Time sequence analysis method for clinical medical information data based on AI analysis

Abstract

The invention discloses a clinical medical information data time sequence analysis method based on AI analysis, and relates to the technical field of clinical data analysis. The method comprises the following steps: performing first time sequence analysis through unified effective data of various characteristics of the hospital electronic medical record; performing first threshold judgment analysis on the clinical medical information data time sequence analysis fluctuation indication value; performing second time sequence analysis through unified effective data of various characteristics of the hospital electronic medical record; performing second threshold judgment analysis on the clinical medical information data time sequence analysis fluctuation comprehensive value; carrying out data correction and notifying related personnel; and ending the time sequence analysis and storing the data. According to the method, the effect of improving the accuracy of the standardized data after the clinical medical information data is standardized is achieved by collecting and standardizing the data, analyzing and evaluating the data exception through the time sequence, and correcting and storing the data, and the problem that the accuracy of the standardized data exception detection after the clinical medical information data is standardized is insufficient in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of clinical data analysis, and in particular to a time series analysis method for clinical medical information data based on AI analysis. Background Art

[0002] With the rapid development of information technology, medical informatization has become the current trend. The popularization of electronic medical records and electronic health records has provided a basis for the analysis and utilization of medical data. At the same time, the explosive growth of medical data and different data exchange standards have led to difficulties in data interaction, making traditional data analysis methods difficult to cope with. New technical means are urgently needed to deal with it.

[0003] Existing time series analysis methods for clinical medical information data are implemented through the following technologies, including: machine learning algorithms, natural language processing, and data mining technology.

[0004] For example, the invention patent publication number CN110335647B discloses a clinical data standardization system and standardized data collection method, including a standardized data collection platform, a data auditing platform, and a business system. The standardized data collection platform includes an admission and exclusion criteria configuration module, a follow-up plan configuration module, a scientific research metadata management module, a scientific research case management module, a standardized maintenance module, a collection module, a service interface module, and an eCRF management module. The data collection method involves a doctor's workstation generating a business application form and business system data based on the patient's follow-up information. The data is then cleaned and converted in a standardized pool, and after standardization, is automatically populated into the disease database of the eCRF management module.

[0005] For example, the invention patent announcement with announcement number: CN112786126B discloses a method, device, electronic device, and medium for analyzing clinical trial data in a time series, including: the method for analyzing clinical trial data in a time series includes: extracting stage information of different medical stages in a case report form, constructing a medical data dictionary based on the stage information, and the case report form is used to store the clinical trial data; determining a medical stage to be bound that matches a target case report form based on different analysis dimensions; binding the medical data dictionary to the target case report form based on the medical stage to be bound, and configuring the binding result into a multidimensional information table based on time series.

[0006] However, in the process of implementing the technical solutions of the invention in the embodiments of the present application, the present application found that the above technology has at least the following technical problems: In the existing technology, clinical medical information data is difficult to standardize, lacks interoperability, and standardization is cumbersome. In particular, for clinical medical information data from electronic clinical medical systems across different hospitals for long-term time series analysis, data standardization is very difficult after the clinical medical information data are mapped to each other. Mapping errors are prone to occur in the early stages of data standardization, and the error rate is high, which poses a hidden danger to the subsequent unified processing of clinical medical information data after standardization. There is also a problem of insufficient accuracy in detecting abnormalities in standardized data after clinical medical information data is standardized. Summary of the Invention

[0007] The embodiments of the present application provide a time series analysis method for clinical medical information data based on AI analysis, thereby solving the problem in the prior art of insufficient accuracy in detecting abnormalities in standardized data after the standardization of clinical medical information data, and achieving the effect of improving the accuracy of standardized data after the standardization of clinical medical information data.

[0008] The embodiment of the present application provides a time series analysis method for clinical medical information data based on AI analysis, comprising the following steps: S1. A central server collects electronic medical record data from different hospitals, extracts clinical medical information data from different hospitals, converts the data into a unified format, and obtains unified electronic medical record data of the hospitals; S2. Data preprocessing is performed on the unified electronic medical record data of the hospitals to obtain unified valid data of various characteristics of the electronic medical records of the hospitals; S3. A first time series analysis is performed on the unified valid data of various characteristics of the electronic medical records of the hospitals to obtain a fluctuation indicator value of the time series analysis of the clinical medical information data; S4. A first threshold judgment analysis is performed on the fluctuation indicator value of the time series analysis of the clinical medical information data. Determine whether to conduct a second time series analysis based on the analysis result of the first threshold value of the fluctuation indicator value of the time series analysis of clinical medical information data. If yes, proceed to S5, otherwise proceed to S8; S5. Conduct a second time series analysis on the unified valid data of various characteristics of the hospital electronic medical record to obtain a comprehensive value of fluctuations in the time series analysis of clinical medical information data; S6. Conduct a second threshold judgment analysis on the comprehensive value of fluctuations in the time series analysis of clinical medical information data, and determine whether to perform data correction based on the analysis result of the second threshold value of the comprehensive value of fluctuations in the time series analysis of clinical medical information data. If yes, proceed to S7, otherwise proceed to S8; S7. Perform data correction and notify relevant personnel; S8. End the time series analysis and store the data.

[0009] Furthermore, clinical medical information data of different hospitals are extracted therefrom, specifically including: the central server extracts data from different hospital servers through the ETL tool via the data connection interface to obtain the original electronic medical record data of different hospitals; the original electronic medical record data of different hospitals retain the timestamps corresponding to the original electronic medical record data; the central server converts the original electronic medical record data of different hospitals into unified electronic medical record data of hospitals through a predefined data mapping model; the central server classifies the unified electronic medical record data of hospitals into unified electronic medical record data of hospitals of different users through predefined patient master index classification rules; the central server uses different users as data index rules and uses different users as data classification partitions to store unified electronic medical record data of hospitals of different users.

[0010] Furthermore, data preprocessing is performed on the unified hospital electronic medical record data, specifically including: classifying the unified hospital electronic medical record data of the same user into unified hospital electronic medical record data of different characteristics of the same user by predefined clinical medical information features, and recording them as various types of characteristic hospital electronic medical record unified data; performing data anomaly detection on the various types of characteristic hospital electronic medical record unified data by using the Z-score method, and comparing and analyzing the Z-score values ​​of the various types of characteristic hospital electronic medical record unified data with the predefined Z-score detection threshold value; if the Z-score values ​​of the various types of characteristic hospital electronic medical record unified data are greater than the predefined Z-score detection threshold value, then the corresponding various types of characteristic hospital electronic medical record unified data are judged as abnormal values ​​to be detected; The various characteristic unified data of hospital electronic medical records judged as abnormal values ​​to be detected are compared and analyzed with the original data of electronic medical records in the corresponding hospital server: if the various characteristic unified data of hospital electronic medical records judged as abnormal values ​​to be detected are the same as the original data of electronic medical records in the corresponding hospital server, they are judged as non-abnormal values ​​and no longer perform data abnormality detection, and the various characteristic unified data of hospital electronic medical records of non-abnormal values ​​are recorded as the various characteristic unified valid data of hospital electronic medical records; if the various characteristic unified data of hospital electronic medical records judged as abnormal values ​​to be detected are different from the original data of electronic medical records in the corresponding hospital server, they are judged as abnormal values, and the various characteristic unified data of hospital electronic medical records of abnormal values ​​to be detected are replaced with the original data of electronic medical records in the corresponding hospital server and data preprocessing is performed again.

