GIST special disease database intelligent management system

Abstract

The invention relates to the technical field of medical data management, and discloses a GIST special disease database intelligent management system, which comprises a multi-source data acquisition module, a standardized processing module, an intelligent analysis module and a security sharing module, the multi-source data acquisition module is in communication connection with a hospital information system H IS, a laboratory information system L IS, an image archiving and communication system PACS and a patient end application, and is used for acquiring electronic medical records, inspection reports, image data and patient active report data; a standardized data model is built in the standardized processing module, and the standardized processing module is used for performing field mapping and format conversion on the acquired multi-modal data; the intelligent analysis module comprises a machine learning algorithm engine. According to the medical data management system, the multi-source data acquisition module is in communication connection with the hospital information system and the laboratory system, so that the problem that most traditional medical data management systems can only acquire single-source or single-mode data, and consequently disease characteristics cannot be comprehensively reflected is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data management, and in particular to an intelligent management system for a GIST disease database. Background Art

[0002] Gastrointestinal stromal tumors (GISTs) are soft tissue tumors that originate from the mesenchymal tissue of the gastrointestinal tract, most commonly in the stomach and small intestine, with a few occurring in the esophagus, colorectum, and other locations. Most cases are sporadic, while some are associated with genetic syndromes (such as Carney triad). Clinical manifestations vary, ranging from asymptomatic in the early stages to abdominal pain, bleeding, and a mass when the tumor is larger. Approximately 80% to 90% of GISTs harbor mutations in the KIT or PDGFRA genes, key targets for targeted therapy. Treatment primarily involves surgical resection, while patients with advanced or metastatic disease require combined treatment with targeted drugs (such as imatinib and sunitinib). GISTs vary in their malignancy, and risk assessment is based on tumor size, mitotic figures, and other factors. Regular follow-up after surgery is necessary to monitor for recurrence and metastasis. In recent years, with the advancement of precision medicine, diagnostic and treatment capabilities have continued to improve.

[0003] Most traditional medical data management systems can only collect data from a single source or a single modality. Due to the lack of data dimensions and insufficient sample heterogeneity, the data analysis conclusions are one-sided and cannot fully reflect the characteristics of the disease. Summary of the Invention

[0004] To remedy the above shortcomings, the present invention provides an intelligent management system for the GIST disease database, aiming to improve the problem that traditional medical data management systems can only collect data from a single source or a single modality. Due to the lack of data dimensions and insufficient sample heterogeneity, the data analysis conclusions are one-sided and cannot fully reflect the characteristics of the disease.

[0005] In a first aspect, the present invention provides the following technical solutions: an intelligent management system for a GIST disease database, comprising a multi-source data acquisition module, a standardized processing module, an intelligent analysis module, and a security sharing module;

[0006] The multi-source data acquisition module is connected to the hospital information system HIS, laboratory information system LIS, image archiving and communication system PACS and patient-side applications respectively, and is used to collect electronic medical records, test reports, imaging data and patient-reported data;

[0007] The standardization processing module has a built-in standardized data model for performing field mapping and format conversion on the collected multimodal data;

[0008] The intelligent analysis module includes a machine learning algorithm engine for constructing a GIST molecular-clinical phenotype association model, and the machine learning algorithm engine includes a data fusion unit for fusing multimodal data;

[0009] The secure sharing module uses blockchain technology to implement data storage and establishes a cross-center data sharing channel through a federated learning framework.

[0010] By adopting the above technical solution, the multi-source data acquisition module is used to connect with the hospital information system, laboratory system, imaging system and patient-side applications to collect electronic medical records, test reports, imaging data and patient-reported data, and then integrate multiple types of data such as clinical diagnosis and treatment, testing, imaging characteristics and patient subjective feedback across platforms, thereby improving the problem that most traditional medical data management systems can only collect single-source or single-modal data, and due to the lack of data dimensions and insufficient sample heterogeneity, the data analysis conclusions are one-sided and cannot fully reflect the characteristics of the disease.

[0011] Preferably, the standardized data model adopts the OMOP public data model, and the standardized processing module performs natural language processing on the unstructured medical record text. The natural language processing process includes the following steps: first, the text is segmented and POS tagged, and then entity features are extracted through the BiLSTM-CRF model. The loss function of the BiLSTM-CRF model is:

[0012]

[0013] Among them, n is the number of samples, T is the sequence length, and θ is the model parameter. is the input feature of the i-th sample at the t-th moment, y t For the corresponding label.

