Remote intelligent home peritoneal dialysis system, peritoneal dialysis control method, terminal and medium

Abstract

The application discloses a remote intelligent home peritoneal dialysis system, a peritoneal dialysis control method, a terminal and a medium, and the system comprises a patient terminal device, a remote server and a mobile terminal. The patient terminal device is used for collecting and transmitting multidimensional data. The remote server is used for collecting the multidimensional data collected by the patient terminal device, and analyzing and processing the multidimensional data by using an AI model, and outputting early warning information, prediction results and treatment schemes. The mobile terminal is used for acquiring the early warning information, the prediction results and the treatment schemes pushed by the remote server, and generating visual data dashboards. The remote intelligent home peritoneal dialysis system provided by the application is beneficial to helping patients clearly know their own health conditions, and at the same time, doctors can monitor patients in an all-round way, which is beneficial to realizing the trinity of all-dimensional monitoring, AI intelligent decision and doctor-patient closed-loop management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent medicine, and in particular to a remote intelligent home peritoneal dialysis system, a peritoneal dialysis control method, a terminal and a medium. BACKGROUND

[0002] Peritoneal dialysis (PD) is an important way of end-stage renal disease replacement therapy. Home peritoneal dialysis can be divided into manual peritoneal dialysis and automatic peritoneal dialysis using an automated peritoneal dialysis machine (APD). However, current home peritoneal dialysis cannot be truly at home, and most patients still need to go back and forth to the hospital regularly to assess the treatment effect and peritoneal dialysis complications, which greatly consumes the patient's time and energy. And some patients with poor compliance have problems such as insufficient self-management and not going to the hospital for examination in time, which greatly reduces their quality of life and survival rate.

[0003] With the development of technology, automatic peritoneal dialysis is accepted by more and more patients due to its many advantages. The current APD machine is equipped with networking function, which can realize automatic recording of peritoneal dialysis treatment data and data transmission between the doctor's end. However, simple treatment data cannot enable the doctor to fully grasp the patient's treatment effect and health status, so as to realize remote individualized prescription adjustment and early warning of related complications; patients also cannot scientifically know their health status at home and communicate with doctors through effective ways. The problems of low home peritoneal dialysis treatment compliance, difficulty in monitoring the health status of patients and delayed response to emergency situations still need to be solved.

[0004] Therefore, the prior art still has defects. SUMMARY

[0005] The technical problem to be solved by the present application is to provide a remote intelligent home peritoneal dialysis system, a peritoneal dialysis control method, a terminal and a medium to solve the above defects of the prior art.

[0006] In a first aspect, the present application provides a remote intelligent home peritoneal dialysis system, wherein the system comprises:

[0007] A patient terminal device is configured to collect and transmit multi-dimensional data, and the patient terminal device comprises any one or more of an automated peritoneal dialysis machine, a human body composition analyzer, a non-invasive hemoglobin detector, a home biochemical analyzer, a body weight scale and a sphygmomanometer.

[0008] A remote server configured to collect multi-dimensional data collected by the patient terminal device, analyze and process the multi-dimensional data using an AI model, and output early warning information, prediction results, and treatment plans;

[0009] A mobile terminal configured to obtain early warning information, prediction results, and treatment plans pushed by the remote server, and generate a visual data dashboard.

[0010] In an implementation, the multi-dimensional data includes any one or more of ultrafiltration volume and drainage time recorded by a fully automatic peritoneal dialysis machine, whole-body fluid volume, fat volume, and muscle volume evaluated by a human composition analyzer, hemoglobin volume detected by a non-invasive hemoglobin detector, electrolytes, creatinine, urea, and body weight recorded by a home biochemical analyzer, daily pre- and post-dialysis body weight recorded by a body weight scale, and daily blood pressure recorded by a sphygmomanometer.

[0011] In an implementation, the mobile terminal includes a patient mobile terminal and a doctor mobile terminal, the patient mobile terminal obtains health reports and adverse event early warnings pushed by the remote server in real time through an APP, and the doctor mobile terminal is configured to provide a visual data dashboard to dynamically display patient multi-dimensional health index trends, mark abnormal data, and trigger early warning prompts.