[0011] Furthermore, the specific process of obtaining the fluctuation indicator value of the time series analysis of clinical medical information data is as follows: extracting the International Classification of Diseases code parameters from the unified valid data of various characteristics of the hospital electronic medical record to obtain the International Classification of Diseases code parameter feature set; extracting the International Classification of Diseases code parameters from the unified valid data of various characteristics of the hospital electronic medical record and the original data of the electronic medical record in the corresponding hospital server to obtain the International Classification of Diseases code parameter set; comparing and analyzing the elements in the International Classification of Diseases code parameter feature set with the elements in the International Classification of Diseases code parameter set to obtain the International Classification of Diseases code parameter fluctuation judgment value; comparing the International Classification of Diseases code parameter feature set with the International Classification of Diseases code parameter set in the adjacent clinical A comparative analysis is performed on the medical information timestamps to obtain the relative frequency of change of the International Classification of Diseases codes; the unified valid data of various characteristics of the hospital electronic medical record are classified according to different clinical medical information timestamps in a predefined time window to obtain the values ​​of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record, and the average values ​​in the predefined time window are analyzed and calculated to obtain the average values ​​of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record; a comprehensive analysis is performed on the International Classification of Diseases code parameter fluctuation judgment value, the relative frequency of change of the International Classification of Diseases code, the values ​​of various characteristics of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record and the average values ​​of various characteristics of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record to obtain the fluctuation indication value of the clinical medical information data time series analysis.

[0012] Furthermore, whether to perform a second timing analysis is determined based on the analysis results of the first threshold value of the clinical medical information data timing analysis fluctuation indicator value, specifically including: if the clinical medical information data timing analysis fluctuation indicator value is equal to or greater than the first threshold value of the clinical medical information data timing analysis, then it is determined that a second timing analysis is performed and the process goes to S5; if the clinical medical information data timing analysis fluctuation indicator value is less than the first threshold value of the clinical medical information data timing analysis, then it is determined that a second timing analysis is not performed and the process goes to S8.

[0013] Furthermore, a comprehensive value of fluctuation of the time series analysis of clinical medical information data is obtained, which specifically includes: the predefined periodic time window contains at least one predefined time window; the unified valid data of various characteristics of the hospital electronic medical record are classified according to different predefined periodic time windows, and the unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows are obtained; the unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows are traversed and analyzed respectively, and the average value, maximum value and minimum value of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows are obtained; the unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows are traversed and analyzed by MATLAB software Perform Fourier transform to convert the unified effective data of various characteristics of the hospital electronic medical record from the original time domain data into frequency domain data, and obtain the various characteristic frequency domain data of the hospital electronic medical record under the predefined periodic time window; perform normalized analysis on the amplitude value of the various characteristic frequency domain data of the hospital electronic medical record through MATLAB software, and obtain the maximum amplitude point corresponding to the maximum normalized weight of the Fourier transform frequency domain analysis under the predefined periodic time window; comprehensively analyze the average value, maximum value, minimum value of various characteristics of the unified effective data of various characteristics of the hospital electronic medical record under different predefined periodic time windows and the maximum amplitude point corresponding to the maximum normalized weight of the Fourier transform frequency domain analysis under the predefined periodic time window to obtain the comprehensive value of the fluctuation of the clinical medical information data time series analysis.

[0014] Furthermore, obtaining the comprehensive value of the clinical medical information data time series analysis fluctuation also includes: comprehensively analyzing the comprehensive value of the clinical medical information data time series analysis fluctuation and the clinical medical information data time series analysis fluctuation indicator value to obtain the comprehensive value of the clinical medical information data time series analysis fluctuation; the specific method for obtaining the comprehensive value of the clinical medical information data time series analysis fluctuation is as follows: ; Number the predefined time windows one by one, Indicates the number of the predefined time window, , Represents the total number of predefined time windows; The predefined periodic time windows of the predefined time windows are numbered. Indicates the number of the predefined periodic time window, , Indicates the total number of predefined periodic time windows; Indicates the Periodic fluctuation correction value of clinical medical information data time series analysis in a predefined time window; Indicates the Periodic fluctuation correction value of clinical medical information data time series analysis in a predefined periodic time window; Indicates the The fluctuation indicator value of clinical medical information data time series analysis in a predefined time window.

[0015] Furthermore, whether to perform data correction is determined based on the analysis results of the second threshold value of the comprehensive value of the timing analysis of clinical medical information data. Specifically, if the comprehensive value of the timing analysis of clinical medical information data fluctuations is equal to or greater than the second threshold value of the timing analysis of clinical medical information data, then it is determined that data correction is to be performed and the process goes to S7; if the comprehensive value of the timing analysis of clinical medical information data fluctuations is less than the second threshold value of the timing analysis of clinical medical information data, then it is determined that data correction is not to be performed and the process goes to S8.

[0016] Furthermore, the specific process of correcting the data and notifying relevant personnel is as follows: extracting unified and valid data of various characteristics of the hospital electronic medical record corresponding to the comprehensive fluctuation value of the clinical medical information data time series analysis in different predefined time windows, and recording them as various characteristic problem data of the hospital electronic medical record; performing a one-to-one comparison analysis on the various characteristic problem data of the hospital electronic medical record and the original data of the electronic medical record of the corresponding hospital according to the corresponding timestamps; retaining the different data in the comparative analysis and recombining them in chronological order to obtain the problem data of the hospital electronic medical record; the central server sends the problem data of the hospital electronic medical record to the manual terminal through the intelligent communication equipment, and notifies the relevant personnel to conduct manual inspection to obtain the correction data of the hospital electronic medical record; the relevant personnel return the correction data of the hospital electronic medical record to the central server.

[0017] Furthermore, the specific process of ending the timing analysis and notifying relevant personnel is as follows: the unified valid data of various characteristics of the hospital electronic medical record corresponding to the clinical medical information data timing analysis fluctuation indicator value less than the first threshold of the clinical medical information data timing analysis are classified and partitioned by using the corresponding user as the data index rule and stored in the central server; the unified valid data of various characteristics of the hospital electronic medical record corresponding to the clinical medical information data timing analysis fluctuation comprehensive value less than the second threshold of the clinical medical information data timing analysis are classified and partitioned by using the corresponding user as the data index rule and stored in the central server; the central server sends the stored unified valid data of various characteristics of the hospital electronic medical record to the manual terminal through the intelligent communication equipment with the corresponding user as the data index rule, and notifies relevant personnel for manual review; if the manual review by the relevant personnel confirms that it is correct, the unified valid data of various characteristics of the hospital electronic medical record after manual review will be recorded as the standardized valid data of the hospital electronic medical record; the relevant personnel return the standardized valid data of the hospital electronic medical record to the central server, and overwrite the unified valid data of various characteristics of the hospital electronic medical record before the corresponding manual review.

