Multi-mode kidney patient home peritoneal dialysis management system

Abstract

The invention discloses a multi-mode kidney patient home peritoneal dialysis management system, and relates to the field of management systems, the system comprises a movable integrated dialysis monitoring and guiding instrument, the movable integrated dialysis monitoring and guiding instrument comprises a multi-mode sensing unit comprising a high-definition camera, a microphone, an electronic weightometer and a depth sensor; the intelligent processing unit is internally provided with an operation compliance detection module, a pollution risk early warning module and a self-adaptive guide module; the interactive execution unit comprises a display screen carrying a digital human image, a voice broadcast device and a touch feedback device; the physical supporting unit comprises a liftable foot stool with a roller and an infusion support with an electric height adjusting function; and the cloud platform communication module synchronizes dialysis data with a hospital server in real time, assists a household peritoneal dialysis patient to realize sufficient regular, flow-standardized and strictly sterile high-quality dialysis, and ensures long-term continuous and stable-effect household treatment through systematic supervision.

Description

Technical Field

[0001] This invention relates to management system technology, specifically to a multimodal home peritoneal dialysis management system for kidney patients. Background Technology

[0002] Patients undergoing home peritoneal dialysis are typically discharged after receiving guidance from hospital nurses and practicing on their own, then operate the dialysis system at home as prescribed by their doctor. The advantage of this treatment model is that it maximizes the patient's ability to resume normal work and life. However, patients commonly face challenges such as poor memory retention, lack of proficiency in the procedure, failure to develop good habits, and personal laxity. These problems often lead to complications such as inadequate dialysis (hospital test results not meeting standards), peritonitis, and peritoneal dialysis catheter infection, even endangering their lives. Some patients may even require catheter re-insertion surgery.

[0003] Based on this, the present invention provides a multimodal home peritoneal dialysis management system for kidney patients. Summary of the Invention

[0004] To address the problems mentioned in the background section, this invention provides a multimodal home peritoneal dialysis management system for kidney patients. This system assists patients in achieving high-quality dialysis at home through regular, standardized, and strictly aseptic procedures, and ensures long-term, stable, and effective home treatment through systematic supervision. On one hand, relying on intelligent end-to-end support, it provides real-time guidance and monitoring during patient operation, effectively reinforcing correct operating habits and mitigating the risk of errors. On the other hand, focusing on continuous incentives, the system continuously accompanies patients throughout their long home dialysis journey, helping them maintain operational enthusiasm, consolidate good habits, and play a vital role in encouragement and support, thus becoming a stable and healthy intelligent partner for patients.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a multimodal home peritoneal dialysis management system for kidney patients, comprising a portable integrated dialysis monitoring and guidance device, wherein the portable integrated dialysis monitoring and guidance device includes:

[0006] Multimodal sensing unit: includes a high-definition camera, microphone, electronic scale, and depth sensor;

[0007] Intelligent processing unit: Built-in operation compliance detection module, pollution risk early warning module and adaptive guidance module;

[0008] Interactive execution unit: equipped with a digital human-like display screen, voice broadcaster, and touch feedback device;

[0009] Physical support unit: height-adjustable stand with casters, electrically adjustable IV stand;

[0010] Cloud platform communication module: Synchronizes dialysis data with the hospital server in real time.

[0011] Furthermore, the operational compliance detection module includes:

[0012] The instrument localization submodule is based on an improved YOLOv7 target detection model, with the following loss function:

[0013]

[0014] Where, x i ,y i To predict the center coordinates of the bounding box (in pixels), the value range is [0, image width - 1] × [0, image height - 1].

[0015] These are the coordinates of the center of the actual bounding box (in pixels);

[0016] ω i h i To predict the bounding box width and height (in pixels), the constraint is ω. i >0,h i >0;

[0017] λ coord ,λ size λ is the weighting coefficient, with a default value of λ. coord =5,λ size =1;

[0018] Action compliance scoring submodule: The scoring formula for the trajectory of the disinfectant swab wiping path is as follows:

[0019]

[0020] Where, Ρt=(p z ,p y The coordinates of the real-time operation trajectory points are in millimeters, and the acquisition frequency is 30Hz.

