Peritoneal dialysis device

Abstract

The invention relates to medical instruments, in particular to a peritoneal dialysis device which comprises a dialysate management unit, a sensor unit, a processor, a storage device, a peritoneal transfer model, an intelligent alarm management unit and a communication unit, and the dialysate management unit is configured to be used for perfusion of dialysate into the abdominal cavity of a patient and drainage of the dialysate from the abdominal cavity of the patient; the sensor unit is configured to be used for monitoring physiological parameters of a patient and dialysate characteristic parameters and receiving symptom information and comfort level scores input by the patient, the storage stores instructions executed by the processor, and the instructions enable the processor to be capable of dynamically adjusting treatment parameters of the dialysate management unit in real time in combination with a pre-trained peritoneal transfer model. The perfusion speed, the drainage speed or the dialysate temperature of the dialysate management unit are dynamically adjusted, so that the comfort level score of the patient is maintained to be above a preset threshold value, treatment parameters can be dynamically adjusted in real time to optimize the treatment effect, peritonitis risks are early warned, and the comfort level of the patient is improved.

Description

Technical Field

[0001] The present invention relates to medical appliances, and more particularly, to a peritoneal dialysis device. Background Art

[0002] Patients with end-stage renal disease (ESRD) need to undergo renal replacement therapy, and peritoneal dialysis (PD) is widely used due to its advantages such as convenient home operation. However, the current peritoneal dialysis technology still faces many challenges. First, the effectiveness of peritoneal dialysis is closely related to the peritoneal transport function of patients. Currently, it mainly relies on the periodic peritoneal equilibration test (PET), which requires patients to go to the hospital, and the result analysis depends on the doctor's experience. It is difficult to achieve real-time and personalized function evaluation, resulting in a lag in the adjustment of the dialysis regimen behind the changes in the peritoneal transport function of patients.

[0003] In addition, peritonitis is one of the most serious complications of peritoneal dialysis, and early diagnosis is crucial. Although the turbidity or color change of the drained fluid is an important early sign of peritonitis, currently, it mainly relies on visual inspection by patients or caregivers. This method is highly subjective, difficult to quantify and standardize, and may miss the best intervention time.

[0004] More importantly, patients often experience discomfort or even pain during peritoneal dialysis, which seriously affects treatment compliance. Traditional devices usually adopt fixed perfusion and drainage speeds, and it is difficult to dynamically adjust according to the real-time physiological state and subjective feelings of patients, exacerbating the discomfort. At the same time, during automated peritoneal dialysis (APD), the operating noise and necessary operations of the device may also interfere with the patient's sleep, further reducing the quality of life.

[0005] In summary, the deficiencies of the existing peritoneal dialysis technology in aspects such as peritoneal transport function monitoring, early warning of peritonitis, optimization of patient comfort, and intelligentization of telemedicine together constitute the technical problems that need to be solved urgently in the current peritoneal dialysis field. It is necessary to integrate multidisciplinary technologies to construct a more intelligent and individualized peritoneal dialysis management device. Summary of the Invention

[0006] Therefore, the purpose of the present invention is to provide a peritoneal dialysis device that can dynamically adjust treatment parameters in real time to optimize the treatment effect, early warn of the risk of peritonitis, and improve patient comfort.

[0007] To achieve the above purpose, the present invention provides the following technical solutions:

[0008] A peritoneal dialysis device, comprising:

[0009] A dialysate management unit configured to perform a treatment cycle, the treatment cycle including operations of perfusing dialysate into the patient's abdominal cavity and draining dialysate from the patient's abdominal cavity;

[0010] A sensor unit configured to monitor in real time a patient's physiological parameters, dialysate characteristic parameters, and receive symptom information and comfort scores input by the patient. The sensor unit includes a pressure sensor, a flow sensor, an optical sensor, and a patient input interface;

[0011] A processor communicatively connected to the dialysate management unit and the sensor unit;

[0012] A memory storing instructions executed by the processor, the instructions enabling the processor to perform the following operations:

[0013] Based on the patient's physiological parameters, dialysate characteristic parameters, and symptom information and comfort scores input by the patient monitored in real time by the sensor unit, and in combination with a pre-trained peritoneal transport model, dynamically adjust in real time the treatment parameters of the dialysate management unit to improve solute clearance and ultrafiltration volume. The peritoneal transport model calculates at least one dynamic correction factor by analyzing the patient's intra-abdominal pressure, dialysate volume, and patient activity status monitored in real time. The dynamic correction factor is used to adjust the basic peritoneal transport parameters or ultrafiltration parameters so that the error between the solute clearance predicted by the model and the actual clearance is less than 2%;

[0014] Based on the patient's physiological parameters, dialysate characteristic parameters, and symptom information input by the patient monitored in real time by the sensor unit, use a pre-trained gradient boosting decision tree classification model to analyze whether there are early risk indicators of peritonitis, and generate a graded warning signal when a risk is detected. The graded warning signal includes low, medium, and high risk prompts;

[0015] Based on the patient's intra-abdominal pressure monitored by the sensor unit, the comfort score input by the patient, and the heart rate variability obtained through an external wearable device, dynamically adjust the perfusion rate, drainage rate, or dialysate temperature of the dialysate management unit to maintain the patient's comfort score above a preset threshold.

[0016] The present invention is further configured such that the dynamic correction factor calculated by the peritoneal transport model includes a dynamic effective contact adjustment factor ,

[0017] ;

[0018] Its input includes the patient's intra-abdominal pressure monitored in real time , the volume of dialysate in the abdominal cavity , and the patient's activity status obtained through an integrated accelerometer , where , , are regression coefficients; is a time decay factor, The historical activity time window is 60 minutes.

[0019] The present invention is further configured such that the peritoneal transport model further calculates a dynamic water transport adjustment factor As one of the dynamic correction factors, the is used to adjust the basic ultrafiltration coefficient to obtain an adjusted ultrafiltration coefficient ;

[0020] ;

[0021] Its inputs include the patient's real-time monitored heart rate variability and the extracellular fluid to total body fluid ratio obtained by bioelectrical impedance analysis as well as the body weight change rate in the past 3 days , and the range of the dynamic water transport adjustment factor is limited to 0.8 - 1.2;

[0022] Wherein, ; ; ; 1.0, , , .

[0023] The present invention is further configured such that the optical sensor of the sensor unit is configured to monitor in real time the turbidity and color of the dialysate drained from the patient's abdominal cavity, and the turbidity and color data are used as input features of the gradient boosting decision tree classification model for identifying abnormal patterns of early peritonitis risk.

[0024] The present invention is further configured such that the patient input interface is configured to receive the abdominal pain score and fever grade input by the patient. The abdominal pain score is from 1 to 5, and the fever grade is normal, low fever, and high fever. The structured data of the abdominal pain score and fever grade are used as input features of the gradient boosting decision tree classification model.

[0025] The present invention is further configured such that the processor dynamically adjusts the perfusion speed or drainage speed according to the patient's intra-abdominal pressure monitored by the sensor unit and the comfort score input by the patient through the following rules:

[0026] If ;

[0027] If ;

[0028] If ;

[0029] Otherwise, ;

[0030] Calculate the adjusted speed:

[0031] ;

[0032] The present invention is further configured to: further include an intelligent alarm management unit;

[0033] The processor controls the intelligent alarm management unit to generate corresponding alarm prompts according to the risk level of the hierarchical early warning signal, and the alarm prompts include visual prompts for low risk, visual and auditory prompts for medium risk, and visual, auditory and remote notification prompts for high risk. The present invention is further configured to: The sensor unit is further configured to obtain heart rate variability and skin conductivity from an external wearable device, and the processor dynamically adjusts the dialysate temperature and the device operation volume by fusing heart rate variability, skin conductivity and the comfort score input by the patient through a fuzzy logic control algorithm.

[0034] The present invention is further configured to: The processor obtains the patient's sleep stage information through an external wearable device, specifically including light sleep, deep sleep and rapid eye movement sleep, and reduces the perfusion speed or drainage speed to a preset minimum value during the deep sleep stage to reduce interference with the patient's sleep.

