Intelligent component blood transfusion and coagulation function dynamic management system and method

Abstract

The present application relates to the technical field of clinical blood transfusion management, in particular to an intelligent component blood transfusion and coagulation function dynamic management system and method, the system comprising: a data acquisition module, a decision algorithm module, a doctor's order generation and pushing module, a whole-process tracking module, a feedback recording module and an algorithm self-optimization module; realizing multi-source medical equipment interface adaptation, dynamic sampling and data preprocessing, constructing a quantitative coupling model based on clinical guidelines and expert consensus, and outputting accurate component blood transfusion schemes; being capable of completing intelligent emergency classification and multi-channel priority pushing, and using fusion positioning technology to track blood product circulation in real time. The method realizes management through the closed-loop steps of data acquisition preprocessing, quantitative decision, hierarchical pushing, whole-process tracking and feedback optimization. The present application can improve the accuracy and timeliness of component blood transfusion, reduce blood product waste, and strengthen the collaborative efficiency of multiple departments, thereby providing strong support for the treatment of patients with massive hemorrhage.

Description

Technical Field

[0001] This invention belongs to the field of clinical blood transfusion management technology, specifically relating to an intelligent dynamic management system and method for component blood transfusion and coagulation function. Background Technology

[0002] Massive traumatic hemorrhage is a common acute and critical condition in clinical emergency departments and operating rooms. The core of its treatment lies in rapidly restoring the patient's blood volume and maintaining stable coagulation function. Precise transfusion of blood components (red blood cells, plasma, platelets, cryoprecipitate, etc.) is crucial for improving patient prognosis. In clinical practice, the coagulation function of patients with massive hemorrhage changes rapidly and dynamically with the rate of bleeding, the amount of blood loss, and treatment interventions. It is necessary to adjust the type, dosage, and proportion of blood components transfused in a timely manner based on real-time monitoring data to avoid refractory bleeding due to deterioration of coagulation function. Currently, clinical component transfusion management mainly relies on traditional manual operation and decentralized system collaboration, which presents the following problems: (1) Low efficiency of data collection and sharing: Coagulation function data needs to be manually recorded by bedside monitors or manually retrieved from the laboratory information system (LIS) of the laboratory. Vital signs data and bleeding status data are scattered across different platforms such as monitors and electronic medical record systems (EMR). The lack of a unified data collection interface and real-time fusion mechanism leads to data transmission delays and the accuracy is easily affected by human factors, making it impossible to provide real-time support for blood transfusion decisions.

[0003] (2) Lack of quantitative and dynamic adjustment capabilities in blood transfusion decisions: Existing blood transfusion protocols mostly rely on medical staff to manually formulate them based on clinical experience and static guidelines. A quantitative decision-making model that combines individual patient characteristics (weight, age), real-time bleeding status (bleeding rate, cumulative blood loss) and changes in coagulation function has not been established. This can easily lead to problems such as imbalanced transfusion ratios, insufficient dosage, or excessive transfusion, which can affect the treatment effect and cause waste of blood products.

[0004] (3) The process of applying for and sending medical orders is cumbersome: the application for component blood requires medical staff to manually enter the medical order information, which is then approved at each level and sent to the blood transfusion department. There is a lack of an intelligent priority determination mechanism based on the urgency of the patient's condition, which results in emergency medical orders not being processed first, delays in blood product preparation and transfusion, and missed opportunities for the best treatment.

[0005] (4) Lack of effective tracking of the entire blood product circulation process: The status information of each link from the blood transfusion department to the hospital to clinical transfusion is not transparent, and there is a lack of real-time positioning and time prediction mechanism. It is impossible to detect abnormal situations such as transportation congestion and delay in a timely manner, which further aggravates the risk of transfusion delay.

[0006] (5) Lack of treatment feedback and algorithm optimization mechanism: The coagulation function re-examination data of patients after infusion lacks correlation mapping with the infusion plan and preoperative monitoring data, and a closed loop link of "collection-decision-infusion-feedback" is not formed, which makes it impossible for the decision algorithm to continuously iterate and optimize according to the actual clinical effect, and the decision accuracy is difficult to improve in long-term application.

[0007] In view of this, the present invention is hereby proposed. Summary of the Invention

[0008] In order to solve the above-mentioned technical problems in the prior art, the present invention provides an intelligent dynamic management system and method for component blood transfusion and coagulation function, which solves the problems of insufficient accuracy of component blood transfusion and low efficiency of process coordination in the treatment of traumatic hemorrhage.

[0009] To achieve the above objectives, the technical solution of the present invention is as follows: Firstly, an intelligent dynamic management system for component blood transfusion and coagulation functions includes: The data acquisition module is used to establish communication connections with multi-source medical devices and information systems, collect patients' coagulation function data, vital signs data, bleeding status data and basic information data in real time, and preprocess the collected data. The decision algorithm module is communicatively connected to the data acquisition module. It has a built-in multi-parameter coupled decision model trained based on clinical guidelines and expert consensus. It is used to receive preprocessed acquisition data and output component blood transfusion plan through quantitative calculation. The medical order generation and push module is connected to the decision algorithm module and is used to automatically generate medical orders for component blood transfusion according to the component blood transfusion plan. After the doctor reviews and confirms the order online, the urgency level of the medical order is determined and the order is pushed to the blood transfusion department and related terminals according to priority. The end-to-end tracking module communicates with the medical order generation and push module to locate and track the blood components from warehousing and transportation to infusion, collect status data at each stage and synchronize them in real time. The feedback recording module is communicatively connected to the data acquisition module, decision algorithm module, and full-process tracking module, respectively, and is used to collect the patient's follow-up data after infusion, establish the association mapping between the collected data, infusion plan and follow-up data, and form a closed-loop data link.