[0018] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. The present invention achieves the effect of improving the accuracy of standardized data after standardization of clinical medical information data by collecting and standardizing data, time series analysis to evaluate data anomalies, and data correction and storage, thereby solving the problem of insufficient accuracy in detecting anomalies in standardized data after standardization of clinical medical information data in the prior art.

[0019] 2. A hierarchical decision-making mechanism is implemented through phased time series analysis and threshold judgment processes, which helps to identify complex data fluctuation patterns and periodic changes, thereby improving the accuracy of data anomaly detection after medical data standardization.

[0020] 3. Through comparative analysis and manual inspection, automatic identification and correction of data in the early stage and manual omission checking in the later stage can identify and correct errors in the data, thereby increasing the control of data quality and greatly improving the overall efficiency of data processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A flowchart of a method for analyzing clinical medical information data using AI analysis according to an embodiment of the present application; Figure 2 Schematic diagram of the analysis function image of the fluctuation indicator value of the clinical medical information data time series analysis provided in the embodiment of the present application. DETAILED DESCRIPTION

[0022] The embodiments of the present application provide a time series analysis method for clinical medical information data based on AI analysis, thereby solving the problem in the prior art of insufficient accuracy in detecting abnormalities in standardized data after the standardization of clinical medical information data. By collecting and standardizing data, evaluating data abnormalities through time series analysis, and correcting and storing data, the accuracy of standardized data after the standardization of clinical medical information data is improved.

[0023] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0024] like Figure 1As shown, it is a flow chart of the time series analysis method of clinical medical information data based on AI analysis provided by an embodiment of the present application, and the method includes the following steps: S1. The central server collects electronic medical record data of different hospitals, extracts clinical medical information data of different hospitals and converts them into a unified format to obtain unified data of electronic medical records of hospitals; S2. Data preprocessing is performed on the unified data of electronic medical records of hospitals to obtain unified valid data of various characteristics of electronic medical records of hospitals; S3. A first time series analysis is performed on the unified valid data of various characteristics of electronic medical records of hospitals to obtain a fluctuation indicator value of the time series analysis of clinical medical information data; S4. The fluctuation indicator value of the time series analysis of clinical medical information data is subjected to a first threshold judgment. Perform a second time series analysis based on the first threshold of the fluctuation indicator value of the time series analysis of the clinical medical information data. If yes, proceed to S5, otherwise proceed to S8; S5. Perform a second time series analysis on the unified valid data of various characteristics of the hospital electronic medical record to obtain a comprehensive fluctuation value of the time series analysis of the clinical medical information data; S6. Perform a second threshold judgment analysis on the comprehensive fluctuation value of the time series analysis of the clinical medical information data. According to the analysis result of the second threshold of the comprehensive fluctuation value of the time series analysis of the clinical medical information data, determine whether to perform data correction. If yes, proceed to S7, otherwise proceed to S8; S7. Perform data correction and notify relevant personnel; S8. End the time series analysis and store the data.

[0025] Furthermore, clinical medical information data of different hospitals are extracted therefrom, specifically including: the central server extracts data from different hospital servers through the ETL tool via the data connection interface to obtain the original electronic medical record data of different hospitals; the original electronic medical record data of different hospitals retain the timestamps corresponding to the original electronic medical record data; the central server converts the original electronic medical record data of different hospitals into unified electronic medical record data of hospitals through a predefined data mapping model; the central server classifies the unified electronic medical record data of hospitals into unified electronic medical record data of hospitals of different users through predefined patient master index classification rules; the central server uses different users as data index rules and uses different users as data classification partitions to store unified electronic medical record data of hospitals of different users.

[0026] In this embodiment, the hospital server is specifically a hospital information system or an electronic medical record system.

[0027] The central server collects electronic medical record data from different hospitals and extracts clinical medical information data, which specifically includes: Identify the data source, which is the electronic medical record system of each hospital; Extract data from each hospital's electronic medical record system using ETL (extract, transform, load) tools or custom scripts; Convert the extracted raw electronic medical record data of hospitals into a unified format and structure, for example, perform data mapping and conversion, such as mapping the diagnosis codes of different hospitals to a unified International Classification of Diseases code, for example, ICD11; load and store the converted data into the database of the central server.

[0028] Convert data to unified formats and standards, such as healthcare information exchange standards including FHIR.

[0029] The unified EMR data representation is a standardized data set generated through conversion on a central server using a predefined data mapping model. It may include the following example content: patient basic information, medical consultation information, consultation status, diagnosis, treating physician, medication information, test results, hospitalization information, and corresponding timestamps. The specific data content of the unified EMR data is derived from the information contained in the original EMR data of each hospital.

[0030] The data content of the original electronic medical record data of different hospitals corresponds to data categories including the above-mentioned example content.

[0031] The process of converting the extracted raw data from hospital electronic medical records into a unified format and structure involves data mapping and conversion at multiple levels. Possible examples include: mapping drug codes from different hospitals to a unified drug coding system (e.g., ATC, RxNorm); mapping surgical procedure codes from different hospitals to the International Classification of Surgical Procedures (e.g., ICD-PCS, OPS); mapping laboratory test codes from different hospitals to a unified test code system (e.g., LOINC); unifying gender representation (e.g., using "M" and "F"); unifying the date of birth format (e.g., YYYY-MM-DD); converting unstructured text medical records into a structured data format, for example, converting text descriptions in medical records into queryable data fields; and unifying measurement units across different hospitals (e.g., unifying "kilograms" and "pounds").

[0032] It should be noted that the above data rule mapping technology is currently mature and convenient, but the specific data mapping rules can be modified and manually determined according to specific circumstances.

[0033] Establish a patient master index to link patient information across hospitals. This refers to a unique identifier used to identify each patient instance within a specific domain. A patient unique identifier is a unique identifier used in clinical practice to assist in uniquely identifying patient information, visible to all stakeholders within the domain and across domains. Unique identifiers can be loaded and identified using ID cards, social security cards, medical insurance cards, citizen cards, and health cards.

[0034] For some patients, such as those with long-term chronic diseases, their clinical medical information data may span a long period of time and be scattered across multiple hospitals. Therefore, if the data from multiple different hospitals are not unified and aggregated according to the patient's primary index, data will be missing during the data collection stage, reducing the accuracy of subsequent data processing.

