Search

CN-122027289-A - Data security privacy protection method for intelligent medical engineering

CN122027289ACN 122027289 ACN122027289 ACN 122027289ACN-122027289-A

Abstract

The invention relates to the technical field of medical engineering data security, in particular to a data security privacy protection method for intelligent medical engineering, which comprises the following steps of acquiring medical data and communication context information when data transmission requirements exist, wherein the information comprises data types, service scenes, algorithm processing stages and historical communication information; based on data and context information, risk factors such as data sensitivity, algorithm distinguishability, historical data association degree, model dependence degree and the like are selected, privacy risk levels are evaluated by adopting differentiated weights, communication granularity strategies are determined according to the risk levels, data dimensionality is adjusted, differential privacy disturbance is performed on the adjusted data, target algorithm verification performance is input, strategies or disturbance modes are adjusted according to results, and finally safe transmission is performed. According to the method, the medical data privacy risk quantification and control are realized through risk assessment, and the problem that privacy is difficult to assess and manage under the traditional unified transmission strategy is solved.

Inventors

  • CHEN XIAOFEI

Assignees

  • 中国人民解放军联勤保障部队第九八〇医院

Dates

Publication Date
20260512
Application Date
20260212

Claims (8)

  1. 1. The data security privacy protection method for intelligent medical engineering is characterized by comprising the following steps of: s1, when data transmission requirements exist, acquiring medical related data to be communicated, and acquiring communication context information, wherein the communication context information at least comprises a communication data type, a service scene, an algorithm processing stage and historical communication information; S2, based on the medical related data and the communication context information, selecting risk factors at least comprising the sensitivity degree of the communication data, the algorithm distinguishing property, the correlation degree of the historical communication data and the dependence degree of an algorithm model, and evaluating by adopting a differentiated weight distribution rule aiming at different types of medical data to obtain a privacy risk level result; S3, determining a corresponding communication granularity strategy according to the privacy risk level result, and adjusting the dimension or the effective dimension proportion of the medical related data according to the strategy; s4, executing corresponding differential privacy disturbance processing on the medical related data processed by the S3 according to the privacy risk level result; S5, inputting the medical related data processed by the S3 and the S4 into a target medical intelligent algorithm for performance verification, and adjusting an S3 communication granularity strategy or an S4 disturbance processing mode according to the result; S6, performing safe communication transmission on the medical related data adjusted in the S5.
  2. 2. The method for protecting data security privacy for intelligent medical engineering according to claim 1, wherein the step of acquiring the medical related data to be communicated comprises the steps of: under the current business scene, determining a medical related data source corresponding to the data transmission requirement; invoking, receiving or reading medically relevant data from the determined data source; and screening the acquired medical related data, and reserving data content related to the current data transmission requirement to form medical related data to be communicated.
  3. 3. The method for protecting data security privacy for intelligent medical engineering according to claim 1, wherein the evaluating with the differential weight distribution rule for different types of medical data comprises the steps of: s21, based on the communication data type in the communication context information, performing type judgment on the medical related data, and determining the medical data type corresponding to the medical related data; S22, selecting a preset weight distribution rule corresponding to the medical data type according to the medical data type determined in the S21, wherein the weight distribution rule is used for respectively giving weights to the sensitivity degree of communication data, the algorithm distinguishability, the correlation degree of historical communication data and the dependence degree of an algorithm model; S23, respectively carrying out quantization and value taking on the sensitivity degree of the communication data, the algorithm distinguishing property, the correlation degree of the historical communication data and the dependence degree of the algorithm model; S24, weighting calculation is carried out on the quantized values of the risk factors obtained in the S23 according to the weight distribution rule selected in the S22, so that a comprehensive evaluation result is obtained; And S25, matching the comprehensive evaluation result with a preset risk level classification rule, and determining a corresponding privacy risk level result.
  4. 4. The method for protecting data security privacy for intelligent medical engineering according to claim 1, wherein the determining the corresponding communication granularity policy comprises the steps of: acquiring a privacy risk level result obtained by privacy risk assessment; Matching the privacy risk level result with a preset communication granularity mapping rule; and determining a communication granularity strategy corresponding to the privacy risk level result according to the matching result.
  5. 5. A data security privacy protection method for intelligent medical engineering according to claim 1, wherein said adjusting the dimension or the effective dimension ratio of the medically relevant data comprises the steps of: Acquiring an original data dimension set of the medical related data; determining a target dimension set participating in communication in the original data dimension set according to the communication granularity policy; And carrying out dimension reconstruction on the medical related data according to the target dimension set to form adjusted medical related data.
  6. 6. The method for protecting data security privacy for intelligent medical engineering according to claim 1, wherein the performing the corresponding differential privacy perturbation process comprises the steps of: acquiring a privacy risk level result obtained by privacy risk assessment; Determining a differential privacy disturbance rule corresponding to the privacy risk level according to the privacy risk level result; determining perturbation parameters for the current medical-related data based on the differential privacy perturbation rules; Generating random disturbance data according to the disturbance parameters; And (3) the generated random disturbance data is acted on the medical related data processed by the S3 to form medical related data processed by the differential privacy disturbance.
  7. 7. The data security privacy protection method for intelligent medical engineering according to claim 1, wherein the input target medical intelligent algorithm performance verification comprises the following steps: after S3 and S4 are completed, determining an input data form corresponding to the target medical intelligent algorithm; inputting the medical related data processed by the S3 and the S4 into a target medical intelligent algorithm according to the input data form, and executing algorithm processing; obtaining result data output after the algorithm processing; and extracting algorithm output index data for performance verification from the result data.
  8. 8. The method for protecting data security and privacy for intelligent medical engineering according to claim 1, wherein the step of adjusting the S3 communication granularity policy or the S4 perturbation processing mode according to the result comprises the following steps: acquiring algorithm output index data obtained by performance verification in S5; comparing the algorithm output index data with preset judging conditions; When the comparison result does not meet the judging condition, determining to adjust the communication granularity strategy in S3 or the differential privacy disturbance processing mode in S4; and updating the communication granularity strategy or the differential privacy disturbance processing mode according to the determined adjustment mode.

