Medical document writing assistant

Abstract

The invention relates to the field of medical documents, and discloses a medical document writing assistant comprising an end interaction module which constructs a graphical user interface through Python-tkinter, is used for receiving patient identity information and document time range instructions input by medical staff and displaying the generated medical documents, and is in butt joint with a hospital heterogeneous information system through an HL7FHIR standard interface, so that the medical documents can be written in the hospital; and medical data of the specified patient in the corresponding time range is acquired in a sending mode, wherein the medical data comprises an inspection result, an image diagnosis report, a doctor's advice record and vital sign data. Clinical dilemma is broken through full-process automatic design, a front-end interaction module quickly receives an instruction, a data acquisition and standardization module concurrently calls multi-system data based on an HL7FHIR standard, and serial waiting is avoided; the intelligent processing module automatically completes data structuring through a regular expression and a medical pre-training model, the document generation module dynamically fills content based on a template, a standard document can be generated within several seconds, and the efficiency is improved by dozens of times compared with manual efficiency.

Description

Technical Field

[0001] This invention relates to the field of medical documentation technology, specifically to a medical documentation writing assistant. Background Technology

[0002] Medical records refer to various professional documents that medical personnel use to record, organize, and preserve information such as the patient's health status, diagnosis and treatment process, medical decisions, and results during medical activities, in accordance with relevant laws, regulations, industry norms, and professional standards.

[0003] Its core function is to objectively reflect medical practices, provide a basis for clinical diagnosis and treatment, ensure medical quality and safety, and serve as important data for medical dispute resolution, medical research, and teaching.

[0004] However, in the current medical setting, medical documentation is one of the core tasks of medical staff, but the manual processing mode has significant shortcomings. On the one hand, medical staff need to manually retrieve patient data from multiple systems such as laboratory, imaging, and medical orders, and then connect and organize it into standardized documents one by one. A complete document often takes half an hour to an hour, which consumes a large amount of time that is crucial for doctor-patient communication and treatment decisions. On the other hand, manual data entry is prone to problems such as omissions (e.g., missing laboratory units) and data confusion (e.g., misattributing patient information) due to the complexity of the data and operator fatigue. Incorrect documents may lead to biased diagnosis and treatment, increasing medical risks. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a medical document writing assistant, which solves the significant drawbacks of the existing manual processing methods.

[0006] To achieve the above objectives, the present invention provides a medical document writing assistant through the following technical solution, including a front-end interaction module, which uses Python-tkinter to build a graphical user interface to receive patient identity information and document time range instructions input by medical personnel, and to display the generated medical documents; The data acquisition and standardization module is used to receive instructions output by the front-end interaction module, and connect to the heterogeneous information system within the hospital through the HL7FHIR standard interface to concurrently acquire the medical data of a specified patient within a corresponding time range. The medical data includes test results, imaging diagnostic reports, medical order records, and vital sign data. The intelligent processing module receives the raw medical data output by the data acquisition and standardization module, parses semi-structured data using regular expressions, identifies unstructured data using pre-trained models in the medical field, and generates well-structured data. The document generation module, based on the structured data output by the intelligent processing module, calls the preset medical document templates and dynamically fills in the data through the template engine to generate medical documents that conform to clinical standards. The security authentication module is used to control access permissions for the front-end interaction module. When medical staff log in, a signature token containing identity and permission information is generated. Subsequent requests are authenticated through the token, and data access operation logs are recorded.

[0007] Preferably, the data acquisition and standardization module includes an FHIR resource adapter unit; The FHIR resource adapter unit follows the adapter design pattern, which can convert the internal unified data request instructions into standardized FHIRRESTfulAPI calls. At the same time, it integrates a thread pool or asynchronous I / O model to realize concurrent requests from multiple data sources. It obtains medical data through the Python-requests package, and the obtained data is in dictionary or text format.

[0008] Preferably, the structured processing of the raw medical data by the intelligent processing module includes: Semi-structured data processing: Extracting key information, including test items, values, and units, from text such as test reports using pre-compiled regular expressions; Unstructured data processing: The BioBERT / ClinicalBERT pre-trained model is called to identify clinical entities in the text and is connected to standard medical terminology sets such as ICD-10 and SNOMEDCT to map non-standard expressions into standardized codes. Data regularization: Discrete data is organized into a two-dimensional data table using the Python-pandas package, outliers are marked with reference to the normal value range in the knowledge base, and data units are standardized.

[0009] Preferably, the document generation module includes a template library and a dynamic rendering unit; The template library contains pre-set medical document templates for multiple departments and types, with each template defining a static structural framework and dynamic data placeholders. The dynamic rendering unit receives structured data output by the intelligent processing module, and generates documents by filling placeholders through built-in condition judgment and loop logic. The generated medical documents are saved in text format and can be directly used and integrated with electronic medical record systems.