[0014] Preferably, the multimodal data includes genomic data, imaging data and pathological data, and the data fusion unit adopts a feature-level fusion strategy, and the specific formula is F=f genomic ;f radiomic ;f pathomic ]

[0015] Among them, f genomic is the genomic feature vector, f radiomic is the radiomics feature vector, f pathomic is the pathology feature vector, and “;” represents the feature concatenation operation.

[0016] Preferably, the blockchain technology adopts a consortium chain architecture, each node stores the data hash value, and the data is verified by a consensus mechanism when it is uploaded to the chain. The consensus mechanism adopts a practical Byzantine fault-tolerant algorithm. The consensus process includes three stages: pre-preparation, preparation, and submission. The nodes reach consensus through message transmission. The message format is<view,n,d,r,m> , where view is the view number, n is the message sequence number, d is the data hash, r is the node identifier, and m is the message content.

[0017] Preferably, the machine learning algorithm engine includes a survival analysis model, and the survival analysis model adopts a COx proportional hazard model, and the model formula is:

[0018] h(t∣X)=h0(t)exp(β1X1+β2X2+…+β p X p )

[0019] Among them, h(t|X) is the risk function of the individual at time t, h0(t) is the benchmark risk function, X1, X2,…, X p is a covariate,

[0020] β1,β2,…,β p is the regression coefficient.

[0021] Preferably, the patient-side application is provided with a data collection interface, which includes a symptom reporting module and a quality of life scoring module. The symptom reporting module uses a structured form to collect patient symptom information, and the fields of the structured form include symptom name, onset time, duration and severity. The quality of life scoring module uses the EORTCQLQ-C30 scale to collect patient scoring data in the form of multiple-choice questions.

[0022] Preferably, the system architecture adopts a cloud computing platform and a microservice architecture. The cloud computing platform provides elastic computing resources and distributed storage services. The microservice architecture splits the system functions into independent microservices such as data collection services, data processing services, analysis services and sharing services. Each microservice communicates through an API gateway. The API gateway adopts the RESTful API specification and the request format is JSON format.

[0023] In a second aspect, the present invention provides the following technical solution: a management method for a GIST disease database intelligent management system includes the following steps:

[0024] S1. Collect GIST-related data from hospital information systems, laboratory information systems, image archiving and communication systems, and patient-side applications through a multi-source data acquisition module. The data includes electronic medical records, test reports, imaging data, and patient-reported data.

[0025] S2. Use the standardized data model built into the standardization processing module to perform field mapping and format conversion on the collected multimodal data, and perform natural language processing on the unstructured medical record text to extract entity features;

[0026] S3. Build a GIST molecular-clinical phenotype association model using the machine learning algorithm engine of the intelligent analysis module, analyze the standardized data, and generate analysis results.

[0027] S4. Use blockchain technology in the secure sharing module to store data, and establish a cross-center data sharing channel through the federated learning framework to achieve secure data sharing.

[0028] In its third aspect, the invention provides the following technical solution: a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor; when the processor executes the computer program, the management method of the above-mentioned GIST disease database intelligent management system is implemented.

[0029] In a fourth aspect, the present invention provides the following technical solution: a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the management method of the above-mentioned GIST disease database intelligent management system.

[0030] The present invention has the following beneficial effects:

[0031] 1. In the present invention, electronic medical records, test reports, imaging data and patient-reported data are collected through the communication connection between the multi-source data acquisition module and the hospital information system, laboratory system, imaging system and patient-side application, and then multi-type data such as clinical diagnosis and treatment, test and detection, imaging characteristics and patient subjective feedback are integrated across platforms, thereby improving the problem that most traditional medical data management systems can only collect single-source or single-modal data, and due to the lack of data dimensions and insufficient sample heterogeneity, the data analysis conclusions are one-sided and cannot fully reflect the characteristics of the disease.