[0035] Furthermore, data preprocessing is performed on the unified hospital electronic medical record data, specifically including: classifying the unified hospital electronic medical record data of the same user into unified hospital electronic medical record data of different characteristics of the same user by predefined clinical medical information features, and recording them as various types of characteristic hospital electronic medical record unified data; performing data anomaly detection on the various types of characteristic hospital electronic medical record unified data by using the Z-score method, and comparing and analyzing the Z-score values ​​of the various types of characteristic hospital electronic medical record unified data with the predefined Z-score detection threshold value; if the Z-score values ​​of the various types of characteristic hospital electronic medical record unified data are greater than the predefined Z-score detection threshold value, then the corresponding various types of characteristic hospital electronic medical record unified data are judged as abnormal values ​​to be detected; The various characteristic unified data of hospital electronic medical records judged as abnormal values ​​to be detected are compared and analyzed with the original data of electronic medical records in the corresponding hospital server: if the various characteristic unified data of hospital electronic medical records judged as abnormal values ​​to be detected are the same as the original data of electronic medical records in the corresponding hospital server, they are judged as non-abnormal values ​​and no longer perform data abnormality detection, and the various characteristic unified data of hospital electronic medical records of non-abnormal values ​​are recorded as the various characteristic unified valid data of hospital electronic medical records; if the various characteristic unified data of hospital electronic medical records judged as abnormal values ​​to be detected are different from the original data of electronic medical records in the corresponding hospital server, they are judged as abnormal values, and the various characteristic unified data of hospital electronic medical records of abnormal values ​​to be detected are replaced with the original data of electronic medical records in the corresponding hospital server and data preprocessing is performed again.

[0036] In this embodiment, the predefined clinical medical information features include but are not limited to the following example feature contents: patient basic information, medical information, medical status, diagnosis information, treating physician, medication information, examination and test results, hospitalization information and corresponding timestamps.

[0037] The original electronic medical record data of different hospitals and the corresponding unified electronic medical record data of the hospitals are stored in the original electronic medical system of the hospital with the user (patient) as the primary index and the timestamp as the secondary index. Therefore, the predefined clinical medical information features are used to classify the unified electronic medical record data of the hospital for the same user. The data can be directly classified according to the specific content of the predefined clinical medical information features, and divided into unified electronic medical record data of the hospital for the same user with different characteristics. The different characteristics are the specific characteristic contents corresponding to the predefined clinical medical information features.

[0038] The Z-score method in data anomaly detection first calculates the mean and standard deviation corresponding to all types of characteristic hospital electronic medical record unified data for each type of characteristic hospital electronic medical record unified data; the predefined Z-score detection threshold can be directly extracted from the clinical medical information data time series analysis database. Generally, the predefined Z-score detection threshold (absolute value) can be set to 3.

[0039] The Z-score method can avoid serious data errors caused by abnormalities after the original data of electronic medical records of different hospitals are standardized.

[0040] Furthermore, the specific process of obtaining the fluctuation indicator value of the time series analysis of clinical medical information data is as follows: extracting the International Classification of Diseases code parameters from the unified valid data of various characteristics of the hospital electronic medical record to obtain the International Classification of Diseases code parameter feature set; extracting the International Classification of Diseases code parameters from the unified valid data of various characteristics of the hospital electronic medical record and the original data of the electronic medical record in the corresponding hospital server to obtain the International Classification of Diseases code parameter set; comparing and analyzing the elements in the International Classification of Diseases code parameter feature set with the elements in the International Classification of Diseases code parameter set to obtain the International Classification of Diseases code parameter fluctuation judgment value; comparing the International Classification of Diseases code parameter feature set with the International Classification of Diseases code parameter set in the adjacent clinical A comparative analysis is performed on the medical information timestamps to obtain the relative frequency of change of the International Classification of Diseases codes; the unified valid data of various characteristics of the hospital electronic medical record are classified according to different clinical medical information timestamps in a predefined time window to obtain the values ​​of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record, and the average values ​​in the predefined time window are analyzed and calculated to obtain the average values ​​of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record; a comprehensive analysis is performed on the International Classification of Diseases code parameter fluctuation judgment value, the relative frequency of change of the International Classification of Diseases code, the values ​​of various characteristics of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record and the average values ​​of various characteristics of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record to obtain the fluctuation indication value of the clinical medical information data time series analysis.

[0041] In this embodiment, the fluctuation indicator value of the clinical medical information data time series analysis is used to describe the relative fluctuation level of different characteristics of the same user in a predefined time window.

[0042] The specific method for obtaining the fluctuation indicator value of the clinical medical information data time series analysis is as follows: like ,but ;like ,but ; ; If the International Classification of Diseases code parameter set and the International Classification of Diseases code parameter feature set are the same, it means that when the extracted original data is converted into a unified format and structure, the clinical medical information data of the same user is not mixed with the clinical medical information data of other users. However, it cannot be guaranteed whether the clinical medical information data of the same user is out of order in time, and further judgment and analysis are needed.

[0043] The predefined time windows used for timing analysis are numbered one by one. Indicates the number of the predefined time window, , Indicates the total number of predefined time windows.

[0044] According to the various characteristics of the hospital electronic medical record within the predefined time window, the clinical medical information timestamps corresponding to the valid data are unified and numbered respectively. A number indicating the timestamp of clinical medical information. , Indicates the total number of clinical medical information timestamps.

[0045] When the central server extracts data from different hospital servers through the ETL tool via the data connection interface, the clinical medical information timestamps corresponding to the hospital's electronic medical record original data are also extracted, and the electronic medical record original data and the clinical medical information timestamps correspond one to one.

[0046] Indicates the The fluctuation indicator value of clinical medical information data time series analysis in a predefined time window.

[0047] Represents a natural constant.

[0048] Set the International Classification of Diseases code parameter set to set , set the International Classification of Diseases code parameter feature set to the set .

[0049] Indicates the The International Classification of Diseases code parameter fluctuation judgment value for a predefined time window.

[0050] The International Classification of Diseases code parameter fluctuation judgment value is used to describe the judgment value corresponding to the change of the International Classification of Diseases code parameter in a predefined time window through the change of the International Classification of Diseases code parameter.

[0051] The International Classification of Diseases code of the clinical medical information timestamp corresponding to the International Classification of Diseases code parameter feature set is compared with the International Classification of Diseases code of the next adjacent clinical medical information timestamp. If the two are exactly the same, the number of identical International Classification of Diseases codes is increased by one; otherwise, the number of different International Classification of Diseases codes is increased by one, until all International Classification of Diseases codes corresponding to the clinical medical information timestamps in the predefined time window are traversed and analyzed. The value of the number of different International Classification of Diseases codes divided by the number of identical International Classification of Diseases codes is recorded as the relative change frequency of the International Classification of Diseases code.

[0052] Indicates the The relative frequency of changes in International Classification of Disease codes over a predefined time window.

