Disease control investigation paperless method, system and equipment based on multi-modal large model and medium

Abstract

The invention discloses a disease control investigation paperless method, system and equipment based on a multi-modal large model, and a medium, mainly relates to the technical field of disease control investigation, and is used for solving the problems that multi-modal data lacks a unified processing framework, an analysis model is limited to preset rules and shallow feature association, and report generation depends on manual arrangement in an existing scheme. Comprising the steps of obtaining initial text data; when a preset analysis requirement is met, determining the initial data as detection data; obtaining input data; wherein the preset cue word templates comprise a disease association analysis cue word template, a propagation path prediction cue word template and a risk factor identification cue word template; obtaining analysis data corresponding to the multi-modal data by using the input data; wherein the analysis data comprises associated diseases, transmission paths and infection factors; and substituting the analysis data into a preset report template to obtain a report result.

Description

Technical Field

[0001] This application relates to the field of paperless disease control investigation technology, and in particular to a paperless method, system, equipment and medium for disease control investigation based on a multimodal large model. Background Technology

[0002] Current disease control surveys primarily rely on traditional paper questionnaires or single-modal electronic forms, with data collection largely limited to structured text input (such as Excel spreadsheets or basic web forms). Some information systems have implemented cloud data storage, using SSL / TLS encryption for transmission and employing rule engines for simple logical validation (such as mandatory field verification and format checking). The analysis phase typically relies on statistical software (such as SPSS) or machine learning models (such as the random forest algorithm) for single-dimensional analysis, outputting standardized statistical reports. The presentation layer often uses BI tools (such as Tableau) to generate visual charts or present results via PDF reports. While this technology stack has formed a fixed workflow in existing disease control scenarios, each step is often pieced together from independent systems, lacking an end-to-end integrated solution.

[0003] The existing solution has three core flaws: First, it lacks a unified processing framework for multimodal data (such as voice recordings, on-site photos, and video vital signs data), resulting in insufficient utilization of unstructured data; second, the analysis model is limited to preset rules and shallow feature associations, and cannot deeply explore the complex network relationships of disease transmission (such as spatiotemporal cross-infection paths); third, report generation relies on manual arrangement, and the average processing cycle from raw data to decision recommendations is long. Summary of the Invention

[0004] This application provides a paperless method, system, device, and medium for disease control surveys based on a multimodal large model, in order to solve the problems of existing solutions, such as the lack of a unified processing framework for multimodal data, analysis models being limited to preset rules and shallow feature associations, and report generation relying on manual arrangement.

[0005] Firstly, this application provides a paperless method for disease control surveys based on a multimodal large model. The multimodal large model includes: a multimodal data acquisition layer, a data transmission layer, a large model analysis layer, and an application display layer. The method includes: An electronic survey form is generated through the multimodal data acquisition layer and distributed to several acquisition terminals corresponding to the multimodal data acquisition layer; the uploaded multimodal data is then acquired through the acquisition terminals. Through the data transmission layer, using a preset encrypted communication protocol, the multimodal data from the acquisition terminal is uploaded to the cloud server; The large model analysis layer preprocesses the data from the cloud server to obtain initial text data. It then checks whether the initial data meets preset analysis requirements; if so, it is designated as detection data. Based on the data type, the detection data is input into a corresponding preset prompt word template to obtain input data. These preset prompt word templates include: disease association analysis prompt word templates, transmission route prediction prompt word templates, and risk factor identification prompt word templates. Using the input data, analysis data corresponding to the multimodal data is obtained. This analysis data includes: associated diseases, transmission routes, and infection factors. Finally, the analysis data is substituted into a preset report template to obtain the report results. The application presentation layer displays the report results and allows for secondary editing of the report results.

[0006] In one implementation of this application, an electronic survey form is generated through a multimodal data acquisition layer, and the electronic survey form is distributed to several acquisition terminals corresponding to the multimodal data acquisition layer, specifically including: Electronic survey form templates for obtaining data from various modalities through a preset interface; The electronic survey form templates for each modality of data were distributed to each data collection terminal.