[0021] qt=(q x ,q y () represents the coordinates of the standard operating trajectory point (in millimeters);

[0022] σ is the tolerance radius (in millimeters), with a typical value of 10 mm.

[0023] Furthermore, the pollution risk early warning module includes:

[0024] Turbidity dynamic analysis submodule: Extracts the variance ΔS of permeate saturation using the HSV color space. t The probability of infection is:

[0025]

[0026] Where, ΔS t The variance of the saturation of the translucent liquid is calculated using the HSV color space;

[0027] Leukocytes are the peripheral blood white blood cell count of a patient (unit: ×10). 9 / L), the clinical normal range is 4.0-10.0×10 9 / L;

[0028] Δt is the time interval between adjacent dialysis sessions (in hours);

[0029] ω1 and ω2 are clinical weighting coefficients, with empirical values ​​of ω1 = 0.35 and ω2 = 0.65.

[0030] Aseptic Area Contamination Alarm Submodule: When a non-sterile object is detected entering the sterile area, an audible and visual alarm is triggered and the operation process is paused.

[0031] Furthermore, the adaptive guidance module employs a temporal knowledge distillation (TKD) model, with the following loss function:

[0032]

[0033] Among them, y s y t Output probability f for the student / teacher model (t) This is a time-series feature vector;

[0034] The formula for dynamically adjusting the prompt frequency is:

[0035] T remind =T base ·(1+γe ComplianceScore );

[0036] The ComplianceScore is calculated based on a weighted average of on-time performance and step completion rate.

[0037] Furthermore, it also includes a closed-loop optimization module for dialysis effectiveness:

[0038] Dialysis adequacy calculation submodule: Based on the deviation δ=|W-W0| / W0 between the weight of the dialysis fluid W and the doctor's order set amount W0, when δ>15%, a pneumatic additional dialysis prompt is given;

[0039] Infection Risk Report Generation Submodule: Links perfusion fluid turbidity data with patient's historical temperature records to generate a peritoneal infection risk level report.

[0040] Furthermore, the digital human character interaction mechanism includes:

[0041] Dialect Adaptive Engine: Dynamically adjusts speech synthesis through dialect phoneme mapping table Φ(d) → PhonemeSet;

[0042] Achievement Incentive System: After N consecutive error-free operations, a virtual badge is activated, and a 3D rotating animation is generated on the screen.

[0043] Furthermore, dynamic monitoring methods for sterile areas include:

[0044] The boundary of the sterile area is defined using an image segmentation algorithm, and its mask function is:

[0045]

[0046] in, Image gradient;

[0047] k is the edge detection threshold;

[0048] Motion vector of an object crossing a boundary Real-time tracking, when An alert is triggered at any time.

[0049] Furthermore, cloud platform data processing methods include:

[0050] Generate multidimensional dialysis reports, including operation time heatmaps, statistics of high-frequency error steps, and compliance trend curves;

[0051] Blockchain-based electronic evidence storage: On-chain encrypted storage of the operation video hash value H(V) and weighing data H(W).

[0052] Compared with existing technologies, this invention provides a multimodal home peritoneal dialysis management system for kidney patients. This system helps patients achieve high-quality dialysis at home with regular, standardized, and strictly aseptic procedures, and ensures long-term, continuous, and stable home treatment through systematic supervision. On the one hand, relying on intelligent end-to-end support, it provides real-time guidance and monitoring during patient operation, effectively reinforcing correct operating habits and avoiding the risk of errors. On the other hand, focusing on continuous incentives, the system will continuously accompany patients throughout their long home dialysis journey, helping them maintain operational enthusiasm, consolidate good habits, and play an important role in encouragement and support, thus becoming a stable and healthy intelligent partner for patients. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0054] Figure 1 A flowchart of the multimodal home peritoneal dialysis management system for kidney patients provided in an embodiment of the present invention. Detailed Implementation

[0055] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0056] like Figure 1 As shown, this invention provides a multimodal home peritoneal dialysis management system for kidney patients, which includes a portable integrated dialysis monitoring and guidance device. The portable integrated dialysis monitoring and guidance device includes:

[0057] Multimodal sensing unit: includes a high-definition camera, microphone, electronic scale, and depth sensor;

[0058] Intelligent processing unit: Built-in operation compliance detection module, pollution risk early warning module and adaptive guidance module;

[0059] Interactive execution unit: equipped with a digital human-like display screen, voice broadcaster, and touch feedback device;

[0060] Physical support unit: height-adjustable stand with casters, electrically adjustable IV stand;

[0061] Cloud platform communication module: Synchronizes dialysis data with the hospital server in real time.

[0062] Specifically, through systematic supervision, the system ensures long-term, continuous, and stable home treatment. On the one hand, relying on intelligent full-process support, it provides real-time guidance and monitoring during patient operation, effectively reinforcing correct operating habits and avoiding the risk of errors. On the other hand, it focuses on continuous incentives. Throughout the patient's long home dialysis journey, the system will continue to accompany them, helping them maintain their operating enthusiasm, consolidate good habits, and play an important role in encouragement and support, thus becoming an intelligent partner for the patient's stable health.

[0063] Furthermore, the adjustable stand and electric IV stand are suitable for patients of different heights (150-190cm), the roller design meets the needs of mobile treatment, and the triple guidance of tactile (touch feedback), visual (digital human) and auditory (dialect voice) reduces the learning cost for elderly users.

[0064] Specifically, the operational compliance detection module includes:

[0065] The instrument localization submodule is based on an improved YOLOv7 target detection model, with the following loss function:

[0066]

[0067] Where, x i,y i To predict the center coordinates of the bounding box (in pixels), the value range is [0, image width - 1] × [0, image height - 1].

[0068] These are the coordinates of the center of the actual bounding box (in pixels);

[0069] ω i h i To predict the bounding box width and height (in pixels), the constraint is ω. i >0,h i >0;

[0070] λ coord ,λ size λ is the weighting coefficient, with a default value of λ. coord =5,λ size =1;

[0071] Action compliance scoring submodule: The scoring formula for the trajectory of the disinfectant swab wiping path is as follows:

[0072]

[0073] Where, Ρt=(p z ,p y The coordinates of the real-time operation trajectory points are in millimeters, and the acquisition frequency is 30Hz.

[0074] qt=(q x ,q y () represents the coordinates of the standard operating trajectory point (in millimeters);

[0075] σ is the tolerance radius (in millimeters), with a typical value of 10 mm.

[0076] By improving the YOLOv7 model, millimeter-level instrument tracking is achieved to avoid missing critical steps. The wiping path is sampled and scored at a high frequency of 30Hz to ensure that the disinfection range covers 100% of the standard area.

[0077] Specifically, the pollution risk early warning module includes:

[0078] Turbidity dynamic analysis submodule: Extracts the variance ΔS of permeate saturation using the HSV color space. t The probability of infection is:

[0079]

[0080] Where, ΔS t The variance of the saturation of the translucent liquid is calculated using the HSV color space;

[0081] Leukocytes are the peripheral blood white blood cell count of a patient (unit: ×10). 9 / L), the clinical normal range is 4.0-10.0×10 9 / L;

[0082] Δt is the time interval between adjacent dialysis sessions (in hours);

[0083] ω1 and ω2 are clinical weighting coefficients, with empirical values ​​of ω1 = 0.35 and ω2 = 0.65.

[0084] Aseptic Area Contamination Alarm Submodule: When a non-sterile object is detected entering the sterile area, an audible and visual alarm is triggered and the operation process is paused.

[0085] By calculating the weight deviation of the dialysis fluid in real time, the system automatically triggers additional dialysis prompts, ensuring that the dosage of each treatment meets the doctor's orders. Combining the turbidity of the dialysis fluid with the patient's body temperature history, the system generates a quantitative risk report, allowing for the detection of signs of infection before clinical symptoms appear.

[0086] Specifically, the adaptive guidance module employs a time-series knowledge distillation (TKD) model, with the following loss function:

[0087]

[0088] Among them, y s y t Output probability f for the student / teacher model (t) This is a time-series feature vector;

[0089] The formula for dynamically adjusting the prompt frequency is:

[0090] T remind =T base ·(1+γe ComplianceScore );

[0091] The ComplianceScore is calculated based on a weighted average of on-time performance and step completion rate.

[0092] The temporal knowledge distillation model for dialysis scenarios applies the TKD model to dynamic physiological data modeling, solving the prediction drift problem of traditional static models in continuous treatment. The multi-module collaborative closed-loop design forms a complete treatment optimization closed loop from operation guidance (process control) → adequacy assessment (effect feedback) → infection early warning (risk prevention and control), filling the intelligent gap in home peritoneal dialysis.

[0093] This demonstrates that a new intelligent diagnosis and treatment paradigm of "precision dialysis - real-time intervention - proactive prevention and control" has been established, providing a safe and controllable home treatment solution for patients with end-stage renal disease.

[0094] Specifically, it also includes a closed-loop optimization module for dialysis effectiveness:

[0095] Dialysis adequacy calculation submodule: Based on the deviation δ=|W-W0| / W0 between the weight of the dialysis fluid W and the doctor's order set amount W0, when δ>15%, a pneumatic additional dialysis prompt is given;

[0096] Infection Risk Report Generation Submodule: Links perfusion fluid turbidity data with patient's historical temperature records to generate a peritoneal infection risk level report.

[0097] Specifically, the digital human character interaction mechanism includes:

[0098] Dialect Adaptive Engine: Dynamically adjusts speech synthesis through dialect phoneme mapping table Φ(d) → PhonemeSet;

[0099] Achievement Incentive System: After N consecutive error-free operations, a virtual badge is activated, and a 3D rotating animation is generated on the screen.

[0100] Specifically, dynamic monitoring methods for sterile areas include:

[0101] The boundary of the sterile area is defined using an image segmentation algorithm, and its mask function is:

[0102]

[0103] in, Image gradient;

[0104] k is the edge detection threshold;

[0105] Motion vector of an object crossing a boundary Real-time tracking, when An alert is triggered at any time.

[0106] The sterile zone is dynamically delineated by an image segmentation mask function, and real-time alarms are triggered for the motion vector of invading objects. Peripheral blood WBC values ​​are introduced as a variable for calculating the probability of infection, which improves the specificity of the early warning. When non-sterile objects invade, the process is immediately suspended, and risky operations are forcibly interrupted, which improves the response speed compared to traditional manual supervision.

[0107] Specifically, cloud platform data processing methods include:

[0108] Generate multidimensional dialysis reports, including operation time heatmaps, statistics of high-frequency error steps, and compliance trend curves;

[0109] Blockchain-based electronic evidence storage: On-chain encrypted storage of the operation video hash value H(V) and weighing data H(W).

[0110] By automatically generating operation heatmaps and statistics on high-frequency error steps, doctors can quickly locate patients' weaknesses. Operation video hash values ​​and weighing data are stored on the chain, and dialysis data is synchronized to the hospital server in real time. Doctors can adjust prescriptions based on cloud platform data, extending the follow-up visit cycle from 1 week to 1 month.

[0111] The multimodal home peritoneal dialysis management system for kidney patients provided by this invention achieves breakthrough progress in three dimensions: medical safety, patient experience, and resource conservation through a four-fold mechanism of "intelligent perception, dynamic early warning, personalized guidance, and closed-loop optimization," providing a quantifiable and scalable smart healthcare solution for home peritoneal dialysis.

[0112] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A multi-modal kidney patient home peritoneal dialysis management system, characterized in that, The application discloses a mobile integrated dialysis monitoring guide device, which comprises a mobile integrated dialysis monitoring guide device. A multi-modal sensing unit includes a high-definition camera, a microphone, an electronic scale, and a depth sensor. An intelligent processing unit includes an operation compliance detection module, a pollution risk early warning module, and a self-adaptive guidance module. An interactive execution unit includes a display screen with a digital human image, a voice announcer, and a touch feedback device. A physical support unit includes a liftable footrest with a roller and an electrically adjustable height infusion stand. A cloud platform communication module synchronizes dialysis data with a hospital server in real time.

