A multi-parameter adaptive control based hemodialysis ultrafiltration volume regulating system

Abstract

The present application relates to the technical field of hemodialysis, and discloses a hemodialysis ultrafiltration volume adjusting system based on multi-parameter adaptive control, which is used to solve the problem of insufficient prediction and prevention accuracy of dialysis-induced hypotension risk in the traditional system; the present application firstly synchronously collects blood volume, blood pressure and ion concentration multi-parameter physiological data in the whole dialysis process, then constructs a physiological state data set through unified standardization and correlation matching, extracts the related change mode of blood volume and blood pressure and superimposes ion deviation to form a comprehensive risk feature, compares the risk feature with a preset threshold through trend analysis, adaptively adjusts the weight of each risk component, generates a hypotension risk trend index, and accordingly classifies and dynamically adjusts the ultrafiltration rate, and is coupled with real-time physiological data feedback in a closed loop, so as to improve the prediction accuracy and regulation refinement level of dialysis-induced hypotension.

Description

Technical Field

[0001] This invention relates to the field of hemodialysis technology, specifically to a hemodialysis ultrafiltration volume adjustment system based on multi-parameter adaptive control. Background Technology

[0002] In hemodialysis, ultrafiltration rate regulation is a core component, aiming to maintain fluid balance by removing excess fluid from the patient's body. However, traditional ultrafiltration rate regulation methods mainly rely on preset parameters and manual adjustments, which easily leads to the risk of intradialysis-induced hypotension (IDH), posing a significant clinical challenge. While current technologies have introduced multi-parameter monitoring (such as blood volume and blood pressure) to assist adaptive control, the accuracy of IDH prediction and prevention remains insufficient, resulting in increased patient discomfort or complications.

[0003] For example, CN111526792A discloses a closed-loop dialysis treatment method using adaptive ultrafiltration rate. This method adjusts the ultrafiltration pump rate by monitoring changes in blood volume in real time and achieves closed-loop regulation based on multi-parameter feedback (such as total blood volume change and working duration). It emphasizes the setting and dynamic adjustment of the initial ultrafiltration rate to avoid the risk of hypotension. However, this method mainly relies on threshold judgment and reactive feedback, resulting in low accuracy in early prediction of IDH. It cannot effectively integrate prospective models to intervene in potential risks in advance, leading to poor preventive effects in highly variable patients (such as those with cardiovascular disease). Ideally, however, fluctuations in the ultrafiltration rate can easily lead to sudden hypotension events. CN105999447B discloses a control method for hemodialysis and ultrafiltration. This method uses a flow control device combined with a pump and electronic scale to achieve weighing, display, and adjustment of the ultrafiltration volume, and provides auxiliary control based on multiple parameters (such as ion concentration and temperature). This improves the accuracy and reliability of ultrafiltration, but it also has limitations in predicting IDH. It relies solely on preset parameters and real-time monitoring for adjustments, lacking advanced prediction algorithms and failing to accurately predict the trend of blood pressure decline. This results in a passive prevention mechanism, and the incidence of IDH remains high in clinical applications.

[0004] While the aforementioned existing technologies have made progress in multi-parameter adaptive control, they generally suffer from insufficient accuracy in predicting and preventing the risk of dialysis-induced hypotension. Specifically, these methods mostly employ threshold or linear feedback models, which are difficult to handle nonlinear changes and sudden disturbances in patients' physiological parameters, resulting in low prediction sensitivity and delayed preventive response, thereby affecting the safety and efficacy of dialysis treatment. Therefore, a hemodialysis ultrafiltration volume regulation system based on multi-parameter adaptive control is needed to achieve accurate prediction and effective prevention of the risk of dialysis-induced hypotension, thereby improving treatment safety and patient comfort. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a hemodialysis ultrafiltration volume adjustment system based on multi-parameter adaptive control, which solves the problem of insufficient accuracy in predicting and preventing the risk of dialysis-induced hypotension in traditional methods.

[0006] To achieve the goals of improving treatment safety and patient comfort mentioned in the background section, the present invention provides the following technical solution:

[0007] A hemodialysis ultrafiltration volume regulation system based on multi-parameter adaptive control includes:

[0008] Parameter acquisition module: Acquires multi-parameter physiological data of patients during dialysis and synchronously transmits the data to the central control unit;

[0009] Physiological state fusion processing module: Standardizes and normalizes the collected multi-parameter physiological data, correlates and matches blood volume changes with blood pressure fluctuations, and fuses ion concentration data to generate a unified physiological state dataset;

[0010] Comprehensive risk assessment module: Extracts the correlation change patterns of blood volume and blood pressure from the physiological state dataset, superimposes ion concentration deviation factors, and constructs a comprehensive risk feature quantity for determining the risk of dialysis hypotension;

[0011] Dynamic trend analysis module: Performs low blood pressure risk trend analysis on comprehensive risk characteristics, compares the current risk characteristics with preset thresholds, dynamically calibrates the weights of each risk component based on the characteristic deviation, and calculates and generates a low blood pressure risk trend index.

[0012] Adaptive ultrafiltration control module: Automatically adjusts the ultrafiltration parameters of the dialysis machine based on the low blood pressure risk trend index. When a high risk trend is detected, the ultrafiltration rate is reduced, and the ultrafiltration rate is gradually restored after the risk is reduced. The adjustment process is coupled with the patient's real-time physiological feedback.

[0013] Closed-loop feedback control module: continuously monitors the patient's physiological state during the ultrafiltration parameter adjustment process, feeds back the actual execution results and updates the aforementioned processing modules to form a complete closed-loop regulation cycle.

[0014] In a preferred embodiment, acquiring multi-parameter physiological data during the patient's dialysis process and synchronously transmitting the data to the central control unit includes:

[0015] A data acquisition system integrated into the dialysis machine frame is used to collect multi-parameter physiological data of patients during dialysis.

[0016] The data acquisition system consists of a blood volume monitoring sensor, a blood pressure measuring device, and an ion concentration analyzer, which are used to periodically sample changes in blood volume, fluctuations in blood pressure, and target ion concentrations, and add a unified timestamp to the sampled data.

[0017] The analog or digital signals output by each sensor are converted from analog to digital and encapsulated into standard data packets, which are then sent to the central control unit via wired or wireless communication links.

[0018] The central control unit sorts and aligns data from different sources based on the timestamp of the data packets, generates composite data frames containing fields of blood volume, blood pressure, and ion concentration in chronological order, and writes them into the first-in-first-out data buffer.

[0019] In a preferred embodiment, the collected multi-parameter physiological data are standardized and normalized, blood volume changes are correlated and matched with blood pressure fluctuations, and ion concentration data are fused to generate a unified physiological state dataset, including:

[0020] Units were standardized and normalized for blood volume, blood pressure, and ion concentration data;

[0021] Blood volume is sequenced according to the initial baseline, and blood pressure is interpolated and aligned on a unified time axis after validity verification, and relevant time periods are marked according to rules to generate time frame paired data.

[0022] The ion concentration is converted to a normal range and fused with the paired data. The index is calculated according to the weights, a matrix dataset is constructed and output through a unified interface, and the window, threshold and weight parameter table configurations are persisted.

[0023] In a preferred embodiment, the correlation patterns of blood volume and blood pressure changes are extracted from the physiological state dataset, and ion concentration deviation factors are superimposed to construct a comprehensive risk feature quantity for determining the risk of dialysis hypotension, including:

[0024] Based on a unified physiological state dataset, time window scanning is performed on changes in blood volume and fluctuations in blood pressure to identify continuous decreasing segments and label relevant patterns according to preset thresholds.

[0025] Within the identified pattern range, the duration and standardized descent rate of each parameter are calculated, the pattern strength is constructed, and the start and end times and quantitative indicators are recorded.

[0026] The deviation of ion concentration during the model time period is statistically analyzed, an ion deviation factor is generated, and the model intensity is corrected to form a feature description containing enhancement intensity and deviation factor.

