Intelligent CRRT pipeline integrated with multi-parameter monitoring sensor

Abstract

The invention is applicable to the technical field of CRRT pipelines, and provides an intelligent CRRT pipeline integrated with a multi-parameter monitoring sensor, which comprises five modules, namely an artery end, a vein end, a filter unit, a waste liquid end and a displacement liquid / dialysate passage, which are connected through quick connectors; sensor arrays are integrated in the artery end, the vein end, the filter unit, the waste liquid end and the replacement liquid / dialysate access module. The CRRT pipeline can continuously and synchronously collect key physiological and treatment parameter data such as pressure, temperature, bubble existence, blood leakage, solution concentration, blood flow velocity / hematocrit / hemoglobin in the pipeline; and real-time feedback is realized through a display terminal or a remote system. According to the technical scheme, whole-process in-situ, real-time and multi-parameter integrated monitoring is achieved, the intelligent and automatic level is improved, a foundation is laid for closed-loop control, remote monitoring and big data analysis, and the burden of medical staff is relieved. According to the pipeline, the safety, the effectiveness and the intelligent degree of CRRT treatment are improved.

Description

Technical Field

[0001] This invention belongs to the field of CRRT pipeline technology, and particularly relates to an intelligent CRRT pipeline integrating multi-parameter monitoring sensors. Background Technology

[0002] Continuous Renal Replacement Therapy (CRRT), as an important blood purification technology, is widely used in clinical treatment for critically ill patients with acute kidney injury, multiple organ dysfunction syndrome, etc. It plays a key role in maintaining the stability of the patient's internal environment and removing metabolic waste and excess water from the body.

[0003] With the continuous development of medical technology, higher demands are being placed on the intelligence and automation of CRRT treatment. Traditional external sensor monitoring methods cannot meet the needs of real-time feedback and remote monitoring, making it difficult to achieve closed-loop control of treatment and limiting the further development and application of CRRT technology. For example, in telemedicine scenarios, medical staff cannot obtain real-time monitoring data from patients in a timely manner, making it difficult to provide timely guidance and intervention. Therefore, developing an intelligent CRRT circuit capable of achieving in-situ, real-time, and multi-parameter integrated monitoring throughout the entire process to improve monitoring efficiency and accuracy, reduce operational complexity, and enhance intelligence and automation has become a key issue that urgently needs to be addressed in the current CRRT treatment field. To solve the above problems, it is essential to design an intelligent CRRT circuit integrating multi-parameter monitoring sensors. Summary of the Invention

[0004] This invention provides an intelligent CRRT pipeline integrating multi-parameter monitoring sensors, aiming to solve the problems mentioned in the background art.

[0005] The present invention is implemented as follows: an intelligent CRRT tubing integrating multi-parameter monitoring sensors, characterized in that the main body of the tubing is made of medical-grade PVC or silicone material, including five modules: arterial end, venous end, filter unit, waste fluid end, and replacement fluid / dialysis fluid passage, and each module is connected by quick connectors;

[0006] The filter unit includes a filter connected to the arterial end via a blood pump. The arterial end is provided with citrate and heparin lines. A venous reservoir and calcium line are provided between the filter and the venous end. The waste outlet of the filter is connected to the waste end. The venous reservoir is connected to the venous end line of the filter. Blood flows into the reservoir from the inlet. The venous reservoir returns the blood to the patient through a venous puncture needle.

[0007] Sensor arrays are integrated in the arterial end, venous end, filter unit, waste fluid end, and replacement fluid / dialysis fluid pathway module. A main control unit is integrated on the main body of the pipeline. The sensor array is connected to the main control unit through built-in circuitry or wireless means to form a real-time monitoring network. The main control unit integrates a data processing module and an analysis algorithm module.

[0008] Preferably, the intravenous vessel further includes a sampling port, a medication inlet, and an exhaust port. The sampling port is used to draw blood samples to test coagulation function or electrolytes. The medication inlet is used to inject anticoagulants, drugs, or supplemental fluids. The exhaust port is used to remove air from the vessel before treatment to prevent air embolism.

[0009] Preferably, the sensor array includes a pressure sensor and a hematocrit sensor disposed at the arterial end. The pressure sensor is used to monitor the inlet pump pressure. Temperature sensors are disposed before and after the filter (1). Both the front and rear temperature monitoring are monitored by dual temperature sensors, one for control and one for monitoring. The temperature sensors are used to monitor the blood preheating effect. The venous reservoir (3) has a liquid level monitoring sensor to monitor the liquid level in real time. There is a bubble sensor after the liquid level reservoir (3). A flow sensor is located in the citrate pipeline at the arterial end to monitor the citrate flow rate.

