Injection pump control method for abnormal judgment

Abstract

The invention discloses an injection pump control method for abnormity discrimination, and relates to the technical field of medical treatment, and the method comprises the following steps: constructing a double-layer reference system of individual adaptation and scene adaptation, the double-layer reference system comprises an individual pressure reference curve based on the individual physiological characteristics of a patient, and a scene interference tolerance interval based on the mobile scene type; collecting a pipeline pressure dynamic signal of an injection pump, a patient limb movement signal and a scene environment interference signal in real time, and synchronously obtaining characteristic parameters of a currently infused medicine; correcting the individual pressure reference curve based on the drug characteristic parameters to obtain a dynamic reference curve matched with the current infusion drug; the method has the advantages that blood vessel conditions and tolerance degrees of different patients can be accurately matched, missed judgment and false alarm caused by a unified threshold value are avoided, abnormal fluctuation of infusion pressure can be accurately captured, and the risk of pipeline blockage can be early warned in time.

Description

Technical Field

[0001] This invention relates to the field of medical technology, and more specifically, to an infusion pump control method for anomaly detection. Background Technology

[0002] In clinical intravenous infusion therapy, precise control and anomaly detection of the infusion pump are crucial for ensuring infusion safety. Traditional methods lack adaptability to mobile scenarios. Clinically, patients are often in different mobile scenarios such as bedridden, walking, and being transported. The impact of limb movement and environmental vibrations on tubing pressure varies in each scenario, but traditional control methods do not define interference tolerance ranges for each scenario, making it difficult to distinguish pressure fluctuations caused by interference from true anomalies. For example, pressure fluctuations caused by equipment vibration during patient transport are often misjudged as tubing blockages, increasing the workload of medical staff in ineffective interventions.

[0003] Meanwhile, traditional infusion pumps do not dynamically adjust the baseline based on drug characteristics. Due to the differences in the physicochemical properties of different drugs, such as viscosity and osmotic pressure, the pressure performance of the infusion line will be directly affected. Traditional methods use a fixed pressure baseline, which can easily trigger misjudgments when infusing high-viscosity drugs due to natural pressure increases, while low-viscosity drugs may be missed due to abnormal pressure drops. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide an abnormality detection method for injection pump control.

[0005] To achieve the above objectives, the present invention provides the following technical solution: An abnormality detection method for controlling an infusion pump, the method comprising the following steps: A dual-layer benchmark system for individual adaptation and scenario adaptation is constructed. The dual-layer benchmark system includes an individual stress benchmark curve based on the individual physiological characteristics of the patient and a scenario interference tolerance range based on the type of mobile scenario. Real-time acquisition of dynamic signals of tubing pressure of the infusion pump, patient limb movement signals and scene environmental interference signals, and simultaneous acquisition of the characteristic parameters of the currently infused drug; A dynamic baseline curve adapted to the current infused drug is obtained by correcting the individual pressure baseline curve based on drug characteristic parameters. The pressure difference between the pipeline pressure dynamic signal and the dynamic reference curve is calculated to obtain the out-of-range frequency value. The out-of-range frequency value is compared with the reference threshold and interference verification is performed to obtain the deviation judgment result. The interference verification result is obtained by combining the scene environment interference signal and the scene interference tolerance range. The deviation judgment result and the interference verification result are combined to obtain the infusion anomaly judgment result. If the infusion anomaly judgment result is a real anomaly, an alarm and infusion adjustment action are triggered. If the infusion anomaly judgment result is a false anomaly caused by interference, data is collected and continuously monitored. If the infusion anomaly judgment result is normal, the dual-layer reference system is dynamically optimized.

[0006] Preferably, a two-layer benchmark system of individual adaptation and scenario adaptation is constructed, specifically including the following steps: Collect individual physiological characteristic data of patients and combine them with the physicochemical properties of infused drugs to generate an initial individual pressure baseline curve; The mobile scenario types are divided, and interference signal samples during normal infusion under each mobile scenario type are collected. The interference signal samples of each scenario are statistically judged, and the range of pressure fluctuation under different mobile scenario types is determined to obtain the scenario interference tolerance range. The scenario interference tolerance range includes the interference intensity threshold and the interference duration threshold. By linking the individual stress baseline curve with the interference tolerance range of each scenario, a two-layer baseline system of individual adaptation and scenario adaptation is formed.

[0007] Preferably, the real-time acquisition of dynamic signals of the infusion pump tubing pressure, patient limb movement, and environmental interference signals includes the following steps: Pressure data from different cross sections of the pipeline are collected, and the pressure data are fused to generate a dynamic pipeline pressure signal. Data on changes in angle, range of motion, and acceleration of movement of the infused limb are collected and integrated into a signal of the patient's limb movement. Vibration frequency parameters, noise intensity parameters, temperature and humidity parameters of the scene environment are collected, and key interference parameters related to pressure fluctuations are screened to form scene environment interference signals.

[0008] Preferably, the dynamic baseline curve adapted to the current infused drug is obtained by correcting the individual pressure baseline curve based on drug characteristic parameters, specifically including the following steps: Establish a comparison table of drug characteristics and pressure correction coefficients, and match the corresponding correction coefficients in the comparison table according to the characteristic parameters of the currently infused drug. The initial individual pressure baseline curve is adjusted using a linear correction algorithm. The correction formula is as follows: , where P net For the dynamic baseline curve, P init This is the initial individual pressure baseline curve, where K is the correction factor. P represents the baseline pressure offset of the drug; If the drug concentration changes during infusion, a dynamic baseline curve is obtained by matching and correcting the coefficient in real time.

