Information security management method and system based on digital medical treatment

Abstract

The invention relates to the technical field of information security management, and discloses an information security management method and system based on digital medical treatment. The method comprises the following steps: verifying doctor professional qualification, patient authorization range, data sensitivity level and time permission validity in a multi-institution medical data access request, and generating a permission verification result; analyzing ontology semantic attributes and sensitivity characteristics of medical search keywords input by a user, and determining a corresponding hierarchical encryption strategy based on the permission verification result; performing homomorphic encryption transformation on the medical term synonym set according to the hierarchical encryption strategy, and constructing a medical ciphertext database; and receiving a medical query request, executing symptom disease association matching and ciphertext state logical reasoning operation in the medical ciphertext database, and outputting an encrypted state search result. According to the method, the privacy leakage risk caused by search behavior trajectory analysis is effectively prevented, and the security and availability of medical data are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information security management, and particularly relates to an information security management method and system based on digital medical treatment. BACKGROUND

[0002] Various medical institutions such as hospitals, clinics and research institutions need to frequently query and share sensitive medical data such as patient medical records, examination reports and diagnosis and treatment plans among each other in order to support application scenarios such as remote medical treatment, collaborative diagnosis and medical research. However, the existing medical data security management method has significant technical defects and security risks in a multi-institution environment. Traditional permission control mechanisms mostly use simple two-dimensional or three-dimensional permission models, which cannot accurately distinguish the professional qualifications of different doctors, the individualized authorization preferences of patients and the sensitivity differences of data, resulting in problems such as rough permission management and inaccurate access control. SUMMARY

[0003] The main purpose of the present application is to provide an information security management method and system based on digital medical treatment, which effectively prevents the privacy leakage risk brought by search behavior trajectory analysis and improves the security and usability of medical data.

[0004] To achieve the above-mentioned purpose, the present application provides an information security management method based on digital medical treatment, comprising the following steps: verifying the professional qualifications of doctors, the authorization range of patients, the data sensitivity level and the time permission validity in the multi-institution medical data access request, and generating a permission verification result; analyzing the ontological semantic attributes and sensitivity features of the medical search keywords input by the user, and determining the corresponding hierarchical encryption strategy based on the permission verification result; performing homomorphic encryption transformation on the medical term synonym set according to the hierarchical encryption strategy, and constructing a medical ciphertext database; receiving a medical query request and performing symptom-disease association matching and ciphertext state logical reasoning operation in the medical ciphertext database, and outputting an encrypted state search result.

[0005] In combination with the first aspect, in a first implementation manner of the first aspect of the present application, the step of verifying the professional qualifications of doctors, the authorization range of patients, the data sensitivity level and the time permission validity in the multi-institution medical data access request, and generating a permission verification result, comprises: parsing the doctor qualification certificate and professional skill requirement in the multi-institution medical data access request, and determining the professional qualifications of doctors according to the doctor qualification certificate and the professional skill requirement; reading the privacy preferences and data category authorization set by the patient for the professional qualifications of doctors to obtain the authorization range of patients and the data sensitivity level; computing a time interval between the current access time and the time when the permission is initially granted, and performing a time permission validity analysis on the time interval according to the patient authorization range, to obtain a time permission validity; generating a permission verification result by comprehensively considering the doctor professional qualification, the patient authorization range, the data sensitivity level, and the time permission validity.

[0006] In a second implementation manner of the first aspect, the method further includes: inputting the medical search keyword input by the user into a medical ontology knowledge graph to perform part-of-speech tagging and named entity recognition, and generating an ontology semantic attribute; computing a semantic vector cosine similarity between each medical entity in the ontology semantic attribute, and determining a semantic correlation score by combining a medical field weight coefficient and the semantic vector cosine similarity; querying a preset sensitivity label of a corresponding entity in the medical ontology according to the semantic correlation score, and identifying a sensitivity feature; matching the data sensitivity level in the permission verification result with the sensitivity feature, and determining a hierarchical encryption strategy.

[0007] In a third implementation manner of the first aspect, the matching the data sensitivity level in the permission verification result with the sensitivity feature, and determining a hierarchical encryption strategy, includes: comparing the data sensitivity level in the permission verification result with a sensitivity value of the sensitivity feature, and dividing the sensitivity feature into a high sensitivity level, a medium sensitivity level, and a low sensitivity level; matching an encryption algorithm type for the high sensitivity level as a homomorphic encryption mode, an encryption algorithm type for the medium sensitivity level as a symmetric encryption mode, and an encryption algorithm type for the low sensitivity level as a hash processing mode; reading parameter configuration information corresponding to the encryption algorithm type, and dynamically adjusting an encryption parameter value in the parameter configuration information by combining a medical data type and an access frequency, to generate an encryption configuration scheme; integrating the sensitivity level classification result, the encryption algorithm type, and the encryption configuration scheme to form the hierarchical encryption strategy.

[0008] In a fourth implementation manner of the first aspect, the performing homomorphic encryption transformation on the medical term synonym set according to the hierarchical encryption strategy, and constructing a medical ciphertext database, includes: read the medical search keywords in the hierarchical encryption strategy, and retrieve semantic equivalent words in the disease ontology, symptom ontology, drug ontology and examination ontology based on the medical search keywords to generate a medical term synonym set; XOR operation is performed on the synonym hash value, semantic vector and medical field noise of each synonym in the medical term synonym set by using the Paillier homomorphic encryption algorithm to generate synonym ciphertexts; An index table of ciphertext semantic association between the synonym ciphertexts is established, and the ciphertext data is stored in a partitioned manner according to disease categories, symptom groups, drug classifications and examination items to form a hierarchical ciphertext data structure; The hierarchical ciphertext data structure and the index table of ciphertext semantic association are integrated and configured to construct a medical ciphertext database.

[0009] In combination with the first aspect, in a fifth implementation manner of the first aspect of the present application, the medical query request is received and the symptom-disease association matching and ciphertext state logical inference operation are performed in the medical ciphertext database to output an encrypted state search result, which includes: The symptom keyword combination in the received medical query request is parsed, and the symptom keyword combination is converted into a symptom ciphertext query instruction; The disease-symptom association matrix pre-constructed in the medical ciphertext database is retrieved according to the symptom ciphertext query instruction, and a disease association degree ciphertext score is calculated based on the disease-symptom association matrix; The medical logical operation of the ciphertext state is performed based on the disease association degree ciphertext score to obtain a logical inference matching result; The logical inference matching result and the timestamp ciphertext information are integrated to form an encrypted state search result.

