Intelligent infusion safety monitoring system and method
By preprocessing infusion parameter data and generating standardized data through risk identification algorithms, dynamic threshold comparison and safety control are performed, solving the problems of signal distortion and drift in intelligent infusion monitoring systems and improving the accuracy and adaptability of infusion safety monitoring.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUZHOU COLLEGE
- Filing Date
- 2026-02-02
- Publication Date
- 2026-05-01
AI Technical Summary
In existing intelligent infusion monitoring systems, the infusion parameter data collected by multi-parameter sensors cannot detect signal distortion and drift during the data acquisition process in real time, resulting in insufficient accuracy of safety monitoring.
By collecting infusion parameter data, preprocessing it to remove outliers and noise, generating standardized data, using risk identification algorithms to calculate risk indicators, comparing dynamic thresholds based on risk indicator data, generating alarm trigger signals, and implementing safety control through flow rate regulation and bubble removal mechanisms, the data is integrated to generate structured reports and optimize monitoring strategies.
It improves the accuracy and adaptability of infusion safety monitoring, reduces signal distortion and drift, promptly identifies complex abnormal patterns, enhances the real-time and comprehensiveness of alarm response, corrects control deviations, and improves the long-term reliability of the system.
Smart Images
Figure CN121944299A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical infusion monitoring technology, specifically to an intelligent infusion safety monitoring system and method. Background Technology
[0002] The medical infusion monitor is a Class I medical device, belonging to infusion auxiliary electronic equipment. It is suitable for intravenous infusion therapy in medical institutions and field environments. The device uses infrared photoelectric sensors and weighing detection technology, and is equipped with a snap-on "C" type device. It can display the infusion drip rate and remaining fluid volume in real time and present them through a digital tube.
[0003] Currently, due to various interference factors during infusion, including changes in patient position, mechanical vibration of equipment, and fluctuations in ambient temperature, the infusion parameter data collected by the multi-parameter sensors used in intelligent infusion safety monitoring cannot detect signal distortion and drift during the data acquisition process in real time. When systematic errors occur in the sensor output, the basic data for risk analysis becomes unreliable, and the accuracy of safety monitoring cannot be guaranteed.
[0004] Therefore, an intelligent infusion safety monitoring system and method are proposed to solve the above problems. Summary of the Invention
[0005] To address the shortcomings of existing technologies, this invention provides an intelligent infusion safety monitoring system and method, which solves the problems mentioned in the background art, such as the inability to detect signal distortion and drift during the data acquisition process from the multi-parameter sensors used in intelligent infusion safety monitoring, and the inability to guarantee the accuracy of safety monitoring.
[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent infusion safety monitoring system and method, the method comprising the following steps: S1. Collect infusion parameter data, including real-time flow rate, infusion pressure, fluid temperature and patient physiological parameters; S2. The infusion parameter data is preprocessed by removing outliers through a data cleaning unit, reducing interference signals through a noise filtering unit, and then generating standardized infusion data through a standardization conversion unit. S3. Based on the standardized infusion data, perform safety risk analysis, and use risk identification algorithms to calculate flow rate stability, bubble probability and allergic reaction trend to generate risk indicator data. S4. Perform dynamic threshold comparison processing based on the risk indicator data, match the risk indicators with the preset safety threshold library, and generate an alarm trigger signal when the indicator exceeds the limit. S5. Execute safety control actions according to the alarm trigger signal, adjust the infusion rate through the flow rate adjustment unit, trigger the bubble removal mechanism to start, and send alarm information through the notification unit; S6. Integrate infusion parameter data, safety risk analysis results, and safety control action logs, and output a structured infusion safety monitoring report through the report generation unit; S7. Optimize the monitoring report parameters based on user feedback data, and use a machine learning model to adaptively adjust the risk algorithm and security threshold to generate an optimized monitoring strategy.