[0007] In one implementation of this application, the uploaded multimodal data is acquired via a data acquisition terminal, specifically including: Using the data collection terminal, select any one of the several electronic survey templates as the upload template; Data is collected using the data collection components embedded in the selected electronic survey template.

[0008] In one implementation of this application, the initial data is checked to see if it meets preset analysis requirements. If the preset analysis requirements are met, the initial data is determined to be the test data. Specifically, this includes: Extract data keywords from the initial data and compare their similarity with keywords corresponding to related diseases, transmission routes, and infectious factors; If no data is found that is associated with the relevant disease, transmission route and infection factor, it is determined that the preset analysis requirements are not met and the submission is rejected. When data is available that is associated with related diseases, transmission routes, and infection factors, the initial data is determined to be the test data.

[0009] In one implementation of this application, before obtaining the analysis data corresponding to the multimodal data using the input data, the method further includes: Obtain historical sample data; the historical sample data shall include at least: historical input data and labeled analysis data; Historical sample data is input into an interpolation data synthesis algorithm to generate a preset number of new sample data. The newly added sample data and historical sample data are used as training data to input the preset large model to obtain the trained preset large model.

[0010] Secondly, this application provides a paperless disease control survey system based on a multimodal large model, the system comprising: The multimodal data acquisition layer is used to generate electronic questionnaires and distribute them to several acquisition terminals corresponding to the multimodal data acquisition layer; the acquisition terminals are used to acquire the uploaded multimodal data. The data transmission layer is used to upload multimodal data from the acquisition terminal to the cloud server using a preset encrypted communication protocol; The large model analysis layer is used to preprocess data from the cloud server to obtain initial text data; it checks whether the initial data meets the preset analysis requirements, and if so, determines the initial data as test data; based on the data type, it inputs the test data into the corresponding preset prompt word template to obtain input data; the preset prompt word templates include: disease association analysis prompt word template, transmission route prediction prompt word template, and risk factor identification prompt word template; using the input data, it obtains the analysis data corresponding to the multimodal data; the analysis data includes: associated diseases, transmission routes, and infection factors; and it substitutes the analysis data into the preset report template to obtain the report results. The application presentation layer is used to display the report results and to perform secondary editing of the report results.

[0011] In one implementation of this application, the large model analysis layer includes detection data units. Used to extract data keywords from the initial data and compare the similarity between the data keywords and the corresponding keywords of related diseases, transmission routes and infection factors; If no data is found that is associated with the relevant disease, transmission route and infection factor, it is determined that the preset analysis requirements are not met and the submission is rejected. When data is available that is associated with related diseases, transmission routes, and infection factors, the initial data is determined to be the test data.

[0012] In one implementation of this application, the large model analysis layer includes a model training unit. Used to obtain historical sample data; wherein, historical sample data includes at least: historical input data and labeled analysis data; Historical sample data is input into an interpolation data synthesis algorithm to generate a preset number of new sample data. The newly added sample data and historical sample data are used as training data to input the preset large model to obtain the trained preset large model.

[0013] Thirdly, this application provides a paperless disease control survey device based on a multimodal large model, the device comprising: processor; And a memory containing executable code, which, when executed, causes the processor to perform a paperless disease control survey method based on a multimodal large model, as described above.

[0014] Fourthly, this application provides a non-volatile computer storage medium storing computer instructions, which, when executed, implement a paperless method for disease control investigation based on a multimodal large model as described above.

[0015] As can be seen from the above technical solutions, this application has the following advantages: I. Standardized integration and acquisition of multimodal data have been achieved: The architecture design, which unifies the data collection terminal with electronic questionnaires and integrates a multimodal data collection layer, effectively solves the problem of standardized integration of heterogeneous data from multiple sources, such as paper forms, images, and sensor data, in traditional disease control surveys. A pre-defined encrypted communication protocol ensures the integrity and security of data from the terminal to the cloud, significantly improving data transmission efficiency compared to traditional manual aggregation methods. This design enables real-time synchronization of different modalities of data, such as text, images, and spatiotemporal information, to a unified analysis platform, laying a data foundation for subsequent in-depth analysis.