[0010] In combination with the first aspect, in a sixth implementation manner of the first aspect of the present application, the medical logical operation of the ciphertext state is performed based on the disease association degree ciphertext score to obtain a logical inference matching result, which includes: The logical connection words and operators between symptom entities are found based on the disease association degree ciphertext score, and the ciphertext logical operation instruction that needs to be executed is determined according to the logical connection words and the operators; The corresponding homomorphic calculation function is called according to the ciphertext logical operation instruction, and the logical operation ciphertext result of the multi-symptom combination is calculated based on the homomorphic calculation function; The medical term fuzzy matching is performed based on the logical operation ciphertext result to obtain a fuzzy matching correction score, and the logical inference matching result is generated by fusing the logical operation ciphertext result and the fuzzy matching correction score.

[0011] In a seventh implementation form of the first aspect, the information security management method based on digital medical treatment further includes: record the behavior characteristic parameters corresponding to the encrypted state search results, and generate an access behavior track according to the behavior characteristic parameters; based on the word frequency distribution and time pattern characteristics in the access behavior track, randomly select a combination of terms in different medical specialty fields and add time offset and noise parameters to construct a false query sequence; extract the access time stamp in the access behavior track, generate a random delay value using Laplace distribution and superimpose it on the access time stamp, and simultaneously perform Markov chain randomization processing on the time stamp of the false query sequence to obtain a set of time-ambiguous access time stamps; end-to-end encrypt the encrypted state search results and perform XOR operation with the institution identity mask to obtain encrypted search data, package and encapsulate the encrypted search data, the false query sequence and the set of time-ambiguous access time stamps to generate a privacy protection data transmission package; monitor the overall privacy loss value of the privacy protection data transmission package and detect abnormal access patterns to output a security management result.

[0012] In an eighth implementation form of the first aspect, the monitoring of the overall privacy loss value of the privacy protection data transmission package and the detection of abnormal access patterns to output a security management result include: parse the query operation records and sensitivity parameters in the privacy protection data transmission package, and based on the cumulative calculation of the query operation records and the sensitivity parameters, obtain the overall privacy loss value; monitor the access behavior data corresponding to the overall privacy loss value, and generate an abnormal access pattern according to the access behavior data; according to the abnormal access pattern, call a dynamic security policy adjustment mechanism to modify the encryption strength parameter, the false query injection ratio and the privacy noise coefficient, and update the security protection configuration parameter; integrate the overall privacy loss value, the abnormal access pattern and the security protection configuration parameter to output a security management result.

[0013] The application further provides an information security management system based on digital medical treatment, comprising: a verification module configured to verify the professional qualifications of doctors, the authorized range of patients, the data sensitivity level and the time permission validity in a medical data access request of multiple institutions, and generate a permission verification result; an analysis module configured to analyze the ontological semantic attributes and sensitivity characteristics of medical search keywords input by a user, and determine a corresponding hierarchical encryption strategy based on the permission verification result. A homomorphic encryption transformation module is configured to perform homomorphic encryption transformation on a medical term synonym set according to the hierarchical encryption strategy, and construct a medical ciphertext database; An output module is configured to receive a medical query request and perform symptom-disease association matching and ciphertext state logical inference operation in the medical ciphertext database, and output an encrypted state search result.

[0014] To sum up, the technical scheme provided by the present application realizes accurate management of the professional qualifications of doctors, the authorized range of patients, the data sensitivity level and the time permission by constructing a four-dimensional fine-grained permission control matrix, which significantly improves the access control accuracy in a multi-institutional environment compared with traditional permission control. The adaptive hierarchical encryption strategy based on the medical ontology knowledge graph can dynamically select an encryption scheme according to the semantic features of medical terms, overcoming the limitations of uniform encryption methods. The medical semantic-aware homomorphic encryption technology enables the system to perform symptom-disease association matching and logical inference in a ciphertext state, solving the technical bottleneck that traditional encrypted search cannot handle medical professional knowledge. Through multi-dimensional privacy protection mechanisms such as false query injection and timestamp obfuscation, the privacy leakage risk caused by search behavior trajectory analysis is effectively prevented. The differential privacy analysis and dynamic security strategy adjustment mechanism realizes adaptive security situation awareness, which can adjust the protection parameters in real time according to the threat level. The multi-institutional threat intelligence sharing mechanism realizes collaborative security protection under the premise of protecting the privacy of each institution, and overall improves the security and usability of medical data. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is a step schematic diagram of the information security management method based on digital medical treatment in an embodiment of the present application; Figure 2 is a structural block diagram of the information security management system based on digital medical treatment in an embodiment of the present application.

[0016] The implementation, functional features and advantages of the present application will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical scheme and advantages of the present application more clear, the present application will be further described in detail below in conjunction with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0018] With reference to Figure 1 The present embodiment provides an information security management method based on digital medical treatment, comprising the following steps: S1, verifying the professional qualifications of doctors, the authorized range of patients, the data sensitivity level and the time permission in the multi-institutional medical data access request, and generating a permission verification result; Among them, the multi-agency medical data access request is structured and parsed, and the doctor's identity, professional qualification certificate and the professional field requirements of the target medical data related to this request contained therein are extracted. Through the doctor professional matching algorithm, the professional certification items, practice scope and title level in the doctor's qualification certificate are vectorized modeling, and combined with the professional skill requirements set by the department or disease category of the target data, the professional matching degree between the doctor and the target data is calculated to determine whether the doctor has the professional qualification required by this request. Access the patient's privacy authorization record set in advance, read the authorized content matched with the doctor's professional qualification, including the data category allowed to be accessed by the patient in the professional field, the corresponding privacy preference level and the authorized effective time range, and perform semantic deconstruction on these information to obtain the patient's authorization range and the data sensitivity level of the requested data under its authorization system. Get the timestamp of the current request, and compare it with the initial authorization time, combine the time dimension rules set in the patient's authorization, calculate and analyze the time interval through the time decay function model to judge whether the current access behavior is still within the authorized effective period, and obtain the validity judgment result of the time permission. Comprehensive the above four types of information, that is, the professional qualification recognition result of the doctor, the content of the patient's corresponding authorization range, the sensitive level classification result of the requested data and the permission validity state of the access time, through the permission decision engine, the weight weighted evaluation is carried out to form the permission verification result.