[0007] Preferably, the infusion parameter data collected in step S1 includes real-time flow rate, infusion pressure, fluid temperature, and patient physiological parameters, comprising the following steps: S11. Real-time infusion parameters are collected through the multi-parameter sensor unit mounted on the intelligent infusion device, including optical flow rate sensor, piezoelectric pressure sensor, infrared temperature sensor and bioelectrode physiological sensor. S12. The collected raw infusion parameter data is transmitted to the central monitoring platform through the wireless communication unit, and an initial infusion dataset is generated.
[0008] Preferably, the preprocessing of the infusion parameter data in step S2, including removing outliers by a data cleaning unit, reducing interference signals by a noise filtering unit, and generating standardized infusion data by a standardization conversion unit, includes the following steps: S21. Receive the initial infusion dataset, and use the sliding window algorithm to clean the data stream, removing outliers and missing data; S22. Apply a digital filter to reduce noise in the cleaned data to generate smooth infusion data; S23. The smoothed infusion data is converted into standardized infusion data with uniform dimensions through the standardization conversion unit, which facilitates subsequent analysis; The standardized transformation is calculated using the following formula: ; in, To smooth individual data points in infusion data, The mean of historical smoothed infusion data. The standard deviation of historical smoothed infusion data. This is the standardized infusion data.
[0009] Preferably, step S3 involves performing safety risk analysis based on the standardized infusion data, using a risk identification algorithm to calculate flow rate stability, bubble probability, and allergic reaction trends, and generating risk indicator data, including the following steps: S31. Obtain the standardized infusion data and input it into the risk identification engine, which integrates a rule-based algorithm and a neural network model. S32. Conduct multi-dimensional risk analysis, including flow rate stability analysis, bubble probability calculation, and physiological parameter trend prediction. S33. Output risk indicator data, which includes risk level score, anomaly type identifier and confidence level parameter.
[0010] Preferably, step S4 involves performing dynamic threshold comparison processing based on the risk indicator data, matching the risk indicators with a preset safety threshold library, and generating an alarm trigger signal when the indicator exceeds the limit. This includes the following steps: S41. A preset dynamic safety threshold library, which is generated based on historical data and clinical guidelines; S42. Compare the risk indicator data with the dynamic safety threshold in real time, and use fuzzy logic algorithm to handle boundary cases; S43. When the comparison result indicates a high risk, generate an alarm trigger signal; otherwise, continue monitoring.
[0011] Preferably, step S5, which involves executing a safety control action based on the alarm trigger signal, adjusting the infusion rate via the flow rate adjustment unit, triggering the bubble removal mechanism, and sending alarm information via the notification unit, includes the following steps: S51. Analyze the alarm trigger signal to determine the type of control action, including flow rate control, bubble treatment, and medical notification; S52. Control actions are achieved through the actuator unit, where flow rate control uses a stepper motor regulating valve, bubble treatment starts the ultrasonic defoaming device, and medical notifications are sent via SMS and App push notifications. S53. Record the execution log of control actions and feed it back to the data recording system.
[0012] Preferably, step S6 integrates infusion parameter data, safety risk analysis results, and safety control action logs, and outputs a structured infusion safety monitoring report through the report generation unit, including the following steps: S61. Integrate infusion parameter data, risk indicator data, control action logs, and user feedback data; S62. A report generation algorithm is adopted, combined with a template engine, to output a structured report, including a summary, trend charts, and recommended measures. The comprehensive risk assessment in the report generation algorithm is calculated using the following formula: ; in, For comprehensive risk scoring, This is an indicator of flow velocity stability. This is an index of the frequency of bubble occurrence. and These are the preset weighting coefficients; S63. Store the report in a cloud database and support access from multiple terminals.
[0013] Preferably, in step S7, optimizing the monitoring report parameters based on user feedback data and using a machine learning model to adaptively adjust the risk algorithm and security threshold to generate an optimized monitoring strategy includes the following steps: S71. Collect user feedback data, including medical staff evaluations, false alarm records, and effectiveness scores; S72. Use reinforcement learning models to optimize risk identification algorithms and security thresholds, and generate adaptive monitoring strategies. S73. Regularly update system parameters to ensure monitoring accuracy and adaptability.