[0012] In a second aspect, the embodiments of the present application also provide a peritoneal dialysis control method based on the remote intelligent home peritoneal dialysis system described in the above-mentioned scheme, and the method includes:

[0013] Collecting multi-dimensional data collected by a patient terminal device;

[0014] Collecting multi-dimensional data collected by the patient terminal device, and analyzing and processing the multi-dimensional data using an AI model to output early warning information, prediction results, and treatment plans;

[0015] Obtaining early warning information, prediction results, and treatment plans pushed by the remote server, and generating a visual data dashboard.

[0016] In an implementation, the analyzing and processing the multi-dimensional data using an AI model to output early warning information, prediction results, and treatment plans includes:

[0017] Performing multi-task prediction during peritoneal dialysis based on a CNN-Transformer-GRU-Attention joint model to obtain the prediction results and the early warning information;

[0018] Outputting treatment plans through an architecture combining a convolutional neural network and a fully connected layer.

[0019] In an implementation manner, the multi-task prediction in the peritoneal dialysis process based on the CNN-Transformer-GRU-Attention combined model obtains the prediction result and the early warning information, and includes:

[0020] The local features and the time sequence information of the multi-dimensional data are extracted based on the one-dimensional convolution layer in the convolutional neural network, important diagnostic information contained in the local features and dynamic change features in the time sequence information are obtained, and the hidden features of different sudden diseases are extracted by reconstructing the time sequence data through the Transformer model.

[0021] The Transformer model input part is coupled with the GRU layer after the hidden features of different sudden diseases are extracted.

[0022] The attention mechanism is added after the GRU layer, and different weight coefficients are given to the hidden features of different sudden diseases.

[0023] The final output values of different tasks are obtained by mapping through the Softmax function, and the prediction result and the early warning information are obtained according to the final output values, and the prediction result is used to reflect the occurrence probability of predicting different sudden diseases.

[0024] In an implementation manner, the architecture combined with the convolutional neural network and the fully connected layer outputs a treatment scheme, and includes:

[0025] The multi-dimensional data are sequentially subjected to one-dimensional convolution for feature extraction, batch normalization for data standardization, maximum pooling layer for down-sampling, and flattening layer for converting the features into one-dimensional vectors.

[0026] Three treatment parameters, i.e., daily exchange frequency, dialysis cycle time and dialysate concentration, are output through the fully connected layer.

[0027] The treatment scheme is generated according to the daily exchange frequency, the dialysis cycle time and the dialysate concentration.

[0028] In an implementation manner, the method further includes:

[0029] An intelligent consultation platform is integrated, a prescription is formulated or adjusted based on the treatment scheme, and parameters are set through a remote tool.

[0030] In a third aspect, the embodiments of the present application also provide a terminal, wherein the terminal includes a memory, a processor, and a peritoneal dialysis control program of a remote intelligent home peritoneal dialysis system stored in the memory and executable on the processor, and when the processor executes the peritoneal dialysis control program of the remote intelligent home peritoneal dialysis system, the steps of the peritoneal dialysis control method of the remote intelligent home peritoneal dialysis system in any of the above solutions are implemented.

[0031] In a fourth aspect, the embodiments of the present application also provide a computer readable storage medium, wherein the computer readable storage medium stores a peritoneal dialysis control program of a remote intelligent home peritoneal dialysis system, and the peritoneal dialysis control program of the remote intelligent home peritoneal dialysis system implements the steps of the peritoneal dialysis control method of the remote intelligent home peritoneal dialysis system according to any one of the above solutions on the computer readable storage medium.

[0032] Beneficial effects: Compared with the prior art, the present application provides a remote intelligent home peritoneal dialysis system, which comprises a patient terminal device, a remote server and a mobile terminal. The patient terminal device is used to collect and transmit multi-dimensional data. The remote server is used to collect the multi-dimensional data collected by the patient terminal device, analyze and process the multi-dimensional data using an AI model, and output early warning information, prediction results and treatment plans. The mobile terminal is used to obtain the early warning information, prediction results and treatment plans pushed by the remote server, and generate a visual data dashboard. The remote intelligent home peritoneal dialysis system provided by the present application is beneficial to helping patients clearly understand their own health status, while doctors can monitor patients comprehensively, which is beneficial to realizing the trinity of full-dimensional monitoring, AI intelligent decision-making and doctor-patient closed-loop management. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 The system framework diagram of the remote intelligent home peritoneal dialysis system provided by the embodiments of the present application.