[0019] S2, analyze the ontology semantic attributes and sensitivity characteristics of the medical search keywords input by the user, and determine the corresponding hierarchical encryption strategy based on the permission verification result; Specifically, the user input medical search keywords are input into the medical ontology knowledge graph processing module, the part-of-speech tagging and named entity recognition operations of the keywords are completed through the integrated natural language processing component, so as to perform semantic classification on each word item in the input content, and identify whether it belongs to a specific medical entity such as disease name, symptom description, drug item or examination index, and generate an ontology semantic attribute set containing semantic type, hyponym-hypernym relationship and semantic position. A semantic vector calculation model is called to encode multiple medical entities in the ontology semantic attribute, convert them into multi-dimensional vector representation, and calculate the similarity between entities by using the cosine similarity function combined with the medical field special weight coefficient, to obtain a set of semantic correlation scores with field weight adjustment factors, to reflect the internal contact strength of each entity in the medical semantic space. The semantic correlation score indexes the preset sensitivity label in the medical ontology, identifies the sensitivity level of all entities semantically adjacent to the search keywords, and integrates and adjusts the context weight of the sensitivity level to form a sensitivity feature judgment result. The sensitivity feature is matched and compared with the data sensitivity level in the permission verification result, the corresponding relationship between the subject permission level and the data restriction degree in the access request is combined, and the adaptive hierarchical encryption strategy is selected and determined, including whether to enable homomorphic encryption, symmetric encryption or hash desensitization and other encryption strength settings.

[0020] S3, performing homomorphic encryption transformation on the medical term synonym set according to the hierarchical encryption strategy to construct a medical ciphertext database; It should be noted that the medical search keywords are extracted according to the hierarchical encryption strategy, and based on the medical search keywords, the ontology expansion retrieval is performed in the pre-constructed four core semantic networks of disease ontology, symptom ontology, drug ontology and examination ontology. Using semantic inheritance relationship, alias matching rule and entity upper and lower structure, all terms with equivalent meaning or clinical use synonym are searched to construct a medical term synonym set containing official terms, colloquial names, abbreviations and cross-language expressions. Each synonym in the medical term synonym set is semantically coded, the word is hashed to extract its word structure feature identifier, and based on the medical special semantic vector model such as Med2Vec, its multi-dimensional vector expression is obtained, and then combined with the Laplace distribution random noise value sampled from the medical semantic disturbance model, the data block structure for encryption is constructed, and the Paillier homomorphic encryption algorithm is called. The hash value, semantic vector and noise value are combined by exclusive or operation and then encrypted to form a synonym ciphertext representation with operationality. The semantic similarity between the ciphertext semantic vectors is calculated based on the relative position relationship, and these relationships are recorded in the ciphertext semantic association index table in the form of ciphertext. The matching degree between each group of synonyms in the encryption domain is recorded, and all ciphertexts are logically divided according to the multi-dimensional semantic labels such as disease category, symptom group, drug classification and examination item to form a hierarchical ciphertext data structure, which realizes the partition storage in physical storage. The hierarchical ciphertext data structure and the ciphertext semantic association index table are integrated to establish the data configuration mapping and access structure, and the medical ciphertext database is formed.

[0021] S4, receiving a medical query request and performing symptom-disease association matching and ciphertext state logical reasoning operation in the medical ciphertext database, and outputting an encrypted state search result.

[0022] Specifically, the symptom keyword combination in the received medical query request is analyzed, the composite symptoms, concurrent manifestations and descriptive phrases involved in the query are standardized by a medical ontology knowledge graph and a named entity recognition model, and are restored to a structured medical symptom entity set. The homomorphic encryption module is called to map these symptom entities to ciphertext representations respectively, and a symptom ciphertext query instruction with semantic equivalence is generated. According to the symptom ciphertext query instruction, the corresponding disease symptom association matrix is retrieved in the medical ciphertext database. The disease symptom association matrix records the association probability or weight relationship between each symptom and multiple diseases obtained through clinical statistics in the form of ciphertext, and the weighted probability vector between multiple symptom ciphertexts and the target disease ciphertext is calculated through a homomorphic weighting operation function, generating a set of ciphertext scores representing the disease association degree. The ciphertext state logical reasoning calculation is performed on the disease association degree ciphertext score, and the ciphertext logic gate operation library is called to operate and process the logical relationship between the query symptoms, such as the "and" logic which represents that multiple symptoms need to be satisfied at the same time, the "or" logic which aggregates the symptom contribution degree in the form of maximum value, and the "not" logic which performs reverse inhibition through constant offset, so that the combination judgment process matching the medical clinical thinking is completed without decrypting the original content, and the ciphertext logical matching result at the reasoning level is obtained. The logical reasoning matching result and the timestamp ciphertext information carried in the query request are fused and processed, the encrypted time range constraint screening operation is performed, the ciphertext of the event or examination record meeting the time condition is extracted, the logical reasoning result and the time limited result are jointly encapsulated to form the encrypted state search result, and the encrypted form is returned to the query initiator.

[0023] In one example, the professional qualifications of doctors, the authorization range of patients, the data sensitivity level and the time permission validity in the multi-institution medical data access request are verified, and a permission verification result is generated, including: The doctor's qualifications certificate and professional skill requirements in the multi-institution medical data access request are analyzed, and the professional qualifications of doctors are determined according to the doctor's qualifications certificate and professional skill requirements; The privacy preference and data category authorization set by the patient for the professional qualifications of the doctor are read to obtain the authorization range of the patient and the data sensitivity level; The time interval between the current access time and the time when the permission is initially granted is calculated, and the time interval is analyzed for time permission validity according to the authorization range of the patient to obtain the time permission validity; The professional qualifications of doctors, the authorization range of patients, the data sensitivity level and the time permission validity are integrated to generate a permission verification result.

[0024] In this example, after receiving a medical data access request from multiple agencies, the physician identity embedded in the request content, the qualification certificate structured information and the professional skill requirement parameters of the target data are extracted, and the legality and authenticity of the physician qualification certificate are confirmed through the digital signature verification mechanism. The elements such as the type of practice permit, the direction of certified profession, the level of qualification and the length of service in the certificate are extracted, converted into a comparable professional vector structure, and aligned with the professional skill requirements set in the access request item by item. The professional matching algorithm is called to calculate the matching degree score between the current qualification of the physician and the required skill of the target data. If the matching score is higher than the preset threshold, it is determined that the physician has the corresponding data access qualification. The corresponding authorization configuration record of the physician is called from the patient authorization database, the access control strategy set by the patient for the professional direction is retrieved according to the professional identity of the physician, and the data categories allowed to access, the restrictive data labels, the access time window and the privacy preference level set by the patient are extracted. The actual boundary of the authorized range is determined in combination with the data classification label requested by the physician, and the sensitive level of the target data in the industry ontology graph is identified through the medical data sensitivity classification model to obtain the data sensitivity level value. The time management module is called to obtain the current system time stamp, and the difference between the granted time in the patient's initial authorization record and the current access behavior is calculated to obtain the time interval length relative to the authorization granted time. Combined with the time validity period set in the patient's authorization strategy and the corresponding permission decay rule, the permission validity factor corresponding to the current time point is calculated to determine whether the access request is within the authorized validity period, and the time permission validity judgment result is output. The professional qualification judgment result of the physician, the authorization range boundary of the patient, the sensitive level classification result of the request data and the time permission validity parameter calculated are combined, and the permission verification engine is used to perform logical discrimination and weight weighting processing on various factors. According to the four-dimensional fine-grained permission matrix, the access permission verification result is generated, and the access permission verification result is expressed in the form of a structured token to indicate whether the access permission is granted, the authorization level strength and the accessible data range.