[0032] 2. In the present invention, the field mapping and format conversion of multimodal data are performed through the standardized data model built into the standardization processing module (such as the OMOP public data model), and the natural language processing technology is used to perform word segmentation, part-of-speech tagging and entity feature extraction on the unstructured medical record text, thereby eliminating the differences in data definition and format, and converting the disordered text into structured data, thereby improving the traditional data processing methods. Most of the problems are low efficiency in unstructured data processing, high cost of manual labeling and insufficient data standardization. Due to the lack of unified data standards, the comparability of data across centers is poor and the analysis application is limited.

[0033] 3. In the present invention, data hash storage and non-tampering verification are realized through the blockchain technology of the security sharing module (alliance chain architecture + PBFT consensus mechanism), and a cross-center "data available but invisible" sharing channel is established based on the federated learning framework, thereby ensuring the security, traceability and privacy protection of data circulation, thereby improving the traditional cross-institutional data sharing model, which mostly has the problems of insufficient security and high risk of privacy leakage. Due to the lack of a trusted distributed storage and compliance sharing mechanism, multi-center collaboration faces data circulation barriers and ethical risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is the system architecture diagram of the GIST disease database intelligent management system proposed in the present invention;

[0035] Figure 2 This is a schematic diagram of the management method steps of the GIST disease database intelligent management system proposed in the present invention. DETAILED DESCRIPTION

[0036] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0037] Example 1

[0038] Reference Figure 1 In the first embodiment of the present invention, the present invention provides an intelligent management system for a GIST disease database, including a multi-source data acquisition module, a standardization processing module, an intelligent analysis module, and a security sharing module;

[0039] The multi-source data acquisition module is connected to the hospital information system HIS, laboratory information system LIS, image archiving and communication system PACS and patient-side applications to collect electronic medical records, test reports, imaging data and data actively reported by patients;

[0040] The standardization processing module has a built-in standardized data model for field mapping and format conversion of collected multimodal data;

[0041] The intelligent analysis module includes a machine learning algorithm engine for building a GIST molecular-clinical phenotype association model. The machine learning algorithm engine includes a data fusion unit for fusing multimodal data.

[0042] The secure sharing module uses blockchain technology to achieve data storage and establishes a cross-center data sharing channel through the federated learning framework.

[0043] Specifically, the multi-source data acquisition module can integrate hospital information systems, laboratory systems, imaging systems and patient-side data across platforms, collect multiple types of data such as clinical diagnosis and treatment, testing, imaging characteristics and patient active reporting, and provide multi-dimensional data support for the database; through the standardized data model built into the standardization processing module, field mapping and format conversion of multimodal data can be performed to eliminate data definition and format differences, establish unified standards, and achieve structured integration and interoperability of multi-source data; through the machine learning algorithm engine and data fusion unit of the intelligent analysis module, multimodal data can be fused and processed to construct a GIST molecular-clinical phenotype association model, explore potential associations between data, and establish cross-dimensional feature mapping relationships; through the blockchain technology of the security sharing module Realizing data notarization can ensure that data cannot be tampered with and is traceable; establishing a cross-center data sharing channel based on the federated learning framework can achieve "data available but invisible" while protecting privacy, and provide a compliant data circulation mechanism for multi-center collaboration; through the communication connection between the multi-source data acquisition module and the hospital information system, laboratory system, imaging system and patient-side application, electronic medical records, test reports, imaging data and patient-actively reported data are collected, and then multi-type data such as clinical diagnosis and treatment, testing, imaging characteristics and patient subjective feedback are integrated across platforms, thereby improving the traditional medical data management system that can only collect single-source or single-modal data. Due to the lack of data dimensions and insufficient sample heterogeneity, the data analysis conclusions are one-sided and cannot fully reflect the characteristics of the disease.

[0044] The standardized data model uses the OMOP public data model. The standardized processing module performs natural language processing on unstructured medical record text. The natural language processing process includes the following steps: first, the text is segmented and POS tagged, and then the entity features are extracted through the BiLSTM-CRF model. The loss function of the BiLSTM-CRF model is:

[0045] Among them, n is the number of samples, T is the sequence length, and θ is the model parameter. is the input feature of the i-th sample at the t-th moment, y t For the corresponding label.

