Medical information management system

Abstract

The invention discloses a medical information management system. The system comprises a user interaction layer for providing an interaction interface; the encryption service layer is used for identifying the type of medical data by utilizing a ResNet-50 fine-tuned CNN classifier through a dynamic weight fragmentation engine, analyzing a text medical record entity in combination with a BERT-NLP model, processing streaming monitoring data by adopting Kafka Streams, dynamically adjusting a storage strategy according to a fragmentation weight calculation formula, calling an encryption resource library formula according to an extraction code sequence, and storing the encrypted data in a database; meanwhile, a Lattice-based KDF engine is deployed to realize an NTRU anti-quantum encryption algorithm, an HMAC-SHA3-512 dynamic bit extractor is matched to generate a fragment key, a Tent mapping generator and a Lorenz disturbance injector are combined to construct a double-chaos encryption assembly line, and dynamic offset protection is performed on an encryption resource library formula; according to the infrastructure layer, a true random entropy source is provided through a quantum random number generator of an ID Quantique Clavis platform, an HSM hardware security module is integrated to accelerate encryption operation, and a Heda Hashgraph light node is deployed to carry out block chain basic connection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical information management, and particularly relates to a medical information management system. BACKGROUND

[0002] The medical information management system is the core infrastructure of the modern medical system, is a comprehensive information technology solution supported by computer technology and information management methods, fully covers the whole process of medical institution operation, and its essence is to integrate medical information flow, logistics and fund flow through digital means, to build a closed-loop management system from patient registration, clinical diagnosis and treatment, medical examination, drug management to financial settlement, and the core goal is to realize multi-dimensional values of medical quality improvement, operation efficiency optimization, resource rational allocation and data security guarantee through informatization, standardization and intelligentization transformation.

[0003] Through retrieval, Chinese patent No. CN118520485A discloses a smart medical information management method and system, and belongs to the technical field of information management. It contains an encryption resource library, an encryption setting unit, an encryption processing unit, an encryption execution unit, and a decryption extraction unit. The patent uses mathematical formulas and digital serial numbers as the encryption basis, uses complex mathematical operations to generate hidden keys, so that unauthorized users are difficult to crack the encrypted information. The combination of encryption codes and formula extraction codes provides a dynamic encryption mechanism. Different mathematical formula combinations can be used for encryption each time, increasing the unpredictability and complexity of the encryption process. Users can adjust the security level of encryption by setting different lengths of encryption codes and selecting different numbers of formula extraction codes to meet different security needs. The encryption processing unit quickly calculates the substitution value and generates the hidden key through the preset digital extraction model, improving the processing efficiency of encryption and decryption.

[0004] However, the mathematical formula encryption mechanism in the above-mentioned patent is vulnerable. The substitution value is generated by extracting simple digital characters (such as k1, k2 multiple values) and summing. If an attacker obtains part of the plaintext-ciphertext pair, the formula structure can be restored through algebraic methods. The randomness of the formula extraction code completely depends on user settings. If the user uses a weak random number generation algorithm (such as a time seed pseudo-random), the overall anti-cracking ability of the system will be significantly reduced. Therefore, a medical information management system is proposed. SUMMARY

[0005] The purpose of the present application is to solve the problems existing in the prior art and to provide a medical information management system.

[0006] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0007] A medical information management system comprises:

[0008] User interaction layer: provides an interactive interface, handles parameter input, operation audit and result visualization, builds a responsive operation interface based on the React framework, integrates biometric authentication modules and ECharts visualization dashboards;

[0009] Encryption service layer: uses ResNet-50 fine-tuned CNN classifier to identify medical data types through dynamic weight sharding engine, combines BERT-NLP model to parse text medical record entities, uses Kafka Streams to process streaming monitoring data, and dynamically adjusts storage strategy according to sharding weight calculation formula, calls encryption resource library formula according to extraction code sequence, deploys Lattice-based KDF engine to implement NTRU quantum-resistant encryption algorithm, generates sharding key with HMAC-SHA3-512 dynamic bit extractor, constructs double chaos encryption pipeline with Tent mapping generator and Lorenz disturbance injector, and implements dynamic offset protection on encryption resource library formula;

[0010] Infrastructure layer: provides true random entropy source through ID Quantique Clavis platform quantum random number generator, integrates HSM hardware security module to accelerate encryption operations, and deploys Hedera Hashgraph light node for blockchain infrastructure connection;

[0011] Data storage layer: uses Ceph distributed storage cluster to configure erasure code strategy, combines CockroachDB distributed database to manage sharding index, and realizes full-process audit tracking through ELK log chain;

[0012] Blockchain anchoring layer: uses Solidity smart contract to verify sharding hash, stores metadata with IPFS network, and uses Gossip protocol and asynchronous Byzantine fault tolerance mechanism to ensure consensus security.