[0027] The risk feature extraction module updates new data at fixed intervals or triggered by acute changes, and writes the feature vectors into the feature buffer in chronological order for the trend analysis module to call.

[0028] In a preferred embodiment, a hypotension risk trend analysis is performed on the comprehensive risk characteristics. The current risk characteristics are compared with a preset threshold, and the weights of each risk component are dynamically adjusted based on the characteristic deviation. A hypotension risk trend index is then calculated and generated, including:

[0029] Read the comprehensive risk feature vector stored in chronological order and compare it with a preset threshold to obtain the corresponding deviation.

[0030] The weights of the mode enhancement intensity, ion deviation factor, and potential hypotension risk index are adjusted according to the deviation amount, and then the weights are normalized after adjustment.

[0031] Within a preset time window, each component is weighted and summed according to the updated weights, and the resulting sequence is smoothed to obtain a trend index of low blood pressure risk.

[0032] The trend analysis module operates normally within a set period. When the change in the comprehensive risk feature vector exceeds the acute change threshold, an additional fast update is performed.

[0033] In a preferred embodiment, automatically adjusting the ultrafiltration parameters of the dialysis machine based on a low blood pressure risk trend indicator, and reducing the ultrafiltration rate when a high-risk trend is detected, includes:

[0034] Trend analysis results are written to the parameter modification buffer in chronological order, and processing priorities are set according to risk level;

[0035] The control unit reads the treatment plan table and obtains the baseline ultrafiltration rate, upper and lower limits of the ultrafiltration rate, and the reduction coefficient.

[0036] When the risk score reaches the high-risk threshold and meets the criteria for blood volume or circulatory pressure deviation, the ultrafiltration rate is reduced according to a segmented decreasing strategy and sent to the pump controller.

[0037] In a preferred embodiment, the ultrafiltration rate is gradually restored after the risk decreases, and the adjustment process is coupled with real-time physiological feedback from the patient, including:

[0038] When the risk score is below the recovery threshold for several consecutive cycles and the trend is stable, the ultrafiltration rate is gradually increased according to the preset recovery step size and interval, and the step size can be paused or reduced.

[0039] The pump reads back the ultrafiltration rate and the latest blood volume, blood pressure and ion concentration data to perform closed-loop correction. Fine-tuning is performed when the rate deviation exceeds the error threshold.

[0040] Before implementing large-scale rate changes, joint verification of safety boundaries should be performed. When approaching the boundary, the adjustment range should be tightened or the observation interval should be extended, and a slow start and slow stop strategy should be adopted.

[0041] In a preferred embodiment, continuous monitoring of the patient's physiological state during ultrafiltration parameter adjustment includes:

[0042] During the dynamic adjustment of ultrafiltration parameters, multiple physiological data are continuously monitored, and the updated parameter baseline is loaded into the monitoring buffer.

[0043] The execution results after parameter adjustment are fed back to the acquisition module. The actual ultrafiltration rate returned by the pump controller is periodically read and compared with the target ultrafiltration rate to obtain the deviation. Based on this, the delay steady-state deviation or overshoot situation is identified.

[0044] Based on the identification results, the compensation tables for blood pressure channels and ion channels are simultaneously corrected, and the feedback channels are updated at a frequency of seconds.

[0045] Based on the feedback, the newly collected data is incrementally standardized, the standardized values ​​of blood volume and blood pressure are recalculated, and the ion deviation percentage is updated accordingly.

[0046] In a preferred embodiment, the actual execution results are fed back and updated to the aforementioned processing modules to form a complete closed-loop adjustment cycle, including:

[0047] The updated physiological state dataset was reused for key feature extraction and risk trend analysis. Relevant patterns were re-identified through incremental feature extraction, ion data was overlaid, and the generated features were appended to the existing feature vector sequence.

[0048] In the trend analysis step, the instantaneous trend value is calculated based on the incremental characteristics, and the trend curve is updated within the sliding time window to recalculate the corresponding risk score and risk level.

[0049] After completing the risk score and risk level update, the label information indicating whether the current state is high-risk or low-risk will be sent to the dynamic ultrafiltration parameter modification module.

[0050] When a cumulative trend of risk bias is detected over multiple analysis periods, the feedback adjustment magnitude is increased or the trend update interval is shortened. For high-risk patients, the trend analysis iteration cycle is shortened accordingly. At the same time, risk thresholds and grading parameters are configured based on clinical guidelines and historical data.

[0051] In the control unit, the acquisition compensation, standardization update, feature addition and trend refresh are executed sequentially according to the preset trigger cycle, and the above steps are repeated in a loop to form a closed-loop adjustment process of monitoring, analysis, adjustment and re-monitoring.

[0052] Compared with the prior art, the present invention provides a hemodialysis ultrafiltration volume adjustment system based on multi-parameter adaptive control, which has the following beneficial effects:

[0053] 1. This invention, by synchronously collecting physiological data of multiple parameters such as blood volume, blood pressure, and ion concentration throughout the dialysis process, and performing unified standardization and correlation matching within the control unit, constructs a physiological state dataset that reflects the coupling relationship between volume changes, hemodynamic fluctuations, and electrolyte imbalance. Based on this, it further extracts the correlation change patterns of blood volume and blood pressure, superimposes ion deviations to construct comprehensive risk features, and compares these features with preset thresholds through a trend analysis module, adaptively adjusting risk weights to generate a continuously evolving hypotension risk trend index. Then, based on the risk trend, it dynamically adjusts the ultrafiltration rate in stages, connecting the rate reduction strategy in the high-risk stage and the gradual recovery strategy in the low-risk stage with a real-time physiological feedback closed loop. This allows the ultrafiltration parameters to be finely adjusted as the risk evolves, thereby improving the accuracy of predicting and preventing dialysis-induced hypotension risk while ensuring dialysis adequacy, solving the problem of insufficient accuracy in predicting and preventing dialysis-induced hypotension risk in traditional methods.

[0054] 2. This invention divides multi-parameter data acquisition, standardization processing, risk feature extraction, trend analysis, and dynamic adjustment of ultrafiltration volume into clearly defined functional modules. A parameter configuration table centrally manages sampling period, threshold range, weighting coefficients, and deceleration recovery strategies, enabling medical staff to flexibly set individualized control strategies based on different dialysis prescriptions, dry weight assessment results, and patients' past adverse event records. Simultaneously, the system employs incremental updates and periodic triggering mechanisms within the control unit, reducing the computational burden of full recalculation and retaining traceable feature sequences and risk score records for easy post-event review and program optimization. Thus, without altering the existing dialysis machine's basic hardware structure, the system's engineering integrability and clinical operability are improved. Attached Figure Description

[0055] Figure 1 This is a structural diagram of a hemodialysis ultrafiltration volume adjustment system based on multi-parameter adaptive control according to the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] Example 1, Figure 1 A hemodialysis ultrafiltration volume regulation system based on multi-parameter adaptive control is presented, including:

[0058] Parameter acquisition module: Acquires multi-parameter physiological data of patients during dialysis and synchronously transmits the data to the central control unit;

[0059] Physiological state fusion processing module: Standardizes and normalizes the collected multi-parameter physiological data, correlates and matches blood volume changes with blood pressure fluctuations, and fuses ion concentration data to generate a unified physiological state dataset;

[0060] Comprehensive risk assessment module: Extracts the correlation change patterns of blood volume and blood pressure from the physiological state dataset, superimposes ion concentration deviation factors, and constructs a comprehensive risk feature quantity for determining the risk of dialysis hypotension;

[0061] Dynamic trend analysis module: Performs low blood pressure risk trend analysis on comprehensive risk characteristics, compares the current risk characteristics with preset thresholds, dynamically calibrates the weights of each risk component based on the characteristic deviation, and calculates and generates a low blood pressure risk trend index.

[0062] Adaptive ultrafiltration control module: Automatically adjusts the ultrafiltration parameters of the dialysis machine based on the low blood pressure risk trend index. When a high risk trend is detected, the ultrafiltration rate is reduced, and the ultrafiltration rate is gradually restored after the risk is reduced. The adjustment process is coupled with the patient's real-time physiological feedback.