[0025] In one example, the ontological semantic attributes and sensitivity characteristics of the user-input medical search keywords are analyzed, and the corresponding hierarchical encryption strategy is determined based on the permission verification result, including: The user-input medical search keywords are input into the medical ontology knowledge graph for part-of-speech tagging and named entity recognition to generate ontological semantic attributes; The semantic vector cosine similarity between each medical entity in the ontological semantic attributes is calculated, and the semantic correlation score is determined based on the medical field weight coefficient and the semantic vector cosine similarity; The preset sensitivity label of the corresponding entity in the medical ontology is queried according to the semantic correlation score to identify the sensitivity characteristics; Match the data sensitivity level in the permission verification result with the sensitivity feature to determine the hierarchical encryption strategy.

[0026] In this example, the user input medical search keywords are input into the medical ontology knowledge graph, which includes semantic subgraph structures covering disease terms, symptom expressions, drug nouns, and examination items, and integrates a part-of-speech analysis module and a named entity recognition model dedicated to medical language processing. Relying on the part-of-speech analysis module to perform syntactic analysis and semantic structure decomposition on the user input content, the specific medical entities involved in the keywords, their part-of-speech categories (such as nouns, adjectives, verbs, etc.), and the naming category labels of the entities are automatically identified, for example, "high blood pressure" is identified as a disease entity, and "dizziness" is identified as a symptom entity. The ontology semantic attribute set with semantic labels, ontology location information, and entity types is output. The ontology semantic attribute set is vectorized, and a medical semantic embedding model is called to convert each medical entity into a high-dimensional semantic vector representation. The semantic correlation between entities is calculated through cosine similarity in the vector space, and the relationship strength between different entities is adjusted by combining the medical semantic level weight mechanism, for example, the similarity between disease-symptom pairs is given a higher weight, and the similarity between disease-drug pairs is given a medium weight. The weighted semantic correlation score matrix is obtained. Using the semantic correlation score as an index, the corresponding entity's preset sensitivity label is found in the medical ontology database. The preset sensitivity label is a predefined sensitivity level based on clinical scenarios, data protection standards, and patient privacy preferences, for example, mental diseases are marked as "confidential" level, common cold is marked as "public" level, and examination items such as "AIDS antibody detection" are marked as "restricted access". According to the semantic correlation, the semantic transmission path between entities is determined, and the sensitivity feature of the current search keywords is comprehensively derived. The sensitivity feature is matched with the data sensitivity level contained in the permission verification result, the maximum sensitivity level threshold that the user request data can tolerate in the permission context is analyzed, and the appropriate encryption strategy level is selected according to the maximum sensitivity level threshold. If the sensitivity is much higher than the user permission threshold, strong homomorphic encryption is used, if the sensitivity is basically the same, symmetric encryption is enabled, if the sensitivity is lower than the permission lower limit, hash transformation or plaintext index scrambling processing is used, and a structured hierarchical encryption strategy configuration item is generated.

[0027] In one example, matching the data sensitivity level in the permission verification result with the sensitivity feature to determine the hierarchical encryption strategy includes: Comparing the data sensitivity level in the permission verification result with the sensitivity value of the sensitivity feature, and dividing it into high sensitivity level, medium sensitivity level, and low sensitivity level; The high-sensitivity level matching encryption algorithm type is homomorphic encryption mode, the medium-sensitivity level matching encryption algorithm type is symmetric encryption mode, and the low-sensitivity level matching encryption algorithm type is hash processing mode; The parameter configuration information corresponding to the encryption algorithm type is read, and the encryption parameter values in the parameter configuration information are dynamically adjusted in combination with the medical data type and the access frequency to generate an encryption configuration scheme; The sensitivity level classification result, the encryption algorithm type and the encryption configuration scheme are packaged and integrated to form a hierarchical encryption strategy.

[0028] In this example, the data sensitivity level value recorded in the permission verification result and the sensitivity value generated by the sensitivity feature analysis module are extracted respectively, and after dimensionless normalization processing by a standardization function, difference comparison is performed, and according to a preset sensitivity mapping rule, the comparison result is mapped to a specific sensitivity level classification. When the difference between the data sensitivity level and the sensitivity feature value is above the critical range, or both are in the high interval, the medical information is determined to be of high sensitivity level. If the difference is in the middle fluctuation interval or one is high and one is medium, it is classified as medium sensitivity level. If both are low values or the difference is below the threshold, it is identified as low sensitivity level. According to the sensitivity level classification, the corresponding encryption algorithm type is found, wherein the high-sensitivity level is automatically matched to the homomorphic encryption algorithm supporting encrypted state operation, and algorithms such as Paillier and BFV that can retain structural computing capability are preferentially used. The medium-sensitivity level is matched to the symmetric encryption algorithm, and high-efficiency encapsulation schemes such as AES256 are selected. The low-sensitivity level is matched to the one-way hash processing mode, and SHA256 and other irreversible hash functions are used to perform masking processing on entity information, thereby constructing an algorithm matching list. The algorithm configuration management module is called to read the standard parameter configuration template corresponding to each encryption algorithm type, which includes key length, random salt value strategy, encryption vector mode, round number setting and running load upper limit, etc. key items, while dynamically adjusting the parameter values in combination with the type characteristics of the current processed medical data (such as structured text, medical images or historical time series) and the average call frequency of the current processed medical data in the historical access record, such as increasing the number of parallel encryption threads for high access frequency data, shortening the encryption period to improve real-time performance, or increasing the noise disturbance intensity for sensitive image data to enhance the image ciphertext anti-leakage capability, and generating an encryption configuration scheme through a weighted parameter adjustment model. The sensitivity level classification result, the encryption algorithm type and the encryption configuration scheme are packaged and integrated to form a unified hierarchical encryption strategy structure according to the data type dimension, the security level dimension and the permission adaptation dimension. The hierarchical encryption strategy structure includes a sensitivity level label, an encryption algorithm identifier, an algorithm parameter set, a dynamic adjustment flag and an application range pointer.

[0029] In one example, a homomorphic encryption transformation is performed on a medical term synonym set according to a hierarchical encryption strategy, and a medical ciphertext database is constructed, including: Read the medical search keywords in the hierarchical encryption strategy, and retrieve the semantic equivalent words in the disease ontology, symptom ontology, drug ontology and examination ontology based on the medical search keywords, to generate a medical term synonym set; The synonym hash value, semantic vector and medical field noise of each synonym in the medical term synonym set are subjected to an exclusive or operation using the Paillier homomorphic encryption algorithm to generate synonym ciphertexts; A ciphertext semantic association index table is established between the synonym ciphertexts, and the ciphertext data is stored in partitions according to disease categories, symptom groups, drug classifications and examination items to form a hierarchical ciphertext data structure; The hierarchical ciphertext data structure and the ciphertext semantic association index table are integrated and configured to construct a medical ciphertext database.

[0030] In this example, the medical search keywords specified in the hierarchical encryption strategy are parsed, and the medical semantic expansion module is called to perform ontology mapping and synonym relationship retrieval operations in the disease ontology, symptom ontology, drug ontology, and examination ontology four subgraphs. Using the equivalent links, semantic inheritance relationships, hierarchical relationships, and clinical reference chains between entities in the ontology graph, a semantic equivalent vocabulary set is constructed from multiple levels such as standard terminology, colloquial terms, term abbreviations, and language variants. The semantic adjacent terms and professional synonyms in the four types of ontologies are included, and a structured medical term synonym set is generated. Start the encryption encoding process, and perform multiple feature encoding on each synonym in the medical term synonym set. Hash processing is performed on the terms to generate fixed-length structured hash values, and a semantic vector representation of each term is obtained by calling a medical field semantic vector model. A set of semantic perturbation parameters is superimposed on each term under the Laplace distribution by calling a medical-specific noise generator. The hash value, semantic vector, and noise parameter are XOR fused by bit operation, and the result is input into the Paillier homomorphic encryption algorithm module for encryption conversion. The output is the synonym ciphertext. The semantic association between synonym ciphertexts is calculated, and the semantic proximity between each pair of ciphertext terms is evaluated using the vector angle cosine similarity algorithm in the ciphertext space. The association strength between each pair of terms is recorded, and a ciphertext semantic association index table is generated. The ciphertext semantic association index table saves the semantic structure graph between terms in an encrypted state structure, supporting semantic similarity retrieval without decryption. According to the medical category information of the term ciphertext, logical partition storage is performed, and a four-dimensional data hierarchical strategy is constructed. The ciphertext data is classified into four hierarchical storage spaces: disease category, symptom group, drug classification, and examination item. Each layer of space is hashed according to the category label to improve data query efficiency and access boundary isolation, and a hierarchical ciphertext data structure is generated. The hierarchical ciphertext data structure and the ciphertext semantic association index table are fused and configured to form a medical ciphertext database architecture.

[0031] In one example, a medical query request is received and a symptom-disease association matching and ciphertext state logical reasoning operation are performed in the medical ciphertext database, and an encrypted state search result is output, including: Parsing the symptom keyword combination in the received medical query request, and converting the symptom keyword combination into a symptom ciphertext query instruction; Retrieving the pre-constructed disease-symptom association matrix in the medical ciphertext database according to the symptom ciphertext query instruction, and calculating the disease association degree ciphertext score based on the disease-symptom association matrix; Performing medical logical operations on the ciphertext state based on the disease association degree ciphertext score to obtain a logical reasoning matching result; Integrating the logical reasoning matching result and the timestamp ciphertext information to form an encrypted state search result.

[0032] In this example, the text content contained in the medical query request is structured parsed, and a medical natural language understanding module is enabled to extract the symptom keyword combination therein, and a named entity recognition and context semantic analysis technology is used to identify all keyword entries belonging to the medical symptom ontology range, such as "cough", "fever", "chest pain", etc. entities, and judge whether there is a clear logical relationship between them, such as "and", "or", "not" and other composite condition instructions. The identified standardized symptom keywords are input into the ontology mapping module, the synonym extension set is generated through ontology synonym extension, and all keyword entries are jointly constructed into ciphertext feature blocks through hash functions, semantic vectorization functions, and medical specific noise generators, and Paillier homomorphic encryption algorithm is called to complete encryption processing, forming a symptom ciphertext expression structure supporting encrypted state calculation, and encapsulating and constructing a symptom ciphertext query instruction according to the original combination logic relationship and inputting it into the search engine. The structure index module is called in the medical ciphertext database to locate the storage path corresponding to the symptom ciphertext, and the pre-constructed disease symptom association matrix is retrieved, the disease symptom association matrix adopts a ciphertext storage structure to record the clinical co-occurrence weight or statistical probability between each type of disease and each type of symptom, and a homomorphic executable correlation function is designed for each group of symptom and disease combination relationship, the symptom ciphertext query instruction and the ciphertext structure in the disease symptom association matrix are aligned in vector, and the multiplication and addition operations in the encrypted state are used to weight and aggregate the symptom items, and the ciphertext form disease correlation score vector corresponding to each potential related disease is obtained. The disease correlation ciphertext score is input into the encrypted logic reasoning module to perform homomorphic operation based on the logic gate function without any plaintext decryption operation, wherein when multiple symptoms specified in the user query are "and" logic, the minimum value operation in the encrypted domain is performed on the correlation score corresponding to each symptom to simulate the intersection condition, when it is "or" logic, the maximum value operation is performed to capture the broad correlation, and when it is "not" logic, the difference between the fixed value and the ciphertext is performed to generate the reverse exclusion condition. All operations are completed in the ciphertext space, and the disease matching ciphertext result after logic reasoning is obtained. According to the time query condition attached in the request, the time range is extracted or derived, and the time range is executed time vector standardization, and the logic reasoning matching result is compared with the timestamp ciphertext of each case or examination record in the database through the homomorphic time screening function. The range comparison operation in the encrypted domain is performed to exclude records that do not meet the time constraint condition, and only related data within the specified time window is retained. The logic reasoning matching result and the timestamp screening result are encapsulated and integrated in the ciphertext space, and the encrypted state search result is generated by binding the structure annotation, class label embedding and time index in the encrypted state.

[0033] In one example, the medical logic operation in the ciphertext state is performed based on the disease correlation ciphertext score, and the logic reasoning matching result is obtained, including: find the logical connection words and operators between the symptom entities based on the disease correlation ciphertext score, and determine the ciphertext logical operation instructions to be executed according to the logical connection words and operators; invoke the corresponding homomorphic computing function according to the ciphertext logical operation instructions, and calculate the logical operation ciphertext result of the multi-symptom combination based on the homomorphic computing function; perform medical term fuzzy matching based on the logical operation ciphertext result, obtain a fuzzy matching correction score, and fuse the logical operation ciphertext result and the fuzzy matching correction score to generate a logical reasoning matching result.

[0034] In this example, the original query request text is subjected to semantic structure analysis, and the logical connection words and operator information between the symptom entities are extracted, such as operation expression forms such as "and", "or", "not", etc., and these natural language logical expression forms are mapped into structured logical computing symbols. The logical rule analysis module abstracts the semantic relationship between each group of symptom pairs into a standardized logical expression, constructs a logical analysis tree in combination with the position relationship and context word pattern, and converts it into corresponding ciphertext logical operation instructions. The instructions record the ciphertext entities participating in the logical operation, the connection order, the logical operator type, and the logical block nesting level. According to the ciphertext logical operation instructions, the corresponding homomorphic computing function is invoked, which includes a minimum value computing function in the encryption domain for implementing the homomorphic operation of "and" logic, a maximum value function for simulating "or" logic, and a fixed difference value function for performing "not" logical inversion operation. And for the symptom ciphertext groups marked in the above instructions, perform pair-by-pair homomorphic computing operation. Without decrypting any ciphertext data, the multiple symptom ciphertext scores are combined according to the logical tree structure, and a comprehensive ciphertext result expressing the joint logical relationship is output, representing the reasoning strength of each candidate disease under the given symptom logical conditions. In order to enhance the fault tolerance of the reasoning model to non-standard terms, semantic variants or spelling deviations in real clinical expressions, fuzzy matching correction processing is performed between the original query keywords involved in the logical operation ciphertext result and the standard medical terms in the database. By introducing an edit distance homomorphic operation model and a medical abbreviation mapping table, the character difference structure and semantic vector residual error between each query word and its target term are calculated in the encrypted state, to generate a set of fuzzy matching correction scores to quantify the semantic deviation between the current query item and the standard term. The ciphertext computing result obtained by logical reasoning is fused with the above fuzzy matching correction score, and the fusion process adopts a fuzzy correction factor interpolation algorithm based on a homomorphic multiplication weight mechanism. In the ciphertext space, the fuzzy matching correction factor is integrated into the reasoning path weight calculation logic by performing reversible scaling or weighted accumulation operations on the logical strength score, to generate a semantic-enhanced ciphertext reasoning matching result.

[0035] In one example, the information security management method based on digital medicine further comprises: record the behavior characteristic parameters corresponding to the search result of the encrypted state, and generate an access behavior trajectory according to the behavior characteristic parameters; Based on the word frequency distribution and time pattern characteristics in the access behavior trajectory, randomly select term combinations in different medical specialty fields and add time offset and noise parameters to construct a false query sequence; Extract the access timestamp in the access behavior trajectory, generate a random delay value using Laplace distribution and superimpose it on the access timestamp, and simultaneously perform Markov chain randomization processing on the timestamp of the false query sequence to obtain a set of time- confused access timestamps; Perform end-to-end encryption on the encrypted state search result and perform XOR operation with the institution identity mask to obtain encrypted search data, package and encapsulate the encrypted search data, the false query sequence and the set of time-confused access timestamps to generate a privacy protection data transmission package; Monitor the overall privacy loss value of the privacy protection data transmission package and detect abnormal access patterns to output a security management result.

[0036] In this example, after the user completes a ciphertext search, the current encrypted state search result is recorded at the behavior level, and feature variables including search keyword ciphertext, response time, query trigger timestamp, returned result ciphertext structure size, disease category ciphertext distribution, etc. are extracted, a structured behavior feature parameter set is constructed, and multiple search behaviors are recorded continuously along the behavior timeline to generate an access behavior trajectory. The word frequency distribution, keyword combination pattern, domain distribution, and search time interval of the search word are calculated in the access behavior trajectory. The behavior confusion module is called to analyze the pattern features in the behavior trajectory, and combined with the medical specialty ontology vocabulary, a plurality of term combinations are selected in a pseudo-random manner from professional fields that have no semantic overlap with real queries to construct a false query set that does not overlap in content logic and is close in statistical structure. Each false term combination is injected with a time disturbance parameter and a fake context template to match the structure style of a real request, forming a false query sequence with confusion ability. The original access timestamp data in the access behavior trajectory is extracted, and a Laplace noise mechanism is applied to each timestamp to generate a random delay value that meets the ε-differential privacy requirement, which is then weighted and superimposed on the original timestamp. At the same time, a state transition probability model based on Markov chain is introduced to the fitting timestamp of the false query sequence to inject jump and continuity changes while controlling the time interval distribution, realize sequence disturbance of time pattern, and output a set of access time stamps after time confusion. When constructing encrypted return data, the search results are end-to-end encrypted and packaged using a symmetric key, and an agency identity masking mechanism is introduced. An identity masking function based on agency registration random seed is called to generate a mask vector, which is then XORed with the encrypted search data to prevent external servers from identifying the agency source of the ciphertext, enhancing the anonymity of cross-agency searches. The encrypted search data, generated false query sequence, and corresponding time confusion timestamp set are packaged together into a unified data structure, which is constructed as a privacy protection data transmission package and uploaded through an encrypted channel. After the transmission of the privacy package is completed, the privacy loss monitor calculates the cumulative value of the privacy consumption caused by the transmission sequence in real time. By recording the sensitivity risk weight and noise interference intensity of each query, the total amount of ε is accumulated according to the differential privacy budget management function. When the total loss approaches the threshold, the privacy enhancement mechanism is automatically triggered, and abnormal behavior analysis is performed on all transmission package behaviors. Based on features such as access frequency distribution, access time period concentration, and repeated word intensity, an abnormal identification model is inferred to determine whether there is a super-frequency access, suspicious disguise, or automated attack behavior. The security management result is output and a strategy feedback is generated according to the risk level to the encryption control module.

[0037] In one example, the total privacy loss value of the privacy protection data transmission package is monitored and abnormal access patterns are detected, and a security management result is output, including: The query operation record and the sensitivity parameter in the privacy protection data transmission package are analyzed, and an overall privacy loss value is obtained based on the query operation record and the sensitivity parameter accumulation calculation; The access behavior data corresponding to the overall privacy loss value is monitored, and an abnormal access mode is generated according to the access behavior data; According to the abnormal access mode, the dynamic security strategy adjustment mechanism is called to modify the encryption strength parameter, the false query injection ratio and the privacy noise coefficient, and the security protection configuration parameter is updated; The overall privacy loss value, the abnormal access mode and the security protection configuration parameter are integrated, and the security management result is output.

[0038] In this example, each query operation record contained in the privacy protection data transmission package is extracted, including search keyword ciphertext, query timestamp, request initiation source, target data ciphertext label, query frequency and corresponding ciphertext response behavior, and the sensitivity parameter is derived from the preset sensitivity level label in the data ontology and the access sensitivity threshold value generated by the user permission verification module. The mapping relationship between these operation records and sensitivity parameters is established, the ε value and δ value corresponding to each search behavior are evaluated according to the privacy loss definition in the differential privacy framework, the privacy exposure contribution value caused by each interaction is superimposed in the cumulative model by calling the privacy loss accumulation module, the total privacy loss value of the current user or the current device in a specific time window is obtained, and a dynamic curve of the total privacy loss value with time is established to monitor the long-term influence trend of the query sequence on data privacy. Through behavior clustering analysis of the query operation records with accumulated calculated privacy loss value, access frequency, symptom combination change rate, query time distribution, target field repetition, word entropy and other behavior characteristics are extracted, and the access behavior portrait of the user or the terminal is constructed. Based on the abnormal identification algorithm, the abnormal fluctuations, frequency surges, cross-specialty frequent switching, query result empty return rate increase and other patterns in the behavior portrait are modeled and identified to form a set of potential abnormal access pattern labels, such as malicious attack attempts, identity spoofing behavior, unauthorized automated operations or patient information leakage risks. According to the abnormal access mode, a dynamic security policy adjustment mechanism is started to modify the currently effective encryption control parameters, security interference configuration and resource use restrictions in linkage, dynamically improve the encryption strength parameters according to the access sensitivity level, such as switching the symmetric encryption mode originally applicable to medium sensitivity data to the higher-strength homomorphic encryption mode, or increasing the key length and encryption round; for the behavior pattern of the exposed query structure, the proportion of false query injection is appropriately increased, more irrelevant domain terms and higher frequency of fake behavior are introduced to enhance the query confusion ability, and the original configured privacy noise coefficient is adaptively adjusted, the Laplace noise parameter ε value is lowered or other disturbance mechanisms such as Gaussian disturbance and query time disturbance are superimposed according to the access density and sensitivity risk, so that the sensitive data is further blurred in the return path. After the adjustment is completed, the updated encryption strength parameters, false query injection proportion and privacy noise control items are written into the security protection configuration parameter table, and the security protection configuration parameters, the total privacy loss value and the identified abnormal access pattern labels are uniformly packaged to generate a security management result.

[0039] Reference Figure 2 The embodiment provides an information security management system based on digital medical treatment, which comprises: The verification module 21 is configured to verify the professional qualification of doctors, the authorized range of patients, the data sensitivity level and the time permission validity in the multi-institution medical data access request, and generate a permission verification result. The analysis module 22 is configured to analyze the ontology semantic attributes and sensitivity features of the medical search keywords input by the user, and determine a corresponding hierarchical encryption strategy based on the permission verification result. The homomorphic encryption transformation module 23 is configured to perform homomorphic encryption transformation on a medical term synonym set according to the hierarchical encryption strategy, and construct a medical ciphertext database. The output module 24 is configured to receive a medical query request and perform symptom-disease association matching and ciphertext state logical reasoning operation in the medical ciphertext database, and output an encrypted state search result.

[0040] In the embodiment, the specific implementation of each unit in the system embodiment is described above, and will not be repeated here.

[0041] The four-dimensional fine-grained permission control matrix is constructed, the four dimensions of the professional qualification of doctors, the authorized range of patients, the data sensitivity level and the time permission validity are considered at the same time, compared with the traditional single or two-dimensional permission control, the accuracy and flexibility of the permission management are significantly improved, and the technical problem of rough permission control in the multi-institution environment is effectively solved. The adaptive hierarchical encryption mechanism based on the medical ontology knowledge graph can dynamically select an encryption strategy according to the semantic features and sensitivity levels of medical terms, overcome the defect that the traditional unified encryption method cannot distinguish the importance of data, ensure the strong encryption protection of high-sensitive data, and improve the processing efficiency of low-sensitive data. Through the medical semantic perception homomorphic encryption technology, the symptom-disease association matching and logical reasoning operation in the encrypted state are realized, the technical bottleneck that the traditional encrypted search cannot process medical professional knowledge reasoning is solved, and the medical data can support complex medical diagnosis auxiliary functions while maintaining the encrypted state. The present application not only encrypts the search content, but also protects the search behavior, access mode and metadata through false query injection, timestamp confusion, institution identity masking and other technical means, compared with the traditional method of protecting only the content, the privacy leakage risk of inferring user intent through search trajectory analysis is significantly reduced. Through the differential privacy analysis and abnormal access detection mechanism, the privacy loss status and security situation of the system can be monitored in real time, and the encryption strength, confusion parameters and other security strategies can be automatically adjusted according to the threat level, which overcomes the technical limitations of the traditional static security protection that cannot cope with dynamic threats. The present application establishes a multi-institution threat intelligence sharing mechanism, which can realize the cooperative perception and joint protection of the security situation without leaking the sensitive information of each institution, and effectively improves the overall security protection level of the entire medical alliance.

[0042] It is to be understood that the terminology "including", "comprising", or any other variation thereof, is intended to cover a non-exclusive inclusion such that process, system, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, system, article, or apparatus. An element proceeded by "comprises a... " does not, without more constraints, exclude the existence of additional identical elements in the process, system, article, or apparatus that comprises the element.

[0043] The preferred embodiments of the present application have been described above with the intent to enable those skilled in the art to make and use it. Various modifications to these embodiments will occur to those skilled in the art and are intended to be encompassed by the present application. Therefore, it is to be understood that, within the scope of the present application, the application can be practiced otherwise than as specifically described. For example, the order of steps can be varied, or some steps can be omitted or adapted; the use of some reagents can be tailored or made redundant, and other suitable terrestrial or extraterrestrial elements shown and described individually can be used in combination. The scope of the present application is, therefore, indicated and limited only by the following claims.

Claims

1. An information security management method based on digital healthcare, characterized in that, include: Verify the validity of physician professional qualifications, patient authorization scope, data sensitivity level, and time permissions in multi-institutional medical data access requests, and generate permission verification results; Analyze the ontological semantic attributes and sensitivity features of the medical search keywords entered by the user, and determine the corresponding hierarchical encryption strategy based on the permission verification results; Based on the hierarchical encryption strategy, a homomorphic encryption transformation is performed on the medical term synonym set to construct a medical ciphertext database. It receives medical query requests and performs symptom-disease association matching and encrypted state logical reasoning operations in the medical encrypted database, and outputs encrypted state search results.

2. The information security management method based on digital healthcare according to claim 1, characterized in that, The verification process verifies the validity of physician professional qualifications, patient authorization scope, data sensitivity level, and time permissions in multi-institution medical data access requests, generating permission verification results, including: The system analyzes the physician qualification certificates and professional skill requirements in multi-institutional medical data access requests and determines the physician's professional qualifications based on the physician qualification certificates and professional skill requirements. The system reads the patient's privacy preferences and data category authorizations set for the doctor's professional qualifications to obtain the patient's authorization scope and data sensitivity level. Calculate the time interval between the current access time and the initial grant of permissions, and perform a time permission validity analysis on the time interval according to the patient's authorization scope to obtain the time permission validity; Based on the doctor's professional qualifications, the patient's authorized scope, the data sensitivity level, and the validity of the time permission, an permission verification result is generated.

3. The information security management method based on digital healthcare according to claim 1, characterized in that, The analysis of the ontological semantic attributes and sensitivity features of the medical search keywords input by the user, and the determination of the corresponding hierarchical encryption strategy based on the permission verification results, includes: Input the medical search keywords entered by the user into the medical ontology knowledge graph for part-of-speech tagging and named entity recognition, and generate ontology semantic attributes; Calculate the cosine similarity of semantic vectors among medical entities in the ontology semantic attributes, and determine the semantic relevance score by combining the medical domain weight coefficient and the cosine similarity of semantic vectors; Based on the semantic relevance score, query the preset sensitivity labels of the corresponding entities in the medical ontology to identify sensitivity features; Match the data sensitivity level in the permission verification result with the sensitivity feature to determine the hierarchical encryption strategy.

4. The information security management method based on digital healthcare according to claim 3, characterized in that, The step of matching the data sensitivity level in the permission verification result with the sensitivity feature to determine the hierarchical encryption strategy includes: The data sensitivity level in the permission verification result is compared with the sensitivity value of the sensitivity feature, and then classified into high sensitivity level, medium sensitivity level and low sensitivity level. For high sensitivity levels, the encryption algorithm type is homomorphic encryption mode; for medium sensitivity levels, the encryption algorithm type is symmetric encryption mode; and for low sensitivity levels, the encryption algorithm type is hash processing mode. Read the parameter configuration information corresponding to the encryption algorithm type, and dynamically adjust the encryption parameter values ​​in the parameter configuration information in combination with the medical data type and access frequency to generate an encryption configuration scheme; The sensitivity level classification results, the encryption algorithm type, and the encryption configuration scheme are encapsulated and integrated to form a hierarchical encryption strategy.

5. The information security management method based on digital healthcare according to claim 1, characterized in that, The step of performing a homomorphic encryption transformation on the medical term synonym set according to the hierarchical encryption strategy to construct a medical ciphertext database includes: Read the medical search keywords in the hierarchical encryption strategy, and retrieve semantically equivalent terms in the disease ontology, symptom ontology, drug ontology and examination ontology based on the medical search keywords, and generate a set of medical term synonyms; The Paillier homomorphic encryption algorithm is used to perform an XOR operation on the synonym hash value, semantic vector, and medical noise of each synonym in the medical term synonym set to generate synonym ciphertext; Establish a ciphertext semantic association index table between the ciphertexts of the synonyms, and store the ciphertext data in partitions according to disease category, symptom grouping, drug classification and examination items to form a hierarchical ciphertext data structure; The hierarchical encrypted data structure and the encrypted semantic association index table are integrated and configured to construct a medical encrypted database.

6. The information security management method based on digital healthcare according to claim 1, characterized in that, The process of receiving medical query requests and performing symptom-disease association matching and encrypted state logical reasoning operations in the medical encrypted database, and outputting encrypted state search results, includes: Parse the combination of symptom keywords in the received medical query request and convert the combination of symptom keywords into a symptom encrypted query instruction; According to the symptom encrypted query instruction, a pre-constructed disease symptom association matrix is ​​retrieved from the medical encrypted database, and a disease association degree encrypted score is calculated based on the disease symptom association matrix. Based on the encrypted score of the disease correlation, perform medical logic operations on the encrypted state to obtain logical reasoning matching results; The logical reasoning and matching results are integrated with the timestamp ciphertext information and encapsulated to form an encrypted search result.

7. The information security management method based on digital healthcare according to claim 6, characterized in that, The step of performing medical logic operations on the encrypted state based on the encrypted score of the disease correlation degree to obtain logical reasoning matching results includes: Based on the ciphertext score of the disease correlation, find the logical connectors and operators between symptom entities, and determine the ciphertext logical operation instructions to be executed according to the logical connectors and operators; The corresponding homomorphic computation function is invoked according to the ciphertext logical operation instruction, and the ciphertext result of the logical operation of multiple symptom combinations is calculated based on the homomorphic computation function; Based on the encrypted result of the logical operation, perform fuzzy matching of medical terms to obtain a fuzzy matching correction score, and fuse the encrypted result of the logical operation with the fuzzy matching correction score to generate a logical reasoning matching result.

8. The information security management method based on digital healthcare according to claim 1, characterized in that, The information security management method based on digital healthcare also includes: Record the behavioral feature parameters corresponding to the encrypted search results, and generate an access behavior trajectory based on the behavioral feature parameters; Based on the word frequency distribution and time pattern characteristics in the access behavior trajectory, a false query sequence is constructed by randomly selecting a combination of terms from different medical specialties and adding time offset and noise parameters. Extract the access timestamps from the access behavior trajectory, generate random delay values ​​using a Laplace distribution and superimpose them onto the access timestamps. At the same time, randomize the timestamps of the false query sequence using a Markov chain to obtain a set of access timestamps after time confusion. The encrypted search results are end-to-end encrypted and XORed with the organization identity mask to obtain encrypted search data. The encrypted search data, the fake query sequence, and the time-obfuscated access timestamp set are packaged and encapsulated to generate a privacy-protected data transmission packet. Monitor the overall privacy loss value of the privacy-protected data transmission packets and detect abnormal access patterns, then output the security management results.

9. The information security management method based on digital healthcare according to claim 8, characterized in that, The system monitors the overall privacy loss value of the privacy-protected data transmission packets and detects abnormal access patterns, outputting security management results, including: The query operation records and sensitivity parameters in the privacy-preserving data transmission packet are analyzed, and the overall privacy loss value is obtained by cumulative calculation based on the query operation records and sensitivity parameters. Monitor access behavior data corresponding to the overall privacy loss value, and generate abnormal access patterns based on the access behavior data; Based on the abnormal access pattern, the dynamic security policy adjustment mechanism is invoked to modify the encryption strength parameter, the proportion of fake query injection, and the privacy noise coefficient, and the security protection configuration parameters are updated. The overall privacy loss value, the abnormal access pattern, and the security protection configuration parameters are integrated to output the security management result.

10. An information security management system based on digital healthcare, characterized in that, The steps for implementing the information security management method based on digital healthcare according to any one of claims 1 to 9, wherein the information security management system based on digital healthcare comprises: The verification module is used to verify the validity of doctors' professional qualifications, patients' authorization scope, data sensitivity level, and time permissions in multi-institutional medical data access requests, and generate permission verification results. The analysis module is used to analyze the ontological semantic attributes and sensitivity features of the medical search keywords entered by the user, and determine the corresponding hierarchical encryption strategy based on the permission verification results. The homomorphic encryption transformation module is used to perform homomorphic encryption transformation on the medical term synonym set according to the hierarchical encryption strategy to construct a medical ciphertext database. The output module is used to receive medical query requests and perform symptom-disease association matching and encrypted state logical reasoning operations in the medical encrypted database, and output encrypted state search results.

Images