[0010] Preferably, the security authentication module includes an input verification subunit; The input verification subunit uses a preset rule set to perform real-time format verification of the patient identity information and time range input by the front-end interaction module through regular expressions; the login authentication adopts a signature token mechanism, the token contains identity and permission information, and the authentication is completed by carrying it through the HTTP header; the operation log contains the operator, time, content and IP address.

[0011] Preferably, the data acquisition and standardization module further includes a data verification unit; the data verification unit performs integrity checks on the acquired medical data, and when a key field is detected to be missing, it automatically marks it and feeds it back to the front-end interaction module to prompt medical personnel for confirmation. The key fields include, but are not limited to: the item name and value of the test data, the test date, and the examination site and diagnosis conclusion of the imaging report.

[0012] Preferably, the medical terminology standardization step of the intelligent processing module adopts a terminology mapping table to establish the association between non-standard expressions and standard codes, and supports dynamic updates of the terminology table.

[0013] Preferably, the front-end interaction module adopts an MVC architecture to separate interface display, data storage and business logic. The view layer is responsible for rendering the interface for command input and document preview, the model layer stores patient identity information, document templates and other data, and the controller layer coordinates the communication between the front-end interaction module and the data acquisition and standardization module.

[0014] Preferably, the heterogeneous information systems within the hospital include, but are not limited to: Laboratory Information System (LIS), Imaging System (PACS), and Medical Order System (HIS).

[0015] This invention provides a medical document writing assistant. It has the following beneficial effects: 1. This invention overcomes clinical challenges through a fully automated design. The front-end interaction module rapidly receives instructions, while the data acquisition and standardization module concurrently retrieves data from multiple systems based on the HL7FHIR standard, avoiding sequential waiting. The intelligent processing module automatically structures the data using regular expressions and pre-trained medical models, and the document generation module dynamically fills in content based on templates, generating standardized documents within seconds—dozens of times more efficient than manual methods. Simultaneously, the data verification unit automatically intercepts data lacking key fields, and standardized medical terminology ensures consistent expression, eliminating human error at the source and guaranteeing accurate and complete document content.

[0016] 2. This invention, through the FHIR resource adapter following the adapter design pattern, can convert internal requests into standardized API calls, flexibly connecting to heterogeneous systems such as LIS, PACS, and HIS without the need for separate interface development; medical-specific models such as BioBERT / ClinicalBERT can accurately identify clinical entities, and the multi-department document template library covers scenarios such as internal medicine and surgery, enabling customized document generation without requiring automation personnel to understand medical logic. It successfully integrates automation technology with medical needs, promoting the technological upgrade of medical document processing.

[0017] 3. This invention ensures data security through a multi-layered security mechanism: the input validation subunit validates front-end input in real time using regular expressions, intercepting incorrectly formatted data in advance; the signature token mechanism carries identity and authorization information in the HTTP header to ensure the legitimacy of the request source; and the operation log fully records information such as the operator, time, and IP address, meeting security audit requirements. Simultaneously, data interaction employs standardized interfaces and reliable transmission tools, ensuring the stability of cross-system data interaction while complying with medical data privacy protection and compliance management requirements, thus filling the gaps in existing technologies for medical data security management. Attached Figure Description

[0018] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0019] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example: Please see the appendix Figure 1 This invention provides a medical document writing assistant, including a front-end interaction module. A graphical user interface (GUI) is constructed using Python-tkinter to receive patient identity information and document time range instructions input by medical personnel, and to display the generated medical document. The choice of Python-tkinter enables lightweight deployment of the interface, meeting the needs of rapid startup and convenient operation in clinical scenarios. The front-end interaction module adopts an MVC architecture to separate interface display, data storage, and business logic. The view layer is responsible for rendering the interface for instruction input and document preview, while the model layer stores patient identity information, document templates, and other data. Unified data management ensures data consistency when different functional modules are called. The controller layer coordinates communication between the front-end interaction module and the data acquisition and standardization module, formatting front-end input instructions into standardized requests to achieve seamless data flow between modules.

[0021] The data acquisition and standardization module receives instructions from the front-end interaction module and connects to heterogeneous information systems within the hospital via the HL7FHIR standard interface. It concurrently acquires medical data for a specified patient within a corresponding time range. The HL7FHIR standard resolves data format incompatibility issues between different hospital systems, and the concurrent acquisition method avoids efficiency losses caused by serial waiting, improving data acquisition speed. Medical data includes test results, imaging diagnostic reports, medical orders, and vital sign data, covering core data types in clinical diagnosis and treatment, and meeting the basic data requirements of different medical documents. The data acquisition and standardization module includes an FHIR resource adapter unit to adapt and convert internal data requests to external system interfaces. Heterogeneous information systems within the hospital include, but are not limited to, the Laboratory Information System (LIS), the Imaging System (PACS), and the Medical Order System (HIS), covering the main data sources during patient diagnosis and treatment, ensuring comprehensive data collection.