[0016] II. Intelligent analysis capabilities that break through rule limitations: The large-scale model analysis layer upgrades traditional statistical rule-based or shallow machine learning-based analysis methods to large-scale model-driven analysis with semantic understanding and deep reasoning capabilities through three types of professional prompt word templates: disease association analysis, transmission route prediction, and risk factor identification. This mechanism can directly mine complex correlation features hidden in multimodal data (such as the correlation between environmental factors and pathogen variation), automatically outputting high-level analysis results such as associated diseases and transmission routes, significantly reducing reliance on epidemiological expert experience rules, and improving the response speed and accuracy of sudden epidemic assessment.

[0017] III. Fully Automated Report Generation and Flexible Interaction: By intelligently matching preset report templates with the analysis results of large-scale models, end-to-end automated generation of structured reports from raw data is achieved. The application presentation layer supports secondary editing of machine-generated reports, retaining the flexibility of manual review while eliminating the time cost and formatting error risks associated with manual report compilation in traditional solutions. This design is particularly suitable for application scenarios requiring the rapid generation of multiple versions of analysis reports (such as cross-regional joint prevention and control of epidemics), significantly enhancing the information support capabilities for disease control decision-making. Attached Figure Description

[0018] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a paperless disease control survey method based on a multimodal large model provided in an embodiment of this application.

[0020] Figure 2 This is a schematic diagram of the internal structure of a paperless disease control investigation system based on a multimodal large model, provided in an embodiment of this application.

[0021] Figure 3 This is a schematic diagram of the internal structure of a paperless disease control survey device based on a multimodal large model, provided in an embodiment of this application. Detailed Implementation

[0022] The technical solutions of 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.

[0023] Those skilled in the art should understand that the embodiments described below are merely preferred embodiments of this disclosure and do not imply that this disclosure can only be implemented through these preferred embodiments. These preferred embodiments are merely used to explain the technical principles of this disclosure and are not intended to limit the scope of protection of this disclosure. Based on the preferred embodiments provided by this disclosure, all other embodiments obtained by those skilled in the art without creative effort should still fall within the scope of protection of this disclosure.

[0024] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0025] The technical solutions proposed in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0026] The embodiment provides a paperless method for disease control surveys based on a multimodal large model, such as Figure 1 As shown in the embodiments of this application, the method mainly includes the following steps: Step 110: Generate an electronic survey form through the multimodal data acquisition layer, and distribute the electronic survey form to several acquisition terminals corresponding to the multimodal data acquisition layer; acquire the uploaded multimodal data through the acquisition terminals.

[0027] In some embodiments, an electronic questionnaire is generated through the multimodal data acquisition layer, and the electronic questionnaire is distributed to several acquisition terminals corresponding to the multimodal data acquisition layer, specifically including: Electronic survey form templates for obtaining data from various modalities through a preset interface; The electronic survey form templates for each modality of data were distributed to each data collection terminal.

[0028] The uploaded multimodal data is acquired through the acquisition terminal, specifically including: Using the data collection terminal, select any one of the several electronic survey templates as the upload template; Data is collected using the data collection components embedded in the selected electronic survey template.

[0029] This step utilizes a multimodal data acquisition layer to generate and distribute electronic questionnaires and collect multimodal data, offering the following benefits: First, using electronic questionnaire templates unifies the data collection format, standardizing data input across different collection terminals and preventing data quality issues caused by inconsistent collection standards. The mechanism of obtaining and distributing templates through a pre-defined interface makes template updates and maintenance more efficient, reducing manual configuration workload. Second, collection terminals can flexibly select applicable electronic questionnaire templates based on actual needs, combining them with embedded collection components to complete data acquisition. This modular design improves the adaptability of data collection, meeting diverse needs in different scenarios. Simultaneously, the electronic data collection process reduces the cost of using paper forms and minimizes errors that may occur during manual data entry. Centralized collection of multimodal data provides a unified data source for subsequent data processing and analysis, avoiding the complexity of integrating multi-source data. The entire process reduces manual intervention through automation, improving collection efficiency while ensuring data integrity. This solution achieves a balance between data quality control and collection efficiency through standardized data collection templates and flexible terminal adaptation, laying the foundation for the subsequent application of multimodal data.