[0010] A pressure sensor and a blood leakage sensor are installed after the filter (1), the pressure sensor being used to monitor transmembrane pressure TMP; a pressure sensor and a hemoglobin sensor are installed at the venous end, the pressure sensor being used to monitor blood return pressure; a flow sensor and a concentration sensor are installed in the replacement fluid / dialysis fluid pathway;

[0011] The main control unit is integrated into the CRRT device host and communicates with the sensor array through a flexible circuit board or wireless low-power protocol, supporting real-time data processing and algorithm operation.

[0012] Preferably, the sensor surface is coated with a heparinized polyurethane coating.

[0013] Preferably, the analysis algorithm module specifically includes the following steps:

[0014] Step 1: Real-time transmembrane pressure (TMP) data within the pipeline is collected using multi-parameter monitoring sensors integrated into the intelligent CRRT pipeline, and the collected data is transmitted to the main control unit.

[0015] Step 2: Set a sliding window in the analysis algorithm module of the main control unit. The window length is set to 10 seconds. The sliding window slides along the time axis at fixed time intervals. Each time it slides, the window contains the latest collected TMP data within the last 10 seconds.

[0016] Step 3: Set a pressure fluctuation alarm threshold. When the transmembrane pressure (TMP) is greater than 300 mmHg, it is determined to be an abnormal pressure fluctuation.

[0017] Step 4: Within each sliding window, perform statistical analysis on the TMP data within the window to determine whether there are any TMP values ​​exceeding the preset threshold (300 mmHg);

[0018] Step 5: If the TMP value exceeds the preset threshold within the sliding window, the main control unit triggers an alarm signal, which is presented through sound, light, or by sending a notification to relevant terminal devices.

[0019] Step 6: If the pressure drop value (250 mmHg) before and after the filter is detected to exceed the preset threshold within the sliding window, the main control unit will trigger an alarm signal. The alarm signal will be presented through sound, light or by sending a notification to the relevant terminal device.

[0020] Step 7: If the error between the control temperature and the monitoring temperature for blood preheating exceeds 3°C within the sliding window, the main control unit will trigger an alarm signal. The alarm signal will be presented through sound, light, or by sending a notification to relevant terminal devices.

[0021] Preferably, in step four, the method for statistically analyzing the TMP data within the sliding window is as follows:

[0022] Traverse all TMP data points within the sliding window; record the TMP value of each data point and compare it with a preset threshold (300 mmHg); if the TMP value of at least one data point exceeds the preset threshold, it is determined that there is an abnormal pressure fluctuation within the sliding window.

[0023] Preferably, the method further includes the following steps:

[0024] Alarm Log: When an alarm signal is triggered, the main control unit records the alarm time, the TMP value at the time of the alarm, and the relevant TMP data in the sliding window into the storage module for subsequent query and analysis.

[0025] Alarm cancellation judgment: After the alarm signal is triggered, TMP data continues to be collected in real time and analyzed using a sliding window. When no TMP value exceeding the preset threshold is detected in multiple consecutive sliding windows (such as three consecutive sliding windows), the alarm signal is cancelled.

[0026] Preferably, the specific method for determining whether to clear the alarm signal in the alarm clearance determination is as follows:

[0027] After the alarm signal is triggered, the TMP data continues to be analyzed in real time using a new sliding window;

[0028] If all TMP data points in each of the three consecutive sliding windows do not exceed the preset threshold (300 mmHg), the pressure fluctuation is determined to have returned to normal, and the main control unit will deactivate the alarm signal.

[0029] If the TMP value exceeds the preset threshold again within any sliding window before the alarm signal is cleared, the counting will restart until no TMP value exceeds the preset threshold is detected within three consecutive sliding windows.

[0030] Preferably, the analysis algorithm module further includes a coagulation prediction model:

[0031] Data acquisition and preprocessing: The multi-parameter monitoring sensor integrated into the intelligent CRRT pipeline is used to collect the pressure difference ΔP data before and after the filter in real time and record the corresponding continuous treatment time information; the collected ΔP data is denoised to remove outliers and interference signals, and the treatment time data is standardized to meet the model input requirements.

[0032] Constructing an LSTM neural network model: Build an LSTM neural network with a multi-layer structure, setting the number of input layer nodes to correspond to the dimensions of the processed ΔP data and treatment time data, the hidden layer containing multiple LSTM units to capture the time-series features of the data, and the output layer having 1 node to output the predicted value of filter lifespan; determine the training parameters of the model, including the learning rate, number of training epochs, and batch size;

[0033] Model training: The preprocessed historical ΔP data and treatment time data are used as training sample sets and divided into training set and validation set according to a certain ratio; the training set data is input into the LSTM neural network model, the model calculates the predicted value based on the input data, and calculates the prediction error by comparing it with the real filter life data and using the loss function.