[0022] The data acquisition and standardization module also includes a data verification unit. This unit can filter incomplete data in advance to avoid document generation anomalies due to data loss in subsequent processing stages. The data verification unit performs integrity checks on the acquired medical data. When key fields are found to be missing, it automatically marks them and provides feedback to the front-end interaction module to prompt medical staff for confirmation. Through automated verification and real-time feedback, the workload of manually checking for missing data is reduced. Key fields include, but are not limited to: the name and value of the test data, the test date, and the examination site and diagnosis conclusion of the imaging report. These fields are the core carriers of clinical information in medical documents, and ensuring their integrity can guarantee the accuracy and validity of the document content.

[0023] The FHIR resource adapter unit follows the adapter design pattern, converting internally unified data request commands into standardized FHIR RESTful API calls. This design flexibly adapts to the interface requirements of different hospital systems, eliminating the need to develop separate request logic for each system. It also integrates thread pools or asynchronous I / O models to enable concurrent requests from multiple data sources, improving the efficiency of data synchronization across multiple systems through parallel multi-task processing. Medical data is acquired via the Python-requests package in dictionary or text format. This tool reliably interfaces with HTTP, ensuring reliable data transmission. Both formats facilitate subsequent intelligent processing modules for parsing and structured operations.

[0024] The intelligent processing module receives raw medical data from the data acquisition and standardization module. It parses semi-structured data using regular expressions and identifies unstructured data using pre-trained medical models, generating well-structured data. Employing a hybrid processing mechanism, it balances processing efficiency and accuracy for different data formats, meeting the diverse structuring needs of medical data. The intelligent processing module's structuring of raw medical data includes: Semi-structured data processing: Key information, including test items, values, and units, is extracted from texts such as test reports using pre-compiled regular expressions. Regular expressions enable high-speed information extraction from fixed-format text, ensuring the accuracy of core information extraction from test data.

[0025] Unstructured data processing: The BioBERT / ClinicalBERT pre-trained model is used to identify clinical entities in the text and is connected to standard medical terminology sets such as ICD-10 and SNOMEDCT to map non-standard expressions into standardized codes. The medical-specific pre-trained model can improve the accuracy of clinical entity recognition, and the connection to standard terminology sets can realize the unified expression of medical information, which is convenient for cross-scenario data reuse.

[0026] Data organization: Discrete data is organized into a two-dimensional data table using the Python-pandas package. Outliers are marked with reference to the normal value range in the knowledge base, and the data units are standardized. The two-dimensional data table can clearly present the time-series change trend of the data. The standardization of outlier marking and units can reduce the difficulty for medical staff to read the data.

[0027] The medical terminology standardization process in the intelligent processing module employs a terminology mapping table to establish a link between non-standard expressions and standard codes, and supports dynamic updates to the terminology table. The mapping table enables rapid matching and conversion of non-standard terms, and the dynamic update function can adapt to the iterative upgrade needs of the medical terminology system. The document generation module, based on the structured data output by the intelligent processing module, calls the pre-set medical document templates and dynamically fills the data through the template engine to generate medical documents that conform to clinical standards. The template-based generation method can ensure the uniformity of document format and avoid format deviations caused by manual typesetting. The document generation module includes a template library and a dynamic rendering unit. The template library contains pre-built medical document templates for multiple departments and types. Each template defines a static structural framework and dynamic data placeholders. The multi-department and multi-type templates can cover common clinical scenarios such as internal medicine, surgery, and obstetrics and gynecology. The static framework ensures the compliance of the document structure, while the dynamic placeholders provide flexible adaptation space for data filling. The dynamic rendering unit receives structured data output from the intelligent processing module and generates documents by filling placeholders through built-in conditional judgments and loop logic. Conditional judgments can realize personalized content display (such as inserting corresponding monitoring data modules according to the patient's disease type), and loop logic can automatically process multiple repeated data (such as test results at multiple time points). The generated medical documents are saved in text format, which supports direct use and integration with electronic medical record systems. The text format has strong compatibility and is easy to view instantly. Integration with electronic medical record systems can realize unified management of document data within the hospital.