[0053] If the International Classification of Diseases code suddenly changes or has an abnormal combination within a predefined time window, it means that the clinical medical information data of the same user may have data disorder in the time dimension when converted to a unified format and structure. The greater the relative change frequency of the International Classification of Diseases code, the greater the fluctuation indicator value of the clinical medical information data time series analysis.

[0054] The various characteristics corresponding to the unified data of the electronic medical records of each type of hospital are numbered separately. The number representing each type of feature, , Represents the total number of features of each category.

[0055] In an electronic medical record system, features can be a way to categorize and identify different types of data. The following are examples of possible features: 1. Medication prescriptions: medication type, dosage, and frequency of administration (the prescription date is the timestamp of the clinical information corresponding to the prescription); 2. Laboratory test results: test item number, test result value, and reference range (the test date is the timestamp of the clinical information corresponding to the prescription); 3. Medical procedure and surgical records: procedure code, procedure name, and procedure result.

[0056] The average value of the same type of features under different clinical medical information timestamps in a predefined time window is calculated for the unified valid data of various features of the hospital electronic medical record, and the average value of the same type of features is obtained, and then the average value of various types of features is obtained.

[0057] Indicates the of a predefined time window The average value of the class feature.

[0058] Indicates the of a predefined time window The clinical medical information timestamp The value of the class characteristic.

[0059] The original data of the electronic medical record corresponds one-to-one to the clinical medical information timestamp, so the value of the feature obtained after processing also corresponds one-to-one to the clinical medical information timestamp. In a certain predefined time window, the value of the same type of feature corresponds one-to-one to the clinical medical information timestamp; there are three clinical medical information timestamps in a certain predefined time window, which correspond to three values ​​of the same type of feature; for example, in a predefined time window of four months, there are three clinical medical information timestamps on January 11, February 15 and March 17, which correspond to blood glucose values ​​of 5.2mmol / L (molar concentration), 6.2mmol / L (molar concentration) and 5.8mmol / L (molar concentration), respectively, which are three values ​​corresponding to the same type of feature.

[0060] like Figure 2 As shown, it is a schematic diagram of the analysis function image of the fluctuation indicator value of the clinical medical information data time series analysis provided by the embodiment of the present application; setting , the relative frequency of changes in the International Classification of Disease codes was used as the independent variable, and the fluctuation indicator value of the time series analysis of clinical medical information data was used as the dependent variable; Function curve in the figure , ,express and , function curve Indicates when and Schematic curve of the analysis function image of the fluctuation indicator value of the clinical medical information data time series analysis.

[0061] Function curve in the figure , ,express and , function curve Indicates when and Schematic curve of the analysis function image of the fluctuation indicator value of the clinical medical information data time series analysis.

[0062] Function curve in the figure , ,express and , function curve Indicates when and Schematic curve of the analysis function image of the fluctuation indicator value of the clinical medical information data time series analysis.

[0063] From the function curve in the figure And the function curve in the figure It is clear that when Same, but At different times, The larger the value of the feature is, the larger the difference between the feature value and the average value of the feature is. The function curve Relative to the function curve The relative fluctuation level in the predefined time window is greater, so the fluctuation indicator value of the clinical medical information data time series analysis is greater.

[0064] From the function curve in the figure And the function curve in the figure It is clear that when Same, but At different times, The larger the value, the larger the average value of the feature. At this time, the relative difference between the value of the feature and the average value of the feature is smaller, and the function curve Relative to the function curve The relative fluctuation level in the predefined time window is smaller, so the fluctuation indicator value of the clinical medical information data time series analysis is smaller.

[0065] As can be seen from the figure, when the relative frequency of changes in the International Classification of Diseases codes continues to increase, the greater the relative frequency of changes in the International Classification of Diseases codes, the greater the relative ratio of the number of different International Classification of Diseases codes to the number of identical International Classification of Diseases codes. In actual situations, when users have long-term chronic diseases and the corresponding electronic medical record original data of the hospital are too scattered in too many hospitals, data standardization is more difficult, the pressure of data unification is greater, and data disorder is prone to occur during data standardization. At this time, the greater the relative frequency of changes in the International Classification of Diseases codes, the hidden dangers will be buried for the data call after subsequent data standardization, and the hidden risks will increase significantly, which requires timely evaluation and adjustment. Furthermore, whether to perform a second timing analysis is determined based on the analysis results of the first threshold value of the clinical medical information data timing analysis fluctuation indicator value, specifically including: if the clinical medical information data timing analysis fluctuation indicator value is equal to or greater than the first threshold value of the clinical medical information data timing analysis, then it is determined that a second timing analysis is performed and the process goes to S5; if the clinical medical information data timing analysis fluctuation indicator value is less than the first threshold value of the clinical medical information data timing analysis, then it is determined that a second timing analysis is not performed and the process goes to S8.

[0066] In this embodiment, the first threshold of the clinical medical information data time series analysis is directly extracted from the clinical medical information data time series analysis database, and the first threshold of the clinical medical information data time series analysis is used to describe whether the clinical medical information data time series analysis fluctuation indicator value exceeds the acceptable range.

[0067] If the fluctuation indicator value of the clinical medical information data time series analysis exceeds or is equal to the first threshold value of the clinical medical information data time series analysis, it is considered that the data fluctuation is abnormal and further second time series analysis is required.

[0068] Furthermore, a comprehensive value of fluctuation of the time series analysis of clinical medical information data is obtained, which specifically includes: the predefined periodic time window contains at least one predefined time window; the unified valid data of various characteristics of the hospital electronic medical record are classified according to different predefined periodic time windows, and the unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows are obtained; the unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows are traversed and analyzed respectively, and the average value, maximum value and minimum value of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows are obtained; the unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows are traversed and analyzed by MATLAB software Perform Fourier transform to convert the unified effective data of various characteristics of the hospital electronic medical record from the original time domain data into frequency domain data, and obtain the various characteristic frequency domain data of the hospital electronic medical record under the predefined periodic time window; perform normalized analysis on the amplitude value of the various characteristic frequency domain data of the hospital electronic medical record through MATLAB software, and obtain the maximum amplitude point corresponding to the maximum normalized weight of the Fourier transform frequency domain analysis under the predefined periodic time window; comprehensively analyze the average value, maximum value, minimum value of various characteristics of the unified effective data of various characteristics of the hospital electronic medical record under different predefined periodic time windows and the maximum amplitude point corresponding to the maximum normalized weight of the Fourier transform frequency domain analysis under the predefined periodic time window to obtain the comprehensive value of the fluctuation of the clinical medical information data time series analysis.

[0069] In this embodiment, the comprehensive fluctuation value of the time series analysis of clinical medical information data is used to describe the comprehensive relative fluctuation level of different characteristics of the same user in the time series analysis.

[0070] The periodic fluctuation correction value of the time series analysis of clinical medical information data is used to describe the relative fluctuation correction level of different characteristics of the same user in different periods in the time series analysis.