[0035] The present invention is further configured to: further include a communication unit, and the communication unit is configured to encrypt and upload real-time monitoring data, hierarchical early warning signals and treatment parameters to a remote server, and receive treatment parameter adjustment instructions or model update instructions from the remote server to support remote monitoring and clinical data analysis.

[0036] Compared with the deficiencies of the prior art, the beneficial effects of the present invention are:

[0037] It is capable of identifying risk signals in advance based on physiological indicators, dialysis data and even data from external wearable devices (such as HRV, and sleep abnormalities may be related to inflammation or volume abnormalities), enabling patients or medical staff to take intervention measures earlier and effectively reducing the incidence and severity of complications.

[0038] It can intelligently adjust the dialysis cycle and parameters according to the patient's real-time status and individual preferences. For example, when the sleep quality is poor, it can fine-tune the dwell time or drainage speed, or provide personalized prompts or suggestions when the patient feels uncomfortable. The remote guidance function also facilitates patients to obtain professional support at home. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a module schematic diagram of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0040] Refer toFigure 1 This further describes an embodiment of a peritoneal dialysis device of the present invention, including: a dialysate management unit, a sensor unit, a processor, a memory, a peritoneal transport model, an intelligent alarm management unit, and a communication unit.

[0041] The dialysate management unit is responsible for controlling the flow, heating, and filtration of the dialysate and is the core part of implementing the peritoneal dialysis treatment cycle. It usually includes:

[0042] Pump system: A peristaltic pump or a diaphragm pump, used to precisely control the perfusion and drainage speed and flow rate of the dialysate. The pump system can achieve dynamic speed adjustment.

[0043] Valve system: Controls the on / off of different pipelines (fresh dialysate bag, waste liquid bag, pipeline connecting to the patient) to ensure that the fluid flows along the preset path.

[0044] Heating module: Heats the dialysate before perfusion to make its temperature close to body temperature (for example, in the range of 35 - 38 °C) to improve patient comfort and optimize transport. The temperature control of this heating module can be dynamically adjusted according to the comfort algorithm.

[0045] Pipes and connectors: The pipeline system connecting the dialysate bag, waste liquid bag, and the patient catheter, usually using a disposable sterile consumable package.

[0046] Weighing module: A weighing sensor for placing the dialysate bag and the waste liquid bag, used to monitor the change in the volume of the dialysate and assist the flow sensor in volume monitoring.

[0047] Under the control of the processor, the dialysate management unit sequentially executes operation cycles such as perfusion, dwell, and drainage according to a preset or dynamically adjusted treatment plan.

[0048] The sensor unit is responsible for obtaining multi-modal information on the patient, dialysate, and device status and providing a decision-making basis for the processor. The sensor unit includes:

[0049] Pressure sensor: Monitors the pressure in the pipeline or at the connection near the patient's abdominal cavity to obtain real-time intra-abdominal pressure data. This helps to evaluate the degree of abdominal distension and is an important parameter for comfort adjustment and input to the transport model.

[0050] Flow sensor: Monitors the flow rate and cumulative flow of the dialysate in the pipeline to precisely control the perfusion / drainage speed and monitor the volume of the dialysate in the abdominal cavity and the ultrafiltration volume.

[0051] Optical Sensor: Usually located on the drainage pipeline, it is used to monitor the turbidity and color of the drained dialysate in real-time or near real-time. Turbidity can be measured by transmitted or scattered light, and color characteristics can be measured by colorimetry or spectral analysis. These data are key inputs for early identification of the risk of peritonitis.

[0052] Patient Input Interface: Provides an interface for the patient to interact with the device, such as a touch screen or physical buttons. The patient can input symptom information (such as abdominal pain score, fever level) and comfort score through this interface. Symptom information can be designed in the form of a structured questionnaire, such as abdominal pain score (1 - 5) and fever level (normal, low fever, high fever, or specific body temperature value). The comfort score can also be a subjective numerical score (1 - 5).

