the gastroenterology department. A mean calculation is performed based on all the similarities, and the mean calculation result is used as the similarity.

3. The method for analyzing electronic medical records of gastroenterology patients according to claim 1, characterized in that: Recommended segmentations are obtained based on the number of repetitions of each segmentation in each matching cluster and the number of occurrences of the names of historical patients in the disease matching cluster, including: Counting the total number of segmented words in the matching cluster and the total number of electronic medical record data in the matching cluster respectively; Based on the names of the patients who have visited the hospital in the disease matching cluster, the corresponding electronic medical record data is matched in the matching cluster, and the total number of electronic medical records of the matching patients is obtained by counting the total number; The recommendation degree of each segmentation is calculated based on the number of repetitions of each segmentation in the matching cluster, the total number of electronic medical records of the matching patient, the total number of segmentations in the matching cluster, and the total number of electronic medical record data in the matching cluster; The recommended word segmentation is obtained based on the recommendation degree of each word segmentation and the preset threshold.

4. The method for analyzing electronic medical records of gastroenterology patients according to claim 3, characterized in that: The preset threshold is 0.

6.

5. The method for analyzing electronic medical records of gastroenterology patients according to claim 1, characterized in that: The methods for obtaining historical data clusters include: Collect electronic medical record data of all patients with a history of visiting the gastroenterology department; Preprocessing each electronic medical record data to obtain a historical word segmentation array set corresponding to each electronic medical record data; Double clustering is performed on all the historical word segmentation array sets to obtain historical data clusters.

6. The method for analyzing electronic medical records of gastroenterology patients according to claim 5, characterized in that: Double clustering is performed on all the historical word segmentation array sets to obtain historical data cluster sets, including: Clustering the historical word segmentation array set based on the title of the electronic medical record data to obtain a plurality of consultation type clusters; Based on the consultation column of the electronic medical record data, each consultation type cluster is clustered to obtain a plurality of symptom clusters.

7. The method for analyzing electronic medical records of gastroenterology patients according to claim 6, characterized in that: Based on the consultation column of the electronic medical record data, each consultation type cluster is clustered to obtain multiple disease clusters, including: Performing target segmentation conversion processing on each of the historical segmentation array sets one by one to obtain an English abbreviation segmentation array corresponding to each of the electronic medical record data, wherein the target segmentation conversion processing is to convert the target segmentation in the consultation column into an English abbreviation segmentation based on a preset abbreviation dictionary; Based on the English abbreviation word segmentation array and the historical word segmentation array set corresponding to each two electronic medical record data, similarity calculation is performed to obtain the similarity between each two electronic medical record data; Clustering is performed based on the similarities between the electronic medical record data to obtain multiple disease clusters.

8. The method for analyzing electronic medical records of gastroenterology patients according to claim 7, characterized in that: The clustering process is K-means clustering process.

9. The method for analyzing electronic medical records of gastroenterology patients according to claim 1, characterized in that: The preprocessing includes word segmentation and stop word removal.

Images