Description

Data security privacy protection method for intelligent medical engineering Technical Field The invention relates to the technical field of medical engineering data security, in particular to a data security privacy protection method for intelligent medical engineering. Background Medical data typically needs to be collected, stored, and transferred between hospitals, clinics, mobile terminals, edge computing nodes, and central servers or multi-architecture collaboration platforms. The data comprise electronic medical records, medical images, genome information, physiological monitoring data, laboratory detection data and the like of patients, and relate to sensitive information such as personal health conditions, medical history information, genetic characteristics, real-time vital signs and the like, so that the safety requirement is extremely high. With the development of telemedicine, intelligent diagnosis and treatment and personalized medical services, the demands of cross-institution sharing and real-time analysis of medical data are continuously increased, so that the data transmission quantity is greatly increased, and meanwhile, the leakage risk of the data in the transmission and processing processes is also increased. The traditional intelligent medical data transmission mostly adopts a unified data transmission strategy, and the privacy risk is difficult to quantify and control because the data type and the algorithm processing background are not considered. Disclosure of Invention In order to make up for the defects, the invention provides a data security privacy protection method for intelligent medical engineering, which aims to solve the problem that privacy risks are difficult to quantify and control because data types and algorithm processing backgrounds are not considered in the conventional intelligent medical data transmission mostly adopting a unified data transmission strategy. The invention provides a data security privacy protection method for intelligent medical engineering, which comprises the following steps: s1, when data transmission requirements exist, acquiring medical related data to be communicated, and acquiring communication context information, wherein the communication context information at least comprises a communication data type, a service scene, an algorithm processing stage and historical communication information; S2, based on the medical related data and the communication context information, selecting risk factors at least comprising the sensitivity degree of the communication data, the algorithm distinguishing property, the correlation degree of the historical communication data and the dependence degree of an algorithm model, and evaluating by adopting a differentiated weight distribution rule aiming at different types of medical data to obtain a privacy risk level result; S3, determining a corresponding communication granularity strategy according to the privacy risk level result, and adjusting the dimension or the effective dimension proportion of the medical related data according to the strategy; s4, executing corresponding differential privacy disturbance processing on the medical related data processed by the S3 according to the privacy risk level result; S5, inputting the medical related data processed by the S3 and the S4 into a target medical intelligent algorithm for performance verification, and adjusting an S3 communication granularity strategy or an S4 disturbance processing mode according to the result; S6, performing safe communication transmission on the medical related data adjusted in the S5. By adopting the technical scheme, the related medical data and the communication context information thereof are acquired when the data transmission requirement exists, and the privacy risk level is further evaluated based on the risk factors, so that the problem that the privacy risk is difficult to quantify and control because the data type and the algorithm processing background are not considered because the unified data transmission strategy is adopted in the traditional intelligent medical data transmission is solved. Further, the acquiring the medically relevant data to be communicated comprises the steps of: under the current business scene, determining a medical related data source corresponding to the data transmission requirement; invoking, receiving or reading medically relevant data from the determined data source; and screening the acquired medical related data, and reserving data content related to the current data transmission requirement to form medical related data to be communicated. Further, the evaluating with the differential weight distribution rule for different types of medical data comprises the steps of: s21, based on the communication data type in the communication context information, performing type judgment on the medical related data, and determining the medical data type corresponding to the medical related data; S22, s