[0034] Figure 2 The technical roadmap of AI prediction in the peritoneal dialysis control method based on the remote intelligent home peritoneal dialysis system provided by the embodiments of the present application.

[0035] Figure 3 The personalized prescription generation logic diagram in the peritoneal dialysis control method of the remote intelligent home peritoneal dialysis system provided by the embodiments of the present application.

[0036] Figure 4 The function diagram of the mobile terminal APP in the embodiments of the present application.

[0037] Figure 5 The flowchart of the peritoneal dialysis control method based on the remote intelligent home peritoneal dialysis system provided by the embodiments of the present application.

[0038] Figure 6 The principle block diagram of the terminal provided by the embodiments of the present application. DETAILED DESCRIPTION

[0039] For the purposes of the present application, the technical solutions and effects, the following will be further described in detail with reference to the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application, and are not intended to limit the present application.

[0040] The flowchart shown in the drawings is only an example and does not necessarily include all contents and operations or steps, nor does it necessarily execute in the order described. For example, some operations or steps can be further divided, combined or partially merged, so the actual execution order may be changed according to the actual situation.

[0041] It should be understood that the terms used in the present application specification herein are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the present application specification and the appended claims, unless otherwise clear from the context, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0042] It should be understood that in order to facilitate the clear description of the technical solutions of the embodiments of the present application, in the embodiments of the present application, the terms "first", "second" and the like are used to distinguish the same or similar items with basically the same function and effect. For example, the first control information and the second control information are only used to distinguish different control information, and do not limit the order.

[0043] Those skilled in the art can understand that the terms "first", "second" and the like do not limit the quantity and execution order, and the terms "first", "second" and the like do not necessarily mean different.

[0044] It should also be understood that the term "and / or" used in the present application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0045] The remote intelligent home peritoneal dialysis system of the present application comprises: a patient terminal device, a remote server and a mobile terminal. Figure 1As shown, the patient terminal device, remote server and mobile terminal of the embodiment can realize data transmission and interaction. The patient terminal device includes any one or more of a fully automatic peritoneal dialysis machine, a human body composition analyzer, a non-invasive hemoglobin detector, a home biochemical analyzer, a body weight scale and a sphygmomanometer. The patient terminal device is used for collecting and transmitting multi-dimensional data, and the collected multi-dimensional data is uploaded to the remote server. The remote server is deployed in the cloud based on an AI deployment mode, is used to collect the multi-dimensional data collected by the patient terminal device, analyzes and processes the multi-dimensional data using an AI model, outputs early warning information, prediction results and treatment schemes, and feeds back the early warning information, prediction results and treatment schemes to the mobile terminal. The mobile terminal includes a patient mobile terminal and a doctor mobile terminal, and is used to obtain the early warning information, prediction results and treatment schemes pushed by the remote server, and generate a visual data board.

[0046] Specifically, the patient terminal device of the embodiment deploys multiple devices, including an automatic peritoneal dialysis machine, a human body composition analyzer, a non-invasive hemoglobin detector, a home biochemical analyzer, a sphygmomanometer, a weight scale, etc. After the patient uses the automatic peritoneal dialysis machine for treatment every day, the automatic peritoneal dialysis machine records treatment parameters such as ultrafiltration volume and drainage time. Then the patient uses the human body composition analyzer to evaluate body composition such as total body fluid volume, fat mass and muscle mass. The non-invasive hemoglobin detector is used for quantitative detection of hemoglobin, and the hemoglobin volume is a direct correlation index for judging whether there is renal anemia. The sphygmomanometer is used to record daily blood pressure. The body weight scale is used to record the body weight before and after dialysis every day. The thermometer is used to record the body temperature. The home biochemical analyzer is used to detect important indicators such as electrolytes, creatinine and urea by home testing of blood and urine. The above recorded data constitutes the multi-dimensional data for analyzing adverse event early warning and complication prediction.