[0071] The specific method for obtaining the comprehensive fluctuation value of clinical medical information data time series analysis is as follows: ; The predefined periodic time window includes at least one predefined time window.

[0072] Will include The predefined periodic time windows of the predefined time windows are numbered. Indicates the number of the predefined periodic time window, , Indicates the total number of predefined periodic time windows.

[0073] For example, the predefined periodic time window is three months, the predefined time window is two weeks, there are eleven cases where the predefined periodic time window includes a certain predefined time window, and at most eleven different predefined periodic time windows include the same predefined time window.

[0074] Indicates the Periodic fluctuation correction value of clinical medical information data time series analysis in a predefined time window.

[0075] Indicates the The correction value of periodic fluctuation of clinical medical information data time series analysis in a predefined periodic time window.

[0076] Indicates the The first of a predefined periodic time window The average value of the class feature.

[0077] Indicates the The first of a predefined periodic time window The maximum value of the class feature.

[0078] Indicates the The first of a predefined periodic time window Minimum value of the class feature.

[0079] According to the unified effective data of various characteristics of the hospital electronic medical record within the predefined periodic time window, Fourier transform is performed using MATLAB software to convert the unified effective data of various characteristics of the hospital electronic medical record from the original time domain data into frequency domain data, and obtain the various characteristic frequency domain data of the hospital electronic medical record within the predefined periodic time window; the various characteristic frequency domain data of the hospital electronic medical record are normalized in terms of amplitude value using MATLAB software, and the result of dividing the maximum amplitude value by the sum of all amplitude values ​​is used as the maximum normalized weight; the amplitude value represents the intensity of a specific frequency component. The larger the maximum normalized weight, the higher the intensity of the corresponding specific frequency component, which means that the unified effective data of various characteristics of the hospital electronic medical record within the predefined periodic time window are more concentrated on a certain value, which means that the relative volatility of the unified effective data of various characteristics of the hospital electronic medical record within the predefined periodic time window is smaller.

[0080] Indicates the The maximum amplitude point of the Fourier transform frequency domain analysis under a predefined periodic time window corresponds to the maximum normalized weight, The value range is .

[0081] Table 1 shows the periodic fluctuation correction values ​​of the time series analysis of clinical medical information data. The example parameters in Table 1 only take a class of characteristic parameters of a predefined time window under a predefined periodic time window for illustration, as shown in Table 1.

[0082] The average value of a feature over a predefined periodic time window The average value of a feature over a predefined time window The maximum value of a feature over a predefined periodic time window Minimum value of a feature over a predefined periodic time window The maximum amplitude point of the Fourier transform frequency domain analysis in the predefined periodic time window corresponds to the maximum normalized weight Periodic fluctuation correction value for time series analysis of clinical medical information data 7.5 5.5 8.5 4.5 0.8 3.82571400765 7.5 5.5 8.5 4.5 0.5 5.06846187 8.5 5.5 8.5 4.5 0.5 6.18833060983 8.5 5.5 10.5 3.5 0.5 8.21729146551 According to Table 1, if the maximum amplitude point corresponding to the maximum normalized weight of the Fourier transform frequency domain analysis in the predefined periodic time window becomes smaller, it means that the unified valid data of various characteristics of the hospital electronic medical record in the predefined periodic time window is less concentrated on a certain value and is more dispersed, indicating that the relative volatility of the unified valid data of various characteristics of the hospital electronic medical record in the predefined periodic time window is greater; the greater the relative difference between the average value of the characteristics of the predefined periodic time window and the average value of the characteristics of the predefined time window, the higher the relative volatility level of the unified valid data of various characteristics of the hospital electronic medical record in the two different time windows, the predefined periodic time window and the predefined time window.

[0083] Furthermore, obtaining the comprehensive value of the clinical medical information data time series analysis fluctuation also includes: comprehensively analyzing the comprehensive value of the clinical medical information data time series analysis fluctuation and the clinical medical information data time series analysis fluctuation indicator value to obtain the comprehensive value of the clinical medical information data time series analysis fluctuation; the specific method for obtaining the comprehensive value of the clinical medical information data time series analysis fluctuation is as follows: ; Number the predefined time windows one by one, Indicates the number of the predefined time window, , Represents the total number of predefined time windows; The predefined periodic time windows of the predefined time windows are numbered. Indicates the number of the predefined periodic time window, , Indicates the total number of predefined periodic time windows; Indicates the Periodic fluctuation correction value of clinical medical information data time series analysis in a predefined time window; Indicates the Periodic fluctuation correction value of clinical medical information data time series analysis in a predefined periodic time window; Indicates the The fluctuation indicator value of clinical medical information data time series analysis in a predefined time window.

[0084] In this embodiment, in specific circumstances, certain medical indicator data of the user will change with the cycle, and the data volatility obtained by analyzing in a time window of one year, a time window of one quarter, or a time window of one month is different. If all data are evaluated with the same data volatility assessment level, it is easy to misjudge the correct data, which greatly slows down the efficiency of data anomaly detection after data standardization. Therefore, it is necessary to obtain different data volatility levels in different cycles and conduct a comprehensive analysis of the data volatility.

[0085] For example, patients with hypertension may have higher levels in winter and lower levels in summer; patients with hyperthyroidism may have increased thyroid hormone levels in winter; therefore, it is normal for the data fluctuation level to be higher in a specific time window and smaller in other time windows. Different data fluctuation level assessments and corrections need to be made according to different time windows to improve the accuracy of data anomaly detection.

[0086] Furthermore, whether to perform data correction is determined based on the analysis results of the second threshold value of the comprehensive value of the timing analysis of clinical medical information data. Specifically, if the comprehensive value of the timing analysis of clinical medical information data fluctuations is equal to or greater than the second threshold value of the timing analysis of clinical medical information data, then it is determined that data correction is to be performed and the process goes to S7; if the comprehensive value of the timing analysis of clinical medical information data fluctuations is less than the second threshold value of the timing analysis of clinical medical information data, then it is determined that data correction is not to be performed and the process goes to S8.

[0087] In this embodiment, the second threshold of the clinical medical information data time series analysis is directly extracted from the clinical medical information data time series analysis database, and the second threshold of the clinical medical information data time series analysis is used to describe whether the comprehensive value of the clinical medical information data time series analysis fluctuation exceeds an acceptable range.

[0088] If the comprehensive value of the fluctuation of the clinical medical information data time series analysis exceeds or is equal to the second threshold of the clinical medical information data time series analysis, it is considered that the data fluctuation is still abnormal after analysis in different time windows, and data correction is required.