[0014] Preferably, the method further includes step S8: performing remote collaborative processing, realizing multi-device data sharing and expert remote diagnosis through a cloud platform, and generating a collaborative monitoring solution.
[0015] Preferably, the system includes: The data acquisition module collects real-time infusion data through a multi-parameter sensor unit, transmits the data through a communication transmission unit, and outputs standardized infusion data through a data preprocessing unit. The risk analysis module receives the standardized infusion data, analyzes safety threats through the risk identification unit, generates risk indicators using the machine learning prediction unit, and outputs alarm signals through the threshold comparison unit. The safety execution module adjusts the infusion parameters through the control action unit based on the alarm signal, sends an alarm through the alarm notification unit, and obtains user response data through the feedback collection unit. The data management module receives the risk indicators and user response data, archives historical records through the storage unit, creates monitoring reports using the report generation unit, and displays the analysis results through the visualization unit. The optimization learning module optimizes algorithm parameters through the model training unit based on user feedback and monitoring data, updates security policies through the adaptive adjustment unit, and outputs optimization instructions through the evaluation unit. The user interaction module displays the real-time monitoring status through a graphical interface unit, processes user commands through an input receiving unit, and provides operation guidance through a reminder unit.
[0016] Compared with the prior art, the present invention provides an intelligent infusion safety monitoring system and method, which has the following beneficial effects: 1. In this invention, during intelligent infusion safety monitoring, infusion parameter data is collected in real time by a multi-parameter sensor unit, and the infusion parameter data is processed by a data cleaning unit, a noise filtering unit, and a standardization conversion unit to generate standardized infusion data. This reduces distortion and drift during signal acquisition, ensures the accuracy and consistency of the basic data used by the risk analysis module, thereby reducing safety monitoring errors and further improving the accuracy of infusion safety monitoring.
[0017] 2. In this invention, during safety risk analysis, the risk analysis module performs multi-dimensional analysis on standardized infusion data through a risk identification unit and a machine learning prediction unit to generate risk indicator data. The threshold comparison unit then dynamically compares the risk indicator data with a preset safety threshold library. When the indicator exceeds the limit, an alarm trigger signal is generated. This enables timely identification of complex abnormal patterns under multi-parameter coupling and triggers a graded alarm mechanism, thereby avoiding risk misjudgment and response delay, and enhancing the comprehensiveness of anomaly detection and the real-time nature of alarm response.
[0018] 3. In this invention, during the execution of safety control, the safety execution module adjusts the infusion parameters through the control action unit based on the alarm trigger signal, and sends alarm information through the alarm notification unit; at the same time, the optimization learning module dynamically optimizes the risk identification algorithm and safety threshold based on user feedback data and monitoring records through the model training unit and adaptive adjustment unit, generates an optimized monitoring strategy, so that the safety control action matches the actual risk state, can correct control deviations, and improve the adaptive capability and long-term reliability of the entire intelligent infusion safety monitoring system. Attached Figure Description
[0019] Figure 1 This is a flowchart illustrating the steps of an intelligent infusion safety monitoring method according to the present invention; Figure 2 This is a schematic diagram of the architecture of an intelligent infusion safety monitoring system according to the present invention. Detailed Implementation
[0020] 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.
[0021] Please see Figure 1-2 The specific implementation of an intelligent infusion safety monitoring system and method is as follows, and the method includes the following steps: S1. Collect infusion parameter data, including real-time flow rate, infusion pressure, fluid temperature and patient physiological parameters; S2. Preprocess the infusion parameter data, remove outliers through the data cleaning unit, reduce interference signals through the noise filtering unit, and generate standardized infusion data through the standardization conversion unit. S3. Conduct safety risk analysis based on standardized infusion data, and use risk identification algorithms to calculate flow rate stability, bubble probability and allergic reaction trend to generate risk indicator data. S4. Perform dynamic threshold comparison processing based on risk indicator data, match the risk indicators with the preset safety threshold library, and generate an alarm trigger signal when the indicator exceeds the limit. S5. Execute safety control actions according to the alarm trigger signal, adjust the infusion rate through the flow rate adjustment unit, trigger the bubble removal mechanism to start, and send alarm information through the notification unit; S6. Integrate infusion parameter data, safety risk analysis results, and safety control action logs, and output a structured infusion safety monitoring report through the report generation unit; S7. Optimize monitoring report parameters based on user feedback data, and use machine learning models to adaptively adjust risk algorithms and security thresholds to generate optimized monitoring strategies.
[0022] In S1, infusion parameter data is collected, including real-time flow rate, infusion pressure, fluid temperature, and patient physiological parameters. The process includes the following steps: S11. Real-time infusion parameters are acquired through the multi-parameter sensor unit mounted on the intelligent infusion device, including optical flow rate sensor, piezoelectric pressure sensor, infrared temperature sensor, and bioelectrode physiological sensor, to obtain the raw signal vector: ; in, The original signal vector, For the first The raw signal values collected by each sensor For the first The raw signal values collected by each sensor, and so on, For the first The raw signal values collected by each sensor This represents the total number of sensors; Specifically, a synchronization clock unit ensures the timestamps of each sensor signal are aligned, and an analog-to-digital converter converts the analog signals... After being converted into digital signals, they are packaged according to a preset sampling period to form the original signal vector. ; S12. The collected raw infusion parameter data is transmitted to the central monitoring platform through the wireless communication unit, and an initial infusion dataset is generated.
[0023] In S2, the infusion parameter data is preprocessed by removing outliers through a data cleaning unit, reducing interference signals through a noise filtering unit, and then generating standardized infusion data through a standardization conversion unit. This process includes the following steps: S21. Receive the initial infusion dataset and use the sliding window algorithm to clean the data stream, removing outliers and missing data. S22. Apply a digital filter to reduce noise in the cleaned data to generate smooth infusion data; S23. The smoothed infusion data is converted into standardized infusion data with uniform dimensions through the standardization conversion unit, which facilitates subsequent analysis; The standardized transformation is calculated using the following formula: ; in, To smooth individual data points in infusion data, The mean of historical smoothed infusion data. The standard deviation of historical smoothed infusion data. This is the standardized infusion data.
[0024] S3 performs safety risk analysis based on standardized infusion data, using risk identification algorithms to calculate flow rate stability, bubble probability, and allergic reaction trends, generating risk indicator data through the following steps: S31. Obtain standardized infusion data and input it into the risk identification engine, which integrates rule-based algorithms and neural network models. S32. Conduct multi-dimensional risk analysis, including flow stability analysis, bubble probability calculation, and physiological parameter trend prediction, including a comprehensive risk score. Calculate using the following formula: ; in, For flow velocity risk scoring based on flow velocity variance, This is a bubble risk score calculated based on ultrasonic detection signals. This is a physiological risk score calculated based on the deviation of physiological parameters. , , These are the preset weighting coefficients for the corresponding risk scores, and ; Further includes: flow rate risk score The bubble risk score is obtained by calculating the sum of squares of the deviations between the current flow rate and the target flow rate. Physiological risk scores are calculated by extracting characteristic frequency amplitudes through spectral analysis of ultrasound echo signals. It was obtained by calculating the normalized Euclidean distance between the patient's real-time heart rate and blood pressure and the baseline value; S33. Output risk indicator data, which includes risk level score, anomaly type identifier and confidence level parameter.
[0025] S4 performs dynamic threshold comparison processing based on risk indicator data, matches the risk indicators with a preset safety threshold library, and generates an alarm trigger signal when the indicator exceeds the limit, including the following steps: S41. A preset dynamic safety threshold library, which is generated based on historical data and clinical guidelines; S42. The risk indicator data is compared with the dynamic safety threshold in real time, and the boundary conditions are handled by the fuzzy logic algorithm. The fuzzy logic system defines the input variable as the risk score, with fuzzy sets of low, medium, and high, and the fuzzy set of the rate of change as small, medium, and large. The output variable is the alarm level, with fuzzy sets of none, mild, and severe. The membership function adopts the trigonometric function. When the risk score is "high", the membership degree is defined as 1.0 when the score is greater than 0.7, and a linear transition from 0.5 to 0.7. The fuzzy rule base contains 9 rules, including "If the risk score is high and the rate of change is large, then the alarm level is severe". Its alarm trigger judgment function for:
[0026] in, To trigger the alarm, To avoid triggering an alarm, High-risk thresholds are set in the dynamic security threshold library. The rate of change of the comprehensive risk score per unit of time. The rate of change threshold; Specifically, the rate of change By calculating the combined risk score of the current moment and the previous moment. The absolute value of the difference is obtained, and fuzzy logic algorithms are used to process it. near and near In the boundary cases, the risk membership degree is calculated through the membership function. When the risk membership degree exceeds the preset fuzzy threshold, it is determined to be an alarm trigger condition. S43. When the comparison result indicates a high risk, generate an alarm trigger signal; otherwise, continue monitoring.
[0027] S5 executes safety control actions based on the alarm trigger signal, adjusts the infusion rate through the flow rate regulation unit, triggers the air bubble removal mechanism, and sends alarm information via the notification unit, including the following steps: S51. Analyze the alarm trigger signal to determine the type of control action, including flow rate control, bubble treatment, and medical notification; S52. Control actions are achieved through the actuator unit, where flow rate control uses a stepper motor regulating valve, bubble treatment starts the ultrasonic defoaming device, and medical notifications are sent via SMS and App push notifications. S53. Record the execution log of control actions and feed it back to the data recording system.
[0028] S6 integrates infusion parameter data, safety risk analysis results, and safety control action logs, and outputs a structured infusion safety monitoring report through the report generation unit, including the following steps: S61. Integrate infusion parameter data, risk indicator data, control action logs, and user feedback data; S62. A report generation algorithm is adopted, combined with a template engine, to output a structured report, including a summary, trend charts, and recommended measures. The comprehensive risk assessment in the report generation algorithm is calculated using the following formula: ; in, For comprehensive risk scoring, This is an indicator of flow velocity stability. This is an index of the frequency of bubble occurrence. and These are preset weighting coefficients, reflecting the contribution of different indicators; S63. Store the report in a cloud database and support access from multiple terminals.
[0029] In S7, the monitoring report parameters are optimized based on user feedback data. Machine learning models are used to adaptively adjust risk algorithms and security thresholds to generate optimized monitoring strategies, including the following steps: S71. Collect user feedback data, including medical staff evaluations, false alarm records, and effectiveness scores; S72. Use a reinforcement learning model to optimize the risk identification algorithm and safety threshold, and generate an adaptive monitoring strategy. The model weight update formula is as follows: ; in, The model weight vector in the t-th iteration. For learning rate, The target risk value is calculated based on user feedback and actual results. Risk value predicted by the model This represents the gradient of the loss function with respect to the weights. Further includes: target risk value The loss function is calculated by weighting user ratings of the accuracy of historical alarms (0 for false alarms and 1 for accurate alarms) with the actual risk level of the corresponding event. The mean squared error loss function is used, i.e. ,in This represents the number of training samples; S73. Regularly update system parameters to ensure monitoring accuracy and adaptability.
[0030] The method also includes step S8: performing remote collaborative processing, realizing multi-device data sharing and expert remote diagnosis through a cloud platform, and generating a collaborative monitoring solution, wherein the consistency index C of multi-device data synchronization is calculated by the following formula: ; in, As a data consistency indicator, Standardized data vectors integrated for the central monitoring platform. For the standardized data vector of the k-th edge device, The total number of devices participating in the synchronization; Specifically, data vectors and All data include standardized flow rate, pressure, and temperature. Before consistency calculation, time window alignment units are used to ensure that each vector data corresponds to the same monitoring period. When the consistency index... When the data falls below a preset threshold, a data verification command is triggered. The central monitoring platform then initiates a data retransmission request to the device with the lowest consistency and corrects conflicting data using a majority rule.
[0031] The system includes: The data acquisition module collects real-time infusion data through a multi-parameter sensor unit, transmits the data through a communication transmission unit, and outputs standardized infusion data through a data preprocessing unit. The risk analysis module receives standardized infusion data, analyzes safety threats through the risk identification unit, generates risk indicators using the machine learning prediction unit, and outputs alarm signals through the threshold comparison unit. The safety execution module adjusts the infusion parameters through the control action unit based on the alarm signal, sends an alarm through the alarm notification unit, and obtains user response data through the feedback collection unit. The data management module receives risk indicators and user response data, archives historical records through the storage unit, creates monitoring reports using the report generation unit, and displays the analysis results through the visualization unit. The optimization learning module optimizes algorithm parameters through the model training unit based on user feedback and monitoring data, updates security policies through the adaptive adjustment unit, and outputs optimization instructions through the evaluation unit. The user interaction module displays the real-time monitoring status through a graphical interface unit, processes user commands through an input receiving unit, and provides operation guidance through a reminder unit.
[0032] The operation steps of an intelligent infusion safety monitoring system and method are as follows: Step 1: Multi-parameter data acquisition and transmission: The intelligent infusion device synchronously collects real-time infusion parameter data through a multi-parameter sensor unit. Among them, the optical flow rate sensor monitors the drug flow rate, the piezoelectric pressure sensor detects the pressure change in the infusion line, the infrared temperature sensor measures the drug temperature, and the bioelectrode physiological sensor acquires the patient's heart rate, blood oxygenation and other related physiological parameters. Each sensor is synchronized with a clock unit to ensure timestamp alignment, and the analog signals are converted into digital signals by an analog-to-digital converter to form a raw signal vector. Subsequently, the communication transmission unit transmits the packaged raw signal vector to the central monitoring platform through a wireless network to generate an initial infusion dataset, providing a complete data foundation for subsequent processing.
[0033] Step 2: Data Preprocessing and Standardization Transformation The data preprocessing unit cleans and filters the initial infusion dataset. First, a sliding window algorithm is used to identify and remove outliers and missing data. Then, a digital filter is applied to reduce environmental noise and high-frequency interference, generating smooth infusion data. The standardization and transformation unit normalizes the smooth infusion data based on the statistical characteristics of historical data. By subtracting the mean and dividing by the standard deviation, the multi-source heterogeneous graph data is converted into standardized infusion data with unified dimensions, ensuring comparability between different parameters and providing high-quality data input for risk analysis.
[0034] Step 3: Multi-dimensional risk analysis and indicator generation: After receiving standardized infusion data, the risk identification unit simultaneously executes rule-based algorithms and neural network model calculations. The rule-based algorithms perform preliminary screening based on threshold ranges set according to clinical guidelines, while the neural network model identifies complex abnormal patterns through deep learning. Specifically, it performs risk assessment in three dimensions: flow rate stability analysis, bubble probability calculation, and physiological parameter trend prediction. A weighted fusion algorithm is then used to generate a comprehensive risk score. The machine learning prediction unit dynamically updates the risk assessment model parameters based on real-time data and outputs risk indicator data that includes risk level scores, abnormality type identifiers, and confidence parameters.
[0035] Step 4: Dynamic Threshold Comparison and Alarm Decision The threshold comparison unit compares the generated risk indicator data with the dynamic safety threshold library in real time. The dynamic safety threshold library is established based on historical operation data and clinical practice guidelines. It can adaptively adjust the threshold range according to different infusion stages and uses fuzzy logic algorithm to handle boundary conditions. When the comprehensive risk score exceeds the high-risk threshold and the rate of change per unit time exceeds the critical value, the alarm trigger function outputs an alarm trigger signal. The decision-making process comprehensively considers the absolute value of the risk score and the trend of change to ensure that the alarm mechanism is both timely and accurate.
[0036] Step 5: Implementation of safety controls and collection of feedback: After the safety execution module analyzes the alarm trigger signal, the control action unit performs corresponding operations according to the alarm level: a level 1 alarm triggers the stepper motor regulating valve to adjust the infusion flow rate; a level 2 alarm activates the ultrasonic defoaming device to remove air bubbles from the pipeline; and a level 3 alarm simultaneously sends an emergency notification to the medical terminal. The alarm notification unit transmits alarm information through multiple communication channels, and the feedback collection unit synchronously records the control action execution effect and user response data to form a closed-loop control log, providing a basis for system optimization.
[0037] Step Six: Data Integration and System Adaptive Optimization The data management module integrates infusion parameter data, risk indicator data, control action logs, and user feedback data generated throughout the entire process. It generates structured monitoring reports through the report generation unit. The optimization learning module dynamically updates the risk identification model parameters and safety threshold settings based on the collected data using reinforcement learning algorithms. The model training unit continuously optimizes the algorithm weights by comparing the predicted results with the actual effects. The adaptive adjustment unit updates the system parameters based on the optimization results, thereby achieving iterative improvement of the monitoring strategy and ensuring the accuracy and adaptability of the system in long-term operation.
[0038] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A method for intelligent infusion safety monitoring, characterized in that, The method includes the following steps: S1. Collect infusion parameter data, including real-time flow rate, infusion pressure, fluid temperature and patient physiological parameters; S2. The infusion parameter data is preprocessed by removing outliers through a data cleaning unit, reducing interference signals through a noise filtering unit, and then generating standardized infusion data through a standardization conversion unit. S3. Based on the standardized infusion data, perform safety risk analysis, and use risk identification algorithms to calculate flow rate stability, bubble probability and allergic reaction trend to generate risk indicator data. S4. Perform dynamic threshold comparison processing based on the risk indicator data, match the risk indicators with the preset safety threshold library, and generate an alarm trigger signal when the indicator exceeds the limit. S5. Execute safety control actions according to the alarm trigger signal, adjust the infusion rate through the flow rate adjustment unit, trigger the bubble removal mechanism to start, and send alarm information through the notification unit; S6. Integrate infusion parameter data, safety risk analysis results, and safety control action logs, and output a structured infusion safety monitoring report through the report generation unit; S7. Optimize the monitoring report parameters based on user feedback data, and use a machine learning model to adaptively adjust the risk algorithm and security threshold to generate an optimized monitoring strategy.
2. The intelligent infusion safety monitoring method according to claim 1, characterized in that, The infusion parameter data collected in step S1 includes real-time flow rate, infusion pressure, fluid temperature, and patient physiological parameters, and includes the following steps: S11. Real-time infusion parameters are collected through the multi-parameter sensor unit mounted on the intelligent infusion device, including optical flow rate sensor, piezoelectric pressure sensor, infrared temperature sensor and bioelectrode physiological sensor. S12. The collected raw infusion parameter data is transmitted to the central monitoring platform through the wireless communication unit, and an initial infusion dataset is generated.
3. The intelligent infusion safety monitoring method according to claim 1, characterized in that, The preprocessing of the infusion parameter data in step S2, including removing outliers by a data cleaning unit, reducing interference signals by a noise filtering unit, and generating standardized infusion data by a standardization conversion unit, includes the following steps: S21. Receive the initial infusion dataset, and use the sliding window algorithm to clean the data stream, removing outliers and missing data; S22. Apply a digital filter to reduce noise in the cleaned data to generate smooth infusion data; S23. The smoothed infusion data is converted into standardized infusion data with uniform dimensions through the standardization conversion unit, which facilitates subsequent analysis; The standardized transformation is calculated using the following formula: ； in, To smooth individual data points in infusion data, The mean of historical smoothed infusion data. The standard deviation of historical smoothed infusion data. This is the standardized infusion data.
4. The intelligent infusion safety monitoring method according to claim 1, characterized in that, The S3 step involves performing a safety risk analysis based on the standardized infusion data, using a risk identification algorithm to calculate flow rate stability, bubble probability, and allergic reaction trends, and generating risk indicator data, including the following steps: S31. Obtain the standardized infusion data and input it into the risk identification engine, which integrates a rule-based algorithm and a neural network model. S32. Conduct multi-dimensional risk analysis, including flow rate stability analysis, bubble probability calculation, and physiological parameter trend prediction. S33. Output risk indicator data, which includes risk level score, anomaly type identifier and confidence level parameter.
5. The intelligent infusion safety monitoring method according to claim 1, characterized in that, The S4 step involves performing dynamic threshold comparison processing based on the risk indicator data, matching the risk indicators with a preset safety threshold library, and generating an alarm trigger signal when an indicator exceeds the limit. This includes the following steps: S41. A preset dynamic safety threshold library, which is generated based on historical data and clinical guidelines; S42. Compare the risk indicator data with the dynamic safety threshold in real time, and use fuzzy logic algorithm to handle boundary cases; S43. When the comparison result indicates a high risk, generate an alarm trigger signal; otherwise, continue monitoring.
6. The intelligent infusion safety monitoring method according to claim 1, characterized in that, The steps in S5, which involve executing safety control actions based on the alarm trigger signal, adjusting the infusion rate via the flow rate adjustment unit, triggering the bubble removal mechanism, and sending alarm information via the notification unit, include the following: S51. Analyze the alarm trigger signal to determine the type of control action, including flow rate control, bubble treatment, and medical notification; S52. Control actions are achieved through the actuator unit, where flow rate control uses a stepper motor regulating valve, bubble treatment starts the ultrasonic defoaming device, and medical notifications are sent via SMS and App push notifications. S53. Record the execution log of control actions and feed it back to the data recording system.
7. The intelligent infusion safety monitoring method according to claim 1, characterized in that, The S6 process integrates infusion parameter data, safety risk analysis results, and safety control action logs, and outputs a structured infusion safety monitoring report through the report generation unit, including the following steps: S61. Integrate infusion parameter data, risk indicator data, control action logs, and user feedback data; S62. A report generation algorithm is adopted, combined with a template engine, to output a structured report, including a summary, trend charts, and recommended measures. The comprehensive risk assessment in the report generation algorithm is calculated using the following formula: ； in, For comprehensive risk scoring, This is an indicator of flow velocity stability. This is an index of the frequency of bubble occurrence. and These are the preset weighting coefficients; S63. Store the report in a cloud database and support access from multiple terminals.
8. The intelligent infusion safety monitoring method according to claim 1, characterized in that, The steps in S7 to optimize the monitoring report parameters based on user feedback data, and to generate an optimized monitoring strategy by adaptively adjusting the risk algorithm and security threshold using a machine learning model, include the following: S71. Collect user feedback data, including medical staff evaluations, false alarm records, and effectiveness scores; S72. Use reinforcement learning models to optimize risk identification algorithms and security thresholds, and generate adaptive monitoring strategies. S73. Regularly update system parameters to ensure monitoring accuracy and adaptability.
9. The intelligent infusion safety monitoring method according to claim 1, characterized in that, The method also includes step S8: performing remote collaborative processing, realizing multi-device data sharing and remote expert diagnosis through a cloud platform, and generating a collaborative monitoring solution.
10. An intelligent infusion safety monitoring system, used to implement the intelligent infusion safety monitoring method according to any one of claims 1-9, characterized in that, The system includes: The data acquisition module collects real-time infusion data through a multi-parameter sensor unit, transmits the data through a communication transmission unit, and outputs standardized infusion data through a data preprocessing unit. The risk analysis module receives the standardized infusion data, analyzes safety threats through the risk identification unit, generates risk indicators using the machine learning prediction unit, and outputs alarm signals through the threshold comparison unit. The safety execution module adjusts the infusion parameters through the control action unit based on the alarm signal, sends an alarm through the alarm notification unit, and obtains user response data through the feedback collection unit. The data management module receives the risk indicators and user response data, archives historical records through the storage unit, creates monitoring reports using the report generation unit, and displays the analysis results through the visualization unit. The optimization learning module optimizes algorithm parameters through the model training unit based on user feedback and monitoring data, updates security policies through the adaptive adjustment unit, and outputs optimization instructions through the evaluation unit. The user interaction module displays the real-time monitoring status through a graphical interface unit, processes user commands through an input receiving unit, and provides operation guidance through a reminder unit.