[0034] The backpropagation algorithm is used to adjust the network parameters of the model based on the prediction error;

[0035] The training process is iterated until the model's prediction accuracy on the validation set reaches ≥90%, at which point the model training is complete.

[0036] Real-time prediction: During the actual operation of the intelligent CRRT pipeline, the current ΔP data and treatment time data are continuously collected, and the same preprocessing operations as in the training phase are performed; the preprocessed real-time data is input into the trained LSTM neural network model, and the model outputs the filter life prediction value at the current moment.

[0037] Preferably, it also includes a display terminal, which is integrated into the touch screen of the CRRT device to display parameter curves and alarm information in real time. The display terminal data is uploaded to the hospital's HIS system via a 4G / 5G module and can be viewed on mobile devices.

[0038] Compared with related technologies, the intelligent CRRT pipeline with integrated multi-parameter monitoring sensors provided by this invention has the following beneficial effects:

[0039] The intelligent CRRT tubing of this application integrates a variety of miniaturized, high-precision sensors, which can collect key physiological and therapeutic parameters such as pressure, temperature, air bubble presence, blood leakage, solution concentration, blood flow rate, hematocrit, and hemoglobin in the tubing in real time, continuously, and synchronously. This provides medical staff with comprehensive and accurate treatment information, which helps to more accurately judge the condition and adjust the treatment plan.

[0040] By monitoring and accurately analyzing key parameters such as transmembrane pressure in real time, alarm signals can be triggered promptly when pressure fluctuations are abnormal, enabling medical staff to take swift action to prevent the condition from worsening and further improve treatment safety. Real-time, accurate monitoring data helps medical staff adjust treatment parameters, such as blood pump flow rate and replacement / dialysis fluid flow rate, according to the patient's actual condition, ensuring that CRRT treatment can continuously and stably achieve optimal results, improving the efficiency of removing metabolic waste and excess water from the patient's body, and improving the patient's internal environment stability.

[0041] The main control unit integrates data processing and analysis algorithms, enabling automatic processing and analysis of collected data. This includes functions such as pressure fluctuation alarms, alarm recording, and deactivation, reducing errors and workload associated with manual operation and improving the intelligence and automation of treatment. It supports real-time data feedback and remote monitoring, allowing medical staff to access patient treatment data anytime via relevant terminal devices for remote guidance and intervention. This lays the foundation for closed-loop control and big data analysis, further promoting the development of CRRT treatment technology.

[0042] This application's intelligent CRRT tubing seamlessly integrates multiple miniaturized, high-precision sensors in situ at key nodes of the CRRT system, such as the arterial end, venous end, before and after the filter, waste fluid end, and replacement fluid / dialysis fluid inlet points. These sensors connect to the main control unit via built-in circuitry or wirelessly, forming a real-time monitoring network. The main control unit integrates data processing and analysis algorithms, continuously and synchronously collecting key physiological and therapeutic parameters within the tubing, such as pressure, temperature, air bubble presence, blood leakage, solution concentration, blood flow rate / hematocrit / hemoglobin, and providing real-time feedback through a display terminal or remote system. This technical solution achieves in-situ, real-time, multi-parameter integrated monitoring throughout the entire process, significantly improving monitoring efficiency and accuracy, avoiding the delays, errors, and operational complexity associated with traditional external sensors, enhancing intelligence and automation, laying the foundation for closed-loop control, remote monitoring, and big data analysis, and reducing the burden on medical staff. The tubing of this invention significantly improves the safety, effectiveness, and intelligence of CRRT treatment. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of the structure of an intelligent CRRT pipeline integrating multi-parameter monitoring sensors according to the present invention;

[0044] Figure 2 This is a schematic diagram of the operation of an intelligent CRRT pipeline integrating multi-parameter monitoring sensors according to the present invention.

[0045] In the diagram: 1. Filter; 2. Blood pump; 3. Vein chamber. Detailed Implementation

[0046] A preferred embodiment of the intelligent CRRT pipeline with integrated multi-parameter monitoring sensors provided by the present invention is as follows: Figures 1 to 2 As shown:

[0047] A smart CRRT tubing integrating multi-parameter monitoring sensors is disclosed. The main body of the tubing is made of medical-grade PVC or silicone material and includes five modules: arterial end, venous end, filter unit, waste fluid end, and replacement fluid / dialysis fluid pathway. Each module is connected via quick connectors. The filter unit includes a filter 1, which is connected to the arterial end via a blood pump 2. A venous reservoir 3 and a calcium supply line 4 are provided between the filter 1 and the venous end. The waste fluid port of the filter 1 is connected to the waste fluid end. The venous reservoir 3 is connected to the venous end tubing of the filter 1. Blood flows into the reservoir 3 from its inlet and is returned to the patient via a venous puncture needle. Sensor arrays are integrated in the arterial end, venous end, filter unit, and replacement fluid / dialysis fluid pathway module. A main control unit is integrated on the main body of the tubing. The sensor arrays are connected to the main control unit via built-in circuitry or wirelessly to form a real-time monitoring network. The main control unit integrates a data processing module and an analysis algorithm module. The venous chamber 3 also includes a sampling port, a drug delivery port, and an exhaust port. The sampling port is used to draw blood samples to test coagulation function or electrolytes, the drug delivery port is used to inject anticoagulants, drugs, or supplemental fluids, and the exhaust port is used to remove air from the chamber before treatment to prevent air embolism.

[0048] In the filter unit, filter 1 is connected to the arterial end via blood pump 2. Blood from the arterial end is pumped into filter 1 via blood pump 2 for purification. A venous reservoir 3 is installed between filter 1 and the venous end. The purified blood flows into the venous reservoir 3 and is then reinfused into the patient through a venous puncture needle. The venous reservoir 3 is equipped with a sampling port, a medication inlet, and an vent. The sampling port is used to draw blood samples to test coagulation function or electrolytes; the medication inlet can inject anticoagulants, medications, or replacement fluids; and the vent is used to remove air from the reservoir before treatment to prevent air embolism.

[0049] Sensor arrays are integrated in the arterial end, venous end, filter unit, and replacement fluid / dialysis fluid pathway module. The arterial end is equipped with a pressure sensor and a hematocrit sensor to monitor the inlet pump pressure and hematocrit, respectively. A temperature sensor and a bubble sensor are installed before filter 1 to detect the blood preheating effect and the presence of bubbles. A pressure sensor and a leakage sensor are installed after filter 1 to monitor the transmembrane pressure (TMP) and whether leakage has occurred. The venous end is equipped with a pressure sensor and a hemoglobin sensor to monitor the return blood pressure and hemoglobin content. The replacement fluid / dialysis fluid pathway is equipped with a flow sensor and a concentration sensor to monitor the flow rate and solution concentration.

[0050] The sensor array includes a pressure sensor and a hematocrit sensor located at the arterial end, with the pressure sensor used to monitor the inlet pump pressure; a temperature sensor and a bubble sensor located before filter 1, with the temperature sensor used to detect and monitor the blood preheating effect; a pressure sensor and a blood leakage sensor located after filter 1, with the pressure sensor used to monitor the transmembrane pressure (TMP); a pressure sensor and a hemoglobin sensor located at the venous end, with the pressure sensor used to monitor the blood return pressure; and a flow sensor and a concentration sensor located in the replacement fluid / dialysis fluid pathway. The main control unit is integrated into the CRRT device host and communicates with the sensor array via a flexible circuit board or a wireless low-power protocol, supporting real-time data processing and algorithm execution.

[0051] The sensor surface is coated with a heparinized polyurethane. During CRRT treatment, when blood comes into contact with the sensor surface, platelets easily adhere to the sensor surface and aggregate, potentially forming thrombi. Heparin in the heparinized polyurethane heparinized coating has strong anticoagulant activity; it can bind to antithrombin III, enhancing the inhibitory effect of antithrombin III on coagulation factors, thereby effectively inhibiting platelet adhesion and aggregation, and significantly reducing the risk of thrombosis.

[0052] The analysis algorithm module specifically includes the following steps:

[0053] Step 1: Real-time transmembrane pressure (TMP) data within the pipeline is collected using multi-parameter monitoring sensors integrated into the intelligent CRRT pipeline, and the collected data is transmitted to the main control unit.

[0054] Step 2: Set a sliding window in the analysis algorithm module of the main control unit. The window length is set to 10 seconds. The sliding window slides along the time axis at fixed time intervals. Each time it slides, the window contains the latest collected TMP data within the last 10 seconds.

[0055] Step 3: Set a pressure fluctuation alarm threshold. When the transmembrane pressure (TMP) is greater than 300 mmHg, it is determined to be an abnormal pressure fluctuation.

[0056] Step 4: Within each sliding window, perform statistical analysis on the TMP data within the window to determine whether there are any TMP values ​​exceeding the preset threshold (300 mmHg);

[0057] Step 5: If the TMP value exceeds the preset threshold within the sliding window, the main control unit triggers an alarm signal, which is presented through sound, light, or by sending a notification to relevant terminal devices.

[0058] In step four, the method for statistical analysis of the TMP data within the sliding window is as follows:

[0059] Traverse all TMP data points within the sliding window; record the TMP value of each data point and compare it with a preset threshold (300 mmHg); if the TMP value of at least one data point exceeds the preset threshold, it is determined that there is an abnormal pressure fluctuation within the sliding window.

[0060] It also includes the following steps:

[0061] Alarm Log: When an alarm signal is triggered, the main control unit records the alarm time, the TMP value at the time of the alarm, and the relevant TMP data in the sliding window into the storage module for subsequent query and analysis.

[0062] Alarm cancellation judgment: After the alarm signal is triggered, TMP data continues to be collected in real time and analyzed using a sliding window. When no TMP value exceeding the preset threshold is detected in multiple consecutive sliding windows (such as three consecutive sliding windows), the alarm signal is cancelled.

[0063] The specific method for determining whether to clear an alarm signal during alarm clearance is as follows:

[0064] After the alarm signal is triggered, the TMP data continues to be analyzed in real time using a new sliding window;

[0065] If all TMP data points in each of the three consecutive sliding windows do not exceed the preset threshold (300 mmHg), the pressure fluctuation is determined to have returned to normal, and the main control unit will deactivate the alarm signal.

[0066] If the TMP value exceeds the preset threshold again within any sliding window before the alarm signal is cleared, the counting will restart until no TMP value exceeds the preset threshold is detected within three consecutive sliding windows.

[0067] The intelligent CRRT tubing of this application integrates a variety of miniaturized, high-precision sensors, which can collect key physiological and therapeutic parameters such as pressure, temperature, air bubble presence, blood leakage, solution concentration, blood flow rate, hematocrit, and hemoglobin in the tubing in real time, continuously, and synchronously. This provides medical staff with comprehensive and accurate treatment information, which helps to more accurately judge the condition and adjust the treatment plan.

[0068] By real-time monitoring and precise analysis of key parameters such as transmembrane pressure and filter pressure drop, alarm signals can be triggered promptly when pressure fluctuations are abnormal. This allows medical staff to take swift action to prevent the condition from worsening and further improve treatment safety. Real-time and accurate monitoring data helps medical staff adjust treatment parameters, such as blood pump flow rate and replacement / dialysis fluid flow rate, according to the patient's actual condition. This ensures that CRRT treatment can continuously and stably achieve optimal results, improve the efficiency of removing metabolic waste and excess water from the patient's body, and improve the patient's internal environment stability.

[0069] The main control unit integrates data processing and analysis algorithms, enabling automatic processing and analysis of collected data. This includes functions such as pressure fluctuation alarms, alarm recording, and deactivation, reducing errors and workload associated with manual operation and improving the intelligence and automation of treatment. It supports real-time data feedback and remote monitoring, allowing medical staff to access patient treatment data anytime via relevant terminal devices for remote guidance and intervention. This lays the foundation for closed-loop control and big data analysis, further promoting the development of CRRT treatment technology.

[0070] The various modules of the tubing are connected using quick-connect couplings, facilitating assembly and disassembly, reducing operation time and difficulty, and improving the work efficiency of medical staff. The intravenous reservoir is equipped with sampling ports, drug delivery ports, and venting ports to meet various needs during treatment, making operation more convenient and safer.

[0071] The analysis algorithm module also includes a coagulation prediction model:

[0072] Data acquisition and preprocessing: The multi-parameter monitoring sensor integrated into the intelligent CRRT pipeline is used to collect the pressure difference ΔP data before and after the filter in real time and record the corresponding continuous treatment time information; the collected ΔP data is denoised to remove outliers and interference signals, and the treatment time data is standardized to meet the model input requirements.

[0073] Constructing an LSTM neural network model: Build an LSTM neural network with a multi-layer structure, setting the number of input layer nodes to correspond to the dimensions of the processed ΔP data and treatment time data, the hidden layer containing multiple LSTM units to capture the time-series features of the data, and the output layer having 1 node to output the predicted value of filter lifespan; determine the training parameters of the model, including the learning rate, number of training epochs, and batch size;

[0074] Model Training: Preprocessed historical ΔP data and treatment time data are used as the training sample set, divided into a training set and a validation set according to a certain ratio. The training set data is input into the LSTM neural network model. The model calculates the predicted value based on the input data and calculates the prediction error by comparing it with the actual filter life data using a loss function. The loss function includes:

[0075]

[0076] in, Indicates the prediction error. This represents the total number of samples in the training set. Indicates the first training set The cumulative treatment time for each sample, where max represents the maximum value function. This indicates treatment time data. This represents the maximum cumulative treatment time across all samples. Indicates the first training set Predicted filter lifetime for a sample. Indicates the first training set The actual filter lifespan data corresponding to each sample. Indicates the lifespan warning threshold. Indicates an indicator function.

[0077] It should be explained that in this application, This indicates an indicator function, with a value range of 0 or 1, representing the predicted filter life. Greater than the lifespan warning threshold hour, The value is 1; when the filter life prediction value is... Less than the lifespan warning threshold hour, The value is 0.