[0013] The above technical solution further includes:

[0014] Further, the specific steps of the user interaction layer to build a complete interaction system covering parameter verification, security authentication, operation monitoring and exception handling are:

[0015] Dynamic form verification: uses React state hooks and regular expressions to implement dynamic form verification, ensures that encryption codes meet 12-24 digit specifications through input event listening and character counters, uses Set data structure to detect formula extraction code uniqueness and disable duplicate value input;

[0016] Biometric authentication: integrates fingerprint / face recognition SDK to build a two-factor authentication mechanism, configures timeout countdown and three-level failure lock strategy, and automatically downgrades to password verification when biometric authentication fails;

[0017] ECharts visualization dashboard construction: Through ECharts heat map, Sankey diagram and radar chart, the operation audit is visualized, and time series, geographical distribution and risk indicators are mapped into graphical elements. The color gradient and dynamic update are used to display the spatiotemporal heat distribution of encryption / decryption operations and the system security situation;

[0018] Fragmented state monitoring: Real-time state channel is established by using WebSocket, and heartbeat mechanism is used to maintain fragment transmission monitoring. Progress bar and status indicator light are used to feedback transmission progress in real time, and alarm is triggered and abnormal fragments are highlighted when stagnation is detected;

[0019] Interface integration and exception handling: Interface integration is completed through route configuration and error boundary. ARIA attributes are used to enhance accessibility. Global error capture and logging are deployed, and a friendly prompt with error code is displayed when the system is abnormal.

[0020] Further, the encryption service layer utilizes a dynamic weight fragmentation engine to intelligently classify and store medical data and dynamically encrypt and protect the medical data.

[0021] Medical data feature analysis: A CNN classifier based on ResNet-50 fine-tuning is used to identify the modalities of medical images. BERT-NLP model is used to extract diagnostic entities from text medical records. KafkaStreams is used to analyze real-time streaming data and detect abnormal fluctuation patterns.

[0022] Fragment weight calculation: According to the preset storage medium benchmark value, type weight table and privacy sensitivity coefficient matrix, the storage strategy is dynamically generated through the fragment size calculation formula V=I×W×R, where I is the storage medium benchmark value, W is the type weight, and R is the privacy coefficient.

[0023] Dynamic storage strategy adjustment: According to the fragmentation generation rule, the data is automatically cut, and the intelligent routing algorithm is used to consider the storage cost model and performance constraint conditions. The automatic expansion and contraction mechanism is implemented by monitoring the fragment queue length and storage utilization rate.

[0024] Encryption formula calling: According to the user-set formula extraction code sequence, the corresponding mathematical formula is called from the encryption resource library, and the value is extracted in combination with the encryption coded digital character set and the chaotic confusion encryption technology to generate a hidden key for data protection.

[0025] Further, the encryption service layer deploys a Lattice-based KDF engine to implement an NTRU quantum-resistant encryption algorithm, and cooperates with an HMAC-SHA3-512 dynamic bit extractor to generate a fragment key, including the following steps:

[0026] The encryption code is mapped to the NTRU lattice space after preprocessing, and a high-dimensional vector is constructed through a polynomial ring Z[x] / (x n-1 ), where n=503 is a security parameter, and the set of extraction codes generates a disturbance polynomial through a hash operation The coefficients (a i , b i ) are derived from the first 64 bits of the HMAC-SHA3-512 value of the extraction code, the ring learning error (LWE) sampling is implemented, the discrete Gaussian noise is injected in the lattice space, and the intermediate key material is generated;

[0027] Dynamic bit extraction strategy: performing HMAC-SHA3-512 operation on each extraction code to generate a 512-bit digest value, the extraction position is determined by the last 4 bits of the digest value, and a sliding window mechanism is used, when the extraction position exceeds the current slice key length, it is automatically cycled to the starting bit;

[0028] Slice key construction: calculating the result value of each extraction code generated intermediate key according to the formula F'(x+δ), δ is the chaotic offset generated by the Tent mapping, according to the dynamic extraction position, the specified bit is intercepted from the result value, and combined into a 128-bit slice key, and the slice key is XORed with the random number generated by the quantum entropy source.

[0029] Further, the encryption service layer combines the Tent mapping generator and the Lorenz disturbance injector to construct a double chaotic encryption pipeline, and the specific steps of implementing dynamic offset protection on the encryption resource library formula are as follows:

[0030] Double chaotic system initialization: a combination of dynamic Tent mapping parameter configuration and classical Lorenz system parameter setting is used to generate an initial state randomly using a quantum entropy source and implement range constraints, dynamic Tent mapping parameter configuration generates a dynamic μ value, represented as μ(t) = 1 + 0.2·sin(t / 1000), t is the system running time, and μ(t) ∈ [1.01, 1.19] by using the integer function, the classical Lorenz system parameter setting: the classical chaotic parameter set σ = 10 is the Prandtl number, ρ = 28 is the Rayleigh number, and β = 8 / 3 is the geometric parameter;

[0031] Chaotic sequence generation: through Tent mapping iterative calculation and Lorenz system numerical solution, using iterative formula, range folding and fourth-order Runge-Kutta method, a sequence with strong chaotic characteristics is generated, the iterative formula is represented as x n+1 = μ(t)·x n ·(1-x n ), when x n ≥1, x n =1-x n ;

[0032] Dynamic offset protection: Adopting a formula segmentation strategy and chaos offset injection, the mathematical formula is divided into multiple protection intervals and an independent offset is applied to each segment. At the same time, the chaos parameters and offset are adjusted in real time through dynamic parameter updates.

[0033] Furthermore, the encryption service layer includes a quantum entropy management module, which constructs a quantum secure random number generation system through hybrid entropy source scheduling and dynamic bias correction.

[0034] Hybrid entropy source dynamic scheduling: Build a three-level priority queue with quantum entropy as the primary source, system entropy as the secondary source, and pseudo-random numbers as the backup source. Combined with a dynamic switching strategy, the entropy source is automatically selected based on the minimum entropy health and transmission rate.

[0035] Continuous monitoring of entropy source health: When quantum entropy is insufficient, the hybrid system entropy maintains availability and continuously monitors the minimum entropy Hmin value through sliding window analysis to isolate the fault source. The minimum entropy Hmin value calculation formula is expressed as Hmin = -log2(max(P(b i ))), where b i All possible values ​​for an 8-bit byte;

[0036] Anti-bias correction processing: Anti-bias correction uses the Von Neumann extraction algorithm to map the original bit stream into an unbiased sequence, combined with the χ 2 The system verifies dynamic threshold settings, automatically adjusts the significance level based on the shard key length, and improves the effective output rate through four-bit grouping preprocessing, ultimately generating quantum-safe random numbers that pass the NIST SP800-90B test.

[0037] Furthermore, the infrastructure layer builds a security enhancement system by integrating a quantum random number generator, an HSM hardware security module, and a Hedera Hashgraph light node. The specific steps are as follows:

[0038] Quantum random number generator integration: The quantum random number generator deployment uses the IDQuantique Clavis platform to set single-photon detector parameters and establish an entropy acquisition pipeline, maintaining minimum entropy through a three-level entropy pool architecture and dynamic injection strategy;

[0039] HSM hardware security module configuration: completes key lifecycle protection by managing key generation, secure channel establishment, and key hierarchical management;

[0040] Hedera Hashgraph light node deployment: Complete network-wide medical data storage by configuring medical test network parameters and deploying smart storage contracts.

[0041] Further, the blockchain anchoring layer builds a medical data trusted anchoring system through sharding hash verification, metadata IPFS storage and consensus security mechanism, including the following steps:

[0042] Sharding hash verification: double SHA3-512 hash calculation combined with dynamic salt value injection technology is adopted to ensure data integrity, and the access control mechanism is set through the verification logic of the Solidity smart contract deployment to realize on-chain storage, and the event-driven mode is used to generate verification credentials to resist data tampering;

[0043] Metadata IPFS storage: a standardized data structure containing encryption parameters and content identifiers is built, content addressing is realized by using Protobuf serialization combined with CIDv1 multi-hash format, and the strong association between sharding hash and physical storage location is established through the on-chain metadata registration mechanism;

[0044] Consensus security mechanism: the fan-out propagation strategy of Gossip protocol and priority queue management are used to ensure efficient message diffusion, the three-phase voting protocol of asynchronous Byzantine fault-tolerant algorithm is combined to reach a safe consensus in an asynchronous network environment, and dynamic view switching and random number generation technology are used to enhance system robustness.

[0045] The present application has the following beneficial effects:

[0046] In the present application, a double chaotic encryption pipeline is built by combining a Tent mapping generator and a Lorenz disturbance injector, and a dynamic offset protection is implemented on the encryption resource library formula, even if the attacker obtains the plaintext-ciphertext pair, due to the initial value sensitivity of the chaotic system and the nonlinear characteristics of the Lorenz equation, the calculation complexity of restoring the formula structure increases from O(n 2 ) to O(2 n ), in addition, the NTRU quantum-resistant encryption algorithm is implemented by deploying a Lattice-based KDF engine, and the sharding key is generated by a HMAC-SHA3-512 dynamic bit extractor, so that the attacker cannot reverse the formula structure through the hidden key mode of multiple encrypted archives. For example, even if two archives use the same formula to extract the code, due to the existence of the disturbance polynomial and the dynamic window, the difference of the hidden key is more than 85%. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 A system block diagram of a medical information management system is provided for the present application;

[0048] Figure 2 An encryption processing flowchart is provided for the present application;

[0049] Figure 3 A decryption recovery flowchart is provided for the present application. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0051] Please refer to Figure 1 As shown in the figure, the present application is a medical information management system, comprising:

[0052] User interaction layer: provides an interactive interface, undertakes parameter input, operation audit and result visualization functions, builds a responsive operation interface (dynamic form verification (such as encryption code length verification, formula extraction code uniqueness detection)) based on the React framework, integrates a biometric authentication module (two-factor authentication reinforcement) and an ECharts visualization dashboard (showing encryption / decryption operation space-time distribution), and realizes encryption parameter dynamic input, decryption request management and shard state monitoring functions;

[0053] User input→parameter verification→encryption service layer processing→result feedback→log persistent storage;

[0054] Encryption service layer: through the dynamic weight shard engine, the ResNet-50 fine-tuned CNN classifier is used to identify the type of medical data (image / DICOM identification), combined with the BERT-NLP model to parse the text medical record entity (text medical record analysis), using Kafka Streams to process streaming monitoring data (streaming data detection), identify abnormal fluctuation patterns, and dynamically adjust the storage strategy according to the shard weight calculation formula, call the encryption resource library formula according to the extraction code sequence, and deploy the Lattice-based KDF engine to realize the NTRU quantum-resistant encryption algorithm, cooperate with the HMAC-SHA3-512 dynamic bit extractor to generate shard keys, and combine the Tent mapping generator and Lorenz disturbance injector to build a double-chaotic encryption pipeline to implement dynamic offset protection on the encryption resource library formula;

[0055] Wherein the ResNet-50 enhances the medical image feature extraction ability through transfer learning, the BERT-NLP model improves the entity recognition accuracy through medical corpus fine-tuning, the sliding window mechanism of Kafka Streams ensures real-time data stream anomaly detection, the shard weight calculation formula optimizes the balance between storage efficiency and security through three-dimensional parameter dynamic adjustment, and the chaotic encryption technology builds a dual protection system against differential analysis and linear cryptanalysis through variable offset generated by Tent mapping and disturbance injection of Lorenz equation group.

[0056] Feature analysis→weight calculation→key generation→formula protection→fragment encryption→result output;

[0057] Infrastructure layer: True random entropy source provided by the quantum random number generator of the ID Quantique Clavis platform, integrated HSM hardware security module to accelerate encryption operations, and Hedera Hashgraph light node deployed for blockchain infrastructure connection;

[0058] Quantum entropy output→HSM encryption acceleration→blockchain transaction broadcast;

[0059] Data storage layer: Ceph distributed storage cluster is used to configure erasure code strategy, combined with CockroachDB distributed database management sharding index, and ELK log chain is used to realize full-process audit tracking;

[0060] Sharding write→metadata index→log record→persistent storage;

[0061] Blockchain anchoring layer: Use Solidity smart contract to verify sharding hash, use IPFS network to store metadata, and use Gossip protocol and asynchronous Byzantine fault tolerance mechanism to ensure consensus security;

[0062] Quantum tunnel establishment→key negotiation→secure communication→side channel suppression.

[0063] In this embodiment, the encryption process reconstruction is described:

[0064] Parameter setting stage:

[0065] User input: encryption code (12-24 digit number); formula extraction code (number ≤ code length, support manual / random generation);

[0066] System processing: Quantum entropy source generates random number to initialize chaotic system, and encryption service layer automatically marks data type (such as MRI image marked as type 3);

[0067] Encryption processing stage:

[0068] Formula extraction: According to the order of extraction code, call the encryption resource library formula (dynamic formula extension is supported);

[0069] Substitute value calculation: Dynamic number set division (k value is determined by LWE sampling result), assuming that the encryption code is "741852", the number of extraction codes is 3→3 number sets are created, and when k=2, the extracted bits are: 2, 4, 6→number set [4, 8, 2];

[0070] Chaotic formula calculation: Assuming that the original formula is F(x)=3x 2 +5, then the chaotic protection is F'(x+0.17)=3(x+0.17)2 +5+Lorenz_noise;

[0071] Dynamic bit extraction: The result value is "138.42" → the valid digits are dynamically determined by HMAC (e.g., taking the 3rd digit), generating the hidden key: "8"

[0072] Data storage stage:

[0073] Sub-file processing: Dynamic calculation of shard size: text medical record (basic 8KB × 1.2 × 1.1), blockchain anchoring: shard metadata + timestamp are uploaded to the chain, sub-file extraction code generation: the result value is padded with zeros and combined with the code (such as "00741852").

[0074] In one embodiment, the user interaction layer constructs a complete interactive system covering parameter verification, security authentication, operation monitoring and exception handling in the following specific steps:

[0075] Dynamic form validation: React state hooks and regular expressions are used to implement dynamic form validation. Input event monitoring and character counters are used to ensure that the encryption code complies with the 12-24 digit standard. The Set data structure is used to detect the uniqueness of the formula extraction code and disable the input of duplicate values.

[0076] Biometric authentication: Integrate fingerprint / face recognition SDK to build a two-factor authentication mechanism, configure a 30-second timeout countdown and a three-level failure lock policy, and automatically downgrade to password verification when biometric authentication fails;

[0077] ECharts visualization dashboard construction: Visualize operational audits through ECharts heat maps, Sankey diagrams, and radar charts, map time series, geographic distribution, and risk indicators into graphical elements, and use color gradients and dynamic updates to display the spatiotemporal heat distribution of encryption / decryption operations and system security status;

[0078] Shard status monitoring: Use WebSocket to establish a real-time status channel, maintain shard transmission monitoring through a heartbeat mechanism, use a progress bar and status indicator to provide real-time feedback on the transmission progress, and trigger an alarm and highlight the abnormal shard when a 30-second pause is detected;

[0079] Interface integration and exception handling: Complete interface integration through routing configuration and error boundaries, use ARIA attributes to enhance accessibility, and deploy global error capture and logging to display friendly prompts with error codes when the system is abnormal.

[0080] In one embodiment, the encryption service layer uses a dynamic weight sharding engine to intelligently classify, store, and dynamically encrypt medical data.

[0081] Medical data feature analysis: modal recognition of medical images using a CNN classifier based on ResNet-50 fine-tuning, combined with a BERT-NLP model to extract diagnostic entities from text medical records, and real-time analysis of streaming monitoring data for abnormal fluctuation patterns using Kafka Streams;

[0082] Sharding weight calculation: based on the preset storage medium reference value, type weight table and privacy sensitivity coefficient matrix, the storage strategy is dynamically generated through the sharding size calculation formula V = I x W x R, where I is the storage medium reference value, W is the type weight, and R is the privacy coefficient;

[0083] Dynamic storage strategy adjustment: automatically cut data according to the sharding generation rule (volume / time / event trigger), and use intelligent routing algorithm to consider the storage cost model (cost = (storage capacity x unit price) + (access frequency x bandwidth fee)) and performance constraints (high priority data (such as surgical records) must be stored in SSD, low frequency access data (such as discharge summary) can be archived to magnetic tape), while monitoring the sharding queue length and storage utilization to implement automatic scaling mechanism (expansion trigger: when the queue length > 100 or the utilization rate > 80%, start a new storage node; shrinkage trigger: when the utilization rate < 30% for 24 hours, release idle resources);

[0084] Encryption formula calling: according to the formula set by the user, extract the code sequence, call the corresponding mathematical formula from the encryption resource library, and combine the encrypted numerical character set extraction value and chaotic confusion encryption technology to generate a hidden key for data protection.

[0085] In one embodiment, the encryption service layer deploys a Lattice-based KDF engine to implement the NTRU quantum-resistant encryption algorithm, and cooperates with the HMAC-SHA3-512 dynamic bit extractor to generate sharding keys, including the following steps:

[0086] Lattice space mapping and perturbation polynomial generation: after preprocessing, the encryption code is mapped to the NTRU lattice space, and a high-dimensional vector is constructed through the polynomial ring Z[x] / (x n-1 ), where n = 503 is the security parameter, and the extraction code set generates the perturbation polynomial The coefficients (a i , b i ) are derived from the first 64 bits of the HMAC-SHA3-512 value of the extraction code, and the ring learning error (LWE) sampling is implemented to inject discrete Gaussian noise in the lattice space and generate intermediate key material;

[0087] Dynamic bit extraction strategy: Perform HMAC-SHA3-512 operation on each extraction code to generate a 512-bit digest value, and the extraction position is determined by the last 4 bits of the digest value: pos = (hash[3] << 24 | hash[2] << 16 | hash[1] << 8 | hash[0]) % key length, and a sliding window mechanism is used, when the extraction position exceeds the length of the current slice key, automatically cycle to the starting bit;

[0088] Slice key construction: Calculate the result value of the formula F'(x+δ) according to the intermediate key generated by each extraction code, δ is the chaotic offset generated by the Tent mapping, and the specified number of bits is extracted from the result value according to the dynamic extraction position to combine a 128-bit slice key, and the slice key is XORed with the random number generated by the quantum entropy source to eliminate potential side channel leakage.

[0089] Suppose the encryption code is "7418529630" and the extraction code set is ["582", "317", "946"];

[0090] The first extraction code "582" generates a perturbation polynomial P1(x) = 5x3 + 8x2 + 2x + 7 (mod 11 simplified);

[0091] Perform LWE sampling to get intermediate key K1 = 0x3A9F... (256 bits);

[0092] Calculate HMAC-SHA3-512("582") to get digest value 0xD1E5A3C7...;

[0093] Extraction position pos = (0xC7) % 256 = 199, and 8-bit sub-key is obtained by extracting K1[199:207];

[0094] Repeat the above process to generate K2 and K3, and combine them into a slice key after chaotic offset calculation: 0xB7D3...;

[0095] Finally, three slice keys (128 bits x 3) are generated, which are used to encrypt medical images, text medical records and streaming monitoring data, respectively.

[0096] In one embodiment, the encryption service layer combines the Tent mapping generator and the Lorenz perturbation injector to construct a double chaotic encryption pipeline, and the specific steps of dynamically offsetting the encryption resource library formula are implemented.

[0097] Double chaotic system initialization: adopt the combination of dynamic Tent mapping parameter configuration and classical Lorenz system parameter setting, use quantum entropy source to randomly generate initial state and implement range constraint, dynamic Tent mapping parameter configuration generates dynamic μ value, expressed as μ(t) = 1 + 0.2 sin(t / 1000), t is system running time, through the integer function, μ(t) ∈ [1.01, 1.19], when the system starts, x0 = 0.314 is randomly generated by quantum entropy source, classical chaotic parameter set: σ = 10 is the Prandtl number, ρ = 28 is the Rayleigh number, and β = 8 / 3 is the geometric parameter, initial condition: x0 = 0.1, y0 = 0, z0 = 0 (initialized by a true random number generator), which ensures the unpredictability and initialization security of the chaotic sequence;

[0098] Chaotic sequence generation: through Tent mapping iterative calculation and Lorenz system numerical solution, the iteration formula, range folding and fourth-order Runge-Kutta method are used to generate a sequence with strong chaotic characteristics, the iteration formula is expressed as x n+1 = μ(t) x n (1-x n ), when x n ≥ 1, x n = 1-x n is executed, which retains sufficient precision to support encryption requirements;

[0099] Dynamic offset protection: adopt formula segmentation strategy and chaotic offset injection, divide the mathematical formula into multiple protection intervals and apply independent offsets to each segment, and simultaneously adjust the chaotic parameters and offsets in real time through dynamic parameter update (Tent mapping μ(t) is updated every second, Lorenz system resets the initial condition every 1MB data, and offset δ i is bound to the fragment key through a secure hash algorithm), which ensures the continuous effectiveness and security of encryption;

[0100] In this embodiment, the protection formula F(x) = x 2 + 3x + 2 is set:

[0101] Chaotic sequence generation:

[0102] After 1000 iterations of Tent mapping, it stabilizes at x = 0.314159...

[0103] When the Lorenz system runs to t = 2500 ms, x_t = 3.241, y_t = -1.789, and z_t = 22.356;

[0104] Offset calculation:

[0105] δ1 = 0.172531 (taken from the 500th iteration of the Tent map);

[0106] δ2 = 0.632847 (taken from the fractional part of x_t of the Lorenz system);

[0107] δ3 = 2 x 0.172531 x 0.632847 / (0.172531 + 0.632847) = 0.2476;

[0108] Formula transformation:

[0109] F1'(x) = (x + 0.172531) 2 = x 2 + 0.345062 x + 0.029768;

[0110] F2'(x) = 3 x + 0.632847;

[0111] F3'(x) = 2 x (1 + 0.2476) = 2.4952;

[0112] Final protection formula:

[0113] F'(x) = x 2 + 0.345062 x + 0.029768 + 3 x + 0.632847 + 2.4952

[0114] = x 2 + 3.345062 x + 3.157815;

[0115] Verification and recovery:

[0116] Use the same chaotic sequence parameters to calculate δ inversely during decryption i ;

[0117] Restore the original formula by solving the equation set:

[0118] x 2 + 3.345062 x + 3.157815 - δ1 x - δ2 - δ3 · F3(x).

[0119] In an embodiment, the encryption service layer includes a quantum entropy management module, and the quantum entropy management module constructs a quantum secure random number generation system through specific steps of hybrid entropy source scheduling and dynamic bias correction:

[0120] Mixed entropy source dynamic scheduling: A three-level priority queue is constructed, with quantum entropy as the primary source (vacuum fluctuation entropy collected from the ID Quantique Clavis platform, generating a raw bit stream through a single-photon detector), system entropy as the secondary source (mixed CPU temperature sensor noise, disk I / O timing jitter, and network packet interval variation), and pseudo-random numbers as the backup source (Mersenne Twister algorithm is used to generate backup sequences (only activated when the quantum entropy rate is <5 Mbps)). Combined with a dynamic switching strategy, the entropy source is automatically selected based on the minimum entropy health degree (Hmin) and transmission rate (when the quantum entropy health degree Hmin≥7.9 and the rate ≥10 Mbps, quantum entropy is used completely; when 7.5≤Hmin<7.9 or the rate 5-10 Mbps, quantum entropy and system entropy are mixed (proportion 3:1); when Hmin<7.5 or the rate <5 Mbps, system entropy is degraded and an alarm is triggered), ensuring that the theoretical security of quantum entropy is fully utilized when it is sufficient;

[0121] Entropy source health continuous monitoring: When quantum entropy is insufficient, system entropy is mixed to maintain availability, and a sliding window analysis is used to continuously monitor the minimum entropy Hmin value (when Hmin<7.9 is detected for three consecutive times, the faulty entropy source is automatically isolated) to isolate the fault source. The minimum entropy Hmin value calculation formula is represented as Hmin=-log2(max(P(b i ))) where b i is all possible values of an 8-bit byte;

[0122] Anti-bias correction processing: Anti-bias correction uses Von Neumann extraction algorithm to map the original bit stream to an unbiased sequence, combined with χ 2 test dynamic threshold setting, automatically adjusts the significance level according to the length of the fragmented key (p=0.001, critical value=32.0), and through four-bit grouping preprocessing, the effective output rate is improved, finally generating quantum secure random numbers that pass the NIST SP 800-90B test.

[0123] In this embodiment, a 128-bit fragmented key

[0124] Entropy source collection:

[0125] Quantum entropy source output: 0xA3F7B2D1...(128 bits, Hmin=8.1);

[0126] System entropy source output: 0x3C9E5A1F...(128 bits, Hmin=7.6);

[0127] Mixed entropy source selection: Because Hmin≥7.9, quantum entropy is used completely;

[0128] Anti-bias correction:

[0129] Before Von Neumann processing: 01100101 10100110...;

[0130] Sequence after extraction: 1010... (significant bits reduced to 64 bits);

[0131] χ 2 Test: Observed frequency O = (32, 28, 31, 29), expected frequency E = (30, 30, 30, 30);

[0132] Calculate χ 2 =0.133+0.133+0.033+0.033=0.332<32.0, passed the test;

[0133] Key generation:

[0134] Corrected sequence: 0xB7D3E1A8... (64 bits);

[0135] Implement chaos expansion: inject the sequence into the Lorenz system as initial conditions;

[0136] Final key: 0x5F3C7A1E... (128 bits, tested according to NIST SP800-90B).

[0137] In one embodiment, the infrastructure layer builds a security enhancement system by integrating a quantum random number generator, an HSM hardware security module, and a Hedera Hashgraph light node. The specific steps are:

[0138] Quantum random number generator integration: The quantum random number generator deployment uses the ID Quantique Clavis platform to set the single-photon detector parameters (wavelength 1550nm, detection efficiency ≥ 35%) and establish an entropy acquisition pipeline (vacuum fluctuation detection → photoelectric conversion → 16-bit analog-to-digital conversion). Combined with quantum tomography verification to ensure pure quantum state output, the three-level entropy pool architecture (quantum entropy pool Q Pool , system entropy pool S Pool Buffer pool B Pool ) and dynamic injection strategy (when Q Pool When the filling degree is >80%, it will automatically turn to S Pool Injecting entropy) maintains minimum entropy;

[0139] HSM hardware security module configuration: complete the key life cycle protection through the management of key generation (generate 256-bit master key using quantum entropy source), secure channel establishment (communicate with encryption service layer through TLS1.3 protocol) and key hierarchical management (master key MK→derived key encryption key KEK→encrypt data encryption key DEK), combined with hardware acceleration NTRU algorithm (lattice basis operation offload, 17.3 times performance improvement) and side channel protection (random delay instruction, constant time processing) to ensure the security of encryption operation;

[0140] Hedera Hashgraph light node deployment: complete the medical data network storage by configuring medical test network parameters (node account 0.0.12345, gRPC / TLS connection) and deploying intelligent storage contract (record_access function, Byzantine fault tolerance consensus).

[0141] In one embodiment, the blockchain anchoring layer builds a medical data trusted anchoring system through sharding hash verification, metadata IPFS storage and consensus security mechanism, including the following steps:

[0142] Sharding hash verification: adopt double SHA3-512 hash calculation combined with dynamic salt value injection technology to ensure data integrity, verify the logic through Solidity smart contract deployment and set access control mechanism to realize on-chain storage, and use event-driven mode to generate verification credentials to resist data tampering;

[0143] Metadata IPFS storage: build a standardized data structure containing encryption parameters and content identifier, realize content addressing by adopting Protobuf serialization combined with CIDv1 multi-hash format, and establish strong association between sharding hash and physical storage location through on-chain metadata registration mechanism;

[0144] Consensus security mechanism: use fan-out propagation strategy and priority queue management of Gossip protocol to ensure efficient message diffusion, combine three-phase voting protocol (PRE-PREPARE / PREPARE / COMMIT) of asynchronous Byzantine fault tolerance algorithm to reach safe consensus in asynchronous network environment, and supplement with dynamic view switching and random number generation technology to enhance system robustness.

[0145] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various changes, modifications, replacements and modifications can be made to the embodiments without departing from the principles and spirits of the present application, and the scope of the present application is limited by the appended claims and their equivalents.

Claims

1. A medical information management system, characterized in that: include: User interaction layer: provides an interactive interface, undertakes parameter input, operation audit and result visualization functions, builds a responsive operation interface based on the React framework, and integrates the biometric authentication module and ECharts visualization dashboard; Encryption service layer: A dynamic weight sharding engine uses a CNN classifier fine-tuned with ResNet-50 to identify medical data types. A BERT-NLP model is used to parse text medical record entities. Kafka Streams is used to process streaming monitoring data. The storage strategy is dynamically adjusted based on the sharding weight calculation formula. Encryption resource library formulas are called according to the extraction code sequence. A Lattice-based KDF engine is deployed to implement the NTRU quantum-resistant encryption algorithm. This is combined with an HMAC-SHA3-512 dynamic bit extractor to generate sharding keys. A dual-chaos encryption pipeline is constructed by combining a Tent map generator and a Lorenz perturbation injector to implement dynamic offset protection for encryption resource library formulas. Infrastructure layer: The ID Quantique Clavis platform's quantum random number generator provides a true random entropy source, integrates an HSM hardware security module to accelerate cryptographic operations, and deploys a Hedera Hashgraph light node for blockchain infrastructure connectivity. Data storage layer: Ceph distributed storage cluster is used to configure erasure coding strategies, combined with CockroachDB distributed database to manage shard indexes, and ELK log chain is used to achieve full process audit tracking; Blockchain anchoring layer: Use Solidity smart contracts to verify shard hashes, use the IPFS network to store metadata, and adopt the Gossip protocol and asynchronous Byzantine fault tolerance mechanism to ensure consensus security.

2. A medical information management system according to claim 1, characterized in that: The specific steps of constructing a complete interactive system covering parameter verification, security authentication, operation monitoring and exception handling at the user interaction layer are as follows: Dynamic form validation: React state hooks and regular expressions are used to implement dynamic form validation. Input event monitoring and character counters are used to ensure that the encryption code complies with the 12-24 digit standard. The Set data structure is used to detect the uniqueness of the formula extraction code and disable the input of duplicate values. Biometric authentication: Integrate fingerprint / face recognition SDK to build a two-factor authentication mechanism, configure a timeout countdown and a three-level failure lockout policy, and automatically downgrade to password verification when biometric authentication fails; ECharts visualization dashboard construction: Visualize operational audits through ECharts heat maps, Sankey diagrams, and radar charts, map time series, geographic distribution, and risk indicators into graphical elements, and use color gradients and dynamic updates to display the spatiotemporal heat distribution of encryption / decryption operations and system security status; Shard status monitoring: Use WebSocket to establish a real-time status channel, maintain shard transmission monitoring through a heartbeat mechanism, use a progress bar and status indicator to provide real-time feedback on the transmission progress, and trigger an alarm and highlight the abnormal shard when a stagnation is detected; Interface integration and exception handling: Complete interface integration through routing configuration and error boundaries, use ARIA attributes to enhance accessibility, and deploy global error capture and logging to display friendly prompts with error codes when the system is abnormal.

3. A medical information management system according to claim 1, characterized in that: The specific steps of the encryption service layer using the dynamic weight sharding engine to intelligently classify and store medical data and dynamically encrypt and protect it are as follows: Medical data feature analysis: We use a CNN classifier fine-tuned based on ResNet-50 to perform modality recognition on medical images, combined with the BERT-NLP model to extract diagnostic entities from text medical records, and use KafkaStreams to analyze abnormal fluctuation patterns in streaming monitoring data in real time. Shard weight calculation: Based on the preset storage medium baseline value, type weight table, and privacy sensitivity coefficient matrix, a storage policy is dynamically generated using the shard size calculation formula V = I × W × R, where I is the storage medium baseline value, W is the type weight, and R is the privacy coefficient. Dynamic storage policy adjustment: Automatically splits data based on sharding rules, uses intelligent routing algorithms to comprehensively consider storage cost models and performance constraints, and implements automatic scaling by monitoring shard queue lengths and storage utilization. Encryption formula call: According to the formula extraction code sequence set by the user, the corresponding mathematical formula is called from the encryption resource library, and the encrypted digital character set extraction value and chaos obfuscation encryption technology are combined to generate a hidden key for data protection.