[0089] Furthermore, the specific process of correcting the data and notifying relevant personnel is as follows: extracting unified and valid data of various characteristics of the hospital electronic medical record corresponding to the comprehensive fluctuation value of the clinical medical information data time series analysis in different predefined time windows, and recording them as various characteristic problem data of the hospital electronic medical record; performing a one-to-one comparison analysis on the various characteristic problem data of the hospital electronic medical record and the original data of the electronic medical record of the corresponding hospital according to the corresponding timestamps; retaining the different data in the comparative analysis and recombining them in chronological order to obtain the problem data of the hospital electronic medical record; the central server sends the problem data of the hospital electronic medical record to the manual terminal through the intelligent communication equipment, and notifies the relevant personnel to conduct manual inspection to obtain the correction data of the hospital electronic medical record; the relevant personnel return the correction data of the hospital electronic medical record to the central server.

[0090] In this embodiment, after the above judgment, the workload of data comparison and manual correction is greatly reduced. In order to ensure that the data is correct, relevant personnel are required to manually confirm the results of the data correction at the end.

[0091] Furthermore, the specific process of ending the timing analysis and notifying relevant personnel is as follows: the unified valid data of various characteristics of the hospital electronic medical record corresponding to the clinical medical information data timing analysis fluctuation indicator value less than the first threshold of the clinical medical information data timing analysis are classified and partitioned by using the corresponding user as the data index rule and stored in the central server; the unified valid data of various characteristics of the hospital electronic medical record corresponding to the clinical medical information data timing analysis fluctuation comprehensive value less than the second threshold of the clinical medical information data timing analysis are classified and partitioned by using the corresponding user as the data index rule and stored in the central server; the central server sends the stored unified valid data of various characteristics of the hospital electronic medical record to the manual terminal through the intelligent communication equipment with the corresponding user as the data index rule, and notifies relevant personnel for manual review; if the manual review by the relevant personnel confirms that it is correct, the unified valid data of various characteristics of the hospital electronic medical record after manual review will be recorded as the standardized valid data of the hospital electronic medical record; the relevant personnel return the standardized valid data of the hospital electronic medical record to the central server, and overwrite the unified valid data of various characteristics of the hospital electronic medical record before the corresponding manual review.

[0092] In this embodiment, relevant personnel transmit the standardized valid data of the hospital electronic medical records back to the central server, and overwrite the unified valid data of various characteristics of the hospital electronic medical records before the corresponding manual review, so that the final unified valid data of various characteristics of the hospital electronic medical records are standardized data without outliers, which greatly simplifies the subsequent unified processing of data.

[0093] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0094] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0095] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0096] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0097] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0098] Obviously, those skilled in the art may make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if such modifications and variations fall within the scope of the claims and their equivalents, the present invention is intended to include such modifications and variations.

Claims

1. A time series analysis method for clinical medical information data based on AI analysis, characterized in that: The following steps are involved: S1. The central server collects electronic medical record data from different hospitals, extracts clinical medical information data from different hospitals and converts them into a unified format to obtain unified electronic medical record data of the hospitals; S2. Perform data preprocessing on the unified data of the hospital's electronic medical records to obtain unified and valid data of various characteristics of the hospital's electronic medical records; S3. Perform a first time series analysis on the unified valid data of various characteristics of the hospital's electronic medical records to obtain a fluctuation indicator value of the clinical medical information data time series analysis; S4. Perform a first threshold judgment analysis on the fluctuation indicator value of the clinical medical information data timing analysis, and determine whether to perform a second timing analysis based on the fluctuation indicator value of the clinical medical information data timing analysis first threshold judgment analysis result. If yes, proceed to S5, otherwise proceed to S8; S5. Perform a second time series analysis by unifying valid data of various characteristics of the hospital's electronic medical records to obtain a comprehensive value of fluctuations in the time series analysis of clinical medical information data; S6. Perform a second threshold judgment analysis on the comprehensive value of the fluctuation of the clinical medical information data time series analysis, and determine whether to perform data correction based on the result of the second threshold judgment analysis of the comprehensive value of the fluctuation of the clinical medical information data time series analysis. If yes, proceed to S7, otherwise proceed to S8; S7. Make data corrections and notify relevant personnel; S8. Finish the timing analysis and store the data.

2. The method for analyzing clinical medical information data based on AI analysis according to claim 1, characterized in that: The clinical medical information data of different hospitals are extracted, specifically including: The central server uses the ETL tool to extract data from different hospital servers through the data connection interface to obtain the original data of electronic medical records of different hospitals; The original electronic medical record data of different hospitals retains the timestamp corresponding to the original electronic medical record data; The central server converts the original data of electronic medical records of different hospitals into unified data of electronic medical records of hospitals through a predefined data mapping model; The central server classifies the hospital electronic medical record unified data into hospital electronic medical record unified data of different users through predefined patient master index classification rules; The central server uses different users as data index rules and uses different users as data classification partitions to store the unified data of hospital electronic medical records of different users.

3. The method for time series analysis of clinical medical information data based on AI analysis as claimed in claim 1, characterized in that: The data preprocessing of the unified data of the hospital electronic medical record specifically includes: Classify the unified hospital electronic medical record data of the same user into unified hospital electronic medical record data of different characteristics of the same user by pre-defining clinical medical information characteristics, and record them as unified hospital electronic medical record data of different characteristics; The Z-score method is used to detect data anomalies on the unified data of electronic medical records of various characteristic hospitals, and the Z-score values ​​of the unified data of electronic medical records of various characteristic hospitals are compared and analyzed with the predefined Z-score detection threshold. If the Z-score values ​​of the unified data of electronic medical records of various characteristic hospitals are greater than the predefined Z-score detection threshold, the corresponding unified data of electronic medical records of various characteristic hospitals are judged as abnormal values ​​to be detected; Compare and analyze the unified electronic medical record data of various characteristic hospitals that are judged as abnormal values ​​to be detected with the original electronic medical record data in the corresponding hospital server: If the various characteristic hospital electronic medical record unified data determined to be abnormal values ​​to be detected are the same as the electronic medical record original data in the corresponding hospital server, they are determined to be non-abnormal values ​​and no longer perform data abnormality detection, and the various characteristic hospital electronic medical record unified data of non-abnormal values ​​are recorded as various characteristic unified valid data of hospital electronic medical records; If the various characteristic hospital electronic medical record unified data judged as abnormal values ​​to be detected are different from the electronic medical record original data in the corresponding hospital server, they are judged as abnormal values, and the various characteristic hospital electronic medical record unified data of the abnormal values ​​to be detected are replaced by the electronic medical record original data in the corresponding hospital server and the data preprocessing is performed again.