[0078] For example, suppose there are 5 sets of samples in the training set, as shown in the table below:

[0079] sample Actual lifespan (hours) Predicted lifespan (hours) Total treatment time (hours) 1 60 58 30 2 85 82 40 3 90 95 50 4 75 70 20 5 100 105 45

[0080] The lifespan warning threshold is 80 hours. Substituting the five sets of data from the table above into the formula yields:

[0081]

[0082] Using the backpropagation algorithm, the network parameters of the model are adjusted according to the prediction error using the following formula:

[0083]

[0084] in, Indicates the first Network parameters trained in rounds, Indicates the initial network parameters. Indicates An exponential function with base 0. Indicates the attenuation coefficient. Indicates the first Prediction error during rounds of training Indicates the first The prediction error of the training round.

[0085] It needs to be explained that in this application, This indicates the number of training epochs in the training parameters, which can be set to 10 epochs. This represents the attenuation coefficient, which can be set to 0.1 to effectively prevent oscillations caused by sudden changes in network parameters.

[0086] For example, assuming the initial network parameters are 0.001, the training epochs are 3, and the prediction errors are 1.5, 1.3, 1.35, and 1.25 respectively, with a decay coefficient of 0.1, substituting these values ​​into the formula, the network parameters for the first training epoch are: The network parameters for the second round of training are: The network parameters for the third round of training are: .

[0087] The training process is iterated until the model's prediction accuracy on the validation set reaches ≥90%, at which point the model training is complete. In this application, the prediction accuracy is calculated using the following formula:

[0088]

[0089] in, Indicates prediction accuracy. This represents the total number of samples in the validation set. Indicates An exponential function with base 0. Indicates the first in the verification set Predicted filter lifetime for a sample. Indicates the first in the verification set The actual filter lifespan data corresponding to each sample. This indicates the maximum filter life in the verification set. Indicates an indicator function, This indicates the allowable error threshold.

[0090] It needs to be explained that in this application, This indicates the allowable error threshold, which can be set to 2 hours. This indicates an indicator function, with a value range of 0 or 1, representing the predicted filter life. and actual filter life data The error between them is no greater than the allowable error threshold. hour, The value is 1; when the filter life prediction value is... and actual filter life data The error between them is greater than the allowable error threshold. hour, The value is 0.

[0091] For example, assuming the validation set has 3 samples, where the actual filter lifetimes are 70 hours, 85 hours, and 90 hours, and the predicted filter lifetimes are 72 hours, 83 hours, and 95 hours, with an allowable error threshold of 2 hours, then the prediction accuracy is:

[0092]

[0093] Real-time prediction: During the actual operation of the intelligent CRRT pipeline, the current ΔP data and treatment time data are continuously collected, and the same preprocessing operations as in the training phase are performed; the preprocessed real-time data is input into the trained LSTM neural network model, and the model outputs the filter life prediction value at the current moment.

[0094] It also includes a display terminal, which is integrated into the touch screen of the CRRT device. It displays parameter curves and alarm information in real time, and uploads the display terminal data to the hospital's HIS system via a 4G / 5G module, supporting viewing on mobile devices.

[0095] The sensor signal lines use a twisted pair shielding layer, and the main control unit integrates digital filtering algorithms (such as Kalman filtering) to eliminate electromagnetic interference.

[0096] Pressure sensor selection: MEMS piezoresistive sensor (range -500~500mmHg, accuracy ±1mmHg), encapsulated on the inner wall of the pipeline, with blood isolated by a silicone membrane. It also integrates a temperature compensation circuit to eliminate the influence of blood temperature changes on pressure measurement. Blood leakage sensor: Employs an 850nm infrared LED and dual photodiodes (PDs) to identify blood leakage by detecting changes in the intensity of transmitted light (sensitivity 0.03ml / min). Bubble sensor emits 1MHz ultrasound and identifies bubbles by detecting the echo time difference. Emergency blood return is triggered when a single bubble diameter ≥0.1mm or a continuous microbubble rate is limited to 0.7mL / min.

[0097] Hematocrit / hemoglobin sensor: Integrates 530nm (hemoglobin absorption peak) and 850nm (reference wavelength) LEDs to calculate HCT (range 20%~60%) and Hb (range 5~20g / dL) through absorbance ratio. Automatically calibrated with standard solution before each treatment to eliminate individual differences.

[0098] By predicting filter lifespan in real time using a coagulation prediction model, an early warning can be issued before the filter approaches a coagulation state, giving medical staff enough time to take appropriate measures, such as adjusting the anticoagulant dosage or replacing the filter. This effectively avoids treatment interruptions and complications caused by filter coagulation, thus ensuring the safety of patients' lives.