[0053] External Device Interface: Used to receive biofeedback signals and sleep stage information from external wearable devices. It can connect to a heart rate belt, sleep bracelet, or smartwatch worn by the patient via Bluetooth or Wi-Fi to obtain heart rate variability (HRV), skin conductivity, and sleep stage information (such as light sleep, deep sleep, rapid eye movement sleep) identified by the device algorithm.

[0054] Temperature Sensor: Monitors the temperature of the dialysate, especially the temperature of the heated dialysate.

[0055] The processor is the control center of the device, responsible for executing instructions stored in the memory, processing data from the sensor unit, running intelligent algorithms and models, and sending control instructions to the dialysate management unit, communication unit, and alarm management unit. The processor can be a high-performance microcontroller or embedded processor.

[0056] Memory Storage: Operating system and application program instructions; preset treatment plans and parameters; raw data collected by sensors and processed feature data;

[0057] Pre-trained intelligent model parameters: including coefficients, thresholds, and rules of peritoneal transport models, complication risk classification models, and comfort control models, etc.; the patient's historical treatment data, physiological parameter trends, and warning history.

[0058] The communication unit is responsible for the secure data exchange between the device and external systems (such as remote servers, healthcare worker workstations, patient mobile phone apps). It usually includes:

[0059] Wireless Module: Supports communication standards such as Wi-Fi, Bluetooth, and cellular networks (such as 4G / 5G) to enable data upload and instruction reception.

[0060] Security Module: Implements functions such as data encryption and identity authentication to ensure the privacy and secure transmission of patient data, as well as the legitimacy of remote instructions.

[0061] The communication unit encrypts and uploads information such as real-time monitoring data, historical treatment data, model analysis results (such as dynamic correction factors, risk predictions), and warning signals to a remote server, supporting medical staff for remote monitoring and treatment plan evaluation. At the same time, it can receive instructions from the remote server, such as adjusting treatment parameters, updating the intelligent model, conducting remote software diagnosis, or updating.

[0062] The intelligent alarm management unit is responsible for generating alarm prompts of different levels according to the analysis results of the processor to attract the attention of patients or medical staff. The alarm prompts can include:

[0063] Visual prompts: For example, text information is displayed on the screen, color changes (such as green - low risk, yellow - medium risk, red - high risk), and indicator lights flash.

[0064] Auditory prompts: Beeping sounds or voice prompts with different frequencies, volumes, or rhythms.

[0065] Remote notification prompts: Send messages, push notifications, or text messages to the mobile apps of medical staff or patients through the communication unit.

[0066] The alarm management unit triggers corresponding alarm prompt methods according to the risk levels (such as low, medium, and high levels) of the warning signals generated by the processor. Low risk may only trigger screen visual prompts, medium risk triggers visual and auditory prompts, and high risk triggers visual, auditory, and immediately sends a remote notification.

[0067] The core function of the processor is to run a variety of intelligent algorithms to achieve personalized treatment adjustment, risk prediction, and comfort optimization.

[0068] The processor uses the real-time monitored patient physiological parameters and dialysate characteristic parameters, combines with the pre-trained peritoneal transport model, and dynamically adjusts the treatment parameters in real time.

[0069] Basic peritoneal transport model: This is a model that can predict the solute clearance rate and ultrafiltration volume under a given dialysate prescription (such as volume, glucose concentration, dwell time).

[0070] The processor obtains sensor data in real time and calculates the dynamic correction factor:

[0071] Dynamic effective contact adjustment factor : Used to adjust the basic parameters related to the effective peritoneal surface area or solute transport. Its calculation formula:

[0072] ;

[0073] The processor monitors in real time and and receives from the accelerometer Data. The current value is calculated through this formula. The model coefficients , , , and the time window T (T = 60 minutes) are obtained by training with a large amount of clinical data of patients to optimally capture , and the impact of the activity status on effective transportation.