[0028] The security authentication module is used to control access to the front-end interaction module. When medical staff log in, a signature token containing identity and permission information is generated. Subsequent requests are authenticated through the token, and data access operation logs are recorded. Through access control and operation traceability, the security and compliance of medical data access are ensured. The security authentication module includes an input verification subunit. The input verification subunit uses a preset rule set and regular expressions to perform real-time format validation on the patient identity information and time range input from the front-end interaction module. Real-time validation can intercept incorrectly formatted data in advance, preventing invalid requests from entering subsequent processes. Login authentication adopts a signature token mechanism. The token contains identity and permission information and is carried in the HTTP header to complete authentication. The token mechanism can ensure the legitimacy of the request source. The way the HTTP header is carried does not affect the main content of the data transmission. The operation log includes the operator, time, content, and IP address. The complete log information can meet the needs of medical data security auditing and facilitate the tracing of abnormal operation behavior.

[0029] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A medical document writing assistant, characterized in that, include, The front-end interaction module uses Python-tkinter to build a graphical user interface, which receives patient identity information and document time range instructions input by medical staff, and displays the generated medical documents. The data acquisition and standardization module is used to receive instructions output by the front-end interaction module, and connect to the heterogeneous information system within the hospital through the HL7FHIR standard interface to concurrently acquire the medical data of a specified patient within a corresponding time range. The medical data includes test results, imaging diagnostic reports, medical order records, and vital sign data. The intelligent processing module receives the raw medical data output by the data acquisition and standardization module, parses semi-structured data using regular expressions, identifies unstructured data using pre-trained models in the medical field, and generates well-structured data. The document generation module, based on the structured data output by the intelligent processing module, calls the preset medical document templates and dynamically fills in the data through the template engine to generate medical documents that conform to clinical standards. The security authentication module is used to control access permissions for the front-end interaction module. When medical staff log in, a signature token containing identity and permission information is generated. Subsequent requests are authenticated through the token, and data access operation logs are recorded.

2. The medical document writing assistant according to claim 1, characterized in that, The data acquisition and standardization module includes an FHIR resource adapter unit; The FHIR resource adapter unit follows the adapter design pattern, which can convert the internal unified data request instructions into standardized FHIRRESTfulAPI calls. At the same time, it integrates a thread pool or asynchronous I / O model to realize concurrent requests from multiple data sources. It obtains medical data through the Python-requests package, and the obtained data is in dictionary or text format.

3. The medical document writing assistant according to claim 1, characterized in that, The intelligent processing module performs structured processing on the raw medical data, including: Semi-structured data processing: Extracting key information, including test items, values, and units, from text such as test reports using pre-compiled regular expressions; Unstructured data processing: The BioBERT / ClinicalBERT pre-trained model is called to identify clinical entities in the text and is connected to standard medical terminology sets such as ICD-10 and SNOMEDCT to map non-standard expressions into standardized codes. Data regularization: Discrete data is organized into a two-dimensional data table using the Python-pandas package, outliers are marked with reference to the normal value range in the knowledge base, and data units are standardized.

4. The medical document writing assistant according to claim 1, characterized in that, The document generation module includes a template library and a dynamic rendering unit; The template library contains pre-set medical document templates for multiple departments and types, with each template defining a static structural framework and dynamic data placeholders. The dynamic rendering unit receives structured data output by the intelligent processing module, and generates documents by filling placeholders through built-in condition judgment and loop logic. The generated medical documents are saved in text format and can be directly used and integrated with electronic medical record systems.

5. The medical document writing assistant according to claim 1, characterized in that, The security authentication module includes an input verification subunit; The input verification subunit performs real-time format verification of the patient identity information and time range input by the front-end interaction module based on a preset rule set and using regular expressions. Login authentication uses a signature token mechanism. The token contains identity and permission information and is carried in the HTTP header to complete authentication. The operation log includes the operator, time, content, and IP address.

6. The medical document writing assistant according to claim 1, characterized in that, The data acquisition and standardization module also includes a data verification unit; the data verification unit performs integrity checks on the acquired medical data, and when a key field is detected to be missing, it automatically marks it and sends feedback to the front-end interaction module to prompt medical personnel for confirmation. The key fields include, but are not limited to: the name and value of the test data, the test date, and the examination site and diagnosis conclusion of the imaging report.

7. The medical document writing assistant according to claim 1, characterized in that, The medical terminology standardization process of the intelligent processing module uses a terminology mapping table to establish a connection between non-standard expressions and standard codes, and supports dynamic updates of the terminology table.

8. The medical document writing assistant according to claim 1, characterized in that, The front-end interaction module adopts an MVC architecture to separate interface display, data storage and business logic. The view layer is responsible for rendering the interface for command input and document preview, the model layer stores patient identity information, document templates and other data, and the controller layer coordinates the communication between the front-end interaction module and the data acquisition and standardization module.

9. The medical document writing assistant according to claim 1, characterized in that, The heterogeneous information systems within the hospital include, but are not limited to: Laboratory Information System (LIS), Imaging System (PACS), and Medical Order System (HIS).

Images