4. The method for analyzing clinical medical information data based on AI analysis according to claim 1, characterized in that: The specific process of obtaining the clinical medical information data time series analysis fluctuation indication value is as follows: Extract the International Classification of Diseases code parameters from the unified and effective data of various characteristics of the hospital's electronic medical records, and obtain the International Classification of Diseases code parameter feature set; Extracting the International Classification of Diseases code parameters from the unified valid data of various characteristics of the hospital's electronic medical records and the original data of the electronic medical records in the corresponding hospital server to obtain an International Classification of Diseases code parameter set; Compare and analyze the elements in the International Classification of Diseases code parameter feature set with the elements in the International Classification of Diseases code parameter set to obtain the International Classification of Diseases code parameter fluctuation judgment value; Compare and analyze the International Classification of Diseases code parameter feature set with the International Classification of Diseases code parameter set at adjacent clinical medical information timestamps within a predefined time window to obtain the relative change frequency of the International Classification of Diseases code; Classify the unified valid data of various characteristics of the hospital electronic medical record according to different clinical medical information timestamps in a predefined time window, obtain the values ​​of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record, analyze and calculate the average value in the predefined time window, and obtain the average value of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record; The fluctuation judgment value of the International Classification of Diseases code parameters, the relative change frequency of the International Classification of Diseases code, the values ​​of various features of the unified valid data of various features of the hospital electronic medical records, and the average values ​​of various features of the unified valid data of various features of the hospital electronic medical records are comprehensively analyzed to obtain the fluctuation indicator value of the clinical medical information data time series analysis.

5. The method for time series analysis of clinical medical information data based on AI analysis as claimed in claim 1, characterized in that: The step of judging whether to perform a second time series analysis based on the first threshold of the fluctuation indication value of the time series analysis of the clinical medical information data specifically includes: If the fluctuation indication value of the clinical medical information data time series analysis is equal to or greater than the first threshold value of the clinical medical information data time series analysis, it is determined that the second time series analysis is to be performed, and the process goes to S5; If the clinical medical information data timing analysis fluctuation indication value is less than the clinical medical information data timing analysis first threshold value, it is determined that the second timing analysis is not to be performed and the process proceeds to S8.

6. The method for time series analysis of clinical medical information data based on AI analysis as claimed in claim 1, characterized in that: The obtaining of the comprehensive value of fluctuations in the time series analysis of clinical medical information data specifically includes: The predefined periodic time window includes at least one predefined time window; Classify the unified valid data of various characteristics of the hospital electronic medical record according to different predefined periodic time windows, and obtain the unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows; The unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows are traversed and analyzed respectively, and the average value, maximum value and minimum value of various characteristics of the unified valid data of various characteristics of the hospital electronic medical record under different predefined periodic time windows are obtained; Through the MATLAB software, various unified and effective data of characteristics of the hospital electronic medical record under different predefined periodic time windows are subjected to Fourier transformation, and various unified and effective data of characteristics of the hospital electronic medical record are converted from original time domain data to frequency domain data, so as to obtain various characteristic frequency domain data of the hospital electronic medical record under the predefined periodic time window; through the MATLAB software, various characteristic frequency domain data of the hospital electronic medical record are subjected to normalization analysis on amplitude value, so as to obtain the maximum normalization weight corresponding to the maximum amplitude point of the Fourier transform frequency domain analysis under the predefined periodic time window; Comprehensively analyze the average value, maximum value, minimum value of various characteristics of hospital electronic medical records under different predefined periodic time windows, and the maximum normalized weight corresponding to the maximum amplitude point of Fourier transform frequency domain analysis under predefined periodic time windows of various characteristics of unified valid data to obtain the comprehensive value of clinical medical information data time series analysis fluctuation.

7. The method for time series analysis of clinical medical information data based on AI analysis as claimed in claim 1, characterized in that: The step of obtaining the comprehensive value of fluctuations in the time series analysis of clinical medical information data also includes: Comprehensively analyzing the clinical medical information data time series analysis fluctuation comprehensive value and the clinical medical information data time series analysis fluctuation indication value to obtain the clinical medical information data time series analysis fluctuation comprehensive value; The specific method for obtaining the comprehensive value of the fluctuation of the time series analysis of clinical medical information data is as follows: ； Number the predefined time windows. represents the number of the predefined time window, , Indicates the total number of predefined time windows; Will include The predefined periodic time windows of the predefined time windows are numbered accordingly. Indicates the number of the predefined periodic time window, , Indicates the total number of predefined periodic time windows; Indicates The correction value of periodic fluctuation of clinical medical information data time series analysis in a predefined time window; Indicates A periodic fluctuation correction value for the time series analysis of clinical medical information data in a predefined periodic time window; Indicates The fluctuation indicator value of clinical medical information data time series analysis in a predefined time window.

8. The method for time series analysis of clinical medical information data based on AI analysis as claimed in claim 1, characterized in that: The step of judging whether to perform data correction according to the second threshold value of the comprehensive value of the time series analysis fluctuation of the clinical medical information data specifically includes: If the comprehensive fluctuation value of the clinical medical information data time series analysis is equal to or greater than the second threshold value of the clinical medical information data time series analysis, it is determined that data correction is to be performed and the process goes to S7; If the comprehensive value of the fluctuation of the clinical medical information data timing analysis is less than the second threshold value of the clinical medical information data timing analysis, it is determined that no data correction is performed and the process enters S8.

9. The method for time series analysis of clinical medical information data based on AI analysis as claimed in claim 1, characterized in that: The specific process of correcting the data and notifying relevant personnel is as follows: Extract the unified valid data of various characteristics of hospital electronic medical records corresponding to the comprehensive value of the fluctuation of clinical medical information data time series analysis in different predefined time windows, and record them as various characteristic problem data of hospital electronic medical records; Compare and analyze the various characteristic problem data of the hospital's electronic medical records with the original data of the corresponding hospital's electronic medical records one by one according to the corresponding timestamps; The different data in the comparative analysis were retained and reassembled in chronological order to obtain the problem data of the hospital electronic medical records; The central server sends the hospital's electronic medical record problem data to the manual terminal through the intelligent communication device, and notifies the relevant personnel to conduct manual inspection to obtain the hospital's electronic medical record correction data; Relevant personnel transmit the hospital's electronic medical record correction data back to the central server.

10. The method for time series analysis of clinical medical information data based on AI analysis as claimed in claim 1, characterized in that: The specific process of completing the timing analysis and notifying relevant personnel is as follows: The unified valid data of various characteristics of hospital electronic medical records corresponding to the clinical medical information data time series analysis fluctuation indicator value less than the first threshold value of the clinical medical information data time series analysis are classified and partitioned by using the corresponding user as the data index rule and stored in the central server; The unified valid data of various characteristics of the hospital electronic medical records corresponding to the clinical medical information data time series analysis fluctuation comprehensive value less than the second threshold of the clinical medical information data time series analysis are classified and partitioned by using the corresponding user as the data index rule and stored in the central server; The central server sends the unified and valid data of various characteristics of the hospital's electronic medical records stored to the manual terminal through intelligent communication equipment with the corresponding user as the data index rule, and notifies the relevant personnel for manual review; If the relevant personnel manually review and confirm that they are correct, the unified and valid data of various characteristics of the hospital electronic medical record after manual review will be recorded as the standardized and valid data of the hospital electronic medical record; Relevant personnel transmit the standardized and valid data of the hospital's electronic medical records to the central server, and cover the unified and valid data of various characteristics of the hospital's electronic medical records before the corresponding manual review.

Images