[0063] Closed-loop feedback control module: continuously monitors the patient's physiological state during the ultrafiltration parameter adjustment process, feeds back the actual execution results and updates the aforementioned processing modules to form a complete closed-loop regulation cycle.

[0064] Acquire multi-parameter physiological data of the patient during dialysis and synchronously transmit the data to the central control unit. The specific implementation is as follows:

[0065] During the initiation phase of hemodialysis treatment, a multi-parameter physiological data acquisition system integrated within the dialysis machine frame is activated. This system consists of a blood volume monitoring sensor, a blood pressure measuring device, and an ion concentration analyzer, used to acquire real-time physiological indicators such as changes in blood volume, blood pressure fluctuations, and electrolyte levels. The blood volume monitoring sensor preferably uses a relative blood volume (RBV) monitor, which is installed at the outlet of the dialysis machine's blood circuit based on optical or ultrasonic principles and is fixedly connected to the patient's extracorporeal blood circulation pathway. It periodically scans changes in hemoglobin concentration through a non-invasive photoelectric detection module and calculates the percentage change curve of blood volume accordingly. In the initial stage of treatment, RBV data is collected at least once per minute with a resolution of not less than 1%, thereby forming a continuous record of blood volume changes.

[0066] The blood pressure measurement device is placed on the non-dialysis side of the patient's upper arm and can use either an intermittent oscillatory blood pressure cuff or a continuous blood pressure monitoring module. During intermittent measurements, a preset timer program controls the cuff to automatically inflate and deflate every 10 seconds, collecting readings of systolic blood pressure, diastolic blood pressure, and mean arterial pressure. During continuous measurements, the integrated sensor unit outputs the blood pressure waveform and corresponding parameters. To reduce the restriction of wired connections on patient movement, the blood pressure measurement device has a built-in Bluetooth or Wi-Fi communication module, wirelessly transmitting the measurement results to the dialysis machine control unit. The device also has a built-in calibration program, combining factory calibration data and periodic self-calibration procedures to control the blood pressure measurement error within ±3 mmHg, ensuring the reliability of the blood pressure data. The physiological limit threshold used to determine whether blood pressure is abnormal is preset by the system according to clinical guidelines, and medical staff can fine-tune it within a safe range through the parameter interface to adapt to different patients' baseline blood pressure levels.

[0067] An ion concentration analyzer is positioned in the dialysate circuit and connected to the dialysate sampler or bypass circulation unit. It periodically samples and detects the concentrations of target ions such as sodium and potassium ions using an ion-selective electrode. Each time the analyzer collects a sample of approximately 0.5 ml from the dialysate, outputting the corresponding concentration value via an electrochemical sensor. The detection range covers the normal physiological range, for example, sodium ions 135–145 mmol / L and potassium ions 3.0–5.5 mmol / L. When the measurement result exceeds the preset range, the analyzer generates an abnormality marker to indicate that the electrolyte state deviates from the baseline. The preset range is set by medical staff in the system parameter interface, referring to the recommended values ​​in the dialysis treatment guidelines, the equipment instruction manual, and the patient's baseline examination results upon admission. It can be dynamically adjusted according to the patient's electrolyte tolerance and treatment plan to adapt to different individual needs. The ion concentration sampling cycle is set to once every 5 seconds to track rapid changes in body fluid electrolytes.

[0068] The aforementioned blood volume monitoring data, blood pressure data, and ion concentration data are acquired in parallel through a unified acquisition control logic. To ensure the time comparability of different data sources, a unified clock is configured within the acquisition system, and a Network Time Protocol (NTP) or equivalent synchronization mechanism is used to align the local time of each sensor to a common time reference, and a millisecond-level timestamp is added to each data record. Furthermore, through this time synchronization mechanism, the RBV change value, blood pressure parameters, and ion concentration measurement results within the same second can be combined into a unified time frame at any time.

[0069] The analog signals output by each sensor are digitized via an analog-to-digital converter, or the digital signals are directly output by the signal processing unit inside the sensor, and then transmitted to the central control unit via wired or wireless means. Wired transmission can use USB, RS-485, or Ethernet interfaces, suitable for fixed connections within the dialysis machine; wireless transmission can use Wi-Fi or Bluetooth protocols, suitable for data exchange with peripherals such as wearable blood pressure devices. To ensure data integrity and anti-interference capabilities during transmission, the system encapsulates each frame of data into a standard data packet. This data packet includes header information, a payload field, and a checksum field. The header records the data type, source device number, and timestamp; the payload records the RBV value, blood pressure parameters, and ion concentration values; and the checksum field performs error detection. In the wireless transmission link, encryption algorithms such as AES-128 can be used to encrypt the data packets to prevent unauthorized interception or tampering during transmission, thus meeting the data security requirements of medical data.

[0070] The central control unit is located within the dialysis machine's control panel. It includes a multi-core processor, data buffer, and communication interface module, used to uniformly receive and integrate multi-source physiological data. Upon receiving data packets from different sensors, the control unit first sorts the data according to the timestamps carried in the packet header, aligns the data from different sources along the time axis, and generates composite data frames at the second or finer time granularity. Each composite frame includes at least one blood volume change indicator, a set of blood pressure indicators, and a set of ion concentration indicators. Subsequently, the control unit performs basic validity checks on each indicator. When it detects that blood pressure data exceeds the physiological limit threshold, ion concentration values ​​significantly exceed the sensor's range, or a certain field is missing, the corresponding record is marked as abnormal, and the data is filled in using linear interpolation or by preserving the previous value based on valid data from adjacent time points to maintain the continuity and integrity of the time series. The upper and lower limits used to determine whether the ion concentration exceeds the range are preset according to the sensor specifications and can be updated in the equipment maintenance interface.

[0071] To meet the real-time transmission and processing needs of large amounts of data during dialysis, the data buffer of the central control unit is designed with a first-in, first-out (FIFO) queue structure, with a capacity to store at least ten minutes of complete multi-parameter data streams. For example, at a sampling frequency of one frame per second, the buffer can store no less than 600 frames of composite data, each frame including fields such as timestamp, percentage change in RBV, systolic blood pressure, diastolic blood pressure, mean arterial pressure, sodium ion concentration, and potassium ion concentration. The control unit writes the latest data into the buffer according to a predetermined refresh cycle and releases the earliest data or transfers it to local non-volatile memory when the buffer is full, so as to perform historical backtracking analysis when needed. By optimizing the data queue management and processing scheduling strategy, the end-to-end delay from sensor sampling to data integration can be controlled within 500ms, thereby meeting the response requirements for real-time control during dialysis.

[0072] In clinical use, after selecting the treatment mode through the dialysis machine interface, the multi-parameter acquisition system is immediately started and enters continuous operation. The central control unit receives and processes physiological data from various sensors in real time, and uses the processed data as the basic input for subsequent processing modules. For patients assessed as being at high risk of hypotension, the sampling cycle of blood pressure and blood volume can be adjusted through the configuration interface to enhance the response to rapid fluctuations. Thus, the established multi-parameter acquisition link can provide continuous, synchronous, and configurable physiological data input during dialysis treatment.

[0073] The collected multi-parameter physiological data were standardized and normalized, blood volume changes were correlated and matched with blood pressure fluctuations, and ion concentration data were fused to generate a unified physiological state dataset. The specific implementation is as follows:

[0074] Based on the multi-source physiological data initially integrated by the parameter acquisition module, the central control unit calls the standardization processing submodule to perform unit unification, scale normalization, and format standardization operations on the blood volume change data, blood pressure data, and ion concentration data in sequence. For the blood volume change data, the standardization processing module, based on the relative blood volume percentage representation, uniformly represents the blood volume readings at each time point as a percentage change sequence with the initial blood volume as the baseline, and completes the numerical conversion in the control unit using floating-point arithmetic, so that the conversion error is controlled within 0.1%, thereby providing a consistent volume dimension basis for statistical normalization.

[0075] After completing the percentage conversion, the standardization module performs statistical normalization on the blood volume change data. For the blood volume percentage change sequence within the sliding time window, it calculates the mean and standard deviation of the window, and performs a subtraction of the mean and division by the standard deviation for each data point, converting the blood volume change into a standardized indicator with a mean of zero and a variance of one, thereby reducing the absolute level differences between different patients and between different stages of the same patient. The sliding window length is taken from the data within the past 5 minutes and can be set through the system configuration interface to adapt to different dialysis protocols. The normalization operation is completed within the data buffer without accessing external storage, which helps maintain the real-time performance of data processing.

[0076] Subsequently, the standardization processing module performs a standardization process similar to that for blood volume on the blood pressure fluctuation data. The control unit first checks the validity of the raw readings of systolic blood pressure, diastolic blood pressure, and mean arterial pressure, removing or marking abnormal readings that clearly exceed the physiological limit threshold as invalid. After removing outliers, the module calculates the mean and standard deviation of systolic blood pressure, diastolic blood pressure, and mean arterial pressure within the same time window, performs the same normalization operation as for blood volume on each data point, and uniformly expresses each blood pressure index as a dimensionless standardized value, so that the blood pressure data is statistically comparable to the blood volume standardized data.

[0077] After the blood volume change data and blood pressure fluctuation data are standardized, the standardization processing module performs time alignment and correlation matching between the two. The control unit, based on the timestamps recorded earlier, interpolates and resamples data from different sources along a unified time axis, generating paired data points at a time granularity of one frame per second. When a discrepancy is found between the blood pressure sampling time and the blood volume sampling time, the gap is filled by linear interpolation or by maintaining the most recent valid value, ensuring that standardized blood volume change values ​​and blood pressure indicators are obtained at every time point. Based on this, the standardization processing module uses preset logical rules to determine the correlation between blood volume and blood pressure. For example, when the standardized blood volume value continuously decreases within a certain time window and the decrease exceeds a preset blood volume change threshold, it checks whether the standardized mean arterial pressure value within the same window shows a downward trend, marking a highly correlated period where blood volume and blood pressure decrease simultaneously. The blood volume change threshold is preset in the system based on clinical consensus and dialysis treatment experience, and can be adjusted by medical staff within a safe range through a parameter interface to adapt to different patients' tolerance levels for blood volume fluctuations. Through the above processing, a preliminary joint dataset containing standardized blood volume data, standardized blood pressure data, and corresponding related labels is formed.

[0078] Based on the established correlation between blood volume and blood pressure, the standardized processing module incorporates ion concentration data into a unified processing framework. For ion concentration data, such as sodium and potassium ion levels, the current measurement value is first converted into a deviation index relative to the normal physiological range set earlier. For sodium ion concentration at each moment, the median of the normal range is used as a reference. The difference between the current value and the median is calculated and divided by the range width to obtain a dimensionless standardized value reflecting the degree of deviation. Potassium ion concentration is processed in the same way. The normal physiological range is set by medical staff in the system parameter interface with reference to the dialysis treatment guidelines, equipment manual recommendations, and the patient's baseline examination results upon admission. It can be adjusted during treatment based on laboratory test results and clinical assessments, so that the standardized index of ion concentration can accurately characterize the electrolyte balance status of different patients.

[0079] After blood volume, blood pressure, and ion concentration are standardized, the standardization processing module performs multi-parameter fusion in the fusion buffer within the control unit. For each time frame, the module constructs a vector containing standardized blood volume, blood pressure, and ion concentration values, and calculates a comprehensive physiological state index based on preset weighting coefficients. For example, the weight of blood volume change is set to 0.5, the weight of blood pressure fluctuation is set to 0.3, and the weight of ion concentration is set to 0.2. By summing the products of each standardized indicator and its corresponding weight, a comprehensive physiological state index is obtained, characterizing the degree of deviation of overall hemodynamics and electrolyte status at the current moment. These weighting coefficients are parameter configurations; their actual values ​​can be set and adjusted in the system according to different dialysis strategies and patient types. Medical staff can increase the weight of blood pressure-related indicators through the configuration interface to emphasize blood pressure stability, or increase the weight of blood volume and ion-related indicators to enhance sensitivity to blood volume control and electrolyte abnormalities.

[0080] After the fusion operation is completed, a unified physiological state dataset is generated and stored in the control unit's memory in matrix form. Each row of the matrix corresponds to a timestamp, and each column corresponds to a standardized physiological parameter or derived index, including standardized blood volume values, standardized blood pressure components, standardized ion concentration values, and comprehensive physiological state indices. Correlation label fields can also be added to mark time periods when blood volume and blood pressure are highly correlated. The number of rows in the matrix increases dynamically with the duration of dialysis treatment, and the number of columns is denoted as N to accommodate the expansion needs of new parameters and features under different implementation methods. The timestamps increment by the number of seconds from the start of dialysis, and the standardized values ​​are retained to three decimal places. The correlation labels use binary or enumeration encoding to quickly determine the correlation of corresponding time periods.

[0081] The standardization processing module provides a unified data access interface within the control unit, outputting the physiological state matrix in a structured manner, enabling subsequent modules to directly conduct analysis based on the standardized indicators. The sliding window length, normalization strategy, blood volume change threshold, physiological limit threshold, normal physiological range, and various fusion weights can all be set through the system configuration interface and persistently stored in the form of configuration files or parameter tables as the basis for standard parameters.

[0082] The correlation patterns between blood volume and blood pressure were extracted from the physiological state dataset, and ion concentration deviation factors were superimposed to construct a comprehensive risk feature for determining the risk of dialysis hypotension. The specific implementation is as follows:

[0083] The comprehensive risk assessment module is used to extract risk features within the dialysis machine control unit; its operation process is as follows: the module is triggered to run at a configurable predetermined period, for example, once every 30 seconds; at each trigger, it reads the latest data of the physiological state matrix output by the standardized processing module and performs pattern analysis and feature extraction for this period;

[0084] First, the module identifies the correlation pattern between changes in blood volume and fluctuations in blood pressure. The algorithm scans the physiological state matrix row by row in chronological order, tracking the trend of changes in standardized blood volume values ​​within a preset time window to determine whether there are continuous decreasing segments. When it is detected that the standardized blood volume value is lower than the previous frame for multiple consecutive frames, and the cumulative decrease within the window exceeds the preset minimum percentage change threshold for blood volume, the corresponding time period is marked as a candidate pattern for blood volume decrease. Subsequently, within the same time window, it checks whether the standardized values ​​of mean arterial pressure and systolic blood pressure show a decreasing trend, thereby screening for preliminary correlation patterns where blood volume decrease and blood pressure decrease occur simultaneously. The minimum change threshold for blood volume is preset in the system based on clinical consensus and dialysis experience, and medical staff can adjust it within a safe range through the interface to adapt to the sensitivity of different patients to blood volume fluctuations.

[0085] After initial pattern identification, the module performs quantitative analysis on each pattern to generate intensity indicators for subsequent evaluation: First, the pattern duration is calculated based on the pattern's start frame, end frame, and sampling interval. Then, within the time range corresponding to the pattern, the average rate of decrease in blood volume and the average rate of decrease in blood pressure are calculated respectively. The module uses the product of the duration and the average rate of decrease in blood volume as the pattern's base intensity component, and superimposes a correction component related to the average rate of decrease in blood pressure to obtain a comprehensive intensity value reflecting the severity of the pattern, thereby achieving joint quantification of volume and blood pressure changes. The quantification operation is performed in real-time within the central control unit, with intermediate results temporarily stored in memory, and the processing delay is controlled to the millisecond level to avoid affecting the continuity of the dialysis process. After quantification, the system records information such as start time, end time, duration, and comprehensive intensity value for each pattern, and writes it into the pattern feature buffer in vector or record structure form for subsequent use.

[0086] Next, the module overlays ion concentration data onto the identified blood volume and blood pressure-related patterns to construct a comprehensive risk profile that more closely reflects actual physiological conditions. Based on normal physiological ranges and deviation thresholds, the module determines whether the standardized values ​​of sodium or potassium ions at each time point in the physiological state matrix exceed preset deviation thresholds. When the absolute value of the deviation of any ion indicator exceeds the threshold, the corresponding time point is marked as an ion abnormality. For each quantified related pattern, the module calculates the proportion of ion abnormalities occurring within its time range and the average magnitude of ion deviation. The statistical results are combined to form an ion deviation factor, which is used to correct the intensity of the pattern.

[0087] When continuous ion abnormalities occur within a certain mode, or when the proportion of ion abnormalities reaches a preset ratio, the module amplifies the baseline intensity of that mode by multiplying the baseline intensity by an amplification factor greater than 1 and superimposing a correction term proportional to the ion deviation amplitude, thereby enhancing the comprehensive risk signal. Conversely, when ion indicators are within the normal range during that mode, the module maintains the mode intensity unchanged, or moderately lowers the intensity according to a pre-configured strategy to distinguish low-risk modes caused by slight volume changes or blood pressure fluctuations. The amplification factor and correction ratio are stored in the system as configurable parameters, with initial values ​​set to 1.2 or other empirical values. Medical staff can adjust these values ​​in conjunction with dialysis strategies and patient risk levels to achieve stratified differentiated amplification. Through the superposition and correction of ion data, each relevant mode forms an enhanced intensity description that simultaneously reflects the combined effects of blood volume changes, blood pressure fluctuations, and electrolyte abnormalities, thereby reducing the bias caused by a single parameter and demonstrating the contribution of multi-factor coupling to risk assessment.

[0088] After completing the ion superposition and intensity correction of all relevant modes, the comprehensive risk assessment module is used to form the basis for hypotension risk assessment: it integrates the enhanced features of multiple modes in chronological order, calculates the average or maximum intensity of the overlapping interval for modes that are adjacent or partially overlapping, so as to avoid the same risk event being counted repeatedly; and merges highly similar modes into a single event when necessary.

[0089] For each processing cycle, the module generates at least one set of comprehensive risk feature vectors, preferably containing three-dimensional components: mode enhancement intensity, ion deviation factor, and a potential hypotension risk index calculated based on the two. The risk index converts mode enhancement intensity and ion deviation factor into a unified risk scale for representation through a preset mapping relationship. The dimension of the feature vector can be expanded according to additional parameters, such as adding derived indicators such as heart rate and pulse pressure, to make the risk characterization more comprehensive. The generated feature vectors are written into the risk feature buffer in chronological order and passed to the downstream trend analysis module through a predefined interface.

[0090] To ensure real-time performance and stability while maintaining computational efficiency, the risk assessment module employs a combined periodic and event-triggered operating strategy. The module performs routine pattern recognition and feature updates on recent data at fixed intervals (approximately 30 seconds, adjustable), while simultaneously monitoring for acute events. When blood volume or blood pressure changes exceed a preset acute threshold, an additional rapid feature extraction is immediately initiated to capture the dynamics of sudden risks. The acute change threshold is preset in the system based on clinical experience, and healthcare professionals can adjust it within a safe range to achieve a balance between system sensitivity and false alarm rate. Through the combination of periodic analysis and instantaneous response mechanisms, the risk assessment module maintains a keen awareness of IDH risk evolution while controlling computational burden. After processing by this module, the system can mine related patterns of blood volume and blood pressure decline based on physiological state datasets and incorporate abnormal ion concentrations as regulatory factors into the quantitative results, thereby forming a comprehensive hypotension risk characteristic that can characterize the effects of volume changes, hemodynamic fluctuations, and electrolyte imbalances.

[0091] A hypotension risk trend analysis is performed on the comprehensive risk characteristics. The current risk characteristics are compared with preset thresholds, and the weights of each risk component are dynamically adjusted based on the characteristic deviation. A hypotension risk trend index is then calculated and generated. The specific implementation is as follows:

[0092] The dynamic trend analysis module is scheduled and operated by the central control unit. Each analysis cycle reads the most recent batch of comprehensive risk feature vectors from the risk feature buffer as input data for the current cycle, ensuring seamless integration with the risk feature output of the previous stage in terms of time and data structure. The risk feature vectors include at least components such as pattern enhancement intensity, ion deviation factor, and potential hypotension risk index, and are stored in the trend analysis buffer in chronological order. The trend analysis buffer can adopt a circular structure to retain the most recent feature sequence, such as historical data of at least 30 minutes, to support rolling calculations and trend updates. The trend analysis process first compares each comprehensive risk feature vector in the current cycle with pre-configured risk thresholds. These risk thresholds include an intensity threshold for determining whether pattern enhancement intensity is abnormal, and a risk index threshold for determining whether the potential hypotension risk index is abnormal. These two types of thresholds can be set identically or separately. Threshold parameters can be configured in the system based on dialysis-related clinical guidelines, historical case statistics, and empirical data, and can be adjusted by healthcare professionals within a safe range through the interface to adapt to the risk management strategies of different institutions and patient groups. Furthermore, the system maps the potential low blood pressure risk index to a range of 0 to 100 points and sets multiple graded thresholds in the configuration table to distinguish between low-risk, medium-risk, and high-risk levels; the trend analysis module reads the currently valid threshold configurations during runtime and uses them to determine the risk level of each feature vector.

[0093] During the specific comparison process, the trend analysis module examines each feature vector within the current period one by one: when any risk component exceeds the corresponding risk threshold, the feature vector is marked as exceeding the threshold state, and the deviation magnitude is calculated. The deviation magnitude can be the difference between the current value and the threshold, or the percentage excess relative to the threshold, to quantify the degree of deviation; for feature vectors that have not exceeded the threshold but are close to the threshold, they are marked as warning state to reflect the edge risk in the comprehensive assessment; based on the marking results and deviation calculation, the trend analysis module generates a deviation vector for each feature vector. The deviation vector includes at least the mode enhancement intensity deviation and the risk index deviation, and can be combined with the ion deviation factor to characterize the impact of electrolyte anomalies on the overall deviation, thereby providing a unified input basis for subsequent risk weight adjustments;

[0094] After obtaining the deviation vector, the trend analysis module adaptively adjusts the weights of each risk component in the trend calculation based on the magnitude of the deviation. In the initial state, the system presets the basic weight ratios of mode intensity, ion factor, and risk index based on the relative influence of parameters such as blood volume, blood pressure, and ions on dialysis-related hypotension (IDH). During module operation, the parameter weight calibration unit dynamically corrects the weights of each component based on the current deviation. For example, when the risk index deviation is consistently high for several consecutive cycles, the trend analysis module increases the weight of the blood pressure-related risk component, gradually enhancing its influence on the overall trend result, and correspondingly reduces the weight of the ion deviation factor to highlight the dominant role of blood pressure fluctuations in the formation of hypotension risk. Weight adjustment is achieved through proportional scaling, and the weights of each component are normalized after each adjustment to maintain a total weight of 1. The weight adjustment strategy and adjustment range support parameterized configuration, allowing medical staff to set initial weights and adjustment steps according to risk level and clinical preferences.

[0095] After obtaining the updated dynamic weights, the trend analysis module calculates the hypotension risk trend index. For each time point, each component of the feature vector at that moment is multiplied by the current weight coefficient and then summed to obtain the instantaneous risk trend value. To reduce the impact of short-term occasional fluctuations, the module smooths the instantaneous trend value sequence, which can be done using a moving average or an exponentially weighted moving average. Within a preset time window, the trend values ​​of multiple adjacent times are combined to generate a smooth risk trend curve. The time window length can be configured on the interface, for example, selecting data from the most recent analysis periods or several minutes. The system reads the current configuration and uses it for trend smoothing calculation. Through weighted calculation and smoothing, a hypotension risk trend index curve that changes continuously over time is finally obtained, so that the curve can be sensitive to short-term risk changes while avoiding interference from single-point anomalies to the overall judgment.

[0096] While generating risk trend indicators, the trend analysis module also comprehensively evaluates trend values ​​within a certain time range to form a risk score for internal control and display reference. At the end of each trend analysis cycle, the module can statistically analyze the smoothed trend values ​​of the most recent period, calculate the average, maximum, or high quantile of the trend values ​​within that interval, and map the statistical results to a single overall risk score. The risk score is mapped to a range of 0 to 100 points and is divided into low, medium, and high risk levels based on the configured risk threshold. The grading thresholds are stored in parameter form, allowing medical staff to adjust them in conjunction with institutional strategies, patient group characteristics, and past event rates. The module records the risk score, risk level, and key statistics calculated in each round, serving as the basis for control decisions by the downstream adaptive ultrafiltration control module. To facilitate internal system access and clinical review, the module generates structured trend analysis result data, including the overall risk score and level, trend indicator values ​​in time series form, and deviation statistics, such as the percentage of time exceeding the risk threshold during the analysis period, the maximum deviation, and the corresponding time period. The results are stored and forwarded in the control unit in JSON or equivalent structured format, facilitating parsing and display by the parameter adjustment module and the human-machine interface. The module is equipped with a dedicated output interface. After completing a new round of trend analysis, the latest results are pushed to the adaptive ultrafiltration control module through event triggering, thereby establishing a close linkage between risk assessment and ultrafiltration adjustment and control.

[0097] The trend analysis module employs a combined periodic and event-triggered update strategy. Periodic triggering performs routine trend updates at fixed time intervals, every 15 to 30 seconds, and supports parameterized configuration to achieve a balance between computational load and risk response speed. Event triggering initiates a rapid analysis process when a sudden change in the comprehensive risk feature vector is detected. When the increase in the potential hypotension risk index exceeds the acute change threshold within a single period, the module immediately performs focused calculations on the latest data and refreshes the risk trend indicators and risk scores within a short period. The acute change threshold can be preset based on clinical experience and historical data and can be adjusted within a safe range to balance system sensitivity and false alarm rate. Through these mechanisms, the trend analysis module can adaptively adjust the calculation weights in real time, continuously updating the risk trend curve and providing a more stable and reliable basis for the dynamic assessment of dialysis hypotension risk.

[0098] The ultrafiltration parameters of the dialysis machine are automatically adjusted based on low blood pressure risk trend indicators. When a high-risk trend is detected, the ultrafiltration rate is reduced, and when the risk decreases, the ultrafiltration rate is gradually restored. The adjustment process is coupled with the patient's real-time physiological feedback. Specifically, the implementation is as follows:

[0099] Based on the trend analysis results generated above, the control unit calls the dynamic ultrafiltration parameter modification module, taking the structured results output by the trend analysis module as input. The trend analysis results include at least the risk score and corresponding risk level calculated within a preset time window, as well as the hypotension risk trend indicators and deviation information arranged in chronological order. The control unit loads the above results into the parameter modification buffer in chronological order. The parameter modification buffer can adopt a queue structure with priority marking, assigning higher scheduling priority to records at high risk levels or with risk scores close to the high risk threshold, so that parameter adjustments can be completed first when high risk occurs, thereby ensuring the continuity and safety of the dialysis process.

[0100] Regarding parameter configuration, before dialysis treatment is initiated, medical staff configure the baseline ultrafiltration rate, minimum allowable ultrafiltration rate, and maximum ultrafiltration rate for each patient based on the dialysis prescription, the patient's dry weight assessment results, and the recommended operating range provided by the dialysis machine manufacturer. They also set the rate reduction coefficient, recovery step size, and related time intervals based on clinical experience. These parameters are stored in the control unit in the form of a treatment plan configuration table. The dynamic ultrafiltration parameter modification module reads these parameters directly during operation and uses them as boundary conditions and adjustment scales for subsequent rate reduction and recovery decisions to ensure that the ultrafiltration parameters are always adjusted within the preset safety range.

[0101] The dynamic ultrafiltration parameter modification module first performs a high-risk detection, scanning the risk score and corresponding risk level within the current analysis window. When the risk score for a certain time period exceeds the preset high-risk threshold and is determined to be at a high-risk level, that time period is defined as a high-risk interval. The high-risk threshold and risk level boundary follow the risk classification configuration mentioned earlier, which is given in the system as parameters and can be adjusted by medical staff through the configuration interface according to institutional strategies and patient population characteristics. After marking the high-risk interval, the module further combines deviation information to determine the source of risk: when the weight of the component related to blood volume and circulatory pressure in the deviation vector is greater than that of the component related to ion deviation, it is confirmed that the current risk of hypotension is mainly caused by changes in body fluids and hemodynamic abnormalities, and a decision to reduce the ultrafiltration rate is triggered accordingly.

[0102] During the high-risk response phase, the dynamic ultrafiltration parameter modification module lowers the current ultrafiltration rate according to a preset rate reduction strategy. When a high-risk zone is detected, the module reads the current baseline ultrafiltration rate and the minimum allowable ultrafiltration rate from the treatment plan configuration table, and calculates the lower limit of the target rate range based on the reduction coefficient, so that the current ultrafiltration rate is reduced by a preset proportion to a safe range close to the lower limit. Furthermore, a segmented reduction method can be used to perform the rate reduction: first, the rate is quickly reduced to a certain proportion of the target rate in a short period of time, and then the target value is gradually approached through several small adjustments to reduce the impact of rate mutations on hemodynamics. The rate reduction command generated by the module is sent to the dialysis machine's pump controller via the control bus. The pump controller completes the rate adjustment within a preset time limit based on the precision actuator. The time limit can be set to a range of several seconds to tens of seconds and can be configured according to equipment performance and clinical acceptance to achieve a balance between response speed and mechanical stability.

[0103] After completing the rate reduction action in the high-risk phase, the dynamic ultrafiltration parameter modification module continues to monitor subsequent risk trends and executes low-risk recovery logic. When the risk score given by the trend analysis module drops below the preset recovery threshold for multiple consecutive processing cycles, and the hypotension risk trend indicator continues to decrease or stabilizes at a low level, the module marks the current state as recoverable. The recovery threshold is set to a safe range value below the high-risk threshold and can be configured by medical staff in the system based on historical case data. To avoid frequent switching between high and low-risk states, the module can require that the ultrafiltration rate recovery process be initiated only after the risk score is simultaneously below the recovery threshold and the components of the deviation vector gradually decrease at several consecutive time points, thereby improving the stability of the adjustment process.

[0104] During the recovery phase, a gradual incremental strategy is employed to increase the ultrafiltration rate. The system configures the maximum allowable recovery rate, single-step recovery step size, and step application interval for the recovery process, with these parameters pre-set in the treatment plan. The dynamic ultrafiltration parameter modification module starts from the current ultrafiltration rate after reduction and increases the rate step size in a fixed or adaptive manner, without exceeding the maximum allowable recovery rate. For example, it increases the rate by a preset increment at fixed intervals until it recovers to the target range that matches the current risk level. If the trend analysis results show that the risk score is approaching the high-risk threshold again during the recovery process, or if the blood pressure-related component in the deviation vector increases significantly again, the module can automatically reduce the recovery step size or temporarily stop the recovery process to prevent a rebound in the risk of hypotension due to excessively rapid recovery. Through the above gradual recovery mechanism, the trajectory of ultrafiltration rate changes can be kept stable, thereby improving the patient's tolerance to parameter adjustments.

[0105] The entire ultrafiltration parameter modification process is coupled in a closed loop with real-time physiological data. After sending a deceleration or recovery command, the dynamic ultrafiltration parameter modification module periodically acquires the actual ultrafiltration rate fed back by the pump controller through the control unit, and simultaneously reads the latest real-time data such as blood volume, blood pressure, and ion concentration output from the physiological acquisition and standardization processing module mentioned above. The module compares the feedback rate with the current target rate. When the deviation exceeds the preset rate error threshold, it automatically generates a fine-tuning command to correct the pump controller, ensuring that the actual ultrafiltration rate closely follows the target adjustment curve. The rate error threshold and fine-tuning amplitude can be set by medical staff in the system parameter table according to the equipment accuracy and clinical goals, so that the closed-loop control maintains sufficient tracking accuracy while avoiding excessively frequent small adjustments.

[0106] Before implementing a significant reduction or recovery step size, the dynamic ultrafiltration parameter modification module can first perform multi-parameter joint verification. The module checks whether the current blood pressure is close to the preset lower limit, whether the blood volume change is still within the acceptable range, and whether the ion deviation is stable near the normal range. When any indicator is found to be close to the safety boundary, the module automatically tightens the ultrafiltration rate adjustment range or extends the observation interval between adjacent adjustments to form a smoother adjustment trajectory. To this end, the module introduces a slow start and slow stop strategy in the control logic to make the ultrafiltration rate change curve present a smooth change shape, reduce the impact of rate jumps on the circulatory system, and ensure the continuity and stability of dialysate flow during the transition from high-risk to low-risk state.

[0107] The patient's physiological state is continuously monitored during the ultrafiltration parameter adjustment process. The actual execution results are fed back and updated to the aforementioned processing modules to form a complete closed-loop regulation cycle. The specific implementation is as follows:

[0108] Once a dynamic adjustment (deceleration or acceleration) of ultrafiltration parameters is completed, the central control unit immediately enters continuous monitoring mode as part of the continuous operation of the closed-loop adjustment process. The system loads the parameter baseline information formed in this round of adjustment into the monitoring buffer. The parameter baseline includes at least the target ultrafiltration rate, the actual execution rate, and the corresponding rate deviation information recorded by timestamp. The monitoring buffer stores the above records in chronological order and can automatically expand its capacity according to the treatment duration to retain the parameter change trajectory and deviation evolution formed throughout the treatment process, which can be easily referenced multiple times in subsequent iterations.

[0109] Continuous monitoring first feeds back the parameter adjustment results to the physiological acquisition module. The control unit periodically reads the actual ultrafiltration rate returned by the pump controller and compares it point by point with the target ultrafiltration rate in the parameter baseline to generate a rate deviation, which is used to identify situations such as execution delay, steady-state deviation, or instantaneous overshoot. Subsequently, the system updates the software compensation parameters or calibration coefficients in the physiological acquisition module according to the deviation. For example, it adjusts the baseline offset used in blood volume monitoring before standardization so that the blood volume readings collected at the new ultrafiltration rate are referenced to the updated baseline in subsequent processing. At the same time, the internal compensation tables of the blood pressure acquisition and ion concentration acquisition channels are synchronously corrected to ensure that the output values ​​of each channel are consistent with the current operating state. The above feedback is transmitted through a two-way communication channel between the control unit and the acquisition module, with the update frequency preferably set to the second level, so that the acquisition module has absorbed the effects of the most recent round of parameter adjustment before the next sampling cycle arrives, thereby improving the continuity of closed-loop regulation.

[0110] After completing the compensation update on the acquisition side, the physiological state fusion processing module performs incremental standardization and matching processing on the newly entered data. It uses the normalization and association rules established earlier, without recalculating all historical data, but only updating the newly added time period and local intervals affected by parameter changes. Specifically, the module first uses the updated baseline and statistics to recalculate the standardization results for the newly added blood volume change data, ensuring numerical continuity with the existing time series. Subsequently, in the blood pressure fluctuation processing stage, while maintaining the original mean and variance framework, the module adds the feedback-generated deviation to the time alignment logic, ensuring the consistency of the correlation between blood volume and blood pressure after parameter adjustments, avoiding systematic shifts caused by changes in the ultrafiltration rate. For ion concentration data, the module updates the deviation percentage according to the latest compensation parameters, ensuring that the standardized ion indicators accurately reflect the current electrolyte state. After the above incremental update processing, the standardization processing module outputs the updated physiological state dataset (matrix form), with its timestamp index consistent with the previous data, which can be directly used as a unified, synchronized, and compensated new input for the subsequent comprehensive risk assessment module.

[0111] The risk feature extraction module performs incremental feature extraction based on the updated physiological state dataset. This module follows the relevant pattern recognition rules and comprehensive risk feature construction rules defined earlier, and rescans the joint change trend of blood volume and blood pressure for newly added time frames and time slices significantly affected by parameter adjustments, identifies new synchronous decline sequences or fluctuation patterns, and calculates the corresponding updated pattern intensity accordingly. For ion concentration components, when new deviations are identified, the module synchronously updates the ion-related superposition factors so that their impact on comprehensive risk features can be corrected in real time with changes in electrolyte state. The module only performs feature calculations on the newly added or affected data, and appends the results to the existing feature vector sequence in the form of feature vectors, keeping historical features unchanged, thereby ensuring the continuity and traceability of the feature sequence on the time axis.

[0112] After receiving the incremental feature vector, the trend analysis module calculates the instantaneous trend value of the new feature vector according to the weight adjustment and trend calculation mechanism established above, and updates the smoothed trend curve within a preset sliding window. Subsequently, based on the updated trend curve, it recalculates the risk score and risk level for the most recent period. If the newly calculated risk score exceeds the high-risk threshold configured above, or if the trend curve shows a significant increase in a short period of time, the trend analysis module sends a risk increase signal to the dynamic ultrafiltration parameter modification module through an event trigger, so as to prepare for new parameter adjustments in advance before the next round of periodic adjustment. If the new trend value continues to decline and stabilizes below the low-risk threshold, this state is used as an auxiliary reference for subsequent recovery strategies. The specific values ​​of the high-risk threshold and the low-risk threshold follow the risk grading parameter configuration mentioned above, which is comprehensively determined by clinical guidelines, historical case statistics, and institutional management strategies, and can be adaptively adjusted within the allowable range through the configuration interface.

[0113] By coordinating incremental standardized updates, incremental risk feature extraction, and incremental updates of trend results, the control unit constructs a continuously executable closed-loop adjustment process. This process runs during the main loop and is triggered according to the trigger cycles set by the feature extraction and trend analysis modules. The cycle is controlled on the order of tens of seconds and can be adjusted according to the patient's condition and equipment capabilities. For example, for patients at high risk of hypotension or those with acute conditions, the iteration cycle can be shortened to about 15 seconds to improve the overall response speed. Within each iteration cycle, the system sequentially completes the acquisition compensation update, standardized matrix update, feature vector appending, and trend analysis refresh, and then returns the latest risk score and risk trend indicators to the dynamic ultrafiltration parameter modification module, thus forming a closed-loop link of monitoring, analysis, adjustment, and remonitoring. When the main loop detects that the deviation shows an accumulating trend over multiple iteration cycles, the magnitude of the feedback compensation can be appropriately increased, or the internal update interval can be shortened, making the system more sensitive to risk changes and avoiding the slow accumulation of hypotension risk during dialysis.

[0114] In this embodiment, a central control unit and a multi-parameter acquisition system are set up inside the dialysis machine to acquire blood volume, blood pressure, and ion concentration data in real time. After analog-to-digital conversion and time alignment, the data are input into the control unit. The control unit standardizes the dimensions and scales, interpolates and aligns blood volume changes and blood pressure fluctuations on a unified time axis, converts ion concentration into a deviation index relative to the normal range, and fuses them according to preset weights to construct a matrix-form physiological state dataset. The risk analysis module identifies the synchronous decline pattern of blood volume and blood pressure within a sliding time window, calculates the pattern intensity, ion deviation factor, and hypotension risk index, and generates a risk trend index updated over time. The dynamic ultrafiltration control module reduces the ultrafiltration rate in high-risk stages based on the risk trend and preset baseline and upper / lower limit ultrafiltration rates, and gradually restores it step by step after the risk is relieved. It also combines pump feedback to correct the actual rate, performing closed-loop adaptive regulation of the dialysis process.

[0115] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0116] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0117] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0118] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0119] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0120] In addition, the functional modules in the embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0121] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0122] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0123] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-parameter adaptive control based ultrafiltration volume regulation system for hemodialysis, characterized in that, include: Parameter acquisition module: Acquires multi-parameter physiological data of patients during dialysis and synchronously transmits the data to the central control unit; Physiological state fusion processing module: Standardizes and normalizes the collected multi-parameter physiological data, correlates and matches blood volume changes with blood pressure fluctuations, and fuses ion concentration data to generate a unified physiological state dataset; Comprehensive risk assessment module: Extracts the correlation change patterns of blood volume and blood pressure from the physiological state dataset, superimposes ion concentration deviation factors, and constructs a comprehensive risk feature quantity for determining the risk of dialysis hypotension; Dynamic Trend Analysis Module: Performs hypotension risk trend analysis on comprehensive risk features, compares the current risk features with preset thresholds, dynamically calibrates the weights of each risk component based on feature deviations, calculates and generates a hypotension risk trend index, reads the comprehensive risk feature vector stored in chronological order, compares it with preset thresholds to obtain the corresponding deviation, adjusts the weights of mode enhancement intensity, ion deviation factor, and potential hypotension risk index according to the deviation, and normalizes each weight after adjustment. Within a preset time window, it performs weighted summation on each component according to the updated weights and smooths the resulting sequence to obtain the hypotension risk trend index. The trend analysis module runs normally within a set period. When the change in the comprehensive risk feature vector exceeds the acute change threshold, an additional fast update is performed. Adaptive ultrafiltration control module: Automatically adjusts the ultrafiltration parameters of the dialysis machine based on the low blood pressure risk trend index. When a high risk trend is detected, the ultrafiltration rate is reduced, and the ultrafiltration rate is gradually restored after the risk decreases. The adjustment process is coupled with the patient's real-time physiological feedback. When the risk score is below the recovery threshold for several consecutive cycles and the trend is stable, the ultrafiltration rate is gradually increased according to the preset recovery step size and interval. It allows pausing or reducing the step size. It reads the ultrafiltration rate from the pump and the latest blood volume, blood pressure, and ion concentration data for closed-loop correction. When the rate deviation exceeds the error threshold, fine-tuning is performed. Before making a large rate change, a joint verification of the safety boundary is performed. When approaching the boundary, the adjustment range is tightened or the observation interval is extended. A slow start and slow stop strategy is adopted. Closed-loop feedback control module: Continuously monitors the patient's physiological state during ultrafiltration parameter adjustment. During dynamic adjustment of ultrafiltration parameters, it continuously monitors multi-parameter physiological data and loads the updated parameter baseline into the monitoring buffer. It feeds back the execution results of parameter adjustment to the acquisition module. By periodically reading the actual ultrafiltration rate returned by the pump controller and comparing it with the target ultrafiltration rate, it obtains the deviation and identifies the delayed steady-state deviation or overshoot. Based on the identification results, it synchronously corrects the compensation tables of the blood pressure channel and ion channel and updates the feedback channel at a frequency of seconds. Based on the feedback, it performs incremental standardization processing on the newly acquired data, recalculates the standardized values ​​of blood volume and blood pressure, and updates the ion deviation percentage accordingly. The actual execution results are fed back and updated to the aforementioned processing modules to form a complete closed-loop regulation cycle.

2. The multi-parameter adaptive control based hemodialysis ultrafiltration volume regulation system according to claim 1, wherein, Acquire multi-parameter physiological data of the patient during dialysis and synchronously transmit the data to the central control unit, including: A data acquisition system integrated into the dialysis machine frame is used to collect multi-parameter physiological data of patients during dialysis. The data acquisition system consists of a blood volume monitoring sensor, a blood pressure measuring device, and an ion concentration analyzer, which are used to periodically sample changes in blood volume, fluctuations in blood pressure, and target ion concentrations, and add a unified timestamp to the sampled data. The analog or digital signals output by each sensor are converted from analog to digital and encapsulated into standard data packets, which are then sent to the central control unit via wired or wireless communication links. The central control unit sorts and aligns data from different sources based on the timestamp of the data packets, generates composite data frames containing fields of blood volume, blood pressure, and ion concentration in chronological order, and writes them into the first-in-first-out data buffer.

3. The hemodialysis ultrafiltration volume adjustment system based on multi-parameter adaptive control according to claim 1, characterized in that, The collected multi-parameter physiological data were standardized and normalized, blood volume changes were correlated and matched with blood pressure fluctuations, and ion concentration data were fused to generate a unified physiological state dataset, including: Units were standardized and normalized for blood volume, blood pressure, and ion concentration data; Blood volume is sequenced according to the initial baseline, and blood pressure is interpolated and aligned on a unified time axis after validity verification, and relevant time periods are marked according to rules to generate time frame paired data. The ion concentration is converted to a normal range and fused with the paired data. The index is calculated according to the weights, a matrix dataset is constructed and output through a unified interface, and the window, threshold and weight parameter table configurations are persisted.

4. The hemodialysis ultrafiltration volume adjustment system based on multi-parameter adaptive control according to claim 1, characterized in that, The correlation patterns between blood volume and blood pressure were extracted from the physiological state dataset, and ion concentration bias factors were superimposed to construct a comprehensive risk feature quantity for determining the risk of dialysis hypotension, including: Based on a unified physiological state dataset, time window scanning is performed on changes in blood volume and fluctuations in blood pressure to identify continuous decreasing segments and label relevant patterns according to preset thresholds. Within the identified pattern range, the duration and standardized descent rate of each parameter are calculated, the pattern strength is constructed, and the start and end times and quantitative indicators are recorded. The deviation of ion concentration during the model time period is statistically analyzed, an ion deviation factor is generated, and the model intensity is corrected to form a feature description containing enhancement intensity and deviation factor. The risk feature extraction module updates new data at fixed intervals or triggered by acute changes, and writes the feature vectors into the feature buffer in chronological order for the trend analysis module to call.

5. The hemodialysis ultrafiltration volume adjustment system based on multi-parameter adaptive control according to claim 1, characterized in that, The ultrafiltration parameters of the dialysis machine are automatically adjusted based on indicators of low blood pressure risk trends, and the ultrafiltration rate is reduced when a high-risk trend is detected, including: Trend analysis results are written to the parameter modification buffer in chronological order, and processing priorities are set according to risk level; The control unit reads the treatment plan table and obtains the baseline ultrafiltration rate, upper and lower limits of the ultrafiltration rate, and the reduction coefficient. When the risk score reaches the high-risk threshold and meets the criteria for blood volume or circulatory pressure deviation, the ultrafiltration rate is reduced according to a segmented decreasing strategy and sent to the pump controller.

6. The hemodialysis ultrafiltration volume adjustment system based on multi-parameter adaptive control according to claim 1, characterized in that, The actual execution results are fed back and updated to the aforementioned processing modules to form a complete closed-loop adjustment cycle, including: The updated physiological state dataset was reused for key feature extraction and risk trend analysis. Relevant patterns were re-identified through incremental feature extraction, ion data was overlaid, and the generated features were appended to the existing feature vector sequence. In the trend analysis step, the instantaneous trend value is calculated based on the incremental characteristics, and the trend curve is updated within the sliding time window to recalculate the corresponding risk score and risk level. After completing the risk score and risk level update, the label information indicating whether the current state is high-risk or low-risk will be sent to the dynamic ultrafiltration parameter modification module. When a cumulative trend of risk bias is detected over multiple analysis periods, the feedback adjustment magnitude is increased or the trend update interval is shortened. For high-risk patients, the trend analysis iteration cycle is shortened accordingly. At the same time, risk thresholds and grading parameters are configured based on clinical guidelines and historical data. In the control unit, the acquisition compensation, standardization update, feature addition and trend refresh are executed sequentially according to the preset trigger cycle, and the above steps are repeated in a loop to form a closed-loop adjustment process of monitoring, analysis, adjustment and re-monitoring.

Images