[0030] Step 120: Upload the multimodal data from the acquisition terminal to the cloud server through the data transmission layer using a preset encrypted communication protocol.

[0031] This step utilizes a pre-defined encrypted communication protocol at the data transmission layer to achieve secure transmission of multimodal data, offering the following advantages: First, the encrypted communication protocol effectively protects the confidentiality and integrity of data during transmission, preventing theft or tampering. Second, the standardized encryption mechanism reduces the risk of data leakage due to transmission security vulnerabilities, while also reducing reliance on the security performance of terminal devices. As an independent module, the data transmission layer is compatible with different data acquisition terminals and cloud servers, improving system scalability. The use of a pre-defined protocol avoids the burden of configuring encryption parameters individually for each transmission, thus improving data transmission efficiency.

[0032] Step 130: Through the large model analysis layer, perform data preprocessing on the data of the cloud server to obtain initial text data; check whether the initial data meets the preset analysis requirements. If it does, determine the initial data as detection data; according to the data type, input the detection data into the corresponding type of preset prompt word template to obtain input data; use the input data to obtain the analysis data corresponding to the multimodal data; substitute the analysis data into the preset report template to obtain the report results.

[0033] The data analyzed includes: associated diseases, transmission routes, and infection factors; the preset prompt word templates include: disease association analysis prompt word template, transmission route prediction prompt word template, and risk factor identification prompt word template.

[0034] The initial data is checked to see if it meets the preset analysis requirements. If it does, the initial data is determined to be the test data, which specifically includes: Extract data keywords from the initial data and compare their similarity with keywords corresponding to related diseases, transmission routes, and infectious factors; If no data is found that is associated with the relevant disease, transmission route and infection factor, it is determined that the preset analysis requirements are not met and the submission is rejected. When data is available that is associated with related diseases, transmission routes, and infection factors, the initial data is determined to be the test data.

[0035] Before using the input data to obtain the analysis data corresponding to the multimodal data, the method also includes: Obtain historical sample data; the historical sample data shall include at least: historical input data and labeled analysis data; Historical sample data is input into an interpolation data synthesis algorithm to generate a preset number of new sample data. The newly added sample data and historical sample data are used as training data to input the preset large model to obtain the trained preset large model.

[0036] To further explain, the training scheme for the multimodal large model mainly involves the prediction training of the large model's analysis layer. The specific training process can be as follows: Prompt template adaptation Based on the data type, a preset prompt template is matched, and structured input data is generated by dynamically populating variables. For example, in e-commerce scenarios, "Please analyze the user sentiment in [text content]" is used as a unified input paradigm.

[0037] The Transformer architecture is used as the base model, and the model is optimized through the following steps: text data is pre-trained using Masked Language Modeling (MLM); cross-modal data is aligned using contrastive learning; attention mechanism is introduced to dynamically weight features of different modalities; and the model is trained by setting hierarchical learning rates (e.g., 0.0001 for text layer and 0.00005 for fusion layer).

[0038] At the same time, an online learning module is deployed. When new data types exceed the coverage of existing templates, the template library expansion process is triggered, and new scenarios are quickly adapted through small sample learning.