2. The multi-modal home peritoneal dialysis management system for kidney patients as claimed in claim 1 wherein, The operation compliance detection module includes: An instrument positioning sub-module based on an improved YOLOv7 target detection model has a loss function as follows: wherein x i ,y i are the predicted bounding box center coordinates (in pixels) with the range [0, image width-1] x [0, image height-1]; is the true bounding box center coordinate (in pixels); ω i , h i are the predicted bounding box width and height in pixels, respectively, with constraints ω i > 0, h i > 0; λ coord ,λ size is a weighting factor, the default value λ coord = 5, λ size = 1 ; An action compliance scoring sub-module has a trajectory scoring formula for disinfectant cotton wiping paths as follows: wherein Ρt= (p z ,p y ) is the real-time operating trajectory point coordinate (in millimeters), and the acquisition frequency is 30 Hz; qt= (q x , t) y ) are the standard operating trajectory point coordinates (in millimeters); σ is a tolerance radius (in millimeters), and a typical value is 10 mm.

3. The multi-modal home peritoneal dialysis management system for kidney patients as claimed in claim 1, wherein, The pollution risk early warning module includes: Turbidity dynamic analysis submodule: extract the transmittance liquid saturation variance ΔS through the HSV color space t The infection risk probability of which is: where ΔS t is the variance of the transmitted liquid saturation, calculated through the HSV color space; leukocyte is the peripheral blood leukocyte count (unit: x 10 9 / L) of the patient, the clinical normal range is 4.0-10.0 x 10 9 / L; Δt is an adjacent dialysis time interval (in hours); ω1 and ω2 are clinical weight coefficients, and empirical values are ω1=0.35 and ω2=0.65; A sterile area pollution alarm sub-module triggers an audible and visual alarm and suspends the operation process when a non-sterile object is detected to enter the sterile area.

4. The multi-modal home peritoneal dialysis management system for kidney patients as claimed in claim 1, wherein, The self-adaptive guidance module adopts a time sequence knowledge distillation model (TKD), and a loss function thereof is as follows: where y s , y t are student / teacher model output probabilities, f (t) is a time-series feature vector; A prompt frequency dynamic adjustment formula is as follows: T remind = T base ·(1 + γe ComplianceScore ); A compliance score is calculated based on an operation punctuality rate and a step completion degree.

5. The multi-modal home peritoneal dialysis management system for kidney patients as claimed in claim 1 wherein, The application also includes a dialysis effect closed-loop optimization module. A dialysis adequacy calculation sub-module is based on a deviation δ=|W-W0| / W0 between a dialysate weight W and a prescribed amount W0, and when δ>15%, a pneumatic additional dialysis prompt is given. An infection risk report generation sub-module associates dialysate turbidity data and patient temperature history records to generate a peritoneal infection risk level report.

6. The multi-modal home peritoneal dialysis management system for kidney patients as claimed in claim 1 wherein, The digital human image interaction mechanism includes: A dialect adaptive engine dynamically adjusts speech synthesis through a dialect phoneme mapping table Φ(d)→PhonemeSet. An achievement incentive system activates a virtual medal after N consecutive error-free operations, and generates a three-dimensional rotating animation on the screen.

7. The multi-modal home peritoneal dialysis management system for kidney patients as claimed in claim 1 wherein, The sterile area dynamic monitoring method includes: An image segmentation algorithm is used to determine the boundary of the sterile area, and a mask function thereof is as follows: wherein is the image gradient; k is an edge detection threshold value. Object motion vector across a border Real-time tracking is performed, when an alert is triggered.

8. The multi-modal home peritoneal dialysis management system for kidney patients as claimed in claim 1, wherein, The cloud platform data processing method includes: A multi-dimensional dialysis report is generated, including an operation time length heat map, a high-frequency error step statistic, and a compliance trend curve. An electronic evidence based on a block chain is stored on a chain by encrypting an operation video hash value H(V) and weighing data H(W).

Images