[0046] Specifically, the OMOP public data model is used as a standardized data model to unify the storage structure and semantic definition of multi-source data, so that data from different sources and formats follow the same specifications, facilitating subsequent integration and analysis. At the same time, the standardized processing module is used to perform natural language processing on unstructured medical record texts. Through word segmentation and part-of-speech tagging, the text is broken down into basic units with grammatical meaning, laying the foundation for information extraction; the BiLSTM-CRF model is used to extract entity features from the annotated text, and the bidirectional learning ability of BiLSTM is used to capture text context information. The label sequence is constrained and optimized in combination with CRF, and the model parameters are continuously adjusted by minimizing the loss function L(θ), thereby accurately identifying disease-related entities in the text. This process realizes the transformation of unstructured medical records from disordered text to structured data, greatly reducing the workload of manual annotation, improving data processing efficiency, enhancing the standardization and consistency of data, and providing high-quality data support for the analysis and application of the GIST disease database.

[0047] Multimodal data includes genomic data, imaging data and pathological data. The data fusion unit adopts the feature-level fusion strategy. The specific formula is F=f genomic ;f radiomic ;f pathomic ]

[0048] Among them, f genomic is the genomic feature vector, f radiomic is the radiomics feature vector, f pathomic is the pathology feature vector, and “;” represents the feature concatenation operation.

[0049] Specifically, the genomic feature vector (f genomic , such as gene mutations, expression profile data), imaging genomics feature vectors (f radiomic , such as tumor morphology, texture features), pathological genomics feature vector (f pathomic, such as cell morphology and histological grade), are spliced by ";" into a unified high-dimensional feature vector F; this process preserves the original feature details of each modality data, avoiding information loss caused by data conversion; single-modal data (such as only genomics) is difficult to comprehensively characterize the molecular-clinical phenotype association of GIST (such as the relationship between gene mutations and imaging features, pathological grades); feature-level fusion enables the model to simultaneously learn the association patterns of data in different dimensions by directly concatenating multi-source features (for example, whether a certain gene mutation is related to specific imaging textures or pathological mitotic figures), solving the problem of insufficient representation ability of single modality; it provides cross-modal feature input for subsequent machine learning algorithms (such as the molecular-clinical phenotype association model in claim 1 and the Cox risk model in claim 5), enabling the model to挖掘更复杂的潜在关联(如基因-影像-病理联合特征与患者预后的关系); the formula F = [f genomic ; f radiomic ; f pathomic clarifies the technical implementation method of feature splicing, ensuring direct interaction of multi-modal data at the feature level, and enhancing the depth and accuracy of data analysis.

[0050] Blockchain technology adopts a consortium chain architecture. Each node stores the data hash value, and the data is verified through a consensus mechanism when it is uploaded to the chain. The consensus mechanism adopts the Practical Byzantine Fault Tolerance algorithm. The consensus process includes three stages: pre-preparation, preparation, and submission. Nodes reach an agreement through message passing. The message format is <view,n,d,r,m>, where view is the view number, n is the message sequence number, d is the data hash, r is the node identifier, and m is the message content.

[0051] Specifically, through hash value storage, the data is converted into a unique fixed-length string. If the data is tampered with, the hash value will change, thereby ensuring data integrity. The consensus process is divided into three stages: pre-preparation, preparation, and submission. Nodes transmit information in the fixed message format <view,n,d,r,m>. The main node first broadcasts the message encapsulating the data hash value. After the slave nodes verify its legality, they confirm it. When more than 2 / 3 of the nodes' confirmations are received, the data is successfully uploaded to the chain. Even in the case of no more than 1 / 3 malicious nodes, the consensus can still be ensured to be correct. The consortium chain only allows authenticated nodes to join, and combined with hash storage, it realizes "data available but invisible", meeting the privacy requirements of medical data. At the same time, the data uploading operation undergoes network-wide consensus, and the records are permanently retained, enabling traceability of abnormal behaviors and enhancing the security and credibility of the system.

[0052] The machine learning algorithm engine includes a survival analysis model. The survival analysis model adopts the COx proportional hazards model, and the model formula is:

[0053] h(t∣X) = h0(t)exp(β1X1 + β2X2 + … + β p X It should be noted that the Chinese phrase "挖掘更复杂的潜在关联" in the original text seems to be an incomplete expression. It might be better to clarify its specific meaning in the context for a more accurate translation. Here, I translated it as "挖掘更复杂的潜在关联" as literally as possible.p )

[0054] Among them, h(t|X) is the risk function of the individual at time t, h0(t) is the benchmark risk function, X1, X2,…, X p is a covariate,

[0055] β1,β2,…,β p is the regression coefficient.

[0056] Specifically, the model takes into account the multidimensional covariates of GIST patients, such as age, gender, tumor size, pathological grade, gene mutation type, treatment method, etc. (i.e. X1, X2…, X p ) into the risk function h(t|X) can comprehensively consider the impact of multiple factors on patient survival; combined with the baseline risk function h0(t), through the regression coefficients β1, β2,…, β p , quantifying the strength of the association between each covariate and the patient's survival risk. A positive regression coefficient indicates that the covariate increases the patient's risk; a negative regression coefficient indicates that the covariate reduces the risk, thus clearly demonstrating the direction and extent of the impact of different factors on survival risk. Using time t as the variable, based on the included covariate data, the risk function h(t|X) for individuals at different time points is calculated, enabling dynamic prediction of patient survival prognosis. This helps clinicians understand the patient's risk profile at different stages of the disease and provides data support for developing personalized treatment plans and follow-up plans. Through model analysis, the key factors affecting GIST patient survival are identified, assisting clinicians in judging disease progression, selecting more appropriate treatment strategies, and evaluating treatment efficacy.

[0057] The patient-side application is equipped with a data collection interface, which includes a symptom reporting module and a quality of life scoring module. The symptom reporting module uses a structured form to collect patient symptom information. The fields of the structured form include symptom name, onset time, duration and severity. The quality of life scoring module uses the EORTCQLQ-C30 scale to collect patient scoring data in the form of multiple-choice questions.

[0058] Specifically, through fixed fields such as symptom name, onset time, duration, and severity, the patient's subjective description is converted into quantifiable structured data, solving the problem that traditional text records are difficult to analyze and providing standardized input for the association study between clinical symptoms and molecular characteristics; the EORTCQLQ-C30 scale is used to collect scoring data in the form of multiple-choice questions, following internationally accepted assessment standards to ensure the standardization and cross-center comparability of quality of life data, and make up for the lack of coverage of patients' subjective experience by objective diagnosis and treatment data; through direct data collection from the patient side, the data source dimension is expanded, and real-time symptom dynamics and quality of life information outside of clinical diagnosis and treatment records are supplemented to build a more complete panoramic data set for GIST patients to meet the needs of precision medicine for "patient-centered" data.

[0059] The system architecture adopts cloud computing platform and microservice architecture. The cloud computing platform provides elastic computing resources and distributed storage services. The microservice architecture splits the system functions into independent microservices such as data collection service, data processing service, analysis service and sharing service. Each microservice communicates through the API gateway. The API gateway adopts the RESTful API specification and the request format is JSON format.

[0060] Specifically, the elastic computing and distributed storage capabilities of the cloud computing platform are utilized to dynamically adjust resource allocation according to data processing and analysis needs, avoiding the fixed resource bottlenecks of traditional architectures and improving the system's flexibility and cost-effectiveness in responding to data volume fluctuations; the microservice architecture splits system functions into independently deployed service units (such as data collection, processing, analysis, and sharing), and each service communicates through the API gateway using the RESTful specification to achieve a highly cohesive and low-coupling system design, improving development efficiency and maintainability; the API gateway uses the RESTful specification and JSON format to ensure standardization and statelessness of cross-service communication, support multi-language and multi-platform access, enhance system scalability and integration capabilities, and provide a technical foundation for cross-institutional collaboration; the deployment model of independent microservices supports fault isolation and rapid recovery, and combined with the load balancing mechanism of cloud computing, it improves the overall availability of the system and ensures the continuity of medical data processing and analysis.

[0061] Example 2:

[0062] Reference Figure 2 In a second embodiment of the present invention, the present invention provides a management method for a GIST disease database intelligent management system, comprising the following steps:

[0063] S1. Collect GIST-related data from hospital information systems, laboratory information systems, image archiving and communication systems, and patient-side applications through a multi-source data acquisition module. The data includes electronic medical records, test reports, imaging data, and patient-reported data.

[0064] S2. Use the standardized data model built into the standardization processing module to perform field mapping and format conversion on the collected multimodal data, and perform natural language processing on the unstructured medical record text to extract entity features;

[0065] S3. Build a GIST molecular-clinical phenotype association model using the machine learning algorithm engine of the intelligent analysis module, analyze the standardized data, and generate analysis results.

[0066] S4. Use blockchain technology in the secure sharing module to store data, and establish a cross-center data sharing channel through the federated learning framework to achieve secure data sharing.

[0067] Specifically, the S1: Multi-source Data Acquisition module collects GIST-related data from multiple systems and the patient side, enabling cross-platform, multi-dimensional data integration. By collecting electronic medical records, test reports, and imaging data from hospital information systems (HIS), laboratory information systems (LIS), and picture archiving and communication systems (PACS), objective data from the clinical diagnosis and treatment process can be obtained. By collecting proactively reported data through patient-side applications, information such as patients' subjective symptoms and quality of life can be supplemented. This can build a complete data set covering clinical diagnosis and treatment, test results, imaging features, and patients' subjective feedback, addressing the issues of insufficient sample heterogeneity and missing data dimensions from a single data source, and providing a comprehensive, multimodal data foundation for GIST-specific disease research. The S2: Standardization Processing module performs field mapping, format conversion, and natural language processing on multimodal data, achieving the following goals: First, using standardized data models for field mapping and format conversion can eliminate differences in the definition and storage format of data from different sources, integrating multimodal data into unified and standardized structured data, resolving compatibility issues caused by inconsistent data standards, and improving data interoperability. Secondly, for unstructured medical record text, natural language processing technology is used for word segmentation and part-of-speech tagging, and entity features are extracted through the BiLSTM-CRF model. This can convert text into structured information, avoiding the high cost and low efficiency of manual tagging, enhancing data analyzability, and providing high-quality, standardized data support for the subsequent intelligent analysis module to build association models, ensuring the accuracy and reliability of data analysis results. S3: The machine learning algorithm engine of the intelligent analysis module analyzes standardized data and constructs a GIST molecular-clinical phenotype association model, enabling in-depth mining and association analysis of multi-dimensional data. The data fusion unit integrates multimodal data features such as genomics, imaging genomics, and pathology genomics, and combines algorithms such as the Cox proportional hazards model to quantitatively analyze the association between molecular-level characteristics (such as gene mutations) and clinical phenotypes (such as symptoms and prognosis). Through model building and data analysis, we can identify key factors affecting the occurrence, development, and prognosis of GIST, predict patient survival outcomes, and provide a data-driven decision-making basis for disease mechanism research and the formulation of precise diagnosis and treatment plans, solving the problem that traditional analysis methods are difficult to reveal complex correlation mechanisms; through the S4: Security Sharing Module, which uses blockchain technology and a federated learning framework, we can achieve full data lifecycle security management and cross-institutional collaboration.Blockchain technology is based on the alliance chain architecture and practical Byzantine fault-tolerant algorithm. It ensures that data cannot be tampered with through hash value storage, and uses the consensus mechanism to achieve multi-node data consistency verification to ensure data authenticity and integrity. The federated learning framework establishes a cross-institutional data sharing channel through encrypted parameter exchange and collaborative modeling without leaking the original data, realizing "data does not move, model moves", which not only meets the requirements of medical data privacy protection, but also breaks through the limitations of data islands, providing safe and reliable technical support for multi-center joint research and collaborative analysis of clinical data, and improving data utilization efficiency and scientific research collaboration capabilities.

[0068] Example 3

[0069] The third embodiment of the present invention is based on the same inventive concept and proposes a computer-readable storage medium. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, it implements the steps of the management method of the GIST disease database intelligent management system of the above embodiment.

[0070] Example 4

[0071] The fourth embodiment of the present invention is based on the same inventive concept. The present invention proposes a computer device, including: a processor and a memory; the processor and the memory communicate with each other; the memory is used to store instructions; the processor is used to execute the instructions in the memory, and execute the management method of the GIST disease database intelligent management system of the above embodiment.

[0072] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0073] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. GIST disease database intelligent management system, characterized by: It includes multi-source data acquisition module, standardization processing module, intelligent analysis module and security sharing module; The multi-source data acquisition module is connected to the hospital information system HIS, laboratory information system LIS, image archiving and communication system PACS and patient-side applications respectively, and is used to collect electronic medical records, test reports, imaging data and patient-reported data; The standardization processing module has a built-in standardized data model for performing field mapping and format conversion on the collected multimodal data; The intelligent analysis module includes a machine learning algorithm engine for constructing a GIST molecular-clinical phenotype association model, and the machine learning algorithm engine includes a data fusion unit for fusing multimodal data; The secure sharing module uses blockchain technology to implement data storage and establishes a cross-center data sharing channel through a federated learning framework.

2. The GIST disease database intelligent management system according to claim 1, characterized in that: The standardized data model adopts the OMOP public data model. The standardized processing module performs natural language processing on the unstructured medical record text. The natural language processing process includes the following steps: first, the text is segmented and part-of-speech tagged, and then entity features are extracted through the BiLSTM-CRF model. The loss function of the BiLSTM-CRF model is: Among them, n is the number of samples, T is the sequence length, and θ is the model parameter. is the input feature of the i-th sample at the t-th moment, y t For the corresponding label.

3. The GIST disease database intelligent management system according to claim 1, characterized in that: The multimodal data includes genomic data, imaging data and pathological data. The data fusion unit adopts a feature-level fusion strategy. The specific formula is F=[f genomic ;f radiomic ;f pathomic ] Among them, f genomic is the genomic feature vector, f radiomic is the radiomics feature vector, f pathomic is the pathology feature vector, and ";" represents the feature concatenation operation.

4. The GIST disease database intelligent management system according to claim 1, characterized in that: The blockchain technology adopts a consortium chain architecture. Each node stores the data hash value. The data is verified through a consensus mechanism when it is uploaded to the chain. The consensus mechanism adopts a practical Byzantine fault-tolerant algorithm. The consensus process includes three stages: pre-preparation, preparation, and submission. The nodes reach consensus through message transmission. The message format is<view,n,d,r,m> , where view is the view number, n is the message sequence number, d is the data hash, r is the node identifier, and m is the message content.

5. The GIST disease database intelligent management system according to claim 1, characterized in that: The machine learning algorithm engine includes a survival analysis model, which adopts the COx proportional hazard model. The model formula is: h(t∣X)=h0(t)exp(β1X1+β2X2+…+β p X p ) Among them, h(t|X) is the risk function of the individual at time t, h0(t) is the benchmark risk function, X1, X2,…, X p is a covariate, β1,β2,…,β p is the regression coefficient.

6. The GIST disease database intelligent management system according to claim 1, characterized in that: The patient-side application is provided with a data collection interface, which includes a symptom reporting module and a quality of life scoring module. The symptom reporting module uses a structured form to collect patient symptom information. The fields of the structured form include symptom name, onset time, duration and severity. The quality of life scoring module uses the EORTCQLQ-C30 scale to collect patient scoring data in the form of multiple-choice questions.

7. The GIST disease database intelligent management system according to claim 1, characterized in that: The system architecture adopts a cloud computing platform and a microservice architecture. The cloud computing platform provides elastic computing resources and distributed storage services. The microservice architecture splits the system functions into independent microservices such as data collection services, data processing services, analysis services, and shared services. Each microservice communicates through an API gateway. The API gateway adopts the RESTful API specification, and the request format is JSON.

8. A management method for the intelligent management system of the GIST disease database, characterized in that: Using the GIST disease database intelligent management system according to any one of claims 1 to 7 comprises the following steps: S1. Collect GIST-related data from hospital information systems, laboratory information systems, image archiving and communication systems, and patient-side applications through a multi-source data acquisition module. The data includes electronic medical records, test reports, imaging data, and patient-reported data. S2. Use the standardized data model built into the standardization processing module to perform field mapping and format conversion on the collected multimodal data, and perform natural language processing on the unstructured medical record text to extract entity features; S3. Build a GIST molecular-clinical phenotype association model using the machine learning algorithm engine of the intelligent analysis module, analyze the standardized data, and generate analysis results. S4. Use blockchain technology in the secure sharing module to store data, and establish a cross-center data sharing channel through the federated learning framework to achieve secure data sharing.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the management method of the GIST disease database intelligent management system according to claim 8 is implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by the processor, it implements the management method of the GIST disease database intelligent management system according to claim 8.

Images