4. A medical information management system according to claim 3, characterized in that: The encryption service layer deploys the Lattice-based KDF engine to implement the NTRU quantum-resistant encryption algorithm, and cooperates with the HMAC-SHA3-512 dynamic bit extractor to generate sharding keys, including the following steps: Lattice space mapping and perturbation polynomial generation: The encrypted code is mapped to the NTRU lattice space after preprocessing, and the perturbation polynomial is generated by the polynomial ring Z[x] / (x n-1 ) constructs a high-dimensional vector, where n = 503 is a security parameter, and the extraction code set generates a perturbation polynomial through hash operation Coefficient (a i , b i ) performing ring learning error sampling on the first 64 bits of the HMAC-SHA3-512 value of the extraction code, injecting discrete Gaussian noise into the lattice space, and generating intermediate key material; Dynamic bit extraction strategy: Perform HMAC-SHA3-512 operation on each extraction code to generate a 512-bit digest value. The extraction position is determined by the last 4 bits of the digest value. A sliding window mechanism is used. When the extraction position exceeds the current shard key length, it automatically loops to the starting bit. Sharding key construction: The intermediate key generated by each extraction code is calculated according to the formula F′(x+δ), where δ is the chaotic offset generated by the Tent mapping. Specified bits are intercepted from the result value according to the dynamic extraction position to form a 128-bit sharding key. The sharding key is then XORed with the random number generated by the quantum entropy source.

5. A medical information management system according to claim 4, characterized in that: The encryption service layer combines the Tent map generator and the Lorenz perturbation injector to build a dual chaos encryption pipeline, and implements the specific steps of dynamic offset protection for the encryption resource library formula; Initialization of the dual chaotic system: A combination of dynamic Tent mapping parameter configuration and classical Lorenz system parameter setting is used. The quantum entropy source is used to randomly generate the initial state and implement range constraints. The dynamic Tent mapping parameter configuration generates a dynamic μ value, expressed as μ(t) = 1 + 0.2 sin(t / 1000), where t is the system running time. The rounding function is used to make μ(t)∈[1.01,1.19]. The classical Lorenz system parameter setting: the classical chaotic parameter set σ = 10 is the Prandtl number, ρ = 28 is the Rayleigh number, and β = 8 / 3 is the geometric parameter. Chaotic sequence generation: Through the iterative calculation of Tent map and the numerical solution of Lorenz system, the iterative formula, range folding and fourth-order Runge-Kutta method are used to generate a sequence with strong chaotic characteristics. The iterative formula is expressed as x n+1 =μ(t)·x n ·(1-x n ), when x n ≥1, execute x n =1-x n ; Dynamic offset protection: Adopting a formula segmentation strategy and chaos offset injection, the mathematical formula is divided into multiple protection intervals and an independent offset is applied to each segment. At the same time, the chaos parameters and offset are adjusted in real time through dynamic parameter updates.

6. A medical information management system according to claim 5, characterized in that: The encryption service layer includes a quantum entropy management module, which constructs a quantum secure random number generation system through hybrid entropy source scheduling and dynamic bias correction. Hybrid entropy source dynamic scheduling: Build a three-level priority queue with quantum entropy as the primary source, system entropy as the secondary source, and pseudo-random numbers as the backup source. Combined with a dynamic switching strategy, the entropy source is automatically selected based on the minimum entropy health and transmission rate. Continuous monitoring of entropy source health: When quantum entropy is insufficient, the hybrid system entropy maintains availability and continuously monitors the minimum entropy Hmin value through sliding window analysis to isolate the fault source. The minimum entropy Hmin value calculation formula is expressed as Hmin = -log2(max(P(b i ))), where b i All possible values ​​for an 8-bit byte; Anti-bias correction processing: Anti-bias correction uses the VonNeumann extraction algorithm to map the original bit stream into an unbiased sequence, combined with the χ 2 The system verifies dynamic threshold settings, automatically adjusts the significance level based on the shard key length, and improves the effective output rate through four-bit grouping preprocessing, ultimately generating quantum-safe random numbers that pass the NIST SP800-90B test.

7. A medical information management system according to claim 1, characterized in that: The infrastructure layer builds a security enhancement system by integrating a quantum random number generator, an HSM hardware security module, and a Hedera Hashgraph light node. The specific steps are as follows: Quantum random number generator integration: The quantum random number generator deployment uses the ID Quantique Clavis platform to set single-photon detector parameters and establish an entropy collection pipeline, maintaining minimum entropy through a three-level entropy pool architecture and dynamic injection strategy; HSM hardware security module configuration: completes key lifecycle protection by managing key generation, secure channel establishment, and key hierarchical management; Hedera Hashgraph light node deployment: Complete network-wide medical data storage by configuring medical test network parameters and deploying smart storage contracts.

8. A medical information management system according to claim 1, characterized in that: The blockchain anchoring layer builds a trusted anchoring system for medical data through shard hash verification, metadata IPFS storage, and consensus security mechanisms, including the following steps: Shard hash verification: Double SHA3-512 hash calculation combined with dynamic salt injection technology ensures data integrity. Verification logic is deployed through Solidity smart contracts and access control mechanisms are set up to achieve on-chain evidence storage. Verification credentials are generated using an event-driven model to resist data tampering. Metadata IPFS storage: Build a standardized data structure containing encryption parameters and content identifiers, use Protobuf serialization with the CIDv1 multi-hash format to achieve content addressing, and establish a strong association between shard hashes and physical storage locations through the on-chain metadata registration mechanism; Consensus security mechanism: The fan-out propagation strategy and priority queue management of the Gossip protocol are used to ensure efficient message diffusion. The three-phase voting protocol based on the asynchronous Byzantine fault-tolerant algorithm is combined to achieve secure consensus in an asynchronous network environment. Dynamic view switching and random number generation technology are used to enhance system robustness.

Images