[0039] This step utilizes a large-scale model analysis layer to achieve intelligent processing and disease analysis of multimodal data, offering the following benefits: First, the introduction of data preprocessing cleans and standardizes the raw data, ensuring input data quality and reducing noise interference with analysis results. A pre-defined detection mechanism for analysis requirements effectively filters out data that meets the analysis criteria, preventing invalid data from entering subsequent processes and improving analysis efficiency. Matching corresponding prompt templates to different data types allows input data to be organized in a structured manner, enhancing the accuracy of the large-scale model's understanding. The application of pre-defined large-scale models automates complex tasks such as disease identification, transmission path prediction, and infection factor analysis, reducing the difficulty and workload of manual analysis. The use of report templates standardizes the presentation of analysis results, facilitating interpretation and application. During model training, historical sample data augmentation techniques generate new sample data, alleviating the problem of insufficient training data and improving the model's generalization ability. The entire process combines standardized templates and automated analysis, improving processing efficiency while ensuring analysis quality. This solution, through its end-to-end design of data preprocessing, intelligent analysis, and result generation, provides a systematic analytical tool for disease-related research and a reliable data foundation for decision support.

[0040] Step 140: Display the report results and perform secondary editing of the report results through the application presentation layer.

[0041] The visualization mechanism in this step presents complex data analysis results in a structured format, making it easy for users to intuitively understand key information such as related diseases, transmission routes, and infection factors. The standardized display format of the report results lowers the barrier to understanding for users with specialized knowledge, enabling them to effectively access information from diverse backgrounds. The secondary editing function allows users to adjust and improve the report content according to their actual needs, enhancing the report's applicability and flexibility.

[0042] As described above, this embodiment effectively solves the problem of standardized integration of heterogeneous data from multiple sources, such as paper forms, images, and sensor data, in traditional disease control surveys by using an architecture design that unifies the data collection terminal with electronic questionnaires and a multimodal data collection layer. The pre-defined encrypted communication protocol ensures the integrity and security of data from the terminal to the cloud, significantly improving data transmission efficiency compared to traditional manual aggregation methods. This design enables real-time synchronization of different modalities of data, such as text, images, and spatiotemporal information, to a unified analysis platform, laying a data foundation for subsequent in-depth analysis.

[0043] The large-scale model analysis layer upgrades traditional statistical rule-based or shallow machine learning-based analysis methods to large-scale model-driven analysis with semantic understanding and deep reasoning capabilities through three types of professional prompt word templates: disease association analysis, transmission route prediction, and risk factor identification. This mechanism can directly mine complex correlation features hidden in multimodal data (such as the correlation between environmental factors and pathogen variation), automatically outputting high-level analysis results such as associated diseases and transmission routes, significantly reducing reliance on epidemiological expert experience rules, and improving the response speed and accuracy of sudden epidemic assessment.

[0044] By intelligently matching preset report templates with the analysis results of large-scale models, end-to-end automated generation of structured reports from raw data is achieved. The application presentation layer supports secondary editing of machine-generated reports, retaining the flexibility of manual review while eliminating the time cost and formatting error risks associated with manual report compilation in traditional solutions. This design is particularly suitable for application scenarios requiring the rapid generation of multiple versions of analysis reports (such as cross-regional joint prevention and control of epidemics), significantly enhancing the information support capabilities for disease control decision-making.

[0045] In addition, this application Figure 2 This application provides an embodiment of a paperless disease control survey system based on a multimodal large model. For example... Figure 2 As shown in the embodiments of this application, the system mainly includes: The multimodal data acquisition layer 210 is used to generate electronic questionnaires and distribute them to several acquisition terminals corresponding to the multimodal data acquisition layer; and to acquire the uploaded multimodal data through the acquisition terminals.

[0046] The data transmission layer 220 is used to upload multimodal data from the acquisition terminal to the cloud server using a preset encrypted communication protocol.

[0047] The large model analysis layer 230 is used to preprocess data from the cloud server to obtain initial text data; it checks whether the initial data meets the preset analysis requirements, and if it does, it determines the initial data as detection data; according to the data type, it inputs the detection data into the corresponding type of preset prompt word template to obtain input data; the preset prompt word templates include: disease association analysis prompt word template, transmission route prediction prompt word template, and risk factor identification prompt word template; using the input data, it obtains the analysis data corresponding to the multimodal data; the analysis data includes: associated diseases, transmission routes, and infection factors; and it substitutes the analysis data into the preset report template to obtain the report results.

