Clean ultrafiltration rate intelligent regulation and control device and method for CRRT treatment

Abstract

The invention discloses a net ultrafiltration rate intelligent regulation and control device and method for CRRT treatment, and relates to the technical field of medical equipment control, and the device comprises a sensor module, a data processing module and a regulation and control module; the sensor module is used for acquiring CRRT treatment associated data of a preset patient; the data processing module is used for predicting a blood volume tolerance curve of a preset patient by utilizing an artificial intelligence technology according to historical CRRT treatment data and CRRT treatment associated data of the preset patient, and generating a net ultrafiltration rate time sequence scheme according to the predicted blood volume tolerance curve and a preset treatment target of the preset patient; and the regulation and control module is used for automatically regulating the rotating speed of a blood pump and the rotating speed of a displacement liquid pump which are used for performing CRRT treatment on the preset patient according to the net ultrafiltration rate time sequence scheme, and regulating the dehydration rate of the preset patient. According to the invention, a more accurate intelligent regulation and control mode can be provided for CRRT treatment.

Description

Technical Field

[0001] This invention relates to the field of medical device control technology, and in particular to an intelligent control device and method for net ultrafiltration rate in CRRT treatment. Background Technology

[0002] CRRT (Continuous Renal Replacement Therapy) refers to the continuous filtration of a patient's blood to remove toxins, excess water, and electrolyte imbalances, and is widely used for blood purification in critically ill patients. The net ultrafiltration rate is a core parameter determining the treatment's effectiveness and safety. Currently, the ultrafiltration rate is primarily set based on clinician experience, which is highly subjective and lacks a dynamic adjustment mechanism based on real-time patient indicators. Furthermore, existing equipment has fixed ultrafiltration parameters, making it difficult to adapt to the patient's diurnal physiological fluctuations. During prolonged treatment, the dynamic balance between the net ultrafiltration rate and changes in the patient's blood volume and hemodynamics is easily disrupted, potentially leading to problems such as hypotension or blood volume overload. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to address the shortcomings of the prior art, specifically by providing an intelligent control device and method for net ultrafiltration rate in CRRT treatment, as detailed below:

[0004] 1) In a first aspect, the present invention provides an intelligent control device for net ultrafiltration rate in CRRT treatment, the specific technical solution of which is as follows:

[0005] It includes a sensor module, a data processing module, and a control module;

[0006] The sensor module is used to: acquire CRRT treatment-related data for a preset patient;

[0007] The data processing module is used to: predict the blood volume tolerance curve of the preset patient based on the patient's historical CRRT treatment data and CRRT treatment-related data, and generate a net ultrafiltration rate time series plan based on the predicted blood volume tolerance curve and the preset treatment target of the preset patient.

[0008] The control module is used to automatically adjust the speed of the blood pump and the speed of the replacement fluid pump used for CRRT treatment of preset patients according to the net ultrafiltration rate timing scheme, and to adjust the dehydration rate of preset patients.

[0009] The beneficial effects of the intelligent net ultrafiltration rate control device for CRRT treatment provided by this invention are as follows:

[0010] On the one hand, it effectively reduces the interference of human experience factors. Specifically, in traditional CRRT treatment, the ultrafiltration rate largely depends on the doctor's experience and is highly subjective. This new approach, however, acquires data through sensors, uses artificial intelligence to predict the blood volume tolerance curve, and generates a net ultrafiltration rate time-series plan based on the curve and treatment goals. This makes the net ultrafiltration rate setting more scientific and objective, reducing human error. On the other hand, it enables dynamic adjustment of the net ultrafiltration rate. Specifically, it can monitor the CRRT treatment-related data of a pre-set patient in real time, dynamically optimize the net ultrafiltration rate based on the patient's real-time status, adapt to the patient's diurnal physiological fluctuations, maintain a dynamic balance between the net ultrafiltration rate and changes in the patient's blood volume and hemodynamics, avoid problems such as hypotension or blood volume overload, improve the safety and effectiveness of treatment, and provide a more precise and personalized intelligent control method for CRRT treatment.

[0011] Based on the above scheme, the intelligent control device for net ultrafiltration rate of the present invention for CRRT treatment can be further improved as follows.

[0012] Furthermore, it also includes an adjustment module, which is used to: determine whether the deviation of the blood volume of the preset patient in one cycle exceeds the preset blood volume deviation threshold when performing CRRT treatment on the preset patient; if so, adjust the net ultrafiltration rate timing scheme.

[0013] The beneficial effects of adopting the above-mentioned further approach are as follows: By introducing an adjustment module, blood volume can be monitored and evaluated in real time during CRRT treatment. Within each cycle of CRRT treatment, it accurately determines whether the blood volume deviation of the pre-set patient exceeds a preset threshold, achieving refined control of the treatment process. Once an excessive deviation is detected, the net ultrafiltration rate timing protocol can be adjusted promptly to ensure the treatment process remains in an ideal state. This dynamic monitoring and adjustment mechanism effectively solves the problem in traditional CRRT treatment where a fixed ultrafiltration rate is difficult to adapt to patient physiological fluctuations. During long-term treatment, the patient's blood volume and hemodynamics constantly change; this approach can make precise adjustments based on real-time data, maintaining a dynamic balance between the net ultrafiltration rate and the patient's physiological state. This significantly reduces the risk of complications such as hypotension and blood volume overload, improving the safety and effectiveness of treatment. It also makes the setting of the net ultrafiltration rate no longer solely dependent on physician experience, enabling personalized treatment based on real-time patient indicators, further enhancing the accuracy and intelligence of CRRT treatment.

[0014] Furthermore, the sensor module is also used to: acquire physiological parameter monitoring data of the preset patient when performing CRRT treatment on the preset patient, predict the blood volume of the preset patient based on the physiological parameter monitoring data, and determine the risk level corresponding to the predicted blood volume.

[0015] The warning module is used to: determine whether to issue a warning based on the risk level corresponding to the predicted blood volume; if so, issue a warning based on the risk level corresponding to the predicted blood volume.

[0016] The beneficial effects of adopting the above-mentioned further solution are: it enables real-time and comprehensive monitoring of the physiological state of the pre-defined patients, allowing for early prediction of potential abnormalities in blood volume and their risk levels, providing crucial information for subsequent treatment decisions. The warning module determines whether to issue a warning based on the risk level corresponding to the predicted blood volume and issues the appropriate warning level, enabling medical staff to promptly detect potential risks in pre-defined patients and intervene in advance, greatly enhancing the safety and reliability of treatment and preventing serious complications caused by blood volume issues.

[0017] Furthermore, CRRT treatment-related data include: hemodynamic parameters, electrolyte concentration, filter transmembrane pressure, blood flow velocity, and filter waste pressure.

[0018] The beneficial effects of adopting the above-mentioned further scheme are that hemodynamic parameters, electrolyte concentration, filter transmembrane pressure, blood flow velocity and filter waste pressure can accurately reflect the physiological and equipment operating status during the CRRT treatment of the preset patients, providing a solid basis for predicting the blood volume tolerance curve and generating the net ultrafiltration rate timing scheme, effectively ensuring the treatment effect and safety.

[0019] 2) In a second aspect, the present invention also provides a method for intelligent control of net ultrafiltration rate for CRRT treatment, the specific technical solution of which is as follows:

[0020] Obtain CRRT treatment-related data for a predefined patient;

[0021] Based on the patient's historical CRRT treatment data and CRRT treatment-related data, and using artificial intelligence technology to predict the patient's blood volume tolerance curve, a net ultrafiltration rate time series plan is generated according to the predicted blood volume tolerance curve and the patient's preset treatment goal.

[0022] Based on the net ultrafiltration rate timing scheme, the speed of the blood pump and the speed of the replacement fluid pump used for CRRT treatment of preset patients are automatically adjusted, and the dehydration rate of preset patients is also adjusted.

[0023] Based on the above scheme, the intelligent control method for net ultrafiltration rate of the present invention for CRRT treatment can be further improved as follows.

[0024] Furthermore, it also includes: when performing CRRT treatment on a preset patient, determining whether the deviation of the preset patient's blood volume within one cycle exceeds a preset blood volume deviation threshold; if so, adjusting the net ultrafiltration rate timing scheme.

[0025] Furthermore, it also includes: displaying in real time the net ultrafiltration rate deviation between the net ultrafiltration rate in the net ultrafiltration rate timing scheme and the actual net ultrafiltration rate when performing CRRT treatment on the preset patient, and determining whether to issue an early warning based on the real-time net ultrafiltration rate deviation, and issuing an early warning if so.