[0009] Preferably, the out-of-range frequency value is obtained by calculating the pressure difference between the pipeline pressure dynamic signal and the dynamic reference curve. The out-of-range frequency value is then compared with the reference threshold and subjected to interference verification to obtain the deviation judgment result. Specifically, this includes the following steps: Calculate the pressure difference between the pipeline pressure dynamic signal and the dynamic reference curve at the same time node, and collect the number of times the pressure difference exceeds the fluctuation range of the dynamic reference curve within at least two sampling periods to obtain the out-of-range frequency value. A preset baseline threshold is used to compare the out-of-range frequency value with the baseline threshold. If the out-of-range frequency value is greater than or equal to the baseline threshold, it is marked as a preliminary anomaly. If the out-of-range frequency value is less than the baseline threshold, it is judged as a false anomaly. Extract key interference parameters from the scene environment interference signals in the preliminary anomaly and compare them with the scene interference tolerance range corresponding to the current scene; If the key interference parameters do not exceed the scene interference tolerance range, the interference is determined to be invalid, and the initial anomaly remains a valid anomaly. If the key interference parameters exceed the scene interference tolerance range, the temporal correlation value between the pipeline pressure dynamic signal and the patient's limb movement signal is determined. If the temporal correlation value is greater than or equal to the preset threshold, the interference is determined to be effective, and the initial abnormality is corrected to a false abnormality; if the temporal correlation value is less than the preset threshold, it is determined to be a valid abnormality.

[0010] Preferably, determining the time-series correlation value between the dynamic signal of tubing pressure and the patient's limb movement signal specifically includes the following steps: Acquire dynamic signal segments of tubing pressure and corresponding signal segments of patient limb movement during the initial abnormality period; The correlation coefficient between dynamic signal segments of pipeline pressure and signal segments of patient limb movement is calculated to obtain the time-series correlation value; If the temporal correlation value is less than the correlation threshold, it is determined to be invalid interference; If the temporal correlation value is greater than or equal to the correlation threshold, it is determined to be a valid interference.

[0011] Preferably, if the infusion anomaly determination result is a true anomaly, an alarm and infusion adjustment actions are triggered, specifically including the following steps: Anomalies are classified into three levels based on the magnitude and duration of pressure deviation: mild anomaly, moderate anomaly, and severe anomaly. Mild abnormality: Triggers a low-intensity audible and visual alarm, controls the infusion pump to maintain the current infusion rate, and continuously monitors pressure changes; Moderate abnormality: Triggers a moderate-intensity audible and visual alarm, sends alarm information to the medical terminal, controls the infusion pump to reduce the current infusion rate, or suspends the infusion; Severe abnormality: Triggers a high-intensity audible and visual alarm, immediately stops the infusion and locks the piston, sends an emergency alarm message to the medical terminal, and collects the complete data chain at the time of the abnormality for traceability and judgment.

[0012] Preferably, the method further includes dynamic optimization of a dual-layer benchmark system of individual adaptation and scene adaptation, specifically including the following steps: Collect all currently valid data after the current infusion is completed; The individual pressure baseline curve is calibrated based on the current valid data. If the average deviation between the actual pressure data and the baseline curve exceeds the preset threshold, the curve fitting parameters are adjusted. Based on mobile scenario data, new data is obtained by filtering the current valid data, and the mobile scenario data is then optimized using the new data to obtain new mobile scenario data. Collect clinical abnormal misjudgment cases, and adjust the preset threshold of scene interference tolerance range and correlation judgment threshold through new mobile scene data.

[0013] Preferably, the effective data includes: patient physiological characteristics, drug parameters, signal data, judgment results, and clinical feedback.

[0014] Compared with existing technologies, this invention has the following beneficial effects: By constructing an individualized pressure baseline curve, which fully combines the physiological characteristics of patients' age, weight, and underlying diseases, it can accurately match the vascular conditions and tolerance levels of different patients, avoiding missed detections and false alarms caused by uniform thresholds. For elderly patients with poor vascular elasticity, it can accurately capture abnormal fluctuations in infusion pressure and provide timely warnings of the risk of tubing blockage. For young patients with active limbs, it can distinguish between pressure fluctuations caused by normal activities and real abnormalities, significantly reducing the probability of false alarms and improving individual adaptability. By dividing different movement scenarios such as bed rest, walking, and transportation and setting exclusive interference tolerance ranges, it can distinguish between pressure fluctuations caused by scenario interference and real abnormalities, reducing the workload of medical staff in ineffective treatments and improving the reliability of abnormality identification in complex clinical scenarios. By establishing the correspondence between drug characteristics and correction coefficients and dynamically adjusting the individualized pressure baseline curve, it can adapt to the infusion pressure characteristics of drugs with different viscosities and osmotic pressures, realize dynamic optimization of drug adaptability, and ensure the safe monitoring of different drug infusions. Attached Figure Description

[0015] Figure 1 This invention provides a schematic diagram of the steps involved in an injection pump control method for anomaly detection. Figure 2 This is a schematic diagram illustrating the steps for obtaining deviation judgment results in an injection pump control method for anomaly detection proposed in this invention. Detailed Implementation

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0017] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0018] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0019] Reference Figures 1-2 As shown.

[0020] This embodiment further illustrates the injection pump control method for anomaly detection proposed in this invention.

[0021] An abnormality detection method for controlling an infusion pump, the method comprising the following steps: A dual-layer benchmark system for individual adaptation and scenario adaptation is constructed. The dual-layer benchmark system includes an individual stress benchmark curve based on the individual physiological characteristics of the patient and a scenario interference tolerance range based on the type of mobile scenario. Real-time acquisition of dynamic signals of tubing pressure of the infusion pump, patient limb movement signals and scene environmental interference signals, and simultaneous acquisition of the characteristic parameters of the currently infused drug; A dynamic baseline curve adapted to the current infused drug is obtained by correcting the individual pressure baseline curve based on drug characteristic parameters. The pressure difference between the pipeline pressure dynamic signal and the dynamic reference curve is calculated to obtain the out-of-range frequency value. The out-of-range frequency value is compared with the reference threshold and interference verification is performed to obtain the deviation judgment result. The interference verification result is obtained by combining the scene environment interference signal and the scene interference tolerance range. The deviation judgment result and the interference verification result are combined to obtain the infusion anomaly judgment result. If the infusion anomaly judgment result is a real anomaly, an alarm and infusion adjustment action are triggered. If the infusion anomaly judgment result is a false anomaly caused by interference, data is collected and continuously monitored. If the infusion anomaly judgment result is normal, the dual-layer reference system is dynamically optimized.

[0022] First, a two-tiered benchmark system is constructed, which adapts to individual needs and scenarios. On the one hand, a unique individual pressure benchmark curve is generated for each patient’s individual physiological characteristics, such as age and underlying diseases, to match the tubing pressure range during normal infusion. On the other hand, for different types of mobility scenarios, such as bedridden, walking, and transport, corresponding scenario interference tolerance ranges are set to clarify the pressure fluctuation thresholds that may occur in different scenarios.

[0023] During the infusion process, multi-dimensional data is collected in real time, including dynamic pressure signals of the infusion pump tubing to reflect the real-time status of the infusion pathway; patient limb activity signals such as limb swinging and changes in body position; environmental interference signals of the current scene; and simultaneously acquiring characteristic parameters of the currently infused drug, such as drug viscosity and flow rate requirements.

[0024] By dynamically correcting the baseline based on the characteristics of the drug, the pre-constructed individual pressure baseline curve is adjusted according to the parameters of the current infusion drug to obtain a dynamic baseline curve that is suitable for the drug. This makes the baseline more in line with the actual needs of the current infusion and avoids misjudgment due to drug differences.

[0025] Anomaly detection is divided into two dimensions: deviation judgment and interference verification. First, the real-time collected pipeline pressure dynamic signal is compared with the corrected dynamic reference curve to determine whether the pressure exceeds the normal range, thus obtaining the deviation judgment result. At the same time, the real-time collected scene environment interference signal is compared with the pre-set scene interference tolerance range to determine whether the current interference is within a reasonable range, thus obtaining the interference verification result.

[0026] The result fusion and response combines the deviation judgment result with the interference verification result to obtain the final infusion abnormality judgment result and corresponding to different processing logics. If it is judged to be a real abnormality, such as tubing blockage or dislodgement, an alarm is immediately triggered and the infusion status is automatically adjusted. If it is judged to be a false abnormality caused by interference, such as pressure fluctuations caused by limb movement when the patient walks, data is continuously collected and monitored without triggering an alarm. If it is judged to be normal, the current effective data is used to dynamically optimize the pre-constructed two-layer benchmark system to make subsequent judgments more accurate.

[0027] Constructing a two-tiered benchmark system for individual adaptation and scenario adaptation includes the following steps: Collect individual physiological characteristic data of patients and combine them with the physicochemical properties of infused drugs to generate an initial individual pressure baseline curve; The mobile scenario types are divided, and interference signal samples during normal infusion under each mobile scenario type are collected. The interference signal samples of each scenario are statistically judged, and the range of pressure fluctuation under different mobile scenario types is determined to obtain the scenario interference tolerance range. The scenario interference tolerance range includes the interference intensity threshold and the interference duration threshold. By linking the individual stress baseline curve with the interference tolerance range of each scenario, a two-layer baseline system of individual adaptation and scenario adaptation is formed.

[0028] Collect individual physiological characteristic data of patients, such as age, weight, and baseline physical condition, and combine this data with the physicochemical properties of the medication to be infused, such as viscosity and osmotic pressure. By combining individual physiological characteristic data with the physicochemical properties of the medication to be infused, an initial individual pressure baseline curve specific to the patient and suitable for the current medication is generated, thus reflecting the normal range that the tubing pressure should be within when the patient is using this medication for infusion.

[0029] The patient movement scenarios are categorized, such as distinguishing between scenarios involving a patient lying still, slow walking, and transport. Interference signal samples generated during normal intravenous infusion are then collected for each scenario, such as tubing pressure fluctuations caused by patient limb movement. The interference signal samples for each scenario are then statistically analyzed to determine the patterns of pressure fluctuations and establish the corresponding pressure fluctuation intervals for different movement scenario types, thus obtaining the scenario interference tolerance interval. The scenario interference tolerance interval includes an interference intensity threshold and an interference duration threshold. The interference intensity threshold represents the maximum reasonable amplitude of pressure fluctuation in that scenario, and the interference duration threshold represents the reasonable duration for which such pressure fluctuations can be sustained in that scenario.

[0030] The generated individual pressure baseline curves are associated and stored with the scene interference tolerance ranges corresponding to each scenario. By combining the individual-level pressure baselines with the interference allowable ranges under different scenarios, a dual-layer baseline system that simultaneously covers individual and scene adaptation is formed, providing a comprehensive and accurate reference standard for abnormal identification in subsequent infusion processes.

[0031] Real-time acquisition of dynamic signals from the infusion pump tubing pressure, patient limb movement, and environmental interference signals includes the following steps: Pressure data from different cross sections of the pipeline are collected, and the pressure data are fused to generate a dynamic pipeline pressure signal. Data on changes in angle, range of motion, and acceleration of movement of the infused limb are collected and integrated into a signal of the patient's limb movement. Vibration frequency parameters, noise intensity parameters, temperature and humidity parameters of the scene environment are collected, and key interference parameters related to pressure fluctuations are screened to form scene environment interference signals.

[0032] Pressure data is collected at different cross-sectional locations along the tubing of the infusion pump, such as pressure values ​​at multiple cross-sections near the pump body and near the patient's puncture site. This pressure data from different cross-sections is then fused and processed to synthesize pressure changes at various locations, ultimately generating a dynamic pressure signal that reflects the real-time pressure status of the entire tubing. This ensures comprehensive monitoring of pressure fluctuations in the infusion pathway.

[0033] The limb receiving the infusion is monitored, and multiple data points are collected during the limb's activity, including changes in limb angles, such as the angle of arm flexion or extension; the range of limb extension and contraction, such as the length range of arm extension or contraction; and acceleration data of limb movement, such as the rate of change of limb speed during activity. These angle, amplitude, and acceleration data are then integrated to form a patient limb activity signal that reflects the patient's limb activity status, thereby capturing the impact of limb movements on the infusion tubing.

[0034] Parameters of the current environment are collected, including vibration frequency parameters (e.g., the vibration frequency when the transport bed moves), noise intensity parameters (e.g., the noise level of equipment operating in the ward), and temperature and humidity parameters (e.g., the temperature and humidity values ​​in the current space). From these collected environmental parameters, key interference parameters related to pipeline pressure fluctuations are selected, such as vibration frequency and other parameters that affect pressure. These key parameters are then integrated to form an environmental interference signal, thereby clarifying the interference of environmental factors on the infusion process.

[0035] The dynamic baseline curve adapted to the current infusion drug is obtained by correcting the individual pressure baseline curve based on drug characteristic parameters. The specific steps include: Establish a comparison table of drug characteristics and pressure correction coefficients, and match the corresponding correction coefficients in the comparison table according to the characteristic parameters of the currently infused drug. The initial individual pressure baseline curve is adjusted using a linear correction algorithm. The correction formula is as follows: , where P net For the dynamic baseline curve, P init This is the initial individual pressure baseline curve, where K is the correction factor. P represents the baseline pressure offset of the drug; If the drug concentration changes during infusion, a dynamic baseline curve is obtained by matching and correcting the coefficient in real time.

[0036] A pre-established table mapping drug properties to pressure correction factors is used. This table maps the physicochemical properties of different drugs, such as viscosity and concentration, to specific correction factors. For example, if a patient's initial individual pressure baseline corresponds to a pressure of 20 kPa, and a high-viscosity antibiotic needs to be infused, the viscosity parameter of that antibiotic is extracted and matched to the corresponding correction factor of 1.2 from the table. Simultaneously, the baseline pressure offset for this drug is determined to be 3 kPa, providing specific parameters for subsequent adjustments to the baseline curve.

[0037] A dynamic baseline curve is generated using a linear correction algorithm. This algorithm combines the initial individual pressure baseline curve with the matched correction parameters, and a dynamic baseline curve suitable for the current drug is calculated using a formula. Continuing the previous example, the pressure P corresponding to the initial individual pressure baseline curve... init The baseline pressure is 20 kPa, the correction factor K is 1.2, and the drug baseline pressure offset ΔP is 3 kPa. Substituting these values ​​into the formula P... net =P init The pressure value of the dynamic baseline curve can be calculated as 27 kPa using ×K+ΔP. This ensures that the baseline curve closely matches the infusion pressure characteristics of high-viscosity drugs, avoiding misinterpretation of abnormal pressure increases caused by high drug viscosity.

[0038] To address changes in drug concentration during infusion, if the drug concentration changes during the infusion process—for example, the concentration of a high-viscosity antibiotic is adjusted from a low to a higher concentration—the system extracts the changed drug concentration parameters in real time and matches the corresponding correction coefficient from the reference table. Assuming the correction coefficient for the higher concentration is updated to 1.4, and the drug baseline pressure offset remains at 3 kPa, the pressure value of the dynamic baseline curve will be updated to 20 × 1.4 + 3 = 31 kPa when resubmitted into the formula. Through real-time matching and calculation, the system ensures that the dynamic baseline curve remains adapted to the characteristics of the current drug throughout the entire infusion process, providing an accurate reference for subsequent anomaly detection.

[0039] The out-of-range frequency value is obtained by calculating the pressure difference between the pipeline pressure dynamic signal and the dynamic reference curve. The deviation judgment result is obtained by comparing the out-of-range frequency value with the reference threshold and performing interference verification. The specific steps include: Calculate the pressure difference between the pipeline pressure dynamic signal and the dynamic reference curve at the same time node, and collect the number of times the pressure difference exceeds the fluctuation range of the dynamic reference curve within at least two sampling periods to obtain the out-of-range frequency value. A preset baseline threshold is used to compare the out-of-range frequency value with the baseline threshold. If the out-of-range frequency value is greater than or equal to the baseline threshold, it is marked as a preliminary anomaly. If the out-of-range frequency value is less than the baseline threshold, it is judged as a false anomaly. Extract key interference parameters from the scene environment interference signals in the preliminary anomaly and compare them with the scene interference tolerance range corresponding to the current scene; If the key interference parameters do not exceed the scene interference tolerance range, the interference is determined to be invalid, and the initial anomaly remains a valid anomaly. If the key interference parameters exceed the scene interference tolerance range, the temporal correlation value between the pipeline pressure dynamic signal and the patient's limb movement signal is determined. If the temporal correlation value is greater than or equal to the preset threshold, the interference is determined to be effective, and the initial abnormality is corrected to a false abnormality; if the temporal correlation value is less than the preset threshold, it is determined to be a valid abnormality.

[0040] First, the pressure difference between the pipeline pressure dynamic signal and the dynamic reference curve at the same time node is calculated, and the frequency of out-of-range values ​​is counted. The system synchronously collects real-time pipeline pressure data and corresponding values ​​of the dynamic reference curve, calculates the pressure difference between the two at the same time node, and then counts the number of times the pressure difference exceeds the fluctuation range within at least two sampling periods. For example, when a patient is currently receiving a high-concentration antibiotic infusion, the pressure value corresponding to the dynamic reference curve is 31 kPa, and the normal fluctuation range of this curve is preset to ±2 kPa. In three consecutive sampling periods, the pipeline pressure dynamic signal is displayed as 32 kPa, 34 kPa, and 35 kPa, respectively. Compared with the dynamic reference value of 31 kPa, the differences are 1 kPa, 3 kPa, and 4 kPa, respectively. Among them, 1 kPa is within the fluctuation range of ±2 kPa, while 3 kPa and 4 kPa are both outside the range. Therefore, the out-of-range frequency value in this stage is 2 times. Through difference calculation and frequency statistics, the abnormal fluctuation trend of pipeline pressure is initially captured, avoiding misjudgment caused by a single accidental fluctuation.

[0041] By initially distinguishing abnormal types through benchmark comparison, a benchmark threshold is first preset. This threshold is a critical value for the frequency of exceeding the range, determined based on clinical data and infusion safety standards. For example, the benchmark threshold is set to 2 times. The obtained frequency of exceeding the range (2 times) is then compared with the benchmark threshold (2 times). If the two values ​​are equal, the current state is marked as a preliminary abnormality. If the frequency of exceeding the range (1 time) is less than the benchmark threshold (2 times), it is directly judged as a false abnormality and temporarily not proceeds to the subsequent verification stage. By comparing the frequency with the threshold, suspected abnormalities requiring further verification are screened out, reducing unnecessary verification processes and improving judgment efficiency.

[0042] Once the system identifies a preliminary anomaly, it extracts the key interference parameters from the corresponding environmental interference signal. The environmental interference signal includes vibration frequency and noise intensity parameters, while the key interference parameters are indicators directly related to pipeline pressure fluctuations. For example, in a scenario where a patient is walking slowly, the corresponding key interference parameter is the pressure fluctuation value caused by limb movement. Simultaneously, the system retrieves the corresponding scene interference tolerance range; for instance, the pressure fluctuation intensity threshold within this range is ±3 kPa. The key interference parameter, such as the 2 kPa pressure fluctuation value caused by the patient walking, is then compared to the ±3 kPa range. If the key interference parameter does not exceed the tolerance range, the interference is deemed invalid, and the preliminary anomaly remains valid, indicating that the anomaly is not caused by environmental interference and requires further investigation.

[0043] If key interference parameters exceed the scene's interference tolerance range, such as pressure fluctuations of 4 kPa during patient walking, exceeding the ±3 kPa tolerance range, the effectiveness of the interference is further determined by the time-series correlation value. The time-series correlation value refers to the degree of overlap between the timing of fluctuations in the pipeline pressure dynamic signal and the timing of the patient's limb movement signal. Its preset threshold is set to 0.8, representing 80% overlap. If the pipeline pressure rises to 34 kPa or 35 kPa, completely coinciding with the moment of significant arm swing, the time-series correlation value reaches 0.9, exceeding the preset threshold of 0.8. In this case, the interference is deemed effective, correcting the initial abnormality to a false abnormality, indicating that the pressure fluctuation is a benign interference caused by the patient's limb movement. If the patient's limb is stationary when the pipeline pressure increases, the time-series correlation value is only 0.4, less than the preset threshold of 0.8. This is considered a valid abnormality, indicating that the pressure fluctuation is not caused by limb movement but may be due to a real malfunction such as pipeline blockage.

[0044] By employing a layered process of difference calculation and frequency statistics, benchmark comparison for initial anomaly judgment, comparison of scene interference parameters and tolerance range, and verification of interference effectiveness by time-series correlation values, a complete judgment logic from initial capture of fluctuations to accurate differentiation of anomaly types has been achieved. This not only avoids false alarms caused by environmental interference but also promptly identifies real infusion anomalies, providing multi-dimensional protection for infusion safety.

[0045] Determining the temporal correlation between dynamic signals of tubing pressure and signals of patient limb movement involves the following steps: Acquire dynamic signal segments of tubing pressure and corresponding signal segments of patient limb movement during the initial abnormality period; The correlation coefficient between dynamic signal segments of pipeline pressure and signal segments of patient limb movement is calculated to obtain the time-series correlation value; If the temporal correlation value is less than the correlation threshold, it is determined to be invalid interference; If the temporal correlation value is greater than or equal to the correlation threshold, it is determined to be a valid interference.

[0046] Continuing with the previous scenario, when a patient receives a high-concentration antibiotic infusion, the system has already flagged an initial anomaly, and key interference parameters exceed the scenario's tolerance range. At this point, the time-series correlation value judgment process is initiated. First, paired signal segments are acquired: the system identifies the time period of the initial anomaly, such as the 10-second sampling period during which the tubing pressure fluctuates between 34 kPa and 35 kPa in the previous case. Simultaneously, dynamic signal segments of the tubing pressure within this time period are extracted, including continuous values ​​of pressure increasing from 31 kPa to 34 kPa and then to 35 kPa, along with corresponding patient limb movement signal segments. Data such as changes in arm angle and extension / retraction amplitude are recorded during this time period, for example, the sequence of arm movements from hanging down to bending at 90 degrees and then extending. This ensures that the dynamic signal segments of the tubing pressure and the corresponding patient limb movement signal segments are within the same time frame, providing matching analytical samples for subsequent correlation calculations and avoiding distortion of correlation judgments due to time period misalignment.

[0047] The temporal correlation value is calculated by determining the correlation coefficient between two types of signal segments. The formula for calculating the correlation coefficient is: , where x i X¯ is the i-th sampled value of the pipeline pressure dynamic signal segment, such as 31 kPa, 34 kPa, 35 kPa, where X¯ is the average pressure value of that segment; y i y represents the i-th sampled value of a segment of the patient's limb movement signal (e.g., arm angles of 0°, 90°, 180°), y¯ is the average limb movement value of that segment, and n is the number of sampling points. Substituting the data from the case study, if the pressure signal is [31, 34, 35] and the limb angle signal is [0, 90, 180], the correlation coefficient r = 0.961 can be calculated, meaning the temporal correlation value is 0.961. If the pressure signal is [31, 34, 35] and the limb angle signal is always 0°, i.e., the patient is still, then the correlation coefficient r = 0, and the temporal correlation value is 0. By quantifying the correlation coefficient, the synchronicity between pressure fluctuations and limb movement is transformed into a specific numerical value, making the correlation judgment more objective and accurate.

[0048] The effectiveness of the interference is determined based on a correlation threshold. A preset correlation threshold is used, for example, 0.8. The calculated temporal correlation value is compared with the threshold. If the temporal correlation value is less than the correlation threshold, for example, a correlation value of 0 when the patient is at rest, it indicates that the pressure fluctuation and limb movement are not synchronized, and it is determined to be an invalid interference, while the initial abnormality remains a valid abnormality. If the temporal correlation value is greater than or equal to the correlation threshold, for example, a correlation value of 0.99 when the arm is moving, it indicates that the pressure fluctuation and limb movement are highly synchronized, and it is determined to be a valid interference, while the initial abnormality is corrected to a false abnormality.

[0049] By pairing signal segments from the same time period, calculating the correlation coefficient to obtain the temporal correlation value, and comparing the correlation value with a threshold to determine interference, the system accurately distinguishes between benign interference caused by limb activity and real anomalies. For example, when administering high-concentration antibiotics, if the temporal correlation value of pressure fluctuation reaches 0.961, the system will clearly identify it as valid interference caused by limb activity; if the correlation value is 0, it will be identified as a real anomaly and trigger subsequent processing, thus avoiding false alarms and ensuring infusion safety.

[0050] If the infusion anomaly assessment result is a genuine anomaly, an alarm and infusion adjustment actions will be triggered, specifically including the following steps: Anomalies are classified into three levels based on the magnitude and duration of pressure deviation: mild anomaly, moderate anomaly, and severe anomaly. Mild abnormality: Triggers a low-intensity audible and visual alarm, controls the infusion pump to maintain the current infusion rate, and continuously monitors pressure changes; Moderate abnormality: Triggers a moderate-intensity audible and visual alarm, sends alarm information to the medical terminal, controls the infusion pump to reduce the current infusion rate, or suspends the infusion; Severe abnormality: Triggers a high-intensity audible and visual alarm, immediately stops the infusion and locks the piston, sends an emergency alarm message to the medical terminal, and collects the complete data chain at the time of the abnormality for traceability and judgment.

[0051] Anomalies are classified into three levels based on the magnitude and duration of pressure deviation: mild, moderate, and severe. The magnitude of the pressure deviation refers to the difference between the dynamic pressure signal and the dynamic reference curve, while the duration is the length of time the deviation persists. The combination of these two factors determines the classification standard for the anomaly level. For example, if the dynamic reference curve pressure is 31 kPa, a mild anomaly is defined as a pressure deviation exceeding the fluctuation range by ±2 kPa, but with a magnitude ≤3 kPa and a duration ≤10 seconds; a moderate anomaly is defined as a deviation of 3 kPa-5 kPa and a duration of 10-30 seconds; and a severe anomaly is defined as a deviation ≥5 kPa and a duration ≥30 seconds.

[0052] For mild anomalies, the system will execute low-intensity intervention actions. For example, if during infusion, the tubing pressure rises to 34 kPa, deviating 3 kPa from the baseline value of 31 kPa, and this state persists for 8 seconds, it meets the criteria for a mild anomaly. At this time, the system will trigger a low-intensity audible and visual alarm, such as a soft beeping sound and a flashing yellow indicator light, while simultaneously controlling the infusion pump to maintain the current infusion rate and continuously monitoring pressure changes. By combining early warning and observation, the system alerts medical staff to potential risks while avoiding excessive adjustments that could disrupt the infusion process, making it suitable for minor fluctuations that do not currently threaten infusion safety.

[0053] If the abnormality level escalates to a moderate abnormality, the system will execute moderate-intensity intervention and notification actions. For example, if the tubing pressure rises to 36 kPa, deviating from the baseline value by 5 kPa, and persists for 20 seconds, reaching the moderate abnormality standard, the system will trigger a moderate-intensity audible and visual alarm. The buzzer volume will be greater than that of a mild abnormality, and the indicator light will flash orange. Simultaneously, an alarm message containing patient information and the current pressure of the infusion medication will be sent to the medical staff terminal, allowing medical personnel to remotely understand the abnormal situation. At the same time, the system will adjust the infusion rate according to the characteristics of the medication. For example, for high-concentration antibiotics, the system will control the infusion pump to reduce the current infusion rate by 20% or directly pause the infusion to avoid the risk of tubing damage or drug overdose caused by continuously rising pressure. This strengthens the early warning level, reduces safety hazards through proactive adjustments, and achieves remote synchronization with the medical staff terminal.

[0054] When the abnormality level reaches severe, the system will execute the highest-intensity emergency intervention. For example, if the tubing pressure rises to 38 kPa, deviating from the baseline value by 7 kPa, and persists for 35 seconds, it meets the criteria for severe abnormality. At this time, the system will trigger a high-intensity audible and visual alarm: the buzzer will change to a continuous, sharp warning sound, the indicator light will be constantly lit red, and the infusion will be immediately paused and the infusion pump piston will be locked, preventing further drug infusion at the hardware level. Simultaneously, the system will send an emergency alarm message to the medical staff terminal, which includes key information about the pressure change curve over time, facilitating rapid assessment by medical personnel. In addition, the system will automatically collect a complete data chain at the time of the abnormality, including dynamic pressure signals, limb activity signals, and scene interference signals, providing comprehensive data for subsequent analysis to trace the cause of the abnormality. Through immediate damage control, comprehensive early warning, and data retention, the system minimizes the harm of severe abnormalities to patients and provides a complete basis for subsequent cause investigation.

[0055] By classifying anomalies into different levels and matching them with corresponding response actions, a hierarchical logic is implemented to accurately handle real anomalies. For example, when administering high-concentration antibiotics, the response is tailored to the severity of each level, from early warning and observation of mild anomalies to adjustment notifications for moderate anomalies and emergency stoppage for severe anomalies. This ensures infusion safety while avoiding unnecessary over-intervention, making infusion monitoring more scientific and efficient.

[0056] The method also includes dynamic optimization of a dual-benchmark system of individual adaptation and scene adaptation, specifically including the following steps: Collect all currently valid data after the current infusion is completed; The individual pressure baseline curve is calibrated based on the current valid data. If the average deviation between the actual pressure data and the baseline curve exceeds the preset threshold, the curve fitting parameters are adjusted. Based on mobile scenario data, new data is obtained by filtering the current valid data, and the mobile scenario data is then optimized using the new data to obtain new mobile scenario data. Collect clinical abnormal misjudgment cases, and adjust the preset threshold of scene interference tolerance range and correlation judgment threshold through new mobile scene data.

[0057] Collect current valid data and summarize all validated valid data during this infusion process, including the patient's individual physiological characteristics data, dynamic signals of tubing pressure at different time periods, limb activity signals under different movement scenarios, drug characteristic parameters, and various data in the abnormality identification process, such as pressure fluctuation data and limb angle change data during the patient's walking scenario in this infusion. These data need to be retained after validity verification as the basic material for optimizing the benchmark system.

[0058] The individual pressure baseline curve is calibrated by comparing the actual pressure data of this infusion with the original individual pressure baseline curve and calculating the average deviation between the two. For example, if the original individual pressure baseline curve has an average pressure of 31 kPa for this drug infusion, while the actual average pressure of this infusion is 32 kPa, the average deviation is 1 kPa. A preset deviation threshold of 0.8 kPa is set. If the average deviation exceeds this threshold (e.g., 1 kPa > 0.8 kPa), the curve fitting parameters are adjusted: the slope and fluctuation range of the original individual pressure baseline curve are corrected using an algorithm, updating it to a new curve that better reflects the actual infusion situation. For example, the average pressure of the baseline curve is adjusted to 32 kPa, and the fluctuation range is adapted to ±2.2 kPa, making the individual baseline curve more accurately match the patient's current physiological state and drug infusion characteristics. If the average deviation does not exceed the threshold, the original curve parameters are maintained, and only the data from this infusion is added to the sample database.

[0059] The mobile scene data is optimized by filtering the current valid data according to the type of mobile scene and extracting new data for the corresponding scene. For example, pressure fluctuation signals and limb activity signals in the patient's walking scene are filtered from the current infusion data to form new data samples for that scene. Then, these new data are integrated into the original mobile scene data. The distribution characteristics of the updated scene data are determined by statistical analysis. For example, if the average pressure fluctuation in the original walking scene is 2 kPa, and the average pressure fluctuation in the new data for that scene is 2.5 kPa, then the pressure fluctuation data of the walking scene is updated to the merged average of 2.3 kPa, resulting in new mobile scene data that better reflects the characteristics of the patient's current activity state.

[0060] We collected abnormal misjudgment cases in this mobile scenario in clinical practice, such as cases where a pressure fluctuation of 3 kPa in a walking scenario was misjudged as a real abnormality. Combining this with new mobile scenario data, we adjusted the preset thresholds for the scenario interference tolerance range and the correlation judgment threshold. For example, the original interference tolerance range pressure threshold for the walking scenario was ±3 kPa, and the correlation judgment threshold was 0.8. Combining the new mobile scenario data, with an average pressure fluctuation of 2.3 kPa and misjudgment cases, we further adjusted the interference tolerance range pressure threshold for this scenario to ±3.5 kPa and fine-tuned the correlation judgment threshold to 0.75. This made the scenario interference tolerance range more suitable for the current scenario data distribution and reduced abnormal misjudgments in subsequent infusions.

[0061] By collecting effective data, calibrating individual baseline curves, optimizing scenario data, and adjusting tolerance ranges, continuous iteration of the dual-layer baseline system has been achieved. For example, after high-concentration antibiotic infusion, the individual pressure baseline curve is more closely aligned with the patient's actual infusion pressure, and the interference tolerance range for walking scenarios is more suitable for the patient's current activity pressure fluctuations. When the patient receives another infusion of the same drug, the accuracy of the baseline system will be significantly improved, reducing false alarms and identifying real abnormalities more promptly, thus gradually increasing the reliability of infusion monitoring with the number of infusions.

[0062] The valid data includes: patient physiological characteristics, drug parameters, signal data, judgment results, and clinical feedback.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for controlling an injection pump to detect anomalies, characterized in that, The method includes the following steps: A dual-layer benchmark system for individual adaptation and scenario adaptation is constructed. The dual-layer benchmark system includes an individual stress benchmark curve based on the individual physiological characteristics of the patient and a scenario interference tolerance range based on the type of mobile scenario. Real-time acquisition of dynamic signals of tubing pressure of the infusion pump, patient limb movement signals and scene environmental interference signals, and simultaneous acquisition of the characteristic parameters of the currently infused drug; A dynamic baseline curve adapted to the current infused drug is obtained by correcting the individual pressure baseline curve based on drug characteristic parameters. The pressure difference between the pipeline pressure dynamic signal and the dynamic reference curve is calculated to obtain the out-of-range frequency value. The out-of-range frequency value is compared with the reference threshold and interference verification is performed to obtain the deviation judgment result. The interference verification result is obtained by combining the scene environment interference signal and the scene interference tolerance range. The deviation judgment result and the interference verification result are combined to obtain the infusion anomaly judgment result. If the infusion anomaly judgment result is a real anomaly, an alarm and infusion adjustment action are triggered. If the infusion anomaly judgment result is a false anomaly caused by interference, data is collected and continuously monitored. If the infusion anomaly judgment result is normal, the dual-layer reference system is dynamically optimized.

2. The method for controlling an injection pump to detect anomalies according to claim 1, characterized in that, Constructing a two-tiered benchmark system for individual adaptation and scenario adaptation includes the following steps: Collect individual physiological characteristic data of patients and combine them with the physicochemical properties of infused drugs to generate an initial individual pressure baseline curve; The mobile scenario types are divided, and interference signal samples during normal infusion under each mobile scenario type are collected. The interference signal samples of each scenario are statistically judged, and the range of pressure fluctuation under different mobile scenario types is determined to obtain the scenario interference tolerance range. The scenario interference tolerance range includes the interference intensity threshold and the interference duration threshold. By linking the individual stress baseline curve with the interference tolerance range of each scenario, a two-layer baseline system of individual adaptation and scenario adaptation is formed.

3. The method for controlling an injection pump to detect anomalies according to claim 2, characterized in that, Real-time acquisition of dynamic signals from the infusion pump tubing pressure, patient limb movement, and environmental interference signals includes the following steps: Pressure data from different cross sections of the pipeline are collected, and the pressure data are fused to generate a dynamic pipeline pressure signal. Data on changes in angle, range of motion, and acceleration of movement of the infused limb are collected and integrated into a signal of the patient's limb movement. Vibration frequency parameters, noise intensity parameters, temperature and humidity parameters of the scene environment are collected, and key interference parameters related to pressure fluctuations are screened to form scene environment interference signals.

4. The method for controlling an injection pump to detect anomalies according to claim 3, characterized in that, The dynamic baseline curve adapted to the current infusion drug is obtained by correcting the individual pressure baseline curve based on drug characteristic parameters. The specific steps include: Establish a comparison table of drug characteristics and pressure correction coefficients, and match the corresponding correction coefficients in the comparison table according to the characteristic parameters of the currently infused drug. The initial individual pressure baseline curve is adjusted using a linear correction algorithm. The correction formula is as follows: , where P net For the dynamic baseline curve, P init This is the initial individual pressure baseline curve, where K is the correction factor. P represents the baseline pressure offset of the drug; If the drug concentration changes during infusion, a dynamic baseline curve is obtained by matching and correcting the coefficient in real time.

5. The method for controlling an injection pump to detect anomalies according to claim 4, characterized in that, The out-of-range frequency value is obtained by calculating the pressure difference between the pipeline pressure dynamic signal and the dynamic reference curve. The deviation judgment result is obtained by comparing the out-of-range frequency value with the reference threshold and performing interference verification. The specific steps include: Calculate the pressure difference between the pipeline pressure dynamic signal and the dynamic reference curve at the same time node, and collect the number of times the pressure difference exceeds the fluctuation range of the dynamic reference curve within at least two sampling periods to obtain the out-of-range frequency value. A preset baseline threshold is used to compare the out-of-range frequency value with the baseline threshold. If the out-of-range frequency value is greater than or equal to the baseline threshold, it is marked as a preliminary anomaly. If the out-of-range frequency value is less than the baseline threshold, it is judged as a false anomaly. Extract key interference parameters from the scene environment interference signals in the preliminary anomaly and compare them with the scene interference tolerance range corresponding to the current scene; If the key interference parameters do not exceed the scene interference tolerance range, the interference is determined to be invalid, and the initial anomaly remains a valid anomaly. If the key interference parameters exceed the scene interference tolerance range, the temporal correlation value between the pipeline pressure dynamic signal and the patient's limb movement signal is determined. If the temporal correlation value is greater than or equal to the preset threshold, the interference is determined to be effective, and the initial abnormality is corrected to a false abnormality; if the temporal correlation value is less than the preset threshold, it is determined to be a valid abnormality.

6. The method for controlling an injection pump to detect anomalies according to claim 5, characterized in that, Determining the temporal correlation between dynamic signals of tubing pressure and signals of patient limb movement involves the following steps: Acquire dynamic signal segments of tubing pressure and corresponding signal segments of patient limb movement during the initial abnormality period; The correlation coefficient between dynamic signal segments of pipeline pressure and signal segments of patient limb movement is calculated to obtain the time-series correlation value; If the temporal correlation value is less than the correlation threshold, it is determined to be invalid interference; If the temporal correlation value is greater than or equal to the correlation threshold, it is determined to be a valid interference.

7. The method for controlling an injection pump to detect anomalies according to claim 6, characterized in that, If the infusion anomaly assessment result is a genuine anomaly, an alarm and infusion adjustment actions will be triggered, specifically including the following steps: Based on the magnitude and duration of pressure deviation, it is classified into mild abnormality, moderate abnormality, and severe abnormality; In case of mild abnormality, a low-intensity audible and visual alarm is triggered, the infusion pump is controlled to maintain the current infusion rate, and pressure changes are continuously monitored; When a moderate abnormality is detected, a medium-intensity audible and visual alarm is triggered, sending alarm information to the medical terminal and controlling the infusion pump to reduce the current infusion rate or suspend the infusion. In case of severe abnormality, a high-intensity audible and visual alarm is triggered, the infusion is immediately stopped and the piston is locked, an emergency alarm message is sent to the medical terminal, and the complete data chain at the time of the abnormality is collected for traceability and judgment.

8. The method for controlling an injection pump to detect anomalies according to claim 7, characterized in that, The method also includes dynamic optimization of a dual-benchmark system of individual adaptation and scene adaptation, specifically including the following steps: Collect all currently valid data after the current infusion is completed; The individual pressure baseline curve is calibrated based on the current valid data. If the average deviation between the actual pressure data and the baseline curve exceeds the preset threshold, the curve fitting parameters are adjusted. Based on mobile scenario data, new data is obtained by filtering the current valid data, and the mobile scenario data is then optimized using the new data to obtain new mobile scenario data. Collect clinical abnormal misjudgment cases, and adjust the preset threshold of scene interference tolerance range and correlation judgment threshold through new mobile scene data.

9. The method for controlling an injection pump to detect anomalies according to claim 8, characterized in that, The valid data includes: patient physiological characteristics, drug parameters, signal data, judgment results, and clinical feedback.

Images