[0048] The large model analysis layer 230 includes detection data units. Used to extract data keywords from the initial data and compare the similarity between the data keywords and the corresponding keywords of related diseases, transmission routes and infection factors; If no data is found that is associated with the relevant disease, transmission route and infection factor, it is determined that the preset analysis requirements are not met and the submission is rejected. When data is available that is associated with related diseases, transmission routes, and infection factors, the initial data is determined to be the test data.

[0049] The large model analysis layer 230 includes model training units. Used to obtain historical sample data; wherein, historical sample data includes at least: historical input data and labeled analysis data; Historical sample data is input into an interpolation data synthesis algorithm to generate a preset number of new sample data. The newly added sample data and historical sample data are used as training data to input the preset large model to obtain the trained preset large model.

[0050] Application presentation layer 240 is used to display report results and to perform secondary editing of the report results.

[0051] The above are method embodiments of this application. Based on the same inventive concept, this application also provides a paperless disease control survey device based on a multimodal large model. Figure 3 As shown, the device includes: a processor; and a memory storing executable code thereon, which, when executed, causes the processor to perform a paperless disease control survey method based on a multimodal large model as described in the above embodiments.

[0052] Specifically, the server generates an electronic survey form through a multimodal data acquisition layer and distributes it to several acquisition terminals corresponding to the multimodal data acquisition layer. The acquisition terminals then acquire the uploaded multimodal data. Through a data transmission layer, using a preset encrypted communication protocol, the multimodal data from the acquisition terminals is uploaded to a cloud server. A large-scale model analysis layer preprocesses the data from the cloud server to obtain initial text data. It checks whether the initial data meets preset analysis requirements; if so, the initial data is designated as detection data. Based on the data type, the detection data is input into a corresponding preset prompt word template to obtain input data. These preset prompt word templates include: disease association analysis prompt word templates, transmission route prediction prompt word templates, and risk factor identification prompt word templates. Using the input data, analysis data corresponding to the multimodal data is obtained. This analysis data includes: associated diseases, transmission routes, and infection factors. The analysis data is then substituted into a preset report template to obtain report results. Finally, through an application display layer, the report results are displayed, and the results can be edited further.

[0053] In addition, this application embodiment also provides a non-volatile computer storage medium storing executable instructions, which, when executed, implement the above-described paperless method for disease control investigation based on a multimodal large model.

[0054] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A paperless method for disease control investigation based on a multi-modal large model, the multi-modal large model comprising: The multi-modal data acquisition layer, the data transmission layer, the large model analysis layer, and the application display layer, characterized in that the method comprises: Through the multi-modal data acquisition layer, an electronic questionnaire is generated, and the electronic questionnaire is distributed to a plurality of collection terminals corresponding to the multi-modal data acquisition layer; through the collection terminal, uploaded multi-modal data is obtained; through the data transmission layer, the multi-modal data of the collection terminal is uploaded to the cloud server using a preset encrypted communication protocol; Through the large model analysis layer, the data of the cloud server is preprocessed to obtain initial text data; it is detected whether the initial data meets the preset analysis requirement, and when the preset analysis requirement is met, the initial data is determined as detection data; according to the data type, the detection data is input into a preset prompt word template of a corresponding type to obtain input data; wherein the preset prompt word template includes: a disease association analysis prompt word template, a transmission path prediction prompt word template, and a risk factor identification prompt word template; using the input data, analysis data corresponding to the multi-modal data is obtained; wherein the analysis data includes: associated diseases, transmission paths, and infection factors; the analysis data is substituted into a preset report template to obtain a report result; Through the application display layer, the report result is displayed and secondary editing of the report result is performed.

2. The multi-modal large model based disease control survey paperless method according to claim 1, wherein, Through the multi-modal data acquisition layer, an electronic questionnaire is generated, and the electronic questionnaire is distributed to a plurality of collection terminals corresponding to the multi-modal data acquisition layer, specifically including: Through a preset interface, an electronic questionnaire template of each modal data is obtained; The electronic questionnaire template of each modal data is distributed to each collection terminal.