[0026] Furthermore, it also includes:

[0027] When performing CRRT on a pre-defined patient, the physiological parameter monitoring data of the pre-defined patient is obtained, and the blood volume of the pre-defined patient is predicted based on the physiological parameter monitoring data, and the risk level corresponding to the predicted blood volume is determined.

[0028] Whether to issue an early warning is determined based on the risk level corresponding to the predicted blood volume. If so, an early warning is issued based on the risk level corresponding to the predicted blood volume.

[0029] Furthermore, CRRT treatment-related data include: hemodynamic parameters, electrolyte concentration, filter transmembrane pressure, blood flow velocity, and filter waste pressure.

[0030] 3) In a third aspect, the present invention also provides an electronic device, the electronic device including a processor coupled to a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor, so that the electronic device realizes any of the above-mentioned intelligent control device for net ultrafiltration rate for CRRT treatment.

[0031] 4) In a fourth aspect, the present invention also provides a computer-readable storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements any of the above-mentioned intelligent control devices for net ultrafiltration rate in CRRT treatment.

[0032] It should be noted that the beneficial effects of the technical solutions of the second to fourth aspects of the present invention and their corresponding possible implementations can be found in the above description of the technical effects of the first aspect and its corresponding possible implementations, and will not be repeated here. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments of the present invention will be briefly introduced below:

[0034] Figure 1 This is a schematic diagram of the structure of an intelligent control device for net ultrafiltration rate in CRRT treatment according to an embodiment of the present invention;

[0035] Figure 2 This is a flowchart illustrating an intelligent control method for net ultrafiltration rate in CRRT treatment according to an embodiment of the present invention.

[0036] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0037] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0038] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0039] like Figure 1 As shown, an intelligent control device for net ultrafiltration rate in CRRT treatment according to an embodiment of the present invention includes a sensor module, a data processing module, and a control module.

[0040] The sensor module is used to: acquire CRRT treatment-related data for a preset patient;

[0041] The data related to CRRT treatment include: hemodynamic parameters, electrolyte concentration, filter transmembrane pressure, blood flow velocity, and filter waste pressure.

[0042] Hemodynamic parameters refer to various physiological parameters related to blood circulation, such as heart rate, blood pressure, cardiac output, peripheral vascular resistance, mean arterial pressure, and central venous pressure. These parameters reflect the functional status of the cardiovascular system and provide a basis for doctors to assess the hemodynamic stability of a pre-selected patient. In CRRT treatment, they help doctors understand whether the patient's circulatory system can tolerate the treatment and whether the treatment has affected the patient's hemodynamic balance.

[0043] Electrolyte concentrations include the concentrations of serum sodium, potassium, chloride, calcium, and magnesium ions. These electrolytes play a crucial role in maintaining normal cellular physiological functions, neuromuscular excitability, and acid-base balance. CRRT treatment can affect electrolyte balance; monitoring electrolyte concentrations allows for the timely detection and correction of electrolyte disturbances, preventing serious complications such as arrhythmias and altered neuromuscular excitability caused by electrolyte abnormalities.

[0044] Transmembrane pressure refers to the pressure difference between the inlet and outlet of the filter, reflecting the filter's filtration resistance. Monitoring transmembrane pressure is crucial for assessing filter permeability and functional status. Excessively high transmembrane pressure may indicate filter blockage or decreased permeability, requiring timely intervention such as filter flushing or replacement to ensure the smooth progress of CRRT treatment.

[0045] Blood flow velocity refers to the speed at which blood flows through the extracorporeal circulation tubing and filters. Appropriate blood flow velocity is a key factor in ensuring the effectiveness of CRRT treatment. Excessively fast blood flow may lead to adverse reactions such as hemolysis and platelet activation, while excessively slow blood flow may affect solute clearance efficiency. By monitoring blood flow velocity, the pump speed can be adjusted in a timely manner to ensure the safety and effectiveness of the treatment.

[0046] The filter waste pressure refers to the pressure at the filter waste liquid outlet, reflecting the resistance to waste liquid discharge. Normal waste pressure facilitates smooth waste liquid discharge, ensuring the normal progress of CRRT treatment. Excessive waste pressure may lead to poor waste liquid drainage, affecting the filter's filtration function and even causing filter damage. Therefore, monitoring filter waste pressure is crucial for timely detection and resolution of waste liquid drainage problems.

[0047] The data processing module is used to: predict the blood volume tolerance curve of the preset patient based on the patient's historical CRRT treatment data and CRRT treatment-related data, and generate a net ultrafiltration rate time series plan based on the predicted blood volume tolerance curve and the preset treatment target of the preset patient.

[0048] Specifically, based on the patient's historical CRRT treatment data and CRRT-related data, and using artificial intelligence technology, the patient's blood volume tolerance curve is predicted.

[0049] 1) First implementation method:

[0050] ① Collect historical CRRT treatment data from the medical information system for pre-defined patients. This historical CRRT treatment data includes: CRRT treatment prescription parameters and equipment operation logs. CRRT treatment prescription parameters include the target ultrafiltration rate, actual ultrafiltration volume, anticoagulation regimen, filter type, and treatment duration. Equipment operation logs include treatment start / stop times and warning event records. Historical CRRT treatment data also includes: structured electronic medical records, bedside ultrasound imaging sequences, and laboratory test time-series data. Structured electronic medical records include intake and output records, vasoactive drug usage, and diagnostic results. Bedside ultrasound imaging sequences include information such as inferior vena cava (IVC) diameter and collapse rate, lung B-lines, and cardiac function. Laboratory test time-series data includes creatinine levels, blood urea nitrogen levels, and lactate levels.

[0051] ② After spatiotemporally aligned preprocessing of historical CRRT treatment data and CRRT treatment-related data of preset patients, hemodynamic parameters were processed through an adaptive sliding window to obtain volume responsiveness characteristics. Bedside ultrasound image sequences were processed using a three-dimensional convolutional network to obtain volume dynamics characteristics. Based on filter transmembrane pressure, blood flow velocity, and filter waste pressure, filter efficiency characteristics were obtained. Among them, volume responsiveness characteristics include: time-domain indices (including stroke volume variability and pulse pressure variability) and frequency-domain power spectrum; volume dynamics characteristics include: inferior vena cava respiratory variability, lung B-line distribution density, and ventricular filling rate of change; filter efficiency characteristics include: filter coagulation risk index and ultrafiltration efficiency coefficient.

[0052] The stroke volume variability (SVV) is explained as follows:

[0053] Stroke volume refers to the amount of blood pumped by the heart with each heartbeat (one ventricular contraction). Stroke volume variability is an important indicator of volume responsiveness. Under specific conditions such as mechanical ventilation, when the patient's pre-set blood volume is insufficient, the amount of venous return to the heart fluctuates with the respiratory cycle, causing stroke volume to fluctuate accordingly, resulting in increased stroke volume variability. For example, in a stable circulatory system, if blood volume is sufficient, the amount of blood pumped by the heart with each beat is relatively stable, and stroke volume variability is small; however, when blood volume is insufficient, this stability of output is disrupted, and stroke volume variability increases, suggesting that blood volume supplementation may be needed to improve circulatory status.

[0054] The pulse pressure variability (PPV) is explained as follows:

[0055] Pulse pressure is the difference between systolic and diastolic blood pressure, primarily reflecting the change in the amount of blood ejected with each heartbeat. Similar to stroke volume variability, pulse pressure variability also changes when blood volume is insufficient due to fluctuations in venous return. Increased pulse pressure variability indicates significant fluctuations in pulse pressure during physiological processes such as the respiratory cycle, suggesting good volume responsiveness, meaning that blood volume supplementation may improve hemodynamic status.

[0056] The frequency domain power spectrum is explained as follows:

[0057] Frequency domain power spectrum: This is a representation of hemodynamic signals converted from the time domain to the frequency domain. It can analyze the energy distribution of the signal across different frequency ranges. For example, for time-series data of hemodynamic parameters, the frequency domain power spectrum can be obtained through methods such as Fourier transform. In the frequency domain, different frequency components may correspond to different physiological mechanisms. Low-frequency components may be related to slowly changing physiological processes such as autonomic nervous system regulation, while high-frequency components may be related to rapidly changing physiological events such as heartbeats. By analyzing the frequency domain power spectrum, we can gain a deeper understanding of the fluctuation characteristics of hemodynamic parameters, thereby extracting feature information related to volume responsiveness.