[0099] Traditional coagulation monitoring methods often have a certain lag, usually only detecting coagulation after it has already occurred and obvious symptoms have appeared. This coagulation prediction model, however, is based on real-time collected data of the pressure difference ΔP before and after the filter, as well as continuous treatment time information. It can detect filter coagulation trends in advance, allowing for proactive measures and reducing unexpected situations during treatment, thus improving the stability and reliability of the treatment.

[0100] It is worth noting that the circuits, electronic components, and modules involved in this invention are all existing technologies, which can be fully implemented by those skilled in the art, and need not be elaborated upon. The content protected by this invention does not involve improvements to the software and methods.

[0101] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the above-mentioned units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be in the form of telecommunications or other forms.

[0102] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the scope of protection of the invention. Obviously, the embodiments described are only some embodiments of the present invention, rather than all embodiments. Based on these embodiments, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in this field can still combine, add, delete or make other adjustments to the features in the various embodiments of the present invention according to the circumstances without conflict, without making creative work, so as to obtain different other technical solutions that do not deviate from the concept of the present invention in essence, and these technical solutions also fall within the scope of protection of the present invention.

Claims

1. A smart CRRT pipeline integrating multi-parameter monitoring sensors, characterized in that, The main body of the tubing is made of medical-grade PVC or silicone material, and includes five modules: arterial end, venous end, filter unit, waste fluid end, and replacement fluid / dialysis fluid access. Each module is connected by quick connectors. The filter unit includes a filter (1), which is connected to the arterial end via a blood pump (2). The arterial end is provided with a citrate line and a heparin line. A venous reservoir (3) and a calcium line (4) are provided between the filter (1) and the venous end. The waste outlet of the filter (1) is connected to the waste outlet. The venous reservoir (3) is connected to the venous end line of the filter (1). Blood flows into the reservoir from the inlet of the venous reservoir (3). The venous reservoir (3) returns the blood to the patient through a venous puncture needle. Sensor arrays are integrated in the arterial end, venous end, waste fluid end, filter unit, and replacement fluid / dialysis fluid pathway module. A main control unit is integrated on the main body of the pipeline. The sensor array is connected to the main control unit through built-in circuitry or wireless means to form a real-time monitoring network. The main control unit integrates a data processing module and an analysis algorithm module.

2. The intelligent CRRT pipeline with integrated multi-parameter monitoring sensors as described in claim 1, characterized in that, The venous chamber (3) also includes a sampling port, a drug injection port and an exhaust port. The sampling port is used to draw blood samples to test coagulation function or electrolytes. The drug injection port is used to inject anticoagulants, drugs or supplements. The exhaust port is used to remove air from the chamber before treatment to prevent air embolism.

3. The intelligent CRRT pipeline with integrated multi-parameter monitoring sensors as described in claim 1, characterized in that, The sensor array includes a pressure sensor and a hematocrit sensor located at the arterial end. The pressure sensor is used to monitor the inlet pump pressure. Temperature sensors are located before and after the filter (1). Both the front and rear temperature monitoring are monitored by dual temperature sensors, one for control and one for monitoring. The temperature sensors are used to monitor the blood preheating effect. The venous reservoir (3) has a liquid level monitoring sensor to monitor the liquid level in real time. There is a bubble sensor after the liquid level reservoir (3). A flow sensor is located in the citrate pipeline at the arterial end to monitor the citrate flow rate. A pressure sensor and a blood leakage sensor are installed after the filter (1), the pressure sensor being used to monitor transmembrane pressure TMP; A pressure sensor and a hemoglobin sensor are installed at the venous end, the pressure sensor being used to monitor the blood return pressure; a flow sensor and a concentration sensor are installed in the replacement fluid / dialysis fluid pathway; The main control unit is integrated into the CRRT device host and communicates with the sensor array through a flexible circuit board or wireless low-power protocol, supporting real-time data processing and algorithm operation.

4. The intelligent CRRT pipeline with integrated multi-parameter monitoring sensors as described in claim 1, characterized in that, The sensor surface is coated with a heparinized polyurethane coating.

5. The intelligent CRRT pipeline with integrated multi-parameter monitoring sensors as described in claim 1, characterized in that, The analysis algorithm module specifically includes the following steps: Step 1: By using multi-parameter monitoring sensors integrated into the intelligent CRRT pipeline, the transmembrane pressure (TMP) and filter downpressure data in the pipeline are collected in real time, and the collected data are transmitted to the main control unit. Step 2: Set a sliding window in the analysis algorithm module of the main control unit. The window length is set to 10 seconds. The sliding window slides along the time axis at fixed time intervals. Each time it slides, the window contains the latest collected TMP data within the last 10 seconds. Step 3: Set a pressure fluctuation alarm threshold. When the transmembrane pressure (TMP) is greater than 300 mmHg, it is determined to be an abnormal pressure fluctuation. Step 4: Within each sliding window, perform statistical analysis on the TMP data to determine if there are any TMP values ​​exceeding the preset threshold (300 mmHg). Step 5: If the TMP value exceeds the preset threshold within the sliding window, the main control unit triggers an alarm signal, which is presented through sound, light, or by sending a notification to relevant terminal devices. Step 6: If the pressure drop (250 mmHg) before and after the filter is detected to exceed the preset threshold within the sliding window, the main control unit will trigger an alarm signal. The alarm signal will be presented through sound, light, or by sending a notification to the relevant terminal device. Step 7: If the error between the control temperature and the monitoring temperature for blood preheating exceeds 3°C within the sliding window, the main control unit will trigger an alarm signal. The alarm signal will be presented through sound, light, or by sending a notification to relevant terminal devices.

6. The intelligent CRRT pipeline with integrated multi-parameter monitoring sensors as described in claim 5, characterized in that, In step four, the method for statistical analysis of the TMP data within the sliding window is as follows: Traverse all TMP data points within the sliding window; record the TMP value of each data point and compare it with a preset threshold (300 mmHg); if the TMP value of at least one data point exceeds the preset threshold, it is determined that there is an abnormal pressure fluctuation within the sliding window.

7. The intelligent CRRT pipeline with integrated multi-parameter monitoring sensors as described in claim 6, characterized in that, It also includes the following steps: Alarm Log: When an alarm signal is triggered, the main control unit records the alarm time, the TMP value at the time of the alarm, and the relevant TMP data in the sliding window into the storage module for subsequent query and analysis. Alarm cancellation judgment: After the alarm signal is triggered, TMP data continues to be collected in real time and analyzed using a sliding window. When no TMP value exceeding the preset threshold is detected in multiple consecutive sliding windows (such as three consecutive sliding windows), the alarm signal is cancelled.

8. The intelligent CRRT pipeline with integrated multi-parameter monitoring sensors as described in claim 7, characterized in that, The specific method for determining whether to clear the alarm signal in the alarm clearance determination is as follows: After the alarm signal is triggered, the TMP data continues to be analyzed in real time using a new sliding window; If, within three consecutive sliding windows, all TMP data points in each window do not exceed the preset threshold (300 mmHg), the pressure fluctuation is determined to have returned to normal, and the main control unit cancels the alarm signal. If the TMP value exceeds the preset threshold again within any sliding window before the alarm signal is cleared, the counting will restart until no TMP value exceeds the preset threshold is detected within three consecutive sliding windows.

9. The intelligent CRRT pipeline with integrated multi-parameter monitoring sensors as described in claim 1, characterized in that, The analysis algorithm module also includes a coagulation prediction model: Data acquisition and preprocessing: The multi-parameter monitoring sensor integrated into the intelligent CRRT pipeline is used to collect the pressure difference ΔP data before and after the filter in real time and record the corresponding continuous treatment time information; the collected ΔP data is denoised to remove outliers and interference signals, and the treatment time data is standardized to meet the model input requirements. Constructing an LSTM neural network model: Build an LSTM neural network with a multi-layer structure, setting the number of input layer nodes to correspond to the dimensions of the processed ΔP data and treatment time data, the hidden layer containing multiple LSTM units to capture the time-series features of the data, and the output layer having 1 node to output the predicted value of filter lifespan; determine the training parameters of the model, including the learning rate, number of training epochs, and batch size; Model training: The preprocessed historical ΔP data and treatment time data are used as training sample sets and divided into training set and validation set according to a certain ratio; the training set data is input into the LSTM neural network model, the model calculates the predicted value based on the input data, and calculates the prediction error by comparing it with the real filter life data and using the loss function. The backpropagation algorithm is used to adjust the network parameters of the model based on the prediction error; The training process is iterated until the model's prediction accuracy on the validation set reaches ≥90%, at which point the model training is complete. Real-time prediction: During the actual operation of the intelligent CRRT tubing, the current ΔP data and treatment time data are continuously collected, and the same preprocessing operations as in the training phase are performed. The preprocessed real-time data is input into the trained LSTM neural network model, and the model outputs the filter lifetime prediction value at the current moment.

10. The intelligent CRRT pipeline with integrated multi-parameter monitoring sensors as described in claim 1, characterized in that, It also includes a display terminal, which is integrated into the touch screen of the CRRT device. It displays parameter curves and alarm information in real time, and uploads the display terminal data to the hospital's HIS system via a 4G / 5G module, supporting viewing on mobile devices.

Images