[0074] Dynamic water transport adjustment factor : Used to adjust the basic ultrafiltration coefficient to obtain the adjusted ultrafiltration coefficient , and its calculation formula:

[0075] ;

[0076] The processor obtains the standardized HRV index s in real time. By analyzing the heart rate variability, it standardizes it to the interval, and obtains the ratio through the BIA sensor (the BIA sensor measures bioelectrical impedance, and the processor calculates the body fluid distribution using a preset algorithm, and the ratio is in the interval), and calculates the weight change rate in the past 3 days based on the recent weight data input by the patient or obtained by connecting a weighing scale (unit: kg / day, range . The model coefficients 1.0, , , are obtained through clinical data training, reflecting the predictive effect of these physiological indicators on the ultrafiltration ability. The function limits the adjustment factor within the range of [0.8, 1.2] to prevent over-adjustment.

[0077] Dynamic adjustment of treatment parameters: The processor updates the parameters of the peritoneal transport model using the calculated dynamic correction factor. Then, based on the prediction of the updated model, the processor can evaluate the current treatment effect, predicted solute clearance rate and ultrafiltration volume in real time or after each cycle. According to the preset treatment goals of the patient (target solute clearance rate, target ultrafiltration volume), the processor dynamically adjusts the operating parameters of the dialysate management unit, such as:

[0078] Dwell time: Dynamically adjusted according to the prediction of solute clearance rate.

[0079] Perfusion volume: Dynamically adjusted according to the prediction of ultrafiltration volume.

[0080] Glucose concentration: Dialysis solution bags with different glucose concentrations are dynamically selected according to the prediction of ultrafiltration requirements.

[0081] Number of cycles: Dynamically adjusted according to the achievement of total clearance and ultrafiltration targets.

[0082] The entire adjustment process aims to make the error between the solute clearance rate predicted by the model and the actual clearance rate less than a preset target (e.g., less than 5%), thereby improving the accuracy and effectiveness of treatment.

[0083] The processor uses multi-modal data input to run a pre-trained gradient boosting decision tree (GBT) classification model to analyze the early risk of peritonitis.

[0084] Model input features:

[0085] Turbidity data of the drained fluid monitored in real-time or near real-time, color data of the drained fluid monitored in real-time or near real-time.

[0086] Structured symptom information input by the patient, including abdominal pain score (e.g., a numerical value from 1 to 5) and fever level (e.g., mapped to a numerical variable: normal = 0, low fever = 1, high fever = 2).

[0087] Other potentially relevant physiological parameters (e.g., body temperature, pulse).

[0088] Pre-trained model: The GBT model is trained on a large amount of historical peritoneal dialysis patient data, which includes the patient's monitoring parameters, symptom information, and labels for whether peritonitis occurred. The model learns to identify abnormal patterns or combinations that may occur before the clinical symptoms of peritonitis are obvious.

[0089] Model output: The GBT model outputs a probability or score indicating the risk of peritonitis. The processor maps this output to a graded warning signal, e.g.:

[0090] Low risk: The model output score is lower than a certain threshold T1.

[0091] Medium risk: The model output score is between T1 and T2.

[0092] High risk: The model output score is higher than T2.

[0093] Warning signal generation: When medium or high risk is detected, the processor generates the corresponding warning signal and notifies the intelligent alarm management unit to trigger a graded alarm. Low risk can only be displayed as a prompt on the device interface.

[0094] The processor integrates objective physiological data and subjective patient input to dynamically adjust operation parameters related to comfort, such as perfusion rate, drainage rate, and dialysis solution temperature.

[0095] Speed adjustment: During the perfusion or drainage phase, the processor adjusts the speed according to the real-time intra-abdominal pressure and the comfort score input by the patient according to the rules.

[0096] Speed adjustment rules:

[0097] If the intra-abdominal pressure is too high or the comfort score is very low, indicating that the patient may feel obvious discomfort and the speed needs to be significantly reduced, so ;

[0098] If the intra-abdominal pressure is on the high side or the comfort score is relatively low, indicating that the patient may feel a certain degree of discomfort and the speed needs to be reduced, so ;

[0099] If the intra-abdominal pressure is low and the comfort score is high, indicating that the patient feels comfortable and the speed can be appropriately increased to shorten the treatment time, so ;

[0100] Otherwise, the intra-abdominal pressure and comfort are within the acceptable range, and the speed remains unchanged, so ;

[0101] Calculate the adjusted speed , limit the speed within the range of 50 - 200 mL / min, and control the dialysate management unit to perform perfusion or drainage. These adjustments can be made in a time-sharing manner within one perfusion / drainage cycle.

[0102] Temperature and device operating volume adjustment: The processor integrates inputs from multiple sources and adjusts the dialysate temperature and device operating volume through a fuzzy logic control algorithm.

[0103] Input fuzzification: Convert continuous or discrete input variables (such as heart rate variability obtained through an external wearable device, skin conductivity, and the comfort score input by the patient) into fuzzy sets. For example, convert the comfort score into fuzzy concepts such as "discomfort", "slight discomfort", "comfort", etc.

[0104] Fuzzy rule base: Establish a set of fuzzy rules based on expert experience or clinical data:

[0105] If the patient is "discomfort" and "skin conductivity is high", then "reduce the temperature" and "reduce the volume".

[0106] If the patient is "slightly discomfort" and "HRV is low", then "slightly reduce the temperature" and "slightly reduce the volume".

[0107] If the patient is "comfortable", then "maintain the temperature" and "maintain the volume".

[0108] Fuzzy inference: Based on the current fuzzy input and the rule base, perform inference to obtain a fuzzy output (e.g., the degree of "lowering the temperature").

[0109] Defuzzification: Convert the fuzzy output into an exact control signal to control the heating module to adjust the dialysis fluid temperature (within the range of 35 - 38 °C) and control the device volume (within the range of 30 - 60 dB). Fuzzy logic is particularly suitable for dealing with these control problems that contain subjective information and have complex relationships.

[0110] The processor receives the patient's sleep stage information from an external wearable device. According to the sleep stage, the processor adjusts the operation timing and parameters of the dialysis fluid management unit to minimize the interference with the patient's sleep, especially during the deep sleep stage.

[0111] The processor continuously receives and identifies the patient's current sleep stage (light sleep, deep sleep, rapid eye movement sleep).

[0112] When the patient enters the deep sleep stage, the processor can perform the following operations:

[0113] Reduce the perfusion / drainage speed: Reduce the current perfusion or drainage speed to a preset minimum value (e.g., 50 mL / min) to reduce the stimulation of the patient caused by the pump noise and flow rate changes.

[0114] Suppress non - emergency alarms: Delay non - emergency device beeps or alarms until the patient enters a lighter sleep stage or wakes up.

[0115] Adjust the cycle interval: If possible, fine - tune the interval between cycles to avoid starting a new perfusion or drainage cycle during the expected peak of deep sleep.

[0116] Example 1 (full function): Dynamic model adjustment + early warning + comfort adjustment + sleep optimization. Dialysis fluid: mainly 1.5% glucose, dynamically switched to 2.5% or 4.25%.

[0117] Example 2 (dynamic model): Dynamic model adjustment. Without early warning, comfort adjustment, sleep optimization. Dialysis fluid: mainly 2.5% glucose, dynamically switched to 4.25%.

[0118] Example 3 (early warning + comfort): Early warning + comfort adjustment + sleep optimization. Without dynamic model adjustment (based on single - time PET). Dialysis fluid: Alternately use 1.5% glucose and icodextrin.

[0119] Control example 1 (fixed high glucose): Traditional APDC. Without dynamic adjustment, early warning, comfort adjustment, sleep optimization. Dialysis fluid: Fixed 2.5% glucose.

[0120] Comparative Example 2 (Fixed Low Glucose): Traditional APDC. Without dynamic adjustment, early warning, comfort adjustment, and sleep optimization. Dialysate: 1.5% glucose fixed.

[0121] Based on the above examples and comparative examples, production comparison table 1 is as follows:

[0122]

[0123]

[0124] Based on Table 1:

[0125] Example 1 (Full Function) and Example 2 (Dynamic Model), by adopting the strategy of dynamically adjusting the peritoneal transport model parameters based on real-time monitoring data, can more accurately predict the real-time peritoneal transport characteristics of patients. This enables the device to dynamically adjust treatment parameters (such as dwell time, dialysate concentration) according to individual differences and physiological changes, thus significantly improving the average weekly Kt / V and average daily ultrafiltration volume, which is superior to the comparative examples with fixed parameters.

[0126] The low error rate of model prediction (solute clearance error < 5%) is the key to achieving precise dynamic adjustment, ensuring the reliability of the prediction results.

[0127] Higher solute clearance rate and ultrafiltration volume contribute to better removal of metabolic wastes and excess water in the body, improve uremia symptoms, more effectively maintain the body fluid balance of patients, and reduce the hospitalization rate caused by volume overload or deficiency.

[0128] 2. Early warning effectively reduces the incidence of peritonitis:

[0129] The gradient boosting decision tree classification model integrated in Example 1 (Full Function) and Example 3 (Early Warning + Comfort), by analyzing the characteristics of the drained fluid (turbidity, color) and the symptoms input by the patient (abdominal pain, fever), can effectively identify the early risk indicators of peritonitis.

[0130] The timely issuance of early warning signals (short time interval from early warning to diagnosis) provides valuable intervention time for medical staff and patients, resulting in a significant reduction in the incidence of peritonitis. High sensitivity and specificity ensure the accuracy of early warning, reducing false alarms and missed alarms.

[0131] 3. Greatly improve patient comfort and treatment compliance:

[0132] Example 1 (Full Function) and Example 3 (Early Warning + Comfort), by dynamically adjusting the perfusion / drainage speed, dialysate temperature, and device operation volume, significantly improve the average comfort score of patients.

[0133] Combined with the sleep stage optimization function, the device reduces the speed during the patient's deep sleep, reducing the interference with sleep and improving the sleep quality.

[0134] Higher comfort and better sleep quality directly lead to a significant reduction in the number of times patients interrupt treatment due to discomfort, thus significantly improving treatment compliance. Good compliance is the key to ensuring the success of long-term peritoneal dialysis treatment.

[0135] 4. Contributes to protecting residual renal function and reducing the risk of other complications:

[0136] Although the specific mechanism still needs further study, more stable fluid management, more accurate solute clearance, and a lower incidence of peritonitis may act together, resulting in a slower decline rate of residual renal function in the patients of Example 1 and Example 3.

[0137] Overall optimization of the treatment process and parameters helps to reduce the incidence of other peritoneal dialysis-related complications (such as exit site infection).

[0138] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any ordinary changes and substitutions made by those skilled in the art within the scope of the technical solution of the present invention should be included in the protection scope of the present invention.

Claims

1. A peritoneal dialysis device, characterized in that, Comprising: A dialysate management unit configured to perform a treatment cycle, the treatment cycle including operations of perfusing dialysate into a patient's peritoneal cavity and draining dialysate from the patient's peritoneal cavity; A sensor unit configured to monitor in real time a patient's physiological parameters, dialysate characteristic parameters, and receive symptom information and comfort scores input by the patient, the sensor unit including a pressure sensor, a flow sensor, an optical sensor, and a patient input interface; A processor communicatively connected to the dialysate management unit and the sensor unit; A memory storing instructions executed by the processor, the instructions enabling the processor to perform the following operations: Based on the patient's physiological parameters, dialysate characteristic parameters, and symptom information and comfort scores input by the patient monitored in real time by the sensor unit, combined with a pre-trained peritoneal transport model, dynamically adjust in real time the treatment parameters of the dialysate management unit to improve solute clearance and ultrafiltration volume, wherein the peritoneal transport model calculates at least one dynamic correction factor by analyzing the patient's intra-abdominal pressure, dialysate volume, and patient activity status monitored in real time, and the dynamic correction factor is used to adjust the basic peritoneal transport parameters or ultrafiltration parameters so that the error between the solute clearance predicted by the model and the actual clearance is less than 2%; Based on the patient's physiological parameters, dialysate characteristic parameters, and symptom information input by the patient monitored in real time by the sensor unit, use a pre-trained gradient boosting decision tree classification model to analyze whether there are early risk indicators of peritonitis, and generate a graded warning signal when a risk is detected, the graded warning signal including low, medium, and high risk prompts; Based on the patient's intra-abdominal pressure monitored by the sensor unit, the comfort score input by the patient, and the heart rate variability obtained through an external wearable device, dynamically adjust the perfusion rate, drainage rate, or dialysate temperature of the dialysate management unit to maintain the patient's comfort score above a preset threshold.