[0047] In the embodiment, the multi-dimensional data has important clinical significance. The ultrafiltration volume and drainage time recorded by the fully automatic peritoneal dialysis machine can be used as peritoneal inflammation complication analysis data. The total body water content recorded by the human body composition analyzer can be used to evaluate the water load of dialysis patients, combined with blood pressure changes, heart rate, to provide data support for early warning of heart failure. Blood pressure and hemoglobin content provide data support for monitoring of renal anemia complications. Electrolytes provide data support for timely early warning of electrolyte disorders. At the same time, the patient's daily dry body weight, total body fluid volume, treatment ultrafiltration volume, blood pressure and biochemical indicators (creatinine, urea and electrolytes) will be used as data sources for individualized treatment schemes. In order to interconnect the data of instruments of different protocols, a multi-mode gateway (i.e. a gateway supporting multiple communication protocols) is used to realize standardized transmission of data, and all the data recorded by the devices are uploaded to the remote server.

[0048] After receiving the multi-dimensional data, the remote server first determines whether the collected multi-dimensional data has obvious abnormalities based on a simple threshold value, and provides a hierarchical early warning, which is sent to the patient mobile terminal and the doctor mobile terminal respectively. Specifically, the present embodiment uses an AI model to provide early warning of related complications, and the specific technical route is as follows Figure 2The embodiment selects a CNN-Transformer-GRU-Attention combined model for multi-task prediction in peritoneal dialysis process, wherein CNN (Convolutional Neural Network) is a convolutional neural network, the core of the Transformer model is an attention mechanism, which allows the model to consider the information of all time steps at the same time when processing sequence data, thereby better capturing long-range dependencies. GRU (Gate Recurrent Unit) is a kind of recurrent neural network. Attention is an attention model. First, one-dimensional convolution is used to extract local features and time sequence information of the input patient data, to obtain important diagnostic information contained in the local features and dynamic change characteristics (such as the change of the collected multi-dimensional data) in the time sequence information. Then, the Transformer model is used to further reconstruct the time sequence data, thereby efficiently extracting deeper and more abstract hidden features of different sudden diseases, better processing time sequence information of a certain length, and effectively capturing long-term dependencies, i.e. capturing the relationship between various important diagnostic information and the change of multi-dimensional data. The Transformer model input part is coupled with the GRU layer in the embodiment. Compared with RNN (Recurrent Neural Network), which is prone to gradient vanishing or explosion problems in long sequences, leading to ineffective learning of long-distance dependencies, and the characteristics of complex structure, large amount of calculation and many parameters of LSTM (Long Short-Term Memory), the GRU introduces a “gate mechanism” to solve the gradient vanishing problem of traditional RNN, and simplifies the complex structure of LSTM, thereby further improving the operation efficiency of the model and shortening the iteration time. The use of the GRU network can make full use of the change of the patient's multi-dimensional data during the peritoneal dialysis process of a certain length, rather than only according to the data of a single historical moment, thereby enabling continuous and accurate identification and prediction of sudden diseases such as heart failure, peritonitis, renal anemia, and electrolyte imbalance during peritoneal dialysis. The data processed by the Transformer layer is fully utilized in the GRU layer, and these features will be determined by the gate structure of the GRU layer whether to remember or forget, so as to model the long-term dependencies between feature sequences from multiple scales, making up for the defect that the Transformer is difficult to capture local dependencies. Since most of the physiological factors of sudden diseases during peritoneal dialysis coincide, but different combinations of physiological factors can distinguish different sudden diseases, in order to better distinguish different actual scenarios, an attention mechanism is added after the GRU layer, and the hidden features of physiological factors of different sudden diseases are assigned different weight coefficients, thereby reflecting their different influence degrees on predicting different sudden diseases.Finally, the output obtained after the calculation of these parts is mapped to the final output value of different tasks through the Softmax function, and the prediction result and the early warning information are obtained according to the final output value, and the prediction result is used to reflect the probability of predicting different sudden diseases. In this embodiment, the case where the output value is the largest is the recognition result of the model on whether a sudden disease will occur or which disease will occur, and the early warning information can be obtained based on the prediction result.

[0049] In addition, the remote server of the embodiment also realizes the formulation of a personalized peritoneal dialysis treatment plan according to multidimensional data such as daily treatment parameters and physiological indicators of a patient, and timely adjusts the treatment parameters (such as the exchange frequency, cycle time, and dialysate concentration) of the full-automatic peritoneal dialysis machine, as shown in the technical route Figure 3 The embodiment adopts a convolutional neural network (CNN) combined with a full connection layer architecture for personalized peritoneal dialysis treatment plan formulation. Specifically, the input layer includes at least five key parameters: dry body weight, ultrafiltration volume, total body fluid volume, biochemical indicators (creatinine, urea, and electrolytes), and patient subjective feelings (using binary classification, 0 representing good and 1 representing bad). The above data are sequentially processed by the CNN layer, including one-dimensional convolution (Convolution1D) feature extraction, batch normalization (BatchNormalization) data standardization, and maximum pooling (MaxPooling) down-sampling; then the features are converted into a one-dimensional vector through the flattening layer (Flatten), and finally three treatment parameters: daily exchange frequency, dialysis cycle time, and dialysate concentration are output through the full connection layer (Dense), to realize the generation of a precise treatment plan based on multidimensional data. The treatment plan generated by the embodiment can be pushed to doctors for review through a mobile terminal, and after confirmation, it is automatically synchronized to the peritoneal dialysis machine for parameter update. The system simultaneously establishes a closed-loop feedback mechanism, and the execution effect of the treatment plan and the data of the patient are fed back to the model to continuously optimize the decision-making accuracy. This scheme significantly improves the timeliness and accuracy of prescription adjustment by combining data-driven intelligent analysis with clinical experience, reduces the risk of complications, and prolongs the stable period of kidney function.

[0050] Further, the APP function diagram of the patient mobile terminal and the doctor mobile terminal of the embodiment is as shown in Figure 4As shown, the mobile terminal APP of the embodiment serves as a man-machine interaction interface, realizing lightweight function deployment and efficient doctor-patient collaboration. The patient mobile terminal obtains the health report and adverse event early warning pushed by the cloud in real time through the APP, so as to timely pay attention to the health status; the doctor mobile terminal APP provides a visual data board, dynamically displays the multi-dimensional health index trend of the patient, marks the abnormal data and triggers the early warning prompt, and at the same time integrates an intelligent consultation platform, quickly formulates or adjusts the prescription based on the treatment scheme pushed by the remote server, and realizes the collaborative decision of the operation such as the dialysis machine parameter setting through the remote tool. Through efficient cooperation with the remote server, the software of the mobile terminal realizes the closed-loop design of “cloud centralized analysis and terminal efficient application”, which significantly improves the complication prevention ability and medical management efficiency of the home dialysis patient. In actual application, the mobile terminal of the embodiment can also send relevant control instructions to the patient terminal device according to the treatment scheme pushed by the remote server, adjust the monitoring frequency of the corresponding device, so as to timely obtain the multi-dimensional data of the user.

[0051] Specifically, the patient mobile terminal of the embodiment can also obtain the multi-dimensional data of the user from the cloud server through the network interface. Then, a cross-platform application is realized by using ReactNative (an open source cross-platform mobile application development framework), and the human body composition data such as total water, muscle mass, extracellular fluid and intracellular fluid ratio, hemoglobin value, dialysis parameters such as ultrafiltration volume, dialysate composition, etc. are displayed in real time. At the same time, trend charts can also be drawn with the help of a data visualization chart library, presenting 7-30 day weight change curves, hemoglobin fluctuation line graphs, body fluid balance trend charts, etc., directly reflecting the dynamic changes of the data. The doctor mobile terminal can pull various data from the remote server based on the background management framework. Data analysis algorithms are used to process the data in multiple dimensions. Patients can be grouped and counted according to specific conditions, for example, “anemia high incidence group” is filtered out. The incidence rate heat map of complications is generated by using geographic information system (GIS) related technology, and the distribution of complications is directly presented. A comparison view can also be established to compare and analyze the current patient data with the historical data, such as analyzing the risk trend of peritoneal function failure, and professional statistical reports are generated to assist doctors in decision-making.

[0052] In the embodiment, after the remote server analyzes and gives a treatment scheme or prescription adjustment suggestion, the doctor can modify the prescription according to the suggestion, and the modification reason needs to be recorded (such as “the patient has too much water in the body, and the dehydration amount needs to be increased”). After confirmation, the prescription is automatically synchronized to the cloud historical prescription library. After confirming the adjusted prescription with the patient, the prescription parameters can be sent to the peritoneal dialysis machine through the Internet of Things. At the same time, the dialysis machine returns a confirmation status (such as “the parameters have been updated”), and the APP records the operation log. The latest updated prescription is used on the peritoneal dialysis machine.

[0053] After receiving the early warning information pushed by the remote server, the patient terminal can perform hierarchical early warning, and the hierarchical early warning rules of the embodiment are as follows:

[0054] First-level early warning (low risk): single or short-term data anomaly (such as slight decrease of hemoglobin, body fluid overload), triggering patient terminal APP pop-up window prompt, pushing health suggestions (such as diet adjustment).

[0055] Second-level early warning (medium risk): continuous data anomaly or critical key indicators (such as continuous 3-day insufficient ultrafiltration, hemoglobin < 10 g / dL), synchronously pushing patient terminal pop-up window and short message, and generating doctor terminal analysis report, suggesting remote evaluation.

[0056] Third-level early warning (high risk): critical value (such as extremely high serum creatinine, electrolyte disorder, sudden decrease of hemoglobin) or device failure. The embodiment can trigger emergency pop-up window, short message and voice call reminder, automatically contact the hospital, generate reconsultation suggestion (such as "need 2-hour emergency treatment"), so as to realize the early warning mode of "pop-up window prompt → short message notification → emergency alarm".

[0057] Since the lifestyle (diet, exercise) of peritoneal dialysis patients needs more refined management to ensure the stability of various indicators of the body, the embodiment can read the physiological indicators of the patient (such as hemoglobin level lower than 10 g / dL) from the remote server in real time based on the automatic decision module of the rule engine, and automatically trigger the diet adjustment suggestion, recommending to increase the intake of red meat, animal liver and other iron-rich foods.

[0058] The remote intelligent home peritoneal dialysis system of the present application has at least the following advantages:

[0059] 1. More comprehensive peritoneal dialysis patient data collection

[0060] For the first time, automatic peritoneal dialysis machines, human component analyzers, non-invasive hemoglobin detectors, biochemical analyzers and other devices are integrated into a unified platform to monitor multi-dimensional health data such as fluid balance, anemia and treatment parameters, breaking through the limitations of single function of existing devices. The health status of peritoneal dialysis patients is systematically mastered from more dimensions. Moreover, most of these devices are non-invasive, greatly reducing the frequency of patients going to the hospital and alleviating the pain of patients during invasive detection.

[0061] 2. Automatic data upload

[0062] Instead of the process of patients writing down various data, the inaccuracy and untimeliness of data are reduced.

[0063] 3. AI-driven data analysis

[0064] (1) Complication dynamic prediction model: Based on multi-modal deep learning, combined with real-time data and historical trends, realize the dynamic risk warning of peritonitis, heart failure and other complications, instead of relying on static threshold, greatly reduce the probability of peritoneal dialysis patients suffering from complications, greatly extend the life cycle of patients.

[0065] (2) Personalized treatment plan generation: Combined with a large number of prescription cases, AI uses daily treatment and body data to evaluate the current treatment plan for each patient and provides prescription modification suggestions, which provides strong support for doctors to better develop professional and effective dialysis plans for each dialysis patient.

[0066] 4. Multifunctional mobile management platform

[0067] The professional peritoneal dialysis management platform for peritoneal dialysis patients integrates important blocks such as data board, report generation, alarm function, information push, etc. It provides a platform for doctors to remotely and real-time monitor patient conditions and provide guidance.

[0068] Based on the above embodiment, the present application also provides a peritoneal dialysis control method based on the remote intelligent home peritoneal dialysis system in the above embodiment. The method of the present embodiment can be applied in a terminal, which can be a mobile phone, a computer or other intelligent product terminal. As shown in the figure, the method of the present embodiment includes the following steps: Figure 5

[0069] Step S100, collecting multi-dimensional data based on patient terminal equipment;

[0070] Step S200, collecting the multi-dimensional data collected by the patient terminal equipment, and using an AI model to analyze and process the multi-dimensional data, outputting warning information, prediction results and treatment plans;

[0071] Step S300, obtaining the warning information, prediction results and treatment plans pushed by the remote server, and generating a visual data board.

[0072] ​In the embodiment, the patient-side device includes any one or more of a fully automatic peritoneal dialysis machine, a human composition analyzer, a non-invasive hemoglobin detector, a home biochemical analyzer, a weight scale, and a sphygmomanometer. The multi-dimensional data includes any one or more of the ultrafiltration amount and the drainage time recorded by the fully automatic peritoneal dialysis machine, the whole-body fluid amount, the fat amount, and the muscle amount evaluated by the human composition analyzer, the hemoglobin amount detected by the non-invasive hemoglobin detector, the electrolyte, the creatinine, the urea analyzed by the home biochemical analyzer, the daily weight before and after dialysis recorded by the weight scale, and the daily blood pressure recorded by the sphygmomanometer. The mobile terminal includes a patient mobile terminal and a doctor mobile terminal. The patient mobile terminal acquires the health report and the adverse event early warning pushed by the remote server in real time through an APP. The doctor mobile terminal is used to provide a visual data board to dynamically display the multi-dimensional health index trend of the patient, mark abnormal data, and trigger an early warning prompt.

[0073] The remote server of the embodiment can perform multi-task prediction in the peritoneal dialysis process based on a CNN-Transformer-GRU-Attention joint model to obtain the prediction result and the early warning information. And output the treatment scheme through the architecture combining the convolutional neural network and the fully connected layer.

[0074] The steps in the peritoneal dialysis control method of the remote intelligent home peritoneal dialysis system of the embodiment are the same as the principles of the modules in the system embodiment, which will not be repeated here.

[0075] Based on the above-mentioned embodiments, the application further provides a terminal. The principle block diagram of the terminal can be as shown in Figure 6 The terminal can include one or more processors 100 (only one is shown in Figure 6 ), a memory 101, and a computer program 102 stored in the memory 101 and executable on the one or more processors 100. For example, a peritoneal dialysis control program of a remote intelligent home peritoneal dialysis system. When the one or more processors 100 execute the computer program 102, the steps in the peritoneal dialysis control method embodiment of the remote intelligent home peritoneal dialysis system can be implemented. Alternatively, when the one or more processors 100 execute the computer program 102, the functions of the modules / units in the peritoneal dialysis control system embodiment of the remote intelligent home peritoneal dialysis system can be implemented, which is not limited here.

[0076] In one embodiment, the processor 100 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic components, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0077] In one embodiment, the memory 101 can be an internal storage unit of the electronic device, such as a hard disk or a memory of the electronic device. The memory 101 can also be an external storage device of the electronic device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 101 can include both the internal storage unit and the external storage device of the electronic device. The memory 101 is used to store computer programs and other programs and data required by the terminal. The memory 101 can also be used to temporarily store data that has been output or will be output.

[0078] Those skilled in the art can understand that, Figure 6 The block diagram shown in the above figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the terminal to which the scheme of the present application is applied. The specific terminal can include more or less components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0079] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, storage, operating database or other medium used in the embodiments of the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0080] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A remote intelligent home peritoneal dialysis system, characterized in that, The system comprises: A patient terminal device for collecting and transmitting multidimensional data, the patient terminal device comprising any one or more of a fully automatic peritoneal dialysis machine, a human body composition analyzer, a non-invasive hemoglobin detector, a home biochemical analyzer, a body weight scale, and a sphygmomanometer; A remote server for collecting the multidimensional data collected by the patient terminal device, analyzing and processing the multidimensional data using an AI model, and outputting early warning information, prediction results, and treatment plans; A mobile terminal for obtaining the early warning information, prediction results, and treatment plans pushed by the remote server and generating a visual data board; The multidimensional data comprises any one or more of the ultrafiltration volume and drainage time recorded by the fully automatic peritoneal dialysis machine, the total body fluid volume, fat mass, and muscle mass evaluated by the human body composition analyzer, the hemoglobin volume detected by the non-invasive hemoglobin detector, the electrolytes, creatinine, urea, and body weight recorded by the body weight scale, and the daily blood pressure recorded by the sphygmomanometer; The mobile terminal comprises a patient mobile terminal and a doctor mobile terminal, the patient mobile terminal obtains the health report and adverse event early warning pushed by the remote server in real time through an APP, and the doctor mobile terminal is used to provide a visual data board to dynamically display the trends of the multidimensional health indicators of the patient, mark abnormal data, and trigger an early warning prompt; The remote server is specifically configured to: Perform multitask prediction during peritoneal dialysis based on a CNN-Transformer-GRU-Attention joint model to obtain the prediction results and the early warning information, and output a treatment plan through an architecture combining a convolutional neural network and a fully connected layer, specifically including: Extracting local features and time series information of the multidimensional data based on a one-dimensional convolutional layer in the convolutional neural network, obtaining important diagnostic information contained in the local features and dynamic change features in the time series information, reconstructing time series data through a Transformer model, extracting hidden features of different sudden diseases, capturing the relationship between various important diagnostic information and changes in the multidimensional data, coupling the Transformer model input part with a GRU layer, the GRU layer using a certain length of changes in the multidimensional data of the patient during peritoneal dialysis, adding an attention mechanism after the GRU layer, and giving different weight coefficients to the hidden features of different sudden diseases, mapping the final output values of different tasks through a Softmax function, obtaining the prediction results and the early warning information according to the final output values, and the prediction results being used to reflect the occurrence probability of predicting different sudden diseases. The input layer of the convolutional neural network comprises five key parameters: dry body weight, ultrafiltration volume, total body fluid volume, biochemical indicators, and patient subjective feelings, and the above data are sequentially subjected to one-dimensional convolution for feature extraction, batch normalization for data standardization, maximum pooling layer for down-sampling, and flattening layer for converting the features into a one-dimensional vector; three treatment parameters, i.e., daily exchange frequency, dialysis cycle time and dialysate concentration, are output through a fully connected layer; and the treatment scheme is generated according to the daily exchange frequency, dialysis cycle time and dialysate concentration; The treatment scheme is pushed to the doctor for review through the mobile terminal, and after confirmation, it is automatically synchronized to the peritoneal dialysis machine for parameter updating, and a closed-loop feedback mechanism is established to return the execution effect of the treatment scheme and the patient's data to the CNN-Transformer-GRU-Attention combined model, continuously optimizing the decision-making accuracy; The patient mobile terminal is specifically used for: Obtaining the multi-dimensional data of the user from the remote server through the network interface, realizing cross-platform application by using ReactNative, and real-time displaying the human body composition data, including: total water, muscle mass, extracellular fluid to intracellular fluid ratio, hemoglobin value, and dialysis parameters; meanwhile, trend charts are drawn by means of a data visualization chart library, presenting the 7-30 day weight change curve, hemoglobin fluctuation line chart, and body fluid balance trend chart; After receiving the early warning information pushed by the remote server, the patient mobile terminal performs hierarchical early warning, and the rules of the hierarchical early warning include: First-level early warning, indicating that single or short-term data anomalies occur, at which time a patient-side APP pop-up window is triggered to prompt and push health suggestions; Second-level early warning, indicating that continuous data anomalies or critical indicators occur, at which time the patient-side pop-up window and short message are synchronously pushed, and a doctor-side analysis report is generated to suggest remote evaluation; Third-level early warning, indicating that critical values or equipment failures occur, at which time an emergency pop-up window, short message and voice call reminder are triggered, the system automatically contacts the hospital, and a re-consultation suggestion is generated; The doctor mobile terminal is specifically used for: Based on the background management framework, various data are pulled from the remote server, data analysis algorithms are used to perform multi-dimensional processing on the data, a heat map of complication incidence rate is generated by using geographic information system related technologies, a comparison view is established to compare and analyze the current patient data with the historical data, and professional statistical reports are generated to assist the doctor in decision-making.

2. A peritoneal dialysis control method based on the remote intelligent home peritoneal dialysis system of claim 1, characterized in that, The method comprises: Collecting multi-dimensional data based on the patient-side device; Collecting the multi-dimensional data collected by the patient-side device, and using an AI model to analyze and process the multi-dimensional data, outputting early warning information, prediction results and treatment schemes; Obtaining the early warning information, prediction results and treatment schemes pushed by the remote server, and generating a visual data dashboard. 3.The peritoneal dialysis control method of the remote intelligent home peritoneal dialysis system according to claim 2, characterized in that, The use of an AI model to analyze and process the multi-dimensional data, output early warning information, prediction results and treatment schemes, comprises: Based on the CNN-Transformer-GRU-Attention combined model, multi-task prediction in the peritoneal dialysis process is performