Internal fistula maintenance data sharing and early warning system based on block chain

Abstract

The invention relates to the technical field of medical information, in particular to an internal fistula maintenance data sharing and early warning system based on a block chain, which comprises a data acquisition and encapsulation module, a block chain evidence storage and sharing module and an intelligent early warning module, the data acquisition and encapsulation module acquires internal fistula multi-source heterogeneous data of a patient, and encapsulates the data into a standardized internal fistula event data structure containing event types, anonymization identifiers and the like; the block chain evidence storage and sharing module adopts an alliance chain architecture, encrypts standardized data to generate a hash value, forms a data fingerprint chain in combination with metadata, and manages access authority based on cryptography; the intelligent early warning module calculates a risk coefficient through a risk assessment model, an integrated parameter deviation coefficient and a complication probability coefficient, then judges a risk grade according to a threshold value and triggers a graded early warning action and on-chain records, so that safe sharing and accurate early warning of internal fistula data are realized, and the internal fistula maintenance quality of a hemodialysis patient is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of medical information technology, and more specifically, to a blockchain-based data sharing and early warning system for arteriovenous fistula maintenance. Background Technology

[0002] An arteriovenous fistula (AVF) is a vascular access created through surgical anastomosis of a patient's own artery and vein, or by using other materials and instruments (such as an artificial arteriovenous fistula). It is primarily used for hemodialysis. The AVF is a lifeline for hemodialysis patients, and its maintenance quality directly determines the effectiveness of dialysis and their quality of life. The essence of "AVF maintenance data sharing and early warning" is to transform scattered medical data into a "predictable, collaborative, and traceable" management tool through technological means. This extends the lifespan of the dialysis access and reduces patient suffering at the patient level; improves efficiency and reduces decision-making bias at the clinical level; and promotes standardized and tiered management at the healthcare system level, ultimately achieving proactive protection of this "lifeline."

[0003] Data sharing for arteriovenous fistulas (AVFs) largely relies on the hospital's HIS system to integrate data collected from dialysis equipment, Doppler ultrasound, and angiography, including dialysis blood flow, venous pressure, vessel diameter, lesion site parameters, and AVF blood flow. This data is available for review by hospital departments. Alternatively, data can be shared between doctors and patients or across institutions via remote platforms using wearable devices at home. Early warning systems are primarily based on preset thresholds. When monitored data (such as blood flow and venous pressure) exceeds the normal range, the system marks the abnormality and sends a notification to medical staff. Some wearable devices will directly send alarm signals to patients, while a few solutions combine basic models to optimize data quality and improve the accuracy of early warnings.

[0004] The existing technology has the following technical defects, specifically: 1. Low data sharing efficiency and insufficient privacy protection: Existing arteriovenous fistula data sharing relies on the hospital's HIS system or basic remote platform. Data is scattered in dialysis centers, ultrasound departments, laboratories, etc., forming "data silos". Cross-hospital / cross-department access requires manual aggregation, which is inefficient. Moreover, the lack of reliable evidence storage and fine-grained access control during the sharing process can easily lead to the leakage of patient privacy, making it impossible to balance "sharing" and "security".

[0005] 2. Lack of accuracy in early warning and reliance on fixed thresholds: Existing early warning systems are mostly based on preset fixed thresholds (such as alarms when blood flow is below a certain value), without dynamic adjustment based on historical stable period data. This can easily lead to missed detections for high-risk patients (such as early stenosis signals in diabetic patients being ignored by fixed thresholds) or over-warnings for low-risk patients, resulting in poor clinical adaptability. Summary of the Invention

[0006] The purpose of this invention is to provide a blockchain-based data sharing and early warning system for arteriovenous fistula maintenance, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention aims to provide a blockchain-based arteriovenous fistula maintenance data sharing and early warning system, comprising: a data acquisition and encapsulation module: used to acquire multi-source heterogeneous data of a patient's arteriovenous fistula and encapsulate the multi-source heterogeneous data into a standardized fistula event data structure.

[0008] The blockchain evidence storage and sharing module adopts a consortium blockchain architecture to store the encrypted hash values ​​of the multi-source heterogeneous data of the patient's arteriovenous fistula, form a data fingerprint chain, and manage data access permissions based on cryptographic principles.

[0009] The intelligent early warning module is used to analyze the risk of arteriovenous fistula in patients based on a risk assessment model, obtain the risk coefficient of the patient's arteriovenous fistula, and then determine the risk level of the patient's arteriovenous fistula based on the risk coefficient, triggering early warning actions and recording corresponding to the risk level.

[0010] As a further improvement to this technical solution, the multi-source heterogeneous data of the patient's arteriovenous fistula includes dialysis parameters, ultrasound measurement data, imaging report data, and patient self-evaluation data.

[0011] As a further improvement to this technical solution, the standardized arteriovenous fistula event data structure includes event type, globally unique event identifier, anonymized identifier of associated patient, precise timestamp, digital signature of data provider, off-chain data pointer, and data hash value.

[0012] As a further improvement to this technical solution, the data fingerprint chain is specifically implemented as follows: encrypting multi-source heterogeneous data of the patient's arteriovenous fistula to generate an encrypted data packet, using an encrypted hash algorithm to calculate the unique hash value of the data packet, associating and packaging this hash value with key metadata, including the data generation timestamp, device identifier, an anonymized patient identifier generated by a public key encryption algorithm, and data type encoding, broadcasting this data storage transaction to the blockchain, and verifying it through a practical Byzantine fault-tolerant consensus algorithm to form a data fingerprint chain.

[0013] As a further improvement to this technical solution, the specific implementation method of the patient's arteriovenous fistula risk coefficient is as follows: The patient's parameter deviation coefficient and complication probability coefficient are obtained from the database; these coefficients are then input into the risk coefficient assessment model to calculate the patient's arteriovenous fistula risk coefficient. The model expression is: Where R represents the patient's arteriovenous fistula risk coefficient, S represents the patient's parameter deviation coefficient, P represents the patient's complication probability coefficient, γ1 represents the preset parameter deviation weighting factor in the database, and γ2 represents the preset complication probability weighting factor in the database.

[0014] As a further improvement to this technical solution, the parameter deviation coefficient of the patient is specifically implemented as follows: obtain each key physiological parameter during the patient's historical stable period from the database, train a time series prediction model based on each key physiological parameter during the patient's historical stable period, generate the baseline predicted value and normal fluctuation range of each key physiological parameter, calculate the absolute deviation between the monitored value of each key physiological parameter and the baseline predicted value, divide the deviation of each key physiological parameter by the difference between the upper limit of the normal fluctuation range and the baseline value to obtain the normalized deviation coefficient of each key physiological parameter, and sum them by weight to obtain the patient's parameter deviation coefficient.

[0015] As a further improvement to this technical solution, the patient's complication probability coefficient is specifically implemented as follows: multimodal time-series data of the patient within a specific time window is obtained from the database, standardized, and the standardized data sequence is input into a deep learning model that is collaboratively optimized by a federated learning framework for analysis to obtain the predicted probability value of the patient's arteriovenous fistula failure within a preset future time period, which is recorded as the patient's complication probability coefficient.

[0016] As a further improvement to this technical solution, the specific method for determining the risk level of a patient's arteriovenous fistula is as follows: extract the patient's arteriovenous fistula risk coefficient, input it into the risk level determination model, and output the determination result of the patient's arteriovenous fistula risk level.

[0017] The assessment result of the patient's arteriovenous fistula risk level includes values ​​of 0, 1, and 2. If the assessment result is 0, the patient's arteriovenous fistula risk level is determined to be low risk; if the assessment result is 1, the patient's arteriovenous fistula risk level is determined to be medium risk; and if the assessment result is 2, the patient's arteriovenous fistula risk level is determined to be high risk.

[0018] As a further improvement to this technical solution, the risk level judgment model is expressed as follows: Where A represents the patient's arteriovenous fistula risk level, R represents the patient's arteriovenous fistula risk coefficient, T1 represents the preset risk coefficient boundary threshold between low risk and medium risk in the database, and T2 represents the preset risk coefficient boundary threshold between medium risk and high risk in the database.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Achieve secure and efficient sharing of arteriovenous fistula (AVF) data, breaking the "data silo" problem: Through the blockchain notarization and sharing module, the standardized AVF event data structure is encrypted to generate a data fingerprint chain. Combined with the consortium blockchain architecture and cryptographic permission management, it ensures real-time data sharing across institutions (hospitals, primary dialysis centers) and protects patient privacy through "data usable but not visible" (only sharing hash values ​​and metadata). At the same time, the data fingerprint chain can verify data integrity, prevent tampering, and solve the imbalance between "efficiency" and "security" in the existing sharing.

[0020] 2. Improve the accuracy of early warning and adapt to individual patient differences: The intelligent early warning module trains a time series model based on the patient's historical stable period data to generate individualized baseline prediction values ​​and fluctuation ranges. Combined with a complication probability model optimized by federated learning, it dynamically calculates the risk coefficient. Instead of relying on a fixed threshold, it can accurately capture early risk signals of different patients (such as diabetic patients and elderly patients), reduce the rate of missed judgments and over-warnings, and improve the practicality of clinical early warning. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the system structure connection of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Example: Please refer to Figure 1 As shown, a blockchain-based arteriovenous fistula maintenance data sharing and early warning system is provided, including: a data acquisition and encapsulation module: used to acquire multi-source heterogeneous data of the patient's arteriovenous fistula and encapsulate the multi-source heterogeneous data into a standardized fistula event data structure.

[0025] The standardized arteriovenous fistula event data structure includes event types (used to identify predefined data properties such as "dialysis parameters" and "ultrasound measurements"), globally unique event identifiers (ensuring the non-repeatability of each record), anonymized identifiers associated with patients (achieving data aggregation while protecting privacy), precise timestamps (establishing a time series of events), digital signatures from data providers (ensuring the authenticity and non-repudiation of the data source), off-chain data pointers (pointing to the storage address of large amounts of raw data such as ultrasound images), and data hash values ​​(used to verify the integrity and consistency of the raw data). Implement it according to the following process: First, the system triggers a specific type of "arteriovenous fistula event" instance based on the data source (such as a dialysis machine, ultrasound equipment, medical staff input terminal, or patient APP). Next, the data encapsulation engine automatically generates a globally unique identifier for the event and adds a precise timestamp. Then, the engine assembles the incoming raw data payload (such as {"blood_flow":650, "venous_pressure":180}) with preset metadata (event type, patient anonymous ID, etc.) to form an initial data structure. Subsequently, the system calculates the complete content of this initial data structure (or the original data file it points to) to generate a unique encrypted hash value. Finally, the data provider uses its private key to digitally sign the hash value and key metadata, thus completing the encapsulation of a standardized arteriovenous fistula event containing the data ontology, its integrity proof (hash), and its source proof (signature). This event is then submitted to the blockchain network for notarization.

[0026] In one specific embodiment, the multi-source heterogeneous data of the patient's arteriovenous fistula includes dialysis parameters, ultrasound measurement data, imaging report data, and patient self-assessment data.

[0027] The dialysis parameters include, but are not limited to, blood flow, arterial and venous pressure, dynamic venous pressure, transmembrane pressure, treatment duration, and ultrafiltration volume.

[0028] The ultrasound measurement data includes, but is not limited to, vessel diameter, blood flow velocity, intima thickness, and puncture site location and condition.

[0029] The image report data includes, but is not limited to, ultrasound image report text, angiography report, and evaluation results of magnetic resonance angiography.

[0030] The patient self-assessment data includes, but is not limited to, self-perception during physical examination, puncture site condition, and daily symptoms.

[0031] In one specific embodiment, the standardized arteriovenous fistula event data structure includes an event type, a globally unique event identifier, an anonymized identifier associated with the patient, a precise timestamp, a digital signature of the data provider, an off-chain data pointer, and a data hash value.

[0032] The blockchain evidence storage and sharing module adopts a consortium blockchain architecture to store the encrypted hash values ​​of the multi-source heterogeneous data of the patient's arteriovenous fistula, form a data fingerprint chain, and manage data access permissions based on cryptographic principles.

[0033] The consortium blockchain architecture consists of complete nodes operated by pre-approved and authorized entities, including at least medical institutions (such as top-tier hospitals and dialysis centers), health administrative regulatory agencies, and independent medical imaging data centers, to ensure decentralized trust and stable operation of the system.

[0034] In one specific embodiment, the data fingerprint chain is implemented as follows: encrypting multi-source heterogeneous data of a patient's arteriovenous fistula to generate an encrypted data packet; using an encrypted hash algorithm to calculate the unique hash value of the data packet; associating and packaging this hash value with key metadata, including a data generation timestamp, device identifier, an anonymized patient identifier generated by a public-key encryption algorithm, and data type encoding; broadcasting this data storage transaction to the blockchain; and verifying it through a practical Byzantine fault-tolerant consensus algorithm to form a data fingerprint chain.

[0035] It should be noted that the aforementioned cryptographic hash algorithm and public key encryption algorithm are existing technologies and will not be described in detail here.

[0036] The intelligent early warning module is used to analyze the risk of arteriovenous fistula in patients based on a risk assessment model, obtain the risk coefficient of the patient's arteriovenous fistula, and then determine the risk level of the patient's arteriovenous fistula based on the risk coefficient, triggering early warning actions and recording corresponding to the risk level.

[0037] In one specific embodiment, the patient's arteriovenous fistula risk coefficient is specifically implemented as follows: The patient's parameter deviation coefficient and complication probability coefficient are obtained from the database; these coefficients are then input into the risk coefficient assessment model to calculate the patient's arteriovenous fistula risk coefficient. The model expression is as follows: Where R represents the patient's arteriovenous fistula risk coefficient, S represents the patient's parameter deviation coefficient, P represents the patient's complication probability coefficient, γ1 represents the preset parameter deviation weighting factor in the database, and γ2 represents the preset complication probability weighting factor in the database.

[0038] The pre-set parameter deviation weighting factor and complication probability weighting factor in the database need to be comprehensively set in conjunction with multi-center clinical data statistics, the pathological mechanism of arteriovenous fistula complications, and expert consensus. First, by retrospectively analyzing historical data of large-scale arteriovenous fistula maintenance (covering different patient characteristics, complication types, and corresponding parameter changes), the correlation strength between deviations of each monitored parameter (such as blood flow, stenosis rate, and thrill intensity) from the normal range and the occurrence of complications is statistically analyzed. The higher the correlation, the larger the parameter deviation weighting factor is set. At the same time, based on the clinical probability and severity of arteriovenous fistula complications (thrombosis, stenosis, infection, etc.), combined with the expert team's (nephrology, vascular surgery, radiology) judgment on the risk level of different complications, the probability weighting factor corresponding to high-risk and high-frequency complications (such as thrombosis) is set to a higher value. Subsequently, the two types of weighting factors need to be dynamically optimized through real-time clinical feedback and model iteration (such as including new patient group data and updating complication diagnosis and treatment guidelines) to ensure that they can accurately reflect the actual impact of parameter abnormalities on complication risk and improve the accuracy and clinical suitability of risk assessment.

[0039] In one specific embodiment, the parameter deviation coefficient of the patient is specifically implemented as follows: obtain each key physiological parameter during the patient's historical stable period from the database, train a time series prediction model based on each key physiological parameter during the patient's historical stable period, generate the baseline predicted value and normal fluctuation range of each key physiological parameter, calculate the absolute deviation between the monitored value of each key physiological parameter and the baseline predicted value, divide each key physiological parameter deviation by the difference between the upper limit of the normal fluctuation range and the baseline value to obtain the normalized deviation coefficient of each key physiological parameter, and sum them by weight to obtain the patient's parameter deviation coefficient.

[0040] The training time series prediction model first extracts time-series data of key physiological parameters (such as blood flow, venous pressure, etc.) from the database during the stable period of patients' arteriovenous fistula maintenance history (which must first pass clinical criteria screening, such as no complication records and continuous time periods with key physiological parameter fluctuations ≤ clinically recognized stable thresholds). The time granularity must be uniform, such as once per hour or once per dialysis cycle. Data preprocessing is then performed: outliers are removed (such as extreme values ​​caused by equipment failure, using the 3σ principle or IQR method), missing values ​​are filled (using linear interpolation or forward imputation), and data smoothing is performed (such as using moving averages to reduce short-term noise). Next, the dataset is divided into a training set (for model fitting) and a validation set (for hyperparameter tuning) in a 7:3 ratio. A suitable model for medical time-series data is selected (such as LSTM, ARIMA, or Prophet, with LSTM preferred for capturing...). The model captures potential temporal correlations between parameters. It uses "parameter values ​​over N consecutive time steps during a historical stable period" as input features and "parameter value at the (N+1)th time step" as output labels. The model is trained on the training set, and its prediction accuracy is evaluated using MAE (mean absolute error) and RMSE (root mean square error) on the validation set. Hyperparameters (such as the number of hidden layer nodes and learning rate of LSTM, and the p, d, and q orders of ARIMA) are repeatedly tuned until the model's prediction error on the validation set is below a preset threshold. Finally, the trained model is solidified to generate baseline predicted values ​​and normal fluctuation ranges for subsequent key physiological parameters of the patient. (The normal fluctuation range is determined by calculating the residual standard deviation between the model's predicted values ​​and actual values ​​during the historical stable period, using the formula "baseline predicted value ± k × residual standard deviation," where k is a clinically accepted confidence coefficient, such as 1.96 corresponding to a 95% confidence interval.)

[0041] In one specific embodiment, the complication probability coefficient of the patient is specifically implemented by: obtaining the patient's multimodal time-series data within a specific time window from the database, standardizing the data, inputting the standardized data sequence into a deep learning model co-optimized by a federated learning framework for analysis, and obtaining the predicted probability value of the patient's fistula failure within a preset future time period, which is denoted as the patient's complication probability coefficient.

[0042] The "specific time window" here refers to a fixed period of historical data collection set for the purpose of accurately extracting data related to the patient's arteriovenous fistula function and supporting the analysis of complication prediction models. Its core is to ensure that the multimodal time-series data (such as physiological parameters, imaging features, clinical records, etc.) input into the model have "timeliness, relevance, and completeness" by limiting the time range of the data. This ensures that the data can truly reflect the recent stable state or early trend of changes in the patient's arteriovenous fistula function, and provide reliable data support for predicting the risk of arteriovenous fistula failure in the future within a preset time period.

[0043] The core of calculating the complication probability lies in the intelligent analysis process of standardized data sequences by a deep learning model optimized through federated learning. This analysis process is as follows: First, the model transforms categorical variables (such as physical sign types) in the multimodal data into dense vectors through an embedding layer, while simultaneously aligning the dimensions of continuous variables. Subsequently, these feature vectors are input to a temporal feature extraction module—typically employing a Long Short-Term Memory (LSTM) network or a temporal convolutional network layer. This module, through its recurrent connections or causal convolutional structures, specifically captures the dynamic patterns of parameter evolution over time and their short-term dependencies. After extracting the temporal features, the model performs high-level abstraction through a self-attention mechanism or fully connected layers, identifying complex nonlinear interaction patterns across parameters (such as specific combinations of decreased blood flow and increased venous pressure). These patterns are highly correlated with the pathophysiological process of arteriovenous fistula failure. Finally, the model maps these learned high-risk feature patterns to a continuous probability value between 0 and 1 through a sigmoid output layer. This probability value quantitatively represents the likelihood of complications such as stenosis and thrombosis occurring in a specific future time period inferred from the current multimodal data sequence. Throughout the analysis, the model's weight parameters all originate from periodic collaborative optimization within the federated learning framework, ensuring that it can learn global common patterns without requiring the centralized processing of sensitive original data. In one specific embodiment, the method for determining the risk level of a patient's arteriovenous fistula is as follows: extract the patient's arteriovenous fistula risk coefficient, input it into the risk level determination model, and output the determination result of the patient's arteriovenous fistula risk level.

[0044] The assessment result of the patient's arteriovenous fistula risk level includes values ​​of 0, 1, and 2. If the assessment result is 0, the patient's arteriovenous fistula risk level is determined to be low risk; if the assessment result is 1, the patient's arteriovenous fistula risk level is determined to be medium risk; and if the assessment result is 2, the patient's arteriovenous fistula risk level is determined to be high risk.

[0045] In one specific embodiment, the risk level assessment model is expressed as follows: Where A represents the patient's arteriovenous fistula risk level, R represents the patient's arteriovenous fistula risk coefficient, T1 represents the preset risk coefficient boundary threshold between low risk and medium risk in the database, and T2 represents the preset risk coefficient boundary threshold between medium risk and high risk in the database.

[0046] The pre-set risk coefficient cutoff thresholds for low and medium risk, and for medium and high risk in the database, need to be comprehensively set based on clinical prognostic data of arteriovenous fistula failure, validation results of multi-center risk assessment models, and consensus among clinical experts. First, a retrospective analysis of large-scale arteriovenous fistula patient cohort data is used to statistically determine the actual incidence of arteriovenous fistula failure corresponding to different risk coefficients (e.g., a risk coefficient of 0.3 corresponds to a failure rate of 5% within 3 months, a risk coefficient of 0.6 corresponds to a failure rate of 30%, and a risk coefficient of 0.8 corresponds to a failure rate of 70%). The low-to-medium risk cutoff threshold is determined using "clinically acceptable low failure rate (e.g., ≤10%)" as the standard (e.g., setting a risk coefficient of 0.4 corresponding to a failure rate of 10% as the low-to-medium risk cutoff). The threshold for "high failure rate requiring urgent intervention" is determined based on... The threshold for dividing medium- to high risk is determined based on the standard of "probability (e.g., ≥50%)" (e.g., setting the risk coefficient of 0.7 corresponding to 50% failure rate as the medium- to high threshold). Subsequently, the effectiveness of the threshold is verified using multicenter prospective data. If the degree of agreement between the risk level predicted by the model and the actual failure results (e.g., accuracy, recall) under a certain threshold is lower than the preset clinical standard, the threshold is adjusted in conjunction with the assessment of intervention costs and patient benefits by the expert team (nephrology, vascular surgery) (e.g., fine-tuning the medium- to high threshold from 0.7 to 0.65 to reduce the missed diagnosis of high-risk cases). Finally, the determined threshold is stored in the database, while reserving a dynamic update interface, which can optimize the threshold according to new clinical data, model iteration results, or updates to treatment guidelines to ensure that the threshold always matches the actual clinical needs.

[0047] The system triggers warning actions and records corresponding to risk levels, with a core design centered around "matching intervention intensity to risk levels and ensuring full-process traceability." This is specifically manifested in three dimensions: tiered push notifications of warning information, linkage with clinical collaboration processes, and automatic generation of traceable records. Furthermore, all actions and records are stored on the blockchain to ensure compliance and auditability. Low risk: The warning action only pushes a light prompt (such as an APP pop-up reminder "The current arteriovenous fistula is stable, it is recommended to monitor the tremor at the regular frequency") and the attending physician's workstation (such as a light prompt in the to-do list), without triggering cross-departmental collaboration; the recorded content includes the risk coefficient, the basis for low risk determination (such as the deviation of each parameter being within the normal range), the push time and the receiving status, and is automatically synchronized to the patient's blockchain health record.

[0048] Medium risk: In addition to pushing notifications to patients (such as reminding them that "blood flow has decreased slightly recently and home monitoring needs to be strengthened") and attending physicians, early warning actions also automatically trigger in-hospital collaboration processes through smart contracts (such as pushing examination tasks to the ultrasound department and scheduling an ultrasound follow-up examination of the arteriovenous fistula within one week; simultaneously notifying the nursing station to arrange puncture point assessment); the recorded content adds the collaborative task allocation information (such as the ultrasound department's receiving time and examination appointment time), and the doctor's preliminary assessment opinions. All collaborative node operations are recorded on the blockchain in real time, forming a closed-loop record of tasks.

[0049] High Risk: The early warning action activates the emergency response mechanism, including sending high-risk alerts to patients (such as strong reminders on the APP and SMS notifications such as "High risk of fistula thrombosis, it is recommended to seek medical attention within 24 hours"), and simultaneously sending emergency warnings to attending physicians, vascular surgeons, and the emergency department (such as system pop-ups and telephone notifications). The system automatically retrieves the patient's historical fistula data (such as previous intervention records and ultrasound images) through smart contracts for the collaborative team to review, and generates an emergency priority treatment QR code. The recorded content covers key parameters that trigger high risk (such as a sudden drop in blood flow of 30%), the start time of the emergency collaboration process, the response time of each department, the final intervention plan (such as whether balloon dilation is performed), and the postoperative effect evaluation. All records are tamper-proof and linked to the digital signatures of each operation node to ensure traceability of responsibility.

[0050] It should be noted that it also includes a database.

[0051] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A blockchain-based data sharing and early warning system for arteriovenous fistula maintenance, characterized in that, include: Data acquisition and encapsulation module: used to acquire multi-source heterogeneous data of the patient's arteriovenous fistula and encapsulate the multi-source heterogeneous data into a standardized fistula event data structure; The blockchain evidence storage and sharing module adopts a consortium blockchain architecture to store the encrypted hash values ​​of the multi-source heterogeneous data of the patient's arteriovenous fistula, form a data fingerprint chain, and manage data access permissions based on cryptographic principles. The intelligent early warning module is used to analyze the risk of arteriovenous fistula in patients based on a risk assessment model, obtain the risk coefficient of the patient's arteriovenous fistula, and then determine the risk level of the patient's arteriovenous fistula based on the risk coefficient, triggering early warning actions and recording corresponding to the risk level.

2. The blockchain-based arteriovenous fistula maintenance data sharing and early warning system according to claim 1, characterized in that, The multi-source heterogeneous data of the patient's arteriovenous fistula includes dialysis parameters, ultrasound measurement data, imaging report data, and patient self-assessment data.

3. The blockchain-based arteriovenous fistula maintenance data sharing and early warning system according to claim 1, characterized in that, The standardized arteriovenous fistula event data structure includes event type, globally unique event identifier, anonymized identifier of associated patient, precise timestamp, digital signature of data provider, off-chain data pointer, and data hash value.

4. The blockchain-based arteriovenous fistula maintenance data sharing and early warning system according to claim 1, characterized in that, The data fingerprint chain is specifically implemented as follows: The multi-source heterogeneous data of the patient's arteriovenous fistula is encrypted to generate an encrypted data packet. The unique hash value of the data packet is calculated using an encrypted hash algorithm. This hash value is then associated with key metadata and packaged together. The metadata includes the data generation timestamp, device identifier, an anonymized patient identifier generated by a public key encryption algorithm, and data type encoding. This data storage transaction is broadcast to the blockchain and verified by a practical Byzantine fault-tolerant consensus algorithm to form a data fingerprint chain.

5. The blockchain-based arteriovenous fistula maintenance data sharing and early warning system according to claim 1, characterized in that, The specific method for determining the patient's arteriovenous fistula risk factor is as follows: The parameter deviation coefficient and complication probability coefficient of the patient are obtained from the database. These coefficients are then input into the risk coefficient assessment model to calculate the patient's arteriovenous fistula risk coefficient. The model expression is as follows: Where R represents the patient's arteriovenous fistula risk coefficient, S represents the patient's parameter deviation coefficient, P represents the patient's complication probability coefficient, γ1 represents the parameter deviation weighting factor preset in the database, and γ2 represents the complication probability weighting factor preset in the database.

6. The blockchain-based arteriovenous fistula maintenance data sharing and early warning system according to claim 5, characterized in that, The parameter deviation coefficient of the patient is specifically implemented as follows: Key physiological parameters from the patient's historical stable period are obtained from the database. A time series prediction model is trained based on these key physiological parameters to generate baseline predicted values ​​and normal fluctuation ranges for each key physiological parameter. The absolute deviation between the monitored values ​​and the baseline predicted values ​​of each key physiological parameter is calculated. The deviation of each key physiological parameter is divided by the difference between the upper limit of the normal fluctuation range and the baseline value to obtain the normalized deviation coefficient of each key physiological parameter. The weighted sum is then used to obtain the parameter deviation coefficient for the patient.

7. The blockchain-based arteriovenous fistula maintenance data sharing and early warning system according to claim 5, characterized in that, The specific implementation method for the patient's complication probability coefficient is as follows: Multimodal time-series data of patients within a specific time window are obtained from the database, standardized, and then the standardized data sequence is input into a deep learning model co-optimized by a federated learning framework for analysis. The predicted probability value of arteriovenous fistula failure in patients within a preset future time period is obtained and recorded as the patient's complication probability coefficient.

8. The blockchain-based arteriovenous fistula maintenance data sharing and early warning system according to claim 7, characterized in that, The specific method for determining the risk level of an arteriovenous fistula in a patient is as follows: Extract the patient's arteriovenous fistula risk coefficient, input it into the risk level judgment model, and output the judgment result of the patient's arteriovenous fistula risk level; The assessment result of the patient's arteriovenous fistula risk level includes values ​​of 0, 1, and 2. If the assessment result of the patient's arteriovenous fistula risk level is 0, the patient's arteriovenous fistula risk level is determined to be low risk; if the assessment result of the patient's arteriovenous fistula risk level is 1, the patient's arteriovenous fistula risk level is determined to be medium risk; and if the assessment result of the patient's arteriovenous fistula risk level is 2, the patient's arteriovenous fistula risk level is determined to be high risk.

9. The blockchain-based arteriovenous fistula maintenance data sharing and early warning system according to claim 8, characterized in that, The risk level assessment model is expressed as follows: Where A represents the patient's arteriovenous fistula risk level, R represents the patient's arteriovenous fistula risk coefficient, T1 represents the preset risk coefficient boundary threshold between low and medium risk in the database, and T2 represents the preset risk coefficient boundary threshold between medium and high risk in the database.

Images