3. The multi-modal large model based disease control survey paperless method according to claim 1, wherein, Through the collection terminal, uploaded multi-modal data is obtained, specifically including: Through the collection terminal, any electronic questionnaire template is selected from a plurality of electronic questionnaire templates as an upload template; Through a collection component embedded in the selected electronic questionnaire template, collection data is obtained.

4. The multi-modal large model based disease control survey paperless method according to claim 1, wherein, It is detected whether the initial data meets the preset analysis requirement, and when the preset analysis requirement is met, the initial data is determined as detection data, specifically including: Data keywords in the initial data are extracted, and similarity comparison between the data keywords and corresponding keywords of associated diseases, transmission paths, and infection factors is performed; According to the similarity, it is determined whether the initial data has data associated with the associated diseases, the transmission paths, and the infection factors; When there is no data associated with the associated diseases, the transmission paths, and the infection factors, it is determined that the preset analysis requirement is not met, and the re-filing is rejected; When there is data associated with the associated diseases, the transmission paths, and the infection factors, the initial data is determined as detection data.

5. The multi-modal large model based disease control survey paperless method according to claim 1, wherein, Before using the input data to obtain analysis data corresponding to the multi-modal data, the method further includes: Historical sample data is obtained; wherein the historical sample data at least includes: historical input data, and labeled analysis data; The historical sample data is input into an interpolation data synthesis algorithm to generate a preset number of new sample data; The new sample data and the historical sample data are input into a preset large model as training data to obtain a trained preset large model.

6. A paperless system for disease control investigation based on a multi-modal large model, characterized in that, The system includes: A multi-modal data collection layer is configured to generate an electronic questionnaire, distribute the electronic questionnaire to a plurality of collection terminals corresponding to the multi-modal data collection layer, and obtain multi-modal data uploaded by the collection terminals; A data transmission layer is configured to upload the multi-modal data of the collection terminals to a cloud server using a preset encryption communication protocol; A large model analysis layer is configured to perform data preprocessing on data of the cloud server to obtain initial text data, detect whether the initial data meets preset analysis requirements, determine the initial data as detection data when the initial data meets the preset analysis requirements, input the detection data into a preset prompt word template of a corresponding type according to a data type to obtain input data, wherein the preset prompt word template includes a disease association analysis prompt word template, a transmission path prediction prompt word template, and a risk factor identification prompt word template, and obtain analysis data corresponding to the multi-modal data using the input data, wherein the analysis data includes associated diseases, transmission paths, and infection factors, and obtain a report result by substituting the analysis data into a preset report template; An application display layer is configured to display the report result and perform secondary editing on the report result.

7. The multi-modal large model based disease control survey paperless system according to claim 6, wherein, The large model analysis layer includes a detection data unit, configured to extract data keywords in the initial data and perform similarity comparison between the data keywords and corresponding keywords of the associated diseases, the transmission paths, and the infection factors; when there is no data associated with the associated diseases, the transmission paths, and the infection factors, it is determined that the preset analysis requirements are not met, and the re-filing is rejected; when there is data associated with the associated diseases, the transmission paths, and the infection factors, it is determined that the initial data is the detection data.

8. The multi-modal large model based disease control survey paperless system according to claim 6, wherein, The large model analysis layer includes a model training unit, configured to obtain historical sample data, wherein the historical sample data at least includes historical input data and labeled analysis data, input the historical sample data into an interpolation data synthesis algorithm to generate a preset number of new sample data, and input the new sample data and the historical sample data as training data into a preset large model to obtain a trained preset large model. The device includes: a processor; 9.A paperless device for disease control investigation based on a multi-modal large model, characterized in that, and a memory having executable code stored thereon, which, when executed, causes the processor to perform a multi-modal large model based disease control investigation paperless method according to any one of claims 1-5. having computer instructions stored thereon, which, when executed, implement a multi-modal large model based disease control investigation paperless method according to any one of claims 1-5. ​ 10. A non-transitory computer storage medium, comprising, ​

Images