[0010] Furthermore, the data acquisition module includes: Multi-interface adapter unit: It uses the HL7FHIR interface to establish communication with the bedside coagulation monitor and laboratory information system, uses the MQTT protocol to establish communication with the monitor, and uses the Ethernet interface to establish communication with the electronic medical record system. Dynamic sampling unit: Adjusts the sampling frequency dynamically based on the patient's bleeding rate. At that time, coagulation function data was sampled once every 5 minutes; when bleeding... At that time, the sampling frequency is 1 time / 15 minutes; Data preprocessing unit: employing Outlier removal is performed according to principles, noise filtering is done using the Kalman filter algorithm, and the collected data is mapped to a normalization algorithm. Interval.

[0011] Furthermore, the implementation of the Kalman filter algorithm in the data preprocessing unit includes a state update unit and an error covariance update unit; the state update unit executes the following formula:

[0012] in, for Filtered data values ​​at any time for Predicted value at any time For Kalman gain, The original data collected at time k, The observation matrix; Error covariance update unit execution formula:

[0013] in, for Time error covariance It is the identity matrix. for Time error covariance.

[0014] Furthermore, the decision algorithm module includes: Coagulation function scoring unit: used to convert coagulation function data into a unified score using the Z-score normalization algorithm, and to calculate the total coagulation function score; Dynamic weight allocation unit: Employing the entropy-weighted analytic hierarchy process (AHP), the basic weights of each decision indicator are first determined using AHP, then dynamic adjustment coefficients are calculated based on the indicator change rate, ultimately yielding the final weights of each indicator. The dynamic weight allocation unit includes: The basic weight determination sub-unit is used to collect expert scores using a 1 to 9 scale and calculate the basic weights of coagulation function total score, bleeding velocity, mean arterial pressure, cumulative bleeding volume and age using the analytic hierarchy process. Dynamic adjustment coefficient calculation subunit: used to execute the formula:

[0015] in, This is the dynamic adjustment coefficient for indicator x. To adjust the coefficient, This refers to the change in the indicator over a 30-minute period. This represents the average of the normal reference values ​​for the indicator. Final weighted synthesis subunit: used to execute the formula:

[0016] in, The final weight of indicator x, The basic weights for indicator x; Infusion protocol calculation unit: It is used to calculate the infusion dose of each blood component based on the total coagulation function score, final weight and patient basic information through quantitative formula, and dynamically adjust the infusion ratio of blood components based on the total coagulation function score.

[0017] Furthermore, the infusion scheme calculation unit includes: Dosage calculation subunit: used to calculate the infusion dose for red blood cells, plasma, platelets, and cryoprecipitate using quantitative formulas, wherein the red blood cell infusion dose... The calculation formula is:

[0018] in, For target hematocrit, To measure hematocrit, For the patient's weight; Proportional constraint subunit: Built-in dynamic proportional threshold, used to automatically correct the infusion dose of each blood component when the proportion of blood component output by the infusion scheme calculation unit exceeds the corresponding threshold.

[0019] Furthermore, the medical order generation and push module includes: Medical Order Review and Confirmation Unit: Provides an online review interface for doctors, displaying the basis of the infusion plan, patient data, and urgency level. It supports doctors to manually confirm, modify, or reject draft medical orders. Once a doctor confirms and issues a medical order, it will proceed to the subsequent push and execution process. Emergency Level Determination Unit: Used to calculate an emergency coefficient using a quantitative formula, and classify emergency levels into Level 1, Level 2, and Level 3 based on the emergency coefficient. The calculation formula is:

[0020] in, For the rate of bleeding, The bleeding rate threshold, Mean arterial pressure, The mean arterial pressure threshold. The total score for coagulation function; Medical order formatting unit: used to automatically fill in patient information, blood component information, emergency level and infusion deadline, and generate standardized medical orders; Multi-channel push unit: Real-time push is achieved using the WebSocket protocol. It is used to push medical orders confirmed by doctors to the blood transfusion department and related terminals. Medical orders are pushed through multiple channels such as blood transfusion department terminals, medical staff mobile APP, and transport personnel terminals according to the urgency level.

[0021] Furthermore, the end-to-end tracking module includes: Identification binding unit: used to establish a unique association between the blood product barcode and the RFID tag through a barcode scanner. The RFID tag stores information such as the blood product number, type, dosage, expiration date, and doctor's order number. Real-time positioning unit: Adopting a fusion solution of RFID reader and Wi-Fi 6 fingerprint positioning, RFID readers are deployed at key nodes in the hospital, and combined with the elevator controller interface to obtain the elevator operating floor, so as to realize the indoor positioning of blood products; Time Prediction and Early Warning Unit: Used to calculate the total predicted time for blood product circulation using a quantitative formula. When the predicted time exceeds the allowable time for the corresponding emergency level, a multi-terminal early warning is triggered, and the total predicted circulation time is updated. The calculation formula is:

[0022] in, Time for preparing for outbound shipment For delivery time, This is the time for receipt confirmation.

[0023] Furthermore, the feedback recording module includes: Data association unit: Using the medical order number as the core association key, a three-dimensional mapping relationship is established between input data, output data and feedback data. The input data is the preprocessed collected data, the output data is the component blood transfusion plan and medical order information, and the feedback data is the patient's follow-up data after transfusion. Storage unit: MySQL is used to store structured data, and MongoDB is used to store unstructured data, with composite indexes created; Effect evaluation unit: Used to calculate treatment effect score using a quantitative formula. The calculation formula is:

[0024] in, This represents the difference between the total coagulation function score before and after infusion. This represents the difference in bleeding velocity before and after infusion. This refers to the rate of bleeding before infusion.

[0025] Furthermore, it also includes an algorithm self-optimization module, which is communicatively connected to the feedback recording module and the decision algorithm module, specifically including: Loss Calculation Unit: Used to calculate the loss function based on the treatment effect score from the feedback recording module. The formula is:

[0026] in, The total score for coagulation function after infusion. The overall score for ideal coagulation function; Parameter optimization unit: The gradient descent method is used to optimize the weight parameters of the decision algorithm module. The optimization formula is as follows:

[0027] in, For the updated weights, As the current weight, For learning rate, This is the partial derivative of the loss function with respect to the weights; Verification and Deployment Unit: Used to verify the optimized weight parameters on the test set, and automatically deploy to the production environment when the accuracy improves by ≥3%.

[0028] Secondly, an intelligent method for dynamic management of component blood transfusion and coagulation function includes: S1. Establish communication connections with multi-source medical equipment and information systems to collect patients' coagulation function data, vital signs data, bleeding status data and basic information data in real time, and preprocess the collected data. S2. A multi-parameter coupled decision-making model trained based on clinical guidelines and expert consensus receives the preprocessed data collected in step S1 and outputs a component blood transfusion plan through quantitative calculation. S3. Based on the component blood transfusion plan output in step S2, a component blood application order is automatically generated. After the doctor reviews and confirms the order online, the urgency level is determined and the order is pushed to the blood transfusion department and related terminals according to priority. S4. Track and locate the entire process of blood components from warehousing and transportation to infusion, collect status data at each stage and synchronize it to relevant terminals in real time; S5. Collect follow-up data after patient infusion, and establish a correlation mapping between the data collected in step S1, the component blood infusion plan in step S2, and the follow-up data.

[0029] Compared with existing technologies, the present invention provides an intelligent component blood transfusion and coagulation function dynamic management system and method. The system includes: a data acquisition module, a decision algorithm module, a medical order generation and push module, a full-process tracking module, a feedback recording module, and an algorithm self-optimization module. It achieves multi-source medical device interface adaptation, dynamic sampling and data preprocessing, constructs a quantitative coupling model based on clinical guidelines and expert consensus, and outputs precise component blood transfusion plans. It can complete intelligent emergency triage and multi-channel priority push, and uses RFID and Wi-Fi 6 fusion positioning technology to track blood product circulation in real time. The method achieves management through a closed-loop process of data acquisition and preprocessing, quantitative decision-making, triage push, full-process tracking, and feedback optimization. This invention can improve the accuracy of component blood transfusion and the timeliness of treatment, reduce blood product waste, enhance the efficiency of multi-departmental collaboration, and provide strong support for the treatment of patients with traumatic hemorrhage. Attached Figure Description

[0030] Figure 1 This is an architecture diagram of the intelligent component blood transfusion and coagulation function dynamic management system provided in an embodiment of the present invention; Figure 2 A flowchart of the intelligent component transfusion and coagulation function dynamic management method provided in the embodiments of the present invention. Detailed Implementation

[0031] The technical solution of the present invention will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are not all embodiments of the present invention. All other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0032] It should be noted that, unless otherwise specifically stated, the relative arrangement and numerical expressions of the components and steps described in these embodiments should not be construed as limiting the scope of the invention.

[0033] The following description of exemplary embodiments is merely illustrative and is not intended to limit the invention or its application or use in any way. Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail herein, but where applicable, such techniques, methods, and apparatus should be considered part of this specification.

[0034] Example 1 See Figure 1 , Figure 1 This is an architecture diagram of an intelligent component transfusion and coagulation function dynamic management system proposed in this invention, which may specifically include: M1, the data acquisition module, is used to establish communication connections with multi-source medical devices and information systems, collect patients' coagulation function data, vital sign data, bleeding status data, and basic information data in real time, and preprocess the collected data; specifically including: M11, Multi-interface adapter unit: It uses the HL7FHIR interface to establish communication with the bedside coagulation monitor and laboratory information system, uses the MQTT protocol to establish communication with the monitor, and uses the Ethernet interface to establish communication with the electronic medical record system; Specifically, a long-term connection is established with the bedside coagulation monitor via the HL7FHIRv4 interface, actively pulling coagulation function data every 5 minutes (acute phase) or 15 minutes (stable phase); the monitor data is subscribed to via the MQTT protocol, with a sampling frequency of 1 time / minute; basic information such as patient weight, age, blood type, and past medical history are retrieved from the EMR system via the Ethernet interface, cached locally after the first retrieval, and refreshed synchronously upon update.

[0035] M12, Dynamic Sampling Unit: The sampling frequency is dynamically adjusted according to the patient's bleeding rate. When the bleeding rate is ≥10ml / min, the sampling frequency for coagulation function data is 1 time / 5 minutes; when the bleeding rate is ≥10ml / min, the sampling frequency is 1 time / 5 minutes. At that time, the sampling frequency is 1 time / 15 minutes; the specific steps of the sampling frequency adjustment mechanism include: M121. Basic sampling rules: Preset two thresholds for bleeding velocity (BLV), which correspond to the sampling frequency in the stable phase and the acute phase, respectively. The sampling frequency in the acute phase is no less than once every 5 minutes, and the sampling frequency in the stable phase is no more than once every 15 minutes. M122 Emergency Sampling Trigger: When the 30-minute change rate of any coagulation index (PT / APTT / Fbg / PLT) is ≥20%, the emergency sampling mode is automatically triggered, and the sampling frequency is increased to 1 time / 2 minutes; after the emergency sampling continues for a preset number of times, if the index change rate is ≤10%, it will automatically return to the basic sampling frequency; M123. Sampling anomaly handling: If two consecutive sampling failures or data loss occur, the system will trigger a device fault alert and automatically switch to a backup sampling device (if available) or start the manual data collection and input channel.

[0036] M13, Data Preprocessing Unit: (Using...) Outlier removal is performed according to the principle, noise filtering is performed using the Kalman filter algorithm, and the collected data is mapped to the [0,1] interval using a standardization algorithm.

[0037] Outlier removal is based on In principle, calculate the average of the last 5 data points. and standard deviation When a certain data satisfies If the error occurs, it is marked as abnormal and stored in the error log. At the same time, a "re-collect" command is sent to the bedside device through the interface. If no new data is received within 1 minute, a manual reminder is triggered. In addition, the implementation of the Kalman filter algorithm in the data preprocessing unit includes a state update unit and an error covariance update unit; the state update unit executes the following formula:

[0038] in, for Filtered data values ​​at any time for Predicted value at any time For Kalman gain, The original data collected at time k, The observation matrix; Error covariance update unit execution formula:

[0039] in, for Time error covariance It is the identity matrix. for Time error covariance.

[0040] The linear standardization formula maps all indicators to the [0,1] interval. The specific formula is as follows:

[0041] Among them, each indicator (Minimum value) and (Maximum value) is set based on a reasonable threshold that has been validated in clinical trials, and supports standard switching for multiple population types (adults, children, pregnant women).

[0042] M2, a decision algorithm module, is communicatively connected to the data acquisition module. It contains a built-in multi-parameter coupled decision model trained based on clinical guidelines and expert consensus. This model receives preprocessed acquisition data and outputs a component transfusion plan through quantitative calculation. Specifically, it includes: M21, Coagulation Function Scoring Unit: Used to convert coagulation function data into a unified score using the Z-score standardization algorithm, and calculate the total coagulation function score; refer to Table 1, multiple sets of reference value standards are built-in, automatically matched based on the patient's basic information:

[0043] Extreme value handling: When PT>120s or APTT>180s, the system defaults to PT=0 or APTT=0, directly triggering a Level 1 emergency blood transfusion decision without participating in subsequent weight calculations.

[0044] It has multiple sets of normal reference values ​​for coagulation indicators specific to different population groups, and can automatically match the corresponding reference value range based on the patient's basic information (age, gender, physiological status); for extreme cases where the indicators exceed the upper limit of the instrument's measurement, a default score value is set and the emergency blood transfusion decision logic is directly triggered.

[0045] M22, Dynamic Weight Allocation Unit: Employing the entropy-weighted analytic hierarchy process (AHP), the basic weights of each decision indicator are first determined using AHP, then dynamic adjustment coefficients are calculated based on the rate of change of the indicators, ultimately yielding the final weights of each indicator; specifically including: M221, Basic Weight Determination Subunit: Used to collect expert scores using the 1 to 9 scale method, and calculate the basic weights of coagulation function total score, bleeding velocity, mean arterial pressure, cumulative bleeding volume and age using the analytic hierarchy process; M222, Dynamic Adjustment Coefficient Calculation Subunit: Used to calculate the dynamic adjustment coefficient based on the rate of change of the index. The specific formula is as follows:

[0046] in, This is the dynamic adjustment coefficient for indicator x. To adjust the coefficient, This refers to the change in the indicator over a 30-minute period. This represents the average of the normal reference values ​​for the indicator. M223, Final Weight Synthesis Subunit: Used to calculate the final weights of each indicator. The specific formula is as follows:

[0047] in, The final weight of indicator x, This is the base weight for indicator x; the weight update cycle is 5 minutes. If the change rate of the dynamic adjustment coefficient of an indicator is ≥30%, an immediate update is triggered.

[0048] M23, Infusion Protocol Calculation Unit: Used to calculate the infusion dose of each blood component based on the total coagulation function score, final weight, and patient baseline information using quantitative formulas, and dynamically adjust the infusion ratio of blood components based on the total coagulation function score; specifically including: M231, Dosage Calculation Subunit: Used to calculate the infusion dose for red blood cells (RBC), plasma (FFP), platelets (PLT), and cryoprecipitate (Cryo) using quantitative formulas. The calculation formula is:

[0049] in, For the target value, For actual measurement of product, The target value in the formula (such as Hct target, PLT target, Fbg target) is based on clinical guidelines and can be dynamically adjusted according to the patient's bleeding degree and underlying diseases.

[0050] M232, Proportional Constraint Subunit: Built-in Dynamic Proportional Threshold ( This function is used to automatically adjust the infusion dosage of each blood component when the proportion of blood components output by the infusion protocol calculation unit exceeds the corresponding threshold. If it does not meet the threshold, the dosage is adjusted according to the principle of "prioritizing the protection of key coagulation components" to ensure compliance with the proportion.

[0051] M3, the medical order generation and push module, is communicatively connected to the decision algorithm module. It is used to automatically generate medical orders for blood component transfusion based on the blood component transfusion plan. These orders require online review and confirmation by a doctor before being issued. The urgency level of the medical order is determined, and the order is pushed to the blood transfusion department and relevant terminals according to priority. Specifically, it includes: M31, Medical Order Review and Confirmation Unit: Provides clinicians with a PC / mobile review interface that fully displays the patient's coagulation data, bleeding status, system decision logic, and infusion plan; when a doctor performs confirmation, approval, modification confirmation, or rejection, the system automatically locks the operation permission and leaves a record; only medical orders that have been confirmed by the doctor will enter the emergency level determination and push process; M32, Emergency Level Determination Unit: Used to calculate the emergency coefficient using a quantitative formula, and classify emergency levels into Level 1, Level 2, and Level 3 based on the emergency coefficient. The calculation formula is:

[0052] in, For the rate of bleeding, The bleeding rate threshold, Mean arterial pressure, The mean arterial pressure threshold. The total score is used to determine the coagulation function; medical orders are classified into three emergency levels based on the urgency coefficient, with each level corresponding to a specific infusion time requirement; the emergency level determination results must maintain high consistency with the clinical expert assessment standards. ).

[0053] M33, Medical Order Formatting Unit: Used to automatically fill in patient information, blood component information (type, dosage, proportion), emergency level and infusion deadline, support medical staff to manually modify key fields (such as dosage, infusion time), and automatically retain modification records (including modifier, time, and reason), and generate standardized medical orders; M34, Multi-channel push unit: Uses WebSocket protocol to achieve real-time push, used to push medical orders confirmed by doctors to the blood transfusion department and related terminals, and pushes medical orders according to the urgency level through multiple channels such as blood transfusion department terminal, medical staff mobile APP, and transport personnel terminal.

[0054] M4, the end-to-end tracking module, communicates with the medical order generation and push module to locate and track the entire process of blood component delivery from warehousing and transportation to infusion, collecting status data at each stage and synchronizing it in real time; specifically including: M41, Identification Binding Unit: Used to establish a unique association between the blood product barcode and the RFID tag through a scanner. The RFID tag stores the blood product number, type, dosage, expiration date, and doctor's order number. Once the association information is written, it cannot be modified.

[0055] M42, Real-time Positioning Unit: Adopts a fusion solution of RFID reader and Wi-Fi 6 fingerprint positioning, deploys RFID readers at key nodes in the hospital, and obtains the elevator operating floor through the elevator controller interface to realize indoor positioning of blood products; M43, Time Prediction and Early Warning Unit: Used to calculate the total predicted time for blood product circulation using a quantitative formula. When the predicted time exceeds the allowable time for the corresponding emergency level, a multi-terminal early warning is triggered, and the total predicted circulation time is updated. The calculation formula is:

[0056] in, The preparation time for blood transfusion is dynamically calculated based on the real-time workload of the blood transfusion department. The delivery time is calculated based on the distance between two points and the average delivery speed using the hospital's GIS system. This is the time for receipt confirmation.

[0057] The early warning mechanism triggers a multi-terminal audible and visual warning when the total predicted time exceeds the allowable infusion time for the corresponding emergency level. At the same time, it pushes an alternative transportation route (planned based on real-time population flow data) and sends an early warning notification to the department management personnel.

[0058] The M5 and feedback recording module are communicatively connected to the data acquisition module, decision algorithm module, and end-to-end tracking module, respectively. They are used to collect post-infusion follow-up data from patients, establish a mapping between the collected data, the infusion protocol, and the follow-up data, forming a closed-loop data link. Specifically, this includes: M51, Data Association Unit: Using the medical order number as the core association key, a three-dimensional mapping relationship is established between input data (pre-processed monitoring data, patient basic information), output data (infusion plan, medical order information), and feedback data (post-infusion follow-up data, treatment effect evaluation). The input data is the pre-processed collected data, the output data is the component blood infusion plan and medical order information, and the feedback data is the patient's post-infusion follow-up data. M52, Storage Unit: MySQL is used to store structured data, MongoDB is used to store unstructured data, and a composite index is created; Specifically, this module adopts a hybrid storage scheme of "structured + unstructured". MySQL stores structured data (patient information, coagulation indicators, medical order records, etc.) and establishes a joint index of "patient ID-medical order number" and "medical order number-follow-up time" to improve query efficiency; MongoDB stores unstructured data (blood product circulation trajectory, algorithm logs, etc.) and supports time range retrieval of trajectory data.

[0059] M53, Efficacy Evaluation Unit: Used to calculate the treatment efficacy score using a quantitative formula. The calculation formula is:

[0060] in, This represents the difference between the total coagulation function score before and after infusion. This represents the difference in bleeding velocity before and after infusion. The rate of bleeding before infusion is recorded. The effectiveness score is categorized into three levels: "Excellent," "Good," and "Poor," providing data support for algorithm optimization.

[0061] M6, Algorithm self-optimization module, which is communicatively connected to the feedback recording module and the decision algorithm module, specifically includes: M61, Loss Calculation Unit: Used to calculate the loss function based on the treatment effect score from the feedback recording module. The formula is:

[0062] in, The total score for coagulation function after infusion. The overall score represents the ideal coagulation function; the loss value is adjusted by combining the treatment effect score to improve the targeted optimization.

[0063] M62, Parameter Optimization Unit: This unit optimizes the weight parameters of the decision algorithm module using gradient descent. The optimization formula is as follows:

[0064] in, For the updated weights, As the current weight, For learning rate, denoted as the partial derivative of the loss function with respect to the weights; the optimization algorithm is deployed on a cloud server and runs in batch mode, with each optimization session lasting ≤5 minutes.

[0065] M63, Validation Deployment Unit: Used to validate the optimized weight parameters on the test set, and automatically deploy to the production environment when the accuracy improves by ≥3%.

[0066] The validation criteria require that the optimized parameters be validated on an independent test set (clinical data not used in model training), and that the decision accuracy improvement be ≥3% and the dose error be ≤10% in order to pass the validation.

[0067] Iterative deployment automatically deploys validated parameters to the production environment, retaining historical versions for rollback support; when multiple consecutive patients receive "poor" treatment outcome scores, a manual intervention process is triggered, where experts adjust parameters and then incorporate them into the optimization model.

[0068] Example 2 See Figure 2 , Figure 2 The flowchart of the intelligent component transfusion and coagulation function dynamic management method proposed in this invention can specifically include: S1. Multi-source data acquisition and preprocessing: Establish communication connections with multi-source medical devices and information systems to acquire patients' coagulation function data, vital signs data, bleeding status data, and basic information data in real time, and preprocess the acquired data; specific steps include: S11. Interface Adaptation and Data Acquisition: Establishes a long-term connection with the bedside coagulation monitor via the HL7FHIRv4.0 protocol to collect coagulation function data such as PT, APTT, Fbg, and PLT in real time; subscribes to monitor data using the MQTTv3.1.1 protocol to obtain vital sign data such as MAP, HR, and SpO2; connects to the hospital information system via the HTTP / 2 protocol to retrieve basic information such as patient weight, age, and blood type; adapts to weighing suction devices and gauze weighing equipment to automatically calculate and upload bleeding velocity (BLV) and total blood loss (TBL).

[0069] S12. Dynamic sampling control: Based on the BLV preset dual threshold, the acute phase and the stable phase are divided. The sampling frequency of coagulation data in the acute phase is set to 1 time / 5 minutes, and in the stable phase it is set to 1 time / 15 minutes. When the change rate of any coagulation index in 30 minutes is ≥20%, emergency sampling is triggered once / 2 minutes. After the index stabilizes, the base frequency is restored. S13. Data preprocessing: Abnormal data is removed based on the 3σ principle, outliers are marked and the device is triggered to re-acquire data; the data is denoised using the Kalman filter algorithm, and the filter parameters are configured according to the preset initial error covariance, process noise covariance and observation noise covariance; the linear standardization formula is used to map all indicators to the [0,1] interval to adapt to the reference value standards of multiple population groups.

[0070] S2. Multi-parameter Coupled Decision Making and Infusion Protocol Generation: Based on a multi-parameter coupled decision-making model trained according to clinical guidelines and expert consensus, this model receives the preprocessed data from step S1 and outputs a component blood transfusion protocol through quantitative calculation; specifically including: S21. Coagulation function score: Based on the Z-score normalization algorithm, the individual scores of each coagulation index are calculated and summed to obtain the total coagulation function score (Stotal). In extreme cases where the index exceeds the instrument's measurement limit, a default score is set and an emergency blood transfusion decision is directly triggered. S22. Dynamic Weight Allocation: The basic weights of Stotal, BLV, MAP, TBL, and age are determined using the analytic hierarchy process (AHP) combined with the expert 1-9 scale method. Dynamic adjustment coefficients are calculated based on the 30-minute rate of change of these indicators. The specific formula is as follows:

[0071] in, This is the dynamic adjustment coefficient for indicator x. To adapt the post-training coefficients to different metrics, This refers to the change in the indicator over a 30-minute period. The average value is the normal reference value for the indicators; the final weight of each indicator is calculated using a normalized formula and updated every 5 minutes. If the weight fluctuates too much, an immediate update is triggered. S23. Infusion Protocol Calculation: For red blood cells, plasma, platelets, and cryoprecipitate, the initial dose is calculated by substituting them into the quantitative dose formula. Based on the S total, the dynamic ratio threshold is automatically matched to verify the initial dose ratio. If it does not meet the requirements, it is corrected according to the principle of "prioritizing the protection of key coagulation components" and a compliant component blood infusion protocol is output.

[0072] S3. Order Generation and Priority Push: Based on the component blood transfusion plan output in step S2, automatically generate component blood transfusion request orders, determine the urgency level of the orders, and push them to the blood transfusion department and relevant terminals according to priority; specifically including: S31. Emergency Level Determination: Based on the formula, an emergency coefficient (E) is calculated, and emergency levels are classified into Level 1, Level 2, and Level 3 according to the E value, corresponding to different infusion time requirements; the specific formula is as follows:

[0073] in, For the rate of bleeding, The bleeding rate threshold, Mean arterial pressure, The mean arterial pressure threshold. The total score for coagulation function is used; medical orders are divided into three levels of emergency based on the emergency coefficient, with each level corresponding to a specific infusion time requirement; the emergency level determination results must maintain high consistency with the clinical expert assessment standards (Kappa value ≥ 0.8).

[0074] S32. Medical Order Formatting and Push: Automatically fills in fields such as patient basic information, component blood transfusion information, emergency level, and transfusion deadline to generate standardized medical orders; uses the WebSocket protocol to achieve real-time push through multiple channels, with the blood transfusion department terminal, medical staff mobile APP, and transport personnel terminal presenting reminders in different ways (highlight, pop-up, voice, etc.) according to the emergency level, ensuring priority response to emergency medical orders.

[0075] S4. Blood Product End-to-End Tracking: Tracking and locating blood components throughout the entire process from warehousing and transportation to transfusion, collecting status data at each stage and synchronizing it to relevant terminals in real time; specifically including: S41. Identification Binding: A unique association is established between the blood product barcode and the UHF RFID tag through a scanning device. The tag is filled with information such as the type of blood product, dosage, expiration date, and doctor's order number. Anti-tampering design is adopted to ensure information security. S42. Real-time Positioning and Status Synchronization: Adopting a fusion solution of "RFID reader + Wi-Fi 6 fingerprint positioning," readers are deployed at key nodes within the facility to collect real-time blood product location information. Combined with floor data obtained from the elevator controller interface, this achieves full-area positioning and tracking with high accuracy. ; S43. Time Prediction and Early Warning: The total predicted time for circulation is calculated using a formula, the specific formula of which is:

[0076] in, Time for preparing for outbound shipment For delivery time. This is the time for receipt confirmation.

[0077] Based on real-time dynamic calculation of the workload of the blood transfusion department Solving for distance and average transport speed using a GIS system; when When the allowed time for the corresponding emergency level is exceeded, a multi-terminal warning is triggered and an alternative route is pushed.

[0078] S5. Feedback Recording and Closed-Loop Optimization: Collect follow-up data after patient infusion and establish a correlation mapping between the data collected in step S1, the component blood transfusion protocol in step S2, and the follow-up data. Specifically, this includes: S51. Feedback Data Collection and Association: Collect coagulation indicators, vital signs, bleeding rate and other follow-up data of patients 2 hours / 24 hours after infusion, and establish a three-dimensional mapping relationship of input data (preprocessed monitoring data), output data (infusion plan) and feedback data with "medical order number" as the core association key. S52. Effectiveness Evaluation: The treatment effectiveness score is calculated using a formula and categorized into "Excellent," "Good," and "Poor" levels. The specific formula is as follows:

[0079] in, This represents the difference between the total coagulation function score before and after infusion. This represents the difference in bleeding velocity before and after infusion. The rate of bleeding before infusion is recorded. The effectiveness score is categorized into three levels: "Excellent," "Good," and "Poor," providing data support for algorithm optimization.

[0080] S53. Algorithm Self-Optimization: After completing the treatment of each patient or accumulating 100 cases of data, the loss value is calculated based on the loss function. The specific formula is as follows:

[0081] in, The total score for coagulation function after infusion. The overall score represents the ideal coagulation function; the loss value is adjusted by combining the treatment effect score to improve the targeted optimization.

[0082] The gradient descent method is used to optimize the weight parameters. The specific formula is as follows:

[0083] in, For the updated weights, As the current weight, For learning rate, Let be the partial derivative of the loss function with respect to the weights; the optimization algorithm is deployed on a cloud server and runs in batch mode, with a single optimization time of _____. The optimized parameters are automatically deployed after being validated on an independent test set (accuracy improvement ≥3%). Manual intervention is triggered when multiple consecutive cases receive a "poor" performance score.

[0084] This embodiment achieves real-time fusion of multi-source data, personalized and accurate decision-making, controllable tracking throughout the entire process, and continuous algorithm optimization through the coordinated implementation of each step. It effectively improves the timeliness and accuracy of blood transfusion treatment for patients with massive bleeding, while also improving the efficiency of multi-departmental collaboration. It can be directly applied to clinical scenarios such as emergency departments and operating rooms.

[0085] In summary, the present invention has the following advantages: 1. Real-time fusion and precise preprocessing of multi-source medical data are achieved, solving the problems of scattered data collection and inefficient sharing; human delays and errors in data transmission are eliminated; and comprehensive, real-time, and high-quality multi-dimensional data support is provided for subsequent decision-making through multi-population standard adaptation and dynamic sampling control, solving the problem of insufficient decision-making basis caused by data fragmentation in traditional solutions from the source. 2. Construct a quantitative dynamic decision-making model to improve the accuracy of component blood transfusion, reduce the waste of blood products and the risk of over-transfusion; avoid the problems of insufficient or excessive transfusion; at the same time, through the proportional constraint correction mechanism, ensure that the transfusion plan complies with the clinical treatment guidelines. It has been verified that this can reduce the waste rate of blood products, improve the accuracy of blood transfusion decisions, and achieve personalized and precise blood transfusion. 3. Establish an intelligent priority push mechanism to shorten the medical order response and blood product transfusion cycle, ensure the timeliness of emergency treatment, and effectively solve the problem of delayed processing of emergency medical orders; at the same time, through the hospital's GIS optimal route planning, further shorten the blood product transportation time, reduce the risk of transfusion delay, and seize the golden treatment window for patients with massive bleeding. 4. Achieve full-process visual tracking and anomaly early warning for blood products, improving the controllability of the circulation process; by comparing the threshold of predicted time with the allowable time, identify anomalies such as transportation congestion and delay in advance and trigger multi-terminal early warning, promote timely handling of problems, reduce the blood product transfusion delay rate, and ensure the safety and controllability of the blood transfusion process. 5. Improve the efficiency of multi-departmental collaborative treatment; Compared with the existing technology model where each department's system is independent and the collaboration is inefficient, it can reduce the time cost of cross-departmental data verification and manual communication, and improve the overall collaborative efficiency of the treatment process; Through the automated flow and recording of data throughout the process, it simplifies the operation process of medical staff, reduces their workload, and allows medical staff to focus on core treatment work.

[0086] The above specific embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to examples, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An intelligent dynamic management system for component blood transfusion and coagulation functions, characterized in that, include: The data acquisition module is used to establish communication connections with multi-source medical devices and information systems, collect patients' coagulation function data, vital signs data, bleeding status data and basic information data in real time, and preprocess the collected data. The decision algorithm module is communicatively connected to the data acquisition module. It has a built-in multi-parameter coupled decision model trained based on clinical guidelines and expert consensus. It is used to receive preprocessed acquisition data and output component blood transfusion plan through quantitative calculation. The medical order generation and push module is connected to the decision algorithm module and is used to automatically generate medical orders for component blood transfusion according to the component blood transfusion plan. After the doctor reviews and confirms the order online, the urgency level of the medical order is determined and the order is pushed to the blood transfusion department and related terminals according to priority. The end-to-end tracking module communicates with the medical order generation and push module to locate and track the blood components from warehousing and transportation to infusion, collect status data at each stage and synchronize them in real time. The feedback recording module is communicatively connected to the data acquisition module, decision algorithm module, and full-process tracking module, respectively, and is used to collect the patient's follow-up data after infusion, establish the association mapping between the collected data, infusion plan and follow-up data, and form a closed-loop data link.

2. The intelligent component transfusion and coagulation function dynamic management system according to claim 1, characterized in that, The data acquisition module includes: Multi-interface adapter unit: It uses the HL7FHIR interface to establish communication with the bedside coagulation monitor and laboratory information system, uses the MQTT protocol to establish communication with the monitor, and uses the Ethernet interface to establish communication with the electronic medical record system. Dynamic sampling unit: Adjusts the sampling frequency dynamically based on the patient's bleeding rate. At that time, coagulation function data was sampled once every 5 minutes; when bleeding... At that time, the sampling frequency is 1 time / 15 minutes; Data preprocessing unit: employing Outlier removal is performed according to principles, noise filtering is done using the Kalman filter algorithm, and the collected data is mapped to a normalization algorithm. Interval.

3. The intelligent component transfusion and coagulation function dynamic management system according to claim 2, characterized in that, The implementation of the Kalman filter algorithm in the data preprocessing unit includes a state update unit and an error covariance update unit; the state update unit executes the following formula: in, for Filtered data values ​​at any time for Predicted value at any time For Kalman gain, The original data collected at time k, The observation matrix; Error covariance update unit execution formula: in, for Time error covariance It is the identity matrix. for Time error covariance.

4. The intelligent component transfusion and coagulation function dynamic management system according to claim 1, characterized in that, The decision algorithm module includes: Coagulation function scoring unit: used to convert coagulation function data into a unified score using the Z-score normalization algorithm, and to calculate the total coagulation function score; Dynamic weight allocation unit: Employing the entropy-weighted analytic hierarchy process (AHP), the basic weights of each decision indicator are first determined using AHP, then dynamic adjustment coefficients are calculated based on the indicator change rate, ultimately yielding the final weights of each indicator. The dynamic weight allocation unit includes: The basic weight determination sub-unit is used to collect expert scores using a 1 to 9 scale and calculate the basic weights of coagulation function total score, bleeding velocity, mean arterial pressure, cumulative bleeding volume and age using the analytic hierarchy process. Dynamic adjustment coefficient calculation subunit: used to execute the formula: in, This is the dynamic adjustment coefficient for indicator x. To adjust the coefficient, This refers to the change in the indicator over a 30-minute period. This represents the average of the normal reference values ​​for the indicator. Final weighted synthesis subunit: used to execute the formula: in, The final weight of indicator x, The basic weights for indicator x; Infusion protocol calculation unit: It is used to calculate the infusion dose of each blood component based on the total coagulation function score, final weight and patient basic information through quantitative formula, and dynamically adjust the infusion ratio of blood components based on the total coagulation function score.

5. The intelligent component transfusion and coagulation function dynamic management system according to claim 4, characterized in that, The infusion scheme calculation unit includes: Dosage calculation subunit: used to calculate the infusion dose for red blood cells, plasma, platelets, and cryoprecipitate using quantitative formulas, wherein the red blood cell infusion dose... The calculation formula is: in, For target hematocrit, To measure hematocrit, For the patient's weight; Proportional constraint subunit: Built-in dynamic proportional threshold, used to automatically correct the infusion dose of each blood component when the proportion of blood component output by the infusion scheme calculation unit exceeds the corresponding threshold.

6. The intelligent component transfusion and coagulation function dynamic management system according to claim 1, characterized in that, The medical order generation and push module includes: Medical Order Review and Confirmation Unit: Provides an online review interface for doctors, displaying the basis of the infusion plan, patient data, and urgency level. It supports doctors to manually confirm, modify, or reject draft medical orders. Once a doctor confirms and issues a medical order, it will proceed to the subsequent push and execution process. Emergency Level Determination Unit: Used to calculate an emergency coefficient using a quantitative formula, and classify emergency levels into Level 1, Level 2, and Level 3 based on the emergency coefficient. The calculation formula is: in, For the rate of bleeding, The bleeding rate threshold, Mean arterial pressure, The mean arterial pressure threshold. The total score for coagulation function; Medical order formatting unit: used to automatically fill in patient information, blood component information, emergency level and infusion deadline, and generate standardized medical orders; Multi-channel push unit: Real-time push is achieved using the WebSocket protocol. It is used to push medical orders confirmed by doctors to the blood transfusion department and related terminals. Medical orders are pushed through multiple channels such as blood transfusion department terminals, medical staff mobile APP, and transport personnel terminals according to the urgency level.

7. The intelligent component transfusion and coagulation function dynamic management system according to claim 1, characterized in that, The end-to-end tracking module includes: Identification binding unit: used to establish a unique association between the blood product barcode and the RFID tag through a barcode scanner. The RFID tag stores information such as the blood product number, type, dosage, expiration date, and doctor's order number. Real-time positioning unit: Adopting a fusion solution of RFID reader and Wi-Fi 6 fingerprint positioning, RFID readers are deployed at key nodes in the hospital, and combined with the elevator controller interface to obtain the elevator operating floor, realizing indoor positioning of blood products; Time Prediction and Early Warning Unit: Used to calculate the total predicted time for blood product circulation using a quantitative formula. When the predicted time exceeds the allowable time for the corresponding emergency level, a multi-terminal early warning is triggered, and the total predicted circulation time is updated. The calculation formula is: in, Time for preparing for outbound shipment For delivery time, This is the time for receipt confirmation.

8. The intelligent component transfusion and coagulation function dynamic management system according to claim 1, characterized in that, The feedback recording module includes: Data association unit: Using the medical order number as the core association key, a three-dimensional mapping relationship is established between input data, output data and feedback data. The input data is the preprocessed collected data, the output data is the component blood transfusion plan and medical order information, and the feedback data is the patient's follow-up data after transfusion. Storage unit: MySQL is used to store structured data, MongoDB is used to store unstructured data, and a composite index is created; Effect evaluation unit: Used to calculate treatment effect score using a quantitative formula. The calculation formula is: in, This represents the difference between the total coagulation function score before and after infusion. This represents the difference in bleeding velocity before and after infusion. This refers to the rate of bleeding before infusion.

9. The intelligent component transfusion and coagulation function dynamic management system according to claim 1, characterized in that, It also includes an algorithm self-optimization module, which is communicatively connected to the feedback recording module and the decision algorithm module, and specifically includes: Loss Calculation Unit: Used to calculate the loss function based on the treatment effect score from the feedback recording module. The formula is: in, The total score for coagulation function after infusion. The overall score for ideal coagulation function; Parameter optimization unit: The gradient descent method is used to optimize the weight parameters of the decision algorithm module. The optimization formula is as follows: in, For the updated weights, As the current weight, For learning rate, This is the partial derivative of the loss function with respect to the weights; Verification and Deployment Unit: Used to verify the optimized weight parameters on the test set, and automatically deploy to the production environment when the accuracy improves by ≥3%.

10. A method for intelligent dynamic management of component blood transfusion and coagulation function, characterized in that, include: S1. Establish communication connections with multi-source medical equipment and information systems to collect patients' coagulation function data, vital signs data, bleeding status data and basic information data in real time, and preprocess the collected data. S2. A multi-parameter coupled decision-making model trained based on clinical guidelines and expert consensus receives the preprocessed data collected in step S1 and outputs a component blood transfusion plan through quantitative calculation. S3. Based on the component blood transfusion plan output in step S2, a component blood application order is automatically generated. After the doctor reviews and confirms the order online, the urgency level is determined and the order is pushed to the blood transfusion department and related terminals according to priority. S4. Track and locate the entire process of blood components from warehousing and transportation to infusion, collect status data at each stage and synchronize it to relevant terminals in real time; S5. Collect follow-up data after patient infusion, and establish a correlation mapping between the data collected in step S1, the component blood infusion plan in step S2, and the follow-up data.

Images