[0058] The specific implementation process of using an adaptive sliding window to process hemodynamic parameters and obtain time-domain indices (stroke volume variability, pulse pressure variability) and frequency-domain power spectra is as follows:

[0059] First, hemodynamic parameters from the historical CRRT treatment data of the pre-defined patients are acquired. These data include time-series data such as stroke volume (SV) and pulse pressure (PP). This data is collected from the medical information system and undergoes preliminary processing and cleaning to ensure data quality and integrity. Based on the characteristics and rate of change of the hemodynamic parameters, the initial length and adjustment strategy of the adaptive sliding window are determined. For example, the initial window length can be set to include a certain number of cardiac cycles (e.g., 20-30 cardiac cycles), and the window length is dynamically adjusted according to data changes. If the hemodynamic parameters change drastically, indicating that the system is in an unstable state, the window length can be appropriately shortened to capture the parameter change characteristics more promptly; conversely, if the parameter changes are relatively stable, the window length can be appropriately lengthened to reduce computation and improve feature stability. For stroke volume variability, within each sliding window, the stroke volume value for each cardiac beat is calculated, and then the ratio of the standard deviation to the mean of these SV values ​​is calculated. The formula is SVV = (Standard deviation (SV) / ​​Mean (SV)) × 100%. This process is repeated, and as the sliding window moves across the time series, a series of SVV values ​​are obtained, forming the SVV time series. Similarly, for pulse pressure variability (PPV), the pulse pressure (PP) value corresponding to each heartbeat is calculated within each sliding window, and then the ratio of the standard deviation to the mean of the PP value is calculated, i.e., PPV = (standard deviation (PP) / mean (PP)) × 100%, yielding the PPV time series. Fourier transforms are performed on the hemodynamic parameter time series data within each sliding window. For example, for the stroke volume time series, a Discrete Fourier Transform (DFT) is performed within the window to convert the time-domain signal to a frequency-domain signal. The power spectral density of the frequency-domain signal, i.e., the power of each frequency component, is then calculated. This can be obtained by calculating the square of the amplitude of the frequency-domain signal. Thus, a frequency-domain power spectrum is obtained at each sliding window position. As the sliding window moves, a series of frequency-domain power spectra can be obtained, allowing analysis of the frequency-domain characteristics of hemodynamic parameters at different time points.

[0060] The variability of inferior vena cava respiration is explained as follows:

[0061] The inferior vena cava is an important venous passage in the human body, and its diameter changes with the respiratory cycle. Inferior vena cava respiratory variability reflects the degree of diameter change during respiration. In hemodynamically unstable or hypovolemic conditions, the collapse rate of the inferior vena cava increases, leading to greater respiratory variability. This is an important indicator for assessing volume responsiveness; a larger inferior vena cava respiratory variability suggests that volume supplementation may improve circulatory status.

[0062] The distribution density of B-lines in the lungs is explained as follows:

[0063] Lung B-lines are ultrasound artifacts observed during bedside ultrasound examinations, reflecting the accumulation of interstitial fluid in the lungs. Normally, lung B-lines are few; their number increases in conditions such as pulmonary edema and interstitial lung disease. The distribution density of lung B-lines can be used to assess pulmonary fluid load and is an indicator of the severity of diseases such as heart failure.

[0064] The rate of change in ventricular filling is explained as follows:

[0065] The rate of change of ventricular filling volume refers to the rate at which the volume of the heart's ventricles changes during diastole. It reflects the heart's diastolic function and its ability to adapt to changes in blood volume. The rate of change of ventricular filling volume can provide a basis for assessing cardiac function and developing treatment plans; for example, in diseases such as heart failure, the rate of change of ventricular filling volume may be abnormal.

[0066] The specific process of using a three-dimensional convolutional network to process bedside ultrasound image sequences to obtain inferior vena cava respiratory variability, lung B-line distribution density, and ventricular filling rate of change is as follows:

[0067] Bedside ultrasound image sequences of pre-defined patients are collected from a medical information system. These sequences include ultrasound images of the inferior vena cava, lungs, and heart. The ultrasound image sequences are preprocessed, including noise removal and contrast enhancement, to improve image quality and ensure the accuracy of subsequent feature extraction. A three-dimensional convolutional network architecture is constructed, with the input layer size matching the size of the preprocessed ultrasound image sequences. The three-dimensional convolutional network contains multiple convolutional layers, pooling layers, and fully connected layers. Convolutional layers are used to extract spatial and temporal features from the ultrasound image sequences, pooling layers are used to reduce feature dimensionality and enhance feature robustness, and fully connected layers are used for feature fusion and classification. Ultrasound image sequences containing the inferior vena cava are input into the three-dimensional convolutional network. The convolutional kernels in the convolutional layers slide in three-dimensional space (two spatial dimensions and a temporal dimension) to extract the diameter variation features of the inferior vena cava during the respiratory cycle. Through multiple convolutional and pooling operations, the network can learn the feature representation of inferior vena cava respiratory variability, which is finally output at the fully connected layer. Lung ultrasound image sequences are also input into the three-dimensional convolutional network. Convolutional layers extract B-line features from lung ultrasound images, including the shape, number, and distribution of B-lines. After processing by the network, the distribution density of lung B-lines is obtained, and the lung B-line distribution density is output. Echocardiographic image sequences are input into a three-dimensional convolutional network. The network extracts the variation characteristics of ventricular filling volume during diastole, i.e., the ventricular filling rate of change, through operations such as convolutional and pooling layers. These features are then processed by fully connected layers to obtain the volumetric dynamic feature of the ventricular filling rate of change.

[0068] The "Filter Coagulation Risk Index" is explained as follows:

[0069] During CRRT treatment, filter clotting may occur, which can affect treatment efficacy and filter lifespan. The filter clotting risk index is a quantitative indicator used to assess the likelihood of filter clotting. A higher filter clotting risk index means that blood components inside the filter are more likely to clot, possibly due to factors such as poor blood flow or inadequate anticoagulation regimens.

[0070] The ultrafiltration efficiency coefficient is explained as follows:

[0071] The ultrafiltration efficiency coefficient reflects the efficiency of a filter in ultrafiltration of fluids. In CRRT treatment, ultrafiltration is used to remove excess water and solutes from the body. A higher ultrafiltration efficiency coefficient indicates that the filter can perform ultrafiltration more effectively per unit time, better completing the task of fluid removal, which is crucial for maintaining the patient's pre-set blood volume balance and homeostasis.

[0072] Among them, based on the filter transmembrane pressure, blood flow velocity, and filter waste pressure, the filter coagulation risk index and ultrafiltration efficiency coefficient are obtained, including:

[0073] Data such as filter transmembrane pressure, blood flow velocity, and filter waste pressure are obtained from the CRRT equipment operation log. This data is continuously monitored time-series data and requires preprocessing, including data cleaning (removing outliers and handling missing values) and normalization (converting data of different dimensions to the same scale) to ensure data quality and consistency, making it suitable for subsequent analysis and modeling. Filter transmembrane pressure reflects the pressure difference across the filter; it typically increases when clotting occurs. Low blood flow velocity may cause blood to stagnate within the filter, increasing the risk of clotting. Changes in filter waste pressure may also be related to clotting. By establishing a mathematical model based on filter transmembrane pressure, blood flow velocity, and filter waste pressure, a filter clotting risk index can be calculated. For example, multiple regression analysis can be used, with filter transmembrane pressure, blood flow velocity, and filter waste pressure as independent variables. The model can be trained using a large amount of historical data to determine their weighted relationships with filter clotting risk, thus deriving the formula for the filter clotting risk index. Assuming the model formula is: Filter Coagulation Risk Index = w1 × Filter Transmembrane Pressure + w2 × (1 / Blood Flow Velocity) + w3 × Filter Waste Pressure + b (where w1, w2, and w3 are weighting coefficients, and b is a bias term), the filter coagulation risk index can be calculated by substituting real-time monitored data of filter transmembrane pressure, blood flow velocity, and filter waste pressure. The ultrafiltration efficiency coefficient is mainly related to the ultrafiltration function of the filter. Filter transmembrane pressure is one of the driving factors for ultrafiltration; appropriate transmembrane pressure can promote ultrafiltration. Blood flow velocity affects the contact efficiency between blood and the filter; appropriate blood flow velocity can improve ultrafiltration efficiency. Simultaneously, filter waste pressure also affects ultrafiltration. Similarly, a model based on these three parameters can be established to calculate the ultrafiltration efficiency coefficient. For example, a method similar to the filter coagulation risk index can be used, through multiple regression or other machine learning algorithms, to determine the relationship between filter transmembrane pressure, blood flow velocity, filter waste pressure, and ultrafiltration efficiency. Assuming the model formula is: Ultrafiltration efficiency coefficient = a1 × filter transmembrane pressure + a2 × blood flow rate + a3 × (1 / filter waste pressure) + c (where a1, a2, and a3 are weighting coefficients, and c is a bias term), the ultrafiltration efficiency coefficient can be obtained by substituting real-time data into the formula.

[0074] ③ Input the capacity responsiveness characteristics and ultrafiltration rate time series data into the temporal convolutional network, so that the temporal convolutional network can analyze the short-time lag correlation between ultrafiltration rate changes and capacity responsiveness indicators, and output the ultrafiltration-physiological response pattern;

[0075] The ultrafiltration-physiological response model refers to the short-term lag correlation between changes in ultrafiltration rate and volume responsiveness indicators during CRRT (continuous renal replacement therapy). This model reflects the response of a pre-defined patient's physiological state (such as blood volume and hemodynamics) to changes in ultrafiltration rate during ultrafiltration treatment. For example, when the ultrafiltration rate increases, the pre-defined patient's volume responsiveness indicators (such as stroke volume variability and pulse pressure variability) may change within a certain time lag. The ultrafiltration-physiological response model can reveal this relationship and provide a basis for clinicians to adjust the ultrafiltration rate.

[0076] The specific implementation process of obtaining the ultrafiltration-physiological response mode is as follows:

[0077] Capacity responsiveness features (such as stroke volume variability, pulse pressure variability, and other time-domain indices and frequency-domain power spectra) and ultrafiltration rate time-series data are acquired. These data are collected from a medical information system and preprocessed with spatiotemporal alignment to ensure consistency across time dimensions. A multi-layered temporal convolutional network (TCN) architecture is constructed, containing multiple one-dimensional convolutional layers with causal convolutions. Causal convolutions ensure that the network uses only information from the current and past moments when predicting future outputs, meeting the causal requirements of time series analysis. Convolutional layers are typically followed by activation functions (such as ReLU), normalization layers (such as batch normalization), and pooling layers (such as max pooling) to extract and enhance features. The capacity responsiveness features and ultrafiltration rate time-series data are used as inputs to the temporal convolutional network. These features and data can be input as multi-channel time series data into the network. Each channel corresponds to a specific feature (e.g., one channel corresponds to stroke volume variability, another to ultrafiltration rate, etc.). Temporal convolutional networks extract features within different time windows by performing convolution operations on input data. The convolutional kernels slide along the time axis, capturing the local temporal dependencies between ultrafiltration rate changes and capacity responsiveness indicators. Due to the network's multi-layered structure, each layer can progressively extract higher-level feature representations. In shallow layers, temporal convolutional networks can extract simple temporal patterns, such as the immediate impact of increases or decreases in ultrafiltration rate on capacity responsiveness indicators; in deeper layers, they can capture more complex patterns, such as short-lag associations. After processing by the temporal convolutional network, the ultrafiltration-physiological response pattern is finally obtained in the output layer. This pattern can be represented numerically to quantify the strength and temporal lag relationship between ultrafiltration rate changes and capacity responsiveness indicators. For example, the output can represent the expected change in capacity responsiveness indicators at different time points after an ultrafiltration rate change, thus providing clinicians with predictive information about the expected response of a patient to ultrafiltration treatment.

[0078] Based on the ultrafiltration-physiological response mode, volumetric dynamics characteristics and filter efficiency characteristics, and by modeling the tolerance evolution path across treatment stages through a multi-head self-attention mechanism, long-term causal chain characteristics are output, such as high ultrafiltration rate → IVC collapse → lactate accumulation, etc.

[0079] Among them, long-term causal chain features refer to the features obtained by modeling the tolerability evolution path across treatment stages during CRRT (continuous renal replacement therapy) based on ultrafiltration-physiological response patterns, volumetric dynamics characteristics, and filter efficacy characteristics, through a multi-head self-attention mechanism. It reflects the long-term causal chain between various physiological and treatment-related factors in different treatment stages for a pre-defined patient. For example, a high ultrafiltration rate may lead to inferior vena cava (IVC) collapse, resulting in a series of chain reactions such as lactate accumulation. Long-term causal chain features can capture and quantify this complex causal evolution across stages.

[0080] The specific implementation process for obtaining long-range causal chain features is as follows:

[0081] The input data includes ultrafiltration-physiological response patterns, volumetric dynamics features, and filter efficacy features. These features are extracted from CRRT treatment data at different stages, covering key information about the pre-defined patient's physiological state and treatment efficacy, and have undergone spatiotemporal alignment preprocessing to ensure consistency across time. A multi-head self-attention mechanism model is constructed. This model contains multiple attention heads, each capturing the correlations between different positions in the input feature sequence. These attention heads are computed in parallel, extracting correlation information between features from different perspectives. The outputs of multiple attention heads are then concatenated and linearly transformed to obtain the final output. The input feature sequence is fed into the multi-head self-attention mechanism model. The model calculates the weights between different features at different time points through the attention mechanism, i.e., the degree of influence of each feature on other features at different time points. This process captures the tolerability evolution path across treatment stages. For example, the correlation between early ultrafiltration-physiological response patterns and later volumetric dynamics features, and how these features interact with filter efficacy features to affect the pre-defined patient's tolerability. After processing by the multi-head self-attention mechanism, the model outputs long-range causal chain features. These features are represented as vectors or matrices, containing long-term causal relationships between ultrafiltration-physiological response patterns, volumetric kinetics, and filter efficiency characteristics. For example, the output long-range causal chain features can be represented as a sequence of feature vectors, each corresponding to a time point and containing causal association information with features at previous time points. These features can be used for further analysis and prediction, such as generating blood volume tolerance curves.

[0082] ④ Input the long-range causal chain features into the cascaded prediction network, and generate the blood volume tolerance curve by generating the trend of volume status change and tolerance boundary in the next N hours.

[0083] The cascaded prediction network is a deep learning model architecture used to predict future trends and boundaries based on input features. In this CRRT (Continuous Renal Replacement Therapy) solution, the cascaded prediction network receives long-range causal chain features as input. These features include long-term causal relationships between ultrafiltration-physiological response patterns, volumetric dynamics features, and filter efficacy features. The cascaded prediction network progressively extracts and fuses these features through its internal multi-layered structure to generate a blood volume tolerance curve for a specific future time period (N hours), i.e., predicting the trend of volume status changes and tolerance boundaries for a pre-defined patient. The tolerance boundary can be understood as a critical range of blood volume changes for the pre-defined patient; exceeding this range may have adverse effects on the patient.

[0084] The specific process for obtaining the blood volume tolerance curve is as follows:

[0085] Long-range causal chain features obtained through a multi-head self-attention mechanism are used as input to the cascaded prediction network. These long-range causal chain features integrate complex relationships between ultrafiltration-physiological response patterns, volumetric dynamics features, and filter efficiency features, providing a comprehensive information foundation for prediction. The cascaded prediction network can employ multi-layer neural network structures, such as recurrent neural network (RNN) architectures like Long Short-Term Memory (LSTM) networks or gated recurrent units (GRUs), or it can combine convolutional neural network (CNN) structures to extract local and global features from time series. These network structures can process time-series data and capture temporal dependencies in the data. For example, using an LSTM network, which includes input, forget, and output gates, can effectively remember and forget information in the sequence, thereby better predicting future volume states. In the cascaded prediction network, the input long-range causal chain features are processed through multiple layers of neurons. Each layer extracts feature representations at different levels and passes them to the next layer. For example, the first layer might extract simple feature patterns, such as trends and periodic changes; deeper layers would fuse these simple features to form more complex feature representations, thereby capturing the underlying patterns of volume state changes. After multi-layer processing by the network, the output layer finally generates the volume state change trend and tolerance boundary for the next N hours. The number of neurons and activation function settings in the output layer are adjusted according to the specific prediction target. For example, for predicting the volume state change trend, a time series can be output, representing the predicted blood volume value at each future time point; for predicting the tolerance boundary, a range can be output, representing the upper and lower limits of the preset patient's blood volume tolerance. Combining these prediction results forms the blood volume tolerance curve, providing clinicians with predictive information about the future blood volume state of the preset patient, helping them to formulate more precise treatment plans, such as adjusting treatment parameters like ultrafiltration rate, to ensure that the preset patient's blood volume is maintained within a safe range.

[0086] 2) The second implementation method:

[0087] ① Collect historical CRRT treatment data and CRRT-related data from multiple patients. Historical CRRT treatment data includes CRRT treatment prescription parameters (target ultrafiltration rate, actual ultrafiltration volume, anticoagulation regimen, filter type, and treatment duration), equipment operation logs (treatment start / stop times and warning event records), structured electronic medical records (intake and output records, vasoactive drug usage, and diagnostic results), bedside ultrasound imaging sequences (inferior vena cava IVC diameter and collapse rate, lung B-line data, and cardiac function information), and time-series laboratory test data (creatinine levels, blood urea nitrogen levels, and lactate levels). CRRT-related data includes hemodynamic parameters, electrolyte concentrations, filter transmembrane pressure, blood flow velocity, and filter waste pressure. The collected data underwent preprocessing, including missing value imputation, outlier handling, and normalization, to ensure data quality and consistency. Simultaneously, the data was spatiotemporally aligned to ensure consistency across different data sources in both time and space dimensions.

[0088] ② Divide the preprocessed data into training, validation, and test sets. Based on the training set, construct an LSTM model. The LSTM model contains multiple LSTM layers, each containing multiple LSTM units, used to capture temporal dependencies in the data. After the LSTM layers, a fully connected layer is typically added to map the LSTM layer output to the target output space. Use various features (such as target ultrafiltration rate, actual ultrafiltration volume, hemodynamic parameters, etc.) from historical CRRT treatment data and CRRT-related data as input features to the model. These features, after preprocessing, are input into the LSTM model as vectors. The target output is the blood volume tolerance curve, i.e., the trend of volume status changes and tolerance boundaries over the next N hours. The numerical value of blood volume change, upper and lower tolerance limits, etc., can be used as output variables. Choose an appropriate loss function, such as the mean squared error (MSE) loss function or the mean absolute error (MAE) loss function, to measure the difference between the model's predicted and actual values. Select an optimizer, such as Adam or RMSprop, to update the model parameters to minimize the loss function. The training data is fed into the LSTM model, the gradient of the loss function is calculated using the backpropagation algorithm, and the model parameters are updated using the optimizer. During training, the model's performance is periodically evaluated on the validation set to avoid overfitting. Training stops when the model's performance on the validation set no longer improves.

[0089] ③ Perform the same preprocessing operations on the historical CRRT treatment data and CRRT treatment-related data of the preset patients to ensure that the data format and dimensions are consistent with the training data. Extract the feature data of the preset patients and input them into the trained LSTM model. Based on the input feature data, the model generates the blood volume tolerance curve of the preset patients, that is, the trend of volume status changes and tolerance boundary in the next N hours.

[0090] Specifically, based on the predicted blood volume tolerance curve and the preset treatment goals for the patients, a net ultrafiltration rate time series protocol is generated, including:

[0091] 1) First implementation method:

[0092] ①After obtaining the predicted blood volume tolerance curve of the preset patient, set the preset treatment target, that is, set the target dehydration volume and the target blood sodium concentration;

[0093] ② Extract the safe tolerance boundary threshold of the blood volume tolerance curve, transform the safe tolerance boundary threshold into a dynamic constraint condition for the upper limit of net ultrafiltration rate, and encode it into a set of mathematical inequalities;

[0094] The safe tolerance threshold of the blood volume tolerance curve refers to the range within which the blood volume of a pre-set patient can be safely maintained during CRRT treatment. It includes the upper and lower limits of blood volume. When the blood volume exceeds this range, it may have adverse effects on the pre-set patient, such as heart failure due to volume overload or hypotension due to volume insufficiency. The safe tolerance threshold is predicted based on the pre-set patient's physiological condition and real-time data during treatment, and can provide a safety reference for operations such as ultrafiltration therapy.

[0095] The specific implementation process for extracting the safe tolerance boundary threshold of the blood volume tolerance curve is as follows:

[0096] Key points are identified in the blood volume tolerance curve. This can be achieved by reducing noise interference through curve smoothing (e.g., moving average), followed by first and second derivative analysis to find the extreme points (maximum and minimum values) and inflection points of the curve. These extreme points and inflection points correspond to critical moments in blood volume changes. At these key points, the corresponding upper and lower limits of blood volume are extracted; these values ​​are the safe tolerance boundary thresholds. The extracted thresholds are stored so that they can be subsequently converted into dynamic constraints for the upper limit of net ultrafiltration rate.

[0097] Specifically, the safety tolerance boundary threshold is transformed into a dynamic constraint condition for the upper limit of the net ultrafiltration rate and encoded as a system of mathematical inequalities, namely:

[0098] Based on the patient's current blood volume status and the safety tolerance threshold, dynamic constraints can be established. Assuming the current blood volume is V, the lower limit of the safety tolerance threshold is Vmin, and the net ultrafiltration rate is U, within a short time interval Δt, to ensure the blood volume does not fall below the lower safety limit, VU × Δt ≥ Vmin must be satisfied. Similarly, if the upper limit of blood volume Vmax is considered, constraints on the upper limit can also be obtained. Combining these constraints allows them to be encoded into a system of mathematical inequalities, which guides the rational setting of the net ultrafiltration rate, ensuring the treatment process is conducted within the patient's pre-defined safety tolerance range.

[0099] ③ Define the objective function as follows: the first term is the sum of squares of the cumulative deviations between the dehydration progress and the target dehydration amount, and the second term is the root mean square error between the measured value and the target value of serum sodium concentration.

[0100] The objective function is defined to comprehensively consider dehydration effectiveness and serum sodium concentration control during CRRT (Continuous Renal Replacement Therapy). Optimizing this objective function adjusts treatment parameters to bring them as close as possible to the expected treatment goals. By combining the first and second terms to form the objective function, both dehydration effectiveness and serum sodium concentration control can be comprehensively considered, resulting in safer and more effective CRRT treatment. Optimizing this objective function helps determine appropriate treatment parameters, such as net ultrafiltration rate, to maintain serum sodium concentration within the normal range while meeting dehydration targets.

[0101] The technical terms involved in the first and second items are explained as follows:

[0102] a. Dehydration progress: This refers to the actual amount of water removed during CRRT treatment. It is a variable that accumulates over time, reflecting the cumulative dehydration effect during the treatment process.

[0103] b. Target dehydration amount: This is a target amount of water to be removed, set before treatment based on the patient's condition and physical state. This target value is determined to achieve the patient's fluid balance and alleviate volume overload.

[0104] c. Cumulative Sum of Squared Deviations: Calculate the difference (deviation) between the dehydration progress and the target dehydration amount at each time point, then square these deviation values, and finally sum all the squared deviation values ​​to obtain the cumulative sum of squared deviations. The purpose of this step is to measure the overall deviation between the dehydration progress and the target dehydration amount throughout the entire treatment process, and to try to ensure that the dehydration progress is as close as possible to the expected target dehydration amount.

[0105] d. Measured serum sodium concentration: This is the concentration of sodium ions in the patient's blood actually measured during CRRT treatment using laboratory methods. Serum sodium concentration is an important indicator reflecting the patient's electrolyte balance, and its changes may be closely related to dehydration, ultrafiltration, and other factors during treatment.

[0106] e. Target value: This refers to the desired level of serum sodium concentration, usually set based on the patient's expected normal physiological range and treatment needs. Maintaining serum sodium concentration near the target value is crucial for maintaining normal neuromuscular excitability and cellular function in the intended patient.

[0107] f. Root Mean Square Error: First, calculate the difference (error) between the measured serum sodium concentration and the target value at each time point. Then, square these error values ​​to obtain the squared error. Next, calculate the average of all squared errors, and finally, take the square root of this average to obtain the root mean square error. This term measures the overall difference between the measured serum sodium concentration and the target value. By minimizing this term, the serum sodium concentration can be made closer to the target value, thereby maintaining the preset electrolyte balance for the patient.

[0108] ④ Based on the system of mathematical inequalities and the objective function, and combined with the trained LSTM model, the initial time series scheme of the net ultrafiltration rate that satisfies the safety constraints is obtained.

[0109] The predefined tolerance boundary constraint model and objective function are used as additional inputs or constraints, and are fed into the trained LSTM model along with the preprocessed feature data. The LSTM model, based on patterns and relationships learned from historical data, combined with the current constraints and objective function, predicts an initial time-series scheme for the net ultrafiltration rate that satisfies safety constraints. This initial time-series scheme for the net ultrafiltration rate is a time series that provides suggested values ​​for the net ultrafiltration rate at each time point during the initial stages of treatment, aiming to approach the target dehydration volume and target serum sodium concentration as closely as possible while meeting the blood volume safety tolerance boundary.

[0110] ⑤ Based on the trend of the slope change of the tolerance curve, identify the period of rapid capacity transfer risk, and generate an anti-fluctuation enhancement scheme by superimposing the ultrafiltration rate smooth transition constraint on the corresponding period. Specifically:

[0111] First, the blood volume tolerance curve for the pre-defined patient is obtained. The slope is calculated by numerical differentiation of this curve. When the absolute value of the slope exceeds a preset threshold, it is identified as a period of rapid volume shift risk. This indicates that during these periods, the pre-defined patient's blood volume is highly sensitive to changes in the ultrafiltration rate and is prone to volume fluctuations. Within the identified risk periods, a smooth transition constraint on the ultrafiltration rate is superimposed. This requires that the change in the ultrafiltration rate during increases or decreases cannot exceed a set smooth transition threshold. For example, the hourly change in the ultrafiltration rate is specified to not exceed a certain value to reduce blood volume fluctuations caused by sudden changes in the ultrafiltration rate. The smooth transition constraint on the ultrafiltration rate is combined with the original treatment plan to generate an anti-fluctuation enhancement plan. During the risk periods, the ultrafiltration rate is adjusted according to the smooth transition constraint to make its changes more stable. Simultaneously, the plan is dynamically adjusted based on the pre-defined patient's real-time blood volume monitoring data to ensure the stability of the treatment process.

[0112] ⑥ The initial time series scheme for net ultrafiltration rate is modified using an anti-fluctuation enhancement scheme to generate a net ultrafiltration rate time series scheme, specifically:

[0113] The anti-fluctuation enhancement protocol was integrated with the initial net ultrafiltration rate (NIU) time-series protocol. The anti-fluctuation enhancement protocol included information such as constraints on the smooth transition of ultrafiltration rate during periods of rapid volume shift risk. During integration, consistency between the two protocols over time was ensured to allow for adjustments during corresponding risk periods. The slope trend of the blood volume tolerance curve was analyzed to identify periods of rapid volume shift risk. This was achieved by numerically differentiating the tolerance curve, calculating the absolute value of the slope, and comparing it to a preset threshold. When the slope exceeded the threshold, the period was identified as a risk period. Within the identified risk periods, the initial NIU time-series protocol was adjusted according to the anti-fluctuation enhancement protocol. Specifically, at the beginning of the risk period, the ultrafiltration rate was gradually reduced to smoothly transition to a lower level over a certain time; at the end of the risk period, the ultrafiltration rate was gradually increased to restore the initial protocol level. This smooth transition reduces blood volume fluctuations caused by abrupt changes in ultrafiltration rate. The adjusted ultrafiltration rate values ​​were arranged in chronological order to form a new NIU time-series protocol. This protocol exhibits more stable ultrafiltration rate changes during high-risk periods, reducing the impact on the pre-selected patient's blood volume and improving treatment safety and stability. During actual treatment, key indicators such as blood volume and serum sodium concentration are continuously monitored in the pre-selected patients. Based on real-time data, the net ultrafiltration rate timing protocol is dynamically adjusted to ensure the treatment process remains within a safe range. If fluctuations exceed expectations, the ultrafiltration rate is adjusted promptly to maintain stable blood volume in the pre-selected patients.

[0114] 2) The second implementation method:

[0115] ① A Long Short-Term Memory (LSTM) network was selected as the deep learning model. LSTM can effectively process time series data and capture blood volume tolerance curves and time dependencies during treatment. The preprocessed dataset was divided into training, validation, and test sets. The LSTM model was trained using the training set. The model parameters were adjusted to minimize the difference between predicted and actual values ​​until the model's performance on the validation set stabilized, resulting in a well-trained LSTM model.

[0116] ② Blood volume tolerance curves are predicted for pre-defined patients. Based on pre-defined treatment goals (target dehydration volume and target serum sodium concentration), the predicted curves and treatment goals are input into a trained LSTM model. The model generates a net ultrafiltration rate time series plan based on learned knowledge and patterns. The plan comprehensively considers the patient's safety tolerance boundaries, dehydration needs, and serum sodium concentration control requirements. Finally, clinicians review and evaluate the plan, adjusting it as needed based on the actual situation of the pre-defined patients before clinical application, achieving precise and personalized CRRT treatment.

[0117] The control module is used to automatically adjust the speed of the blood pump and the speed of the replacement fluid pump used for CRRT treatment of preset patients according to the net ultrafiltration rate timing scheme, and to adjust the dehydration rate of preset patients.

[0118] Specifically, based on the net ultrafiltration rate timing scheme, the speed of the blood pump and the speed of the replacement fluid pump used for CRRT treatment of preset patients are automatically adjusted. The specific implementation process is as follows:

[0119] First, the data output interface of the net ultrafiltration rate timing scheme is connected to the control system of the CRRT equipment. This ensures that the ultrafiltration rate data in the scheme can be transmitted to the equipment control system in real time and accurately. This may involve configuring the data communication protocol of the medical device, such as establishing a data transmission channel through medical information standard protocols like DICOM and HL7, or a dedicated API interface provided by the equipment manufacturer. After receiving the net ultrafiltration rate timing scheme data, the CRRT equipment control system parses the data. Based on the performance parameters of the blood pump and replacement fluid pump, the ultrafiltration rate is converted into the corresponding blood pump speed and replacement fluid pump speed. This requires pre-establishing a conversion model between the ultrafiltration rate and pump speed, based on the physical characteristics of the equipment and treatment requirements, such as obtaining a linear or non-linear relationship curve between the ultrafiltration rate and pump speed through experimental calibration. Based on the converted blood pump speed and replacement fluid pump speed, the equipment control system generates corresponding speed control commands. These commands are typically sent to the pump drive motor control unit in the form of digital signals. For example, for a pump using a stepper motor, the control system calculates the required step pulse frequency and number to drive the motor to the target speed. Upon receiving speed control commands, the drive motor control units of the blood pump and replacement fluid pump adjust the pump speed in real time. During treatment, as the net ultrafiltration rate timing scheme changes, the device control system continuously updates the speed control commands to achieve dynamic adjustment of the pump speed. Simultaneously, the control system monitors the actual pump speed in real time, ensuring speed accuracy through a feedback control mechanism. Throughout the automatic adjustment process, the device continuously monitors the patient's vital signs and treatment parameters, such as blood flow, ultrafiltration volume, and transmembrane pressure. If any abnormalities are detected, such as excessive blood pump speed leading to a drop in blood pressure, or abnormal replacement fluid pump speed affecting the replacement fluid flow rate, the device triggers an early warning mechanism and automatically adjusts the pump speed or suspends treatment according to a preset safety strategy to ensure the patient's safety.

[0120] The dehydration rate refers to the total amount of water removed from the patient's body through the filter per unit time, usually measured in milliliters per hour (ml / h) or liters per hour (L / h). It is equal to the difference between the filter waste pump flow rate and the flow rates of the replacement fluid (pre-dilution + post-dilution) and dialysate, reflecting the net ultrafiltration rate of the equipment. Clinical adjustments need to be made dynamically based on the patient's blood volume tolerance curve to avoid hemodynamic fluctuations.

[0121] The specific implementation process, which involves adjusting the preset dehydration rate for patients based on the net ultrafiltration rate timing scheme, is as follows:

[0122] Every five minutes, the net ultrafiltration rate is read from the time-series protocol, and the current rate is compared with the patient's real-time blood pressure, heart rate, and blood volume index. If all three are within the safe window, the original dehydration rate is maintained; if any indicator approaches the threshold, the rate is immediately reduced in 10% increments, and rechecked after one minute until the indicators stabilize. If the indicators continue to deteriorate, dehydration is paused and 50 ml of normal saline is reinfused, restarting at a lower increment after recovery. Conversely, if the indicators are better than the preset upper limit and remain so for ten minutes, the system slowly increases the rate in 5% increments, but never exceeding the daily peak value of the protocol. Each adjustment is recorded and uploaded to the cloud along with physiological data for physician review at any time. If more than five adjustments are made within two hours, the system automatically triggers an alarm, prompting manual intervention and pausing subsequent automated adjustments. The entire process is completed in a closed loop by the embedded chip without manual calculation, ensuring that the daily target dehydration volume is achieved while minimizing hemodynamic fluctuations.

[0123] In another embodiment, the dehydration rate for a preset patient is adjusted according to the net ultrafiltration rate timing scheme, and the specific implementation process is as follows:

[0124] Initially, the net ultrafiltration rate is fixed at 0.5 ml / kg·h as a safety baseline. After confirming "Tolerable?" is "Yes", the process enters a cycle of "increasing by 0.5 ml / kg·h every hour". If the real-time monitored blood pressure, heart rate, and blood volume are all stable, the system increases the rate incrementally until it reaches the daily upper limit of 3.0 ml / kg·h or an intolerance signal appears. When any indicator approaches the warning value, "Tolerable?" immediately flips to "No", the system immediately triggers a "Stop" command, freezes the current rate, and reduces it by 1.5 ml / kg·h within 30 seconds as a buffer; then, a rapid reassessment is performed every minute. If the indicators still do not recover, the rate continues to decrease in increments of 0.5 ml / kg·h until the lowest safe level of 0.5 ml / kg·h is reached; if the condition still worsens, dehydration is paused and 50 ml of normal saline is reinfused. Once hemodynamics stabilize, the rate restarts at 0.5 ml / kg·h. All changes in net ultrafiltration rate and their corresponding timestamps are written to the relevant index in real time, forming a traceable audit chain. If the cumulative number of adjustments exceeds five times within two hours, or if the net ultrafiltration rate oscillates between 3.0 ml / kg·h and 0.5 ml / kg·h for more than two cycles, the system automatically sends a red alert to the on-duty terminal, prompting manual review. From the initial 0.5 ml / kg·h to the upper limit of 3.0 ml / kg·h, each step requires a "tolerable" judgment; any "no" result will activate the "stop-reassess tolerance" sub-process, while a "yes" result allows the rate to continue increasing at a gradient of 0.5 ml / kg·h until the predetermined dehydration rate is reached.

[0125] Optionally, the above technical solution further includes: an adjustment module for: determining whether the deviation of the blood volume of the preset patient in one cycle exceeds a preset blood volume deviation threshold when performing CRRT treatment on the preset patient; if so, adjusting the net ultrafiltration rate timing scheme.

[0126] Real-time monitoring of blood volume changes is crucial when performing CRRT on pre-selected patients. First, a cycle is established to assess blood volume deviation. The duration of a cycle can be 15 minutes or 30 minutes, depending on the specific circumstances. Simultaneously, a blood volume deviation threshold is pre-set based on the pre-selected patient's specific condition and treatment goals. This threshold is typically determined based on the patient's baseline blood volume, hemodynamic stability, and clinician recommendations.

[0127] At the end of each monitoring cycle, the current blood volume data of the preset patient is obtained from the CRRT device or its supporting monitoring system. The current blood volume is compared with the target blood volume in the treatment goals to calculate the blood volume deviation. If the absolute value of the deviation exceeds the preset blood volume deviation threshold, it is determined that the net ultrafiltration rate timing scheme needs to be adjusted.

[0128] When making adjustments, the direction and magnitude of the blood volume deviation should be considered. If the blood volume is higher than the target value and exceeds the threshold, it indicates that the ultrafiltration rate is too low, and the net ultrafiltration rate needs to be appropriately increased to increase the amount of water removed and bring the blood volume closer to the target value. Conversely, if the blood volume is lower than the target value and exceeds the threshold, it may be necessary to reduce the net ultrafiltration rate to decrease the rate of water removal and prevent adverse reactions caused by low blood volume.

[0129] The adjusted net ultrafiltration rate timing protocol should be re-entered into the CRRT device's control system. The device will automatically adjust the speeds of the blood pump and replacement fluid pump according to the new protocol to achieve the adjusted ultrafiltration rate. Simultaneously, the patient's blood volume and related physiological indicators, such as blood pressure and heart rate, should be continuously monitored to ensure the safety and effectiveness of the adjusted treatment protocol. If the adjusted protocol still fails to bring the blood volume deviation back within the threshold range, further evaluation of the patient's condition may be necessary to consider other factors affecting blood volume, such as fluid intake or bleeding, and the treatment protocol should be adjusted accordingly.

[0130] Optionally, the above technical solution also includes a display and warning module, which is used to: display in real time the net ultrafiltration rate deviation between the net ultrafiltration rate in the net ultrafiltration rate timing scheme and the actual net ultrafiltration rate when performing CRRT treatment on the preset patient, and determine whether to issue a warning based on the real-time obtained net ultrafiltration rate deviation, and if so, issue a warning.

[0131] When performing CRRT on pre-designed patients, it is crucial to monitor and display in real time the deviation between the net ultrafiltration rate (NIFR) in the NIFR timing protocol and the actual NIFR. First, a real-time data acquisition system needs to be established to acquire the actual NIFR data during CRRT equipment operation, while simultaneously extracting the target NIFR value from the NIFR timing protocol. The difference between these two values, calculated using dedicated software or a monitoring interface, is the NIFR deviation, which is then displayed in real-time in a visually intuitive graphical or numerical format on the healthcare professional's interface, such as an electronic medical record system or the equipment display screen.

[0132] Simultaneously, a net ultrafiltration rate deviation threshold is set, determined based on the patient's specific condition, treatment goals, and historical data. If the real-time deviation consistently exceeds this threshold, the system will automatically identify a risk. An early warning mechanism is then activated, alerting medical staff through various means such as audible and visual warnings, pop-up notifications, or push notifications. Upon receiving the warning, medical staff immediately respond quickly according to the pre-set emergency procedures, investigating the cause, such as equipment malfunction or abnormal blood volume in the pre-set patient. They promptly take measures such as adjusting ultrafiltration parameters and checking tubing patency to ensure safe and effective treatment. This process forms a closed-loop monitoring system, guaranteeing the accuracy and safety of CRRT treatment.

[0133] Optionally, in the above technical solution, the sensor module is also used to: acquire physiological parameter monitoring data of the preset patient when performing CRRT treatment on the preset patient, predict the blood volume of the preset patient based on the physiological parameter monitoring data of the preset patient, and determine the risk level corresponding to the predicted blood volume.

[0134] When performing CRRT on pre-selected patients, real-time acquisition of physiological parameter monitoring data is a crucial step in accurately predicting blood volume and assessing risk. First, hemodynamic parameters such as stroke volume variability and pulse pressure variability are collected using bedside monitoring devices; these parameters directly reflect volume responsiveness. Simultaneously, respiratory parameters, including inferior vena cava respiratory variability, and laboratory test data such as creatinine and blood urea nitrogen are monitored over time; these data collectively provide a multi-dimensional reference for blood volume prediction.

[0135] The collected data is processed using deep learning models (such as LSTM). These models, trained on extensive historical data, can capture complex patterns of blood volume changes. Real-time monitoring data is input into the model to dynamically predict blood volume trends for pre-defined patients. Based on the predictions and pre-defined safe blood volume thresholds, a risk level is determined: green (normal), yellow (mild deviation), and red (severe deviation). If the predicted blood volume exceeds the safe range, the system automatically triggers an alert, prompting medical staff to intervene promptly. This process ensures the real-time nature and accuracy of CRRT treatment, effectively improving the safety and efficacy of the treatment.

[0136] The warning module is used to: determine whether to issue a warning based on the risk level corresponding to the predicted blood volume; if so, issue a warning based on the risk level corresponding to the predicted blood volume.

[0137] When performing CRRT on pre-selected patients, providing early warning based on the risk level corresponding to the predicted blood volume is a crucial safety measure. First, the system uses a deep learning model (such as LSTM) to predict the blood volume of the pre-selected patient in real time and compares the prediction result with a pre-defined safe blood volume range to determine the risk level. Risk levels are typically divided into three categories: low risk (green), medium risk (yellow), and high risk (red).

[0138] If the predicted blood volume is at a low-risk level, it means the patient's blood volume is within a safe range, no warning is needed, and treatment can continue as planned. When the predicted blood volume enters the medium-risk level, the system will activate an early warning mechanism, reminding medical staff to pay attention to the patient's condition via pop-up notifications or push notifications, and suggesting further monitoring and evaluation. At this time, medical staff should closely observe changes in the patient's physiological parameters and be prepared to intervene if necessary.

[0139] If the predicted blood volume reaches a high-risk level, the system will immediately trigger an emergency alert, quickly attracting the attention of medical staff through audible and visual warnings. At this time, medical staff must respond swiftly and take action according to the pre-set emergency procedures, such as adjusting the ultrafiltration rate, checking the equipment's operational status, or assessing the overall condition of the pre-set patient, to prevent the abnormal blood volume from causing serious harm to the patient.

[0140] The implementation of this early warning mechanism relies on accurate blood volume prediction models and timely monitoring data. Through real-time prediction and risk assessment, healthcare professionals can be given sufficient warning time before problems occur, thereby improving the safety and effectiveness of CRRT treatment, reducing complications caused by abnormal blood volume, and ensuring the safety of patients during the treatment process.

[0141] In the above embodiments, although the steps are numbered S1, S2, etc., they are only specific embodiments given by the present invention. Those skilled in the art can adjust the execution order of S1, S2, etc. according to the actual situation, which is also within the protection scope of the present invention. It can be understood that in some embodiments, some or all of the above embodiments may be included.

[0142] like Figure 2 As shown in the figure, an embodiment of the present invention provides a method for intelligent control of net ultrafiltration rate for CRRT treatment, comprising the following steps:

[0143] S1. Obtain CRRT treatment-related data for preset patients;

[0144] S2. Based on the patient's historical CRRT treatment data and CRRT treatment-related data, and using artificial intelligence technology, the patient's blood volume tolerance curve is predicted. Based on the predicted blood volume tolerance curve and the patient's preset treatment goal, a net ultrafiltration rate time series plan is generated.

[0145] S3. Based on the net ultrafiltration rate timing scheme, automatically adjust the speed of the blood pump and the speed of the replacement fluid pump used for CRRT treatment of the preset patient, and adjust the dehydration rate of the preset patient.

[0146] Optionally, the above technical solution further includes: when performing CRRT treatment on a preset patient, determining whether the deviation of the preset patient's blood volume within one cycle exceeds a preset blood volume deviation threshold; if so, adjusting the net ultrafiltration rate timing scheme.

[0147] Optionally, the above technical solution further includes: displaying in real time the net ultrafiltration rate deviation between the net ultrafiltration rate in the net ultrafiltration rate timing scheme and the actual net ultrafiltration rate when performing CRRT treatment on a preset patient, and determining whether to issue an early warning based on the real-time net ultrafiltration rate deviation; if so, issuing an early warning.

[0148] Optionally, the above technical solution also includes:

[0149] When performing CRRT on a pre-defined patient, the physiological parameter monitoring data of the pre-defined patient is obtained, and the blood volume of the pre-defined patient is predicted based on the physiological parameter monitoring data, and the risk level corresponding to the predicted blood volume is determined.

[0150] Whether to issue an early warning is determined based on the risk level corresponding to the predicted blood volume. If so, an early warning is issued based on the risk level corresponding to the predicted blood volume.

[0151] Optionally, in the above technical solution, the CRRT treatment-related data includes: hemodynamic parameters, electrolyte concentration, filter transmembrane pressure, blood flow velocity, and filter waste pressure.

[0152] It should be noted that the beneficial effects of the intelligent net ultrafiltration rate control method for CRRT treatment provided in the above embodiments are the same as the beneficial effects of the intelligent net ultrafiltration rate control device for CRRT treatment described above, and will not be repeated here. Furthermore, the system provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the system can be divided into different functional modules according to the actual situation to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process is detailed in the method embodiments, and will not be repeated here.

[0153] The modules described in the embodiments of this invention can be implemented in software or hardware. The names of the modules are not, in some cases, limiting the scope of the module itself.

[0154] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-mentioned intelligent control methods for net ultrafiltration rate in CRRT treatment. That is, an electronic device according to an embodiment of the present invention may include, but is not limited to: a processor and a memory; the memory is used to store the computer program; the processor is used to execute the intelligent control method for net ultrafiltration rate in CRRT treatment shown in any embodiment of the present invention by calling the computer program.

[0155] In one alternative embodiment, an electronic device is provided, such as Figure 3 As shown, Figure 3 The illustrated electronic device 4000 includes a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 4004 is not limited to one type, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0156] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. Processor 4001 may also be a combination that implements computational functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0157] Bus 4002 may include a path for transmitting information between the aforementioned components. Bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 4002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The bus 4002 is represented by only one thick line, but this does not mean that there is only one bus or one type of bus.

[0158] The memory 4003 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0159] The memory 4003 stores the application code (computer program) for executing the present invention, and its execution is controlled by the processor 4001. The processor 4001 executes the application code stored in the memory 4003 to implement the content shown in the foregoing method embodiments.

[0160] Among them, electronic devices can also be terminal devices, which can be any device that can install applications, including at least one of smartphones, tablets, laptops, desktop computers, smart speakers, smartwatches, smart TVs, and smart in-vehicle devices.

[0161] It should be noted that, Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0162] An embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the above-mentioned intelligent control methods for net ultrafiltration rate in CRRT treatment.

[0163] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc.

[0164] In an exemplary embodiment, a computer program product or computer program is also provided, comprising computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform any of the above-described intelligent control methods for net ultrafiltration rate in CRRT treatment.

[0165] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0166] It should be understood that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0167] The computer-readable storage medium provided in this invention can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EEPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0168] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0169] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0170] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0171] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0172] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A smart device for controlling the net ultrafiltration rate in CRRT treatment, characterized in that, It includes a sensor module, a data processing module, and a control module; The sensor module is used to: acquire CRRT treatment-related data for a preset patient; The data processing module is used to: predict the blood volume tolerance curve of the preset patient based on the patient's historical CRRT treatment data and CRRT treatment-related data, and generate a net ultrafiltration rate time series scheme based on the predicted blood volume tolerance curve and the preset treatment target of the preset patient. The control module is used to: automatically adjust the speed of the blood pump and the speed of the replacement fluid pump used for CRRT treatment of the preset patient according to the net ultrafiltration rate timing scheme, and adjust the dehydration rate of the preset patient.

2. The intelligent control device for net ultrafiltration rate in CRRT treatment according to claim 1, characterized in that, It also includes an adjustment module, which is used to: determine whether the deviation of the blood volume of the preset patient in one cycle exceeds a preset blood volume deviation threshold when performing CRRT treatment on the preset patient; if so, adjust the net ultrafiltration rate timing scheme.

3. The intelligent control device for net ultrafiltration rate in CRRT treatment according to claim 1, characterized in that, It also includes a display and warning module, which is used to: display in real time the net ultrafiltration rate deviation between the net ultrafiltration rate in the net ultrafiltration rate timing scheme and the actual net ultrafiltration rate when the preset patient is treated with CRRT, and determine whether to issue a warning based on the real-time obtained net ultrafiltration rate deviation, and if so, issue a warning.

4. The intelligent control device for net ultrafiltration rate in CRRT treatment according to claim 3, characterized in that, The sensor module is also used to: acquire physiological parameter monitoring data of the preset patient when performing CRRT treatment on the preset patient, predict the blood volume of the preset patient based on the physiological parameter monitoring data, and determine the risk level corresponding to the predicted blood volume. The display and warning module is used to: determine whether to issue a warning based on the risk level corresponding to the predicted blood volume; if so, issue a warning based on the risk level corresponding to the predicted blood volume.

5. A smart device for controlling the net ultrafiltration rate for CRRT treatment according to any one of claims 1 to 4, characterized in that, The CRRT treatment-related data include: hemodynamic parameters, electrolyte concentration, filter transmembrane pressure, blood flow velocity, and filter waste pressure.

6. A method for intelligent control of net ultrafiltration rate for CRRT treatment, characterized in that, include: Obtain CRRT treatment-related data for a predefined patient; Based on the historical CRRT treatment data and CRRT treatment correlation data of the preset patient, and using artificial intelligence technology to predict the blood volume tolerance curve of the preset patient, a net ultrafiltration rate time series scheme is generated according to the predicted blood volume tolerance curve and the preset treatment target of the preset patient. Based on the net ultrafiltration rate timing scheme, the speed of the blood pump and the speed of the replacement fluid pump used for CRRT treatment of the preset patient are automatically adjusted, and the dehydration rate of the preset patient is also adjusted.

7. The intelligent control method for net ultrafiltration rate in CRRT treatment according to claim 6, characterized in that, Also includes: When performing CRRT on the preset patient, it is determined whether the deviation of the preset patient's blood volume within one cycle exceeds a preset blood volume deviation threshold. If so, the net ultrafiltration rate timing scheme is adjusted.

8. The intelligent control method for net ultrafiltration rate in CRRT treatment according to claim 6, characterized in that, Also includes: The deviation between the net ultrafiltration rate in the net ultrafiltration rate timing scheme and the actual net ultrafiltration rate when performing CRRT treatment on the preset patient is displayed in real time, and an early warning is issued based on the real-time net ultrafiltration rate deviation. If so, an early warning is issued.

9. An electronic device, characterized in that, The device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the intelligent control method for net ultrafiltration rate for CRRT treatment as described in any one of claims 6 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the intelligent control method for net ultrafiltration rate in CRRT treatment as described in any one of claims 6 to 8.

Images