2. The peritoneal dialysis device according to claim 1, characterized in that, The dynamic correction factor calculated by the peritoneal transport model includes a dynamic effective contact adjustment factor , which ； Its inputs include the intra-abdominal pressure of the patient under monitoring , the volume of peritoneal dialysis fluid in the abdominal cavity , and the patient's activity status obtained by the integrated accelerometer , where , , are regression coefficients; is the time decay factor, is the historical activity time window of 60 minutes.

3. The peritoneal dialysis device according to claim 2, characterized in that, The peritoneal transport model also calculates a dynamic water transport adjustment factor As one of the dynamic correction factors, the is used to adjust the basic ultrafiltration coefficient to obtain the adjusted ultrafiltration coefficient ; ； Its input includes the patient's heart rate variability monitored in real time , the ratio of extracellular fluid to total body fluid obtained through bioelectrical impedance analysis , and the weight change rate in the past 3 days , and the range of the dynamic water transport adjustment factor is limited to 0.8 - 1.2; Among them, ; ; ; 1.0, , , .

4. A peritoneal dialysis device according to claim 1, characterized in that, The optical sensor of the sensor unit is configured to monitor in real time the turbidity and color of the dialysate drained from the patient's peritoneal cavity, and the turbidity and color data serve as input features of the gradient boosting decision tree classification model for identifying abnormal patterns of early peritonitis risk.

5. A peritoneal dialysis device according to claim 4, characterized in that, The patient input interface is configured to receive the abdominal pain score and fever grade input by the patient, the abdominal pain score is from 1 to 5, and the fever grade is normal, low fever, or high fever, and the structured data of the abdominal pain score and fever grade serve as input features of the gradient boosting decision tree classification model.

6. The peritoneal dialysis device according to claim 5, characterized in that The processor adjusts the perfusion rate or the drainage rate dynamically according to the intra-abdominal pressure of the patient monitored by the sensor unit and the comfort score input by the patient , by the following rules: If ; If ; If ; Otherwise, ; Calculate the adjusted speed: 。 7. A peritoneal dialysis device according to claim 6, characterized in that, It further includes an intelligent alarm management unit; the processor controls the intelligent alarm management unit to generate corresponding alarm prompts according to the risk level of the graded warning signal, and the alarm prompts include visual prompts for low risk, visual and auditory prompts for medium risk, and visual, auditory, and remote notification prompts for high risk.

8. A peritoneal dialysis device according to claim 7, characterized in that, The sensor unit is further configured to obtain heart rate variability and skin conductivity from an external wearable device, and the processor dynamically adjusts the dialysate temperature and the device operation volume by fusing the heart rate variability, skin conductivity, and the comfort score input by the patient through a fuzzy logic control algorithm.

9. A peritoneal dialysis device according to claim 8, characterized in that, The processor obtains the patient's sleep stage information through an external wearable device, specifically including light sleep, deep sleep, and rapid eye movement sleep, and reduces the perfusion rate or drainage rate to a preset minimum value during the deep sleep stage to reduce interference with the patient's sleep.

10. A peritoneal dialysis device according to claim 1, characterized in that, It further includes a communication unit configured to encrypt and upload real-time monitoring data, graded warning signals, and treatment parameters to a remote server, and receive treatment parameter adjustment instructions or model update instructions from the remote server to support remote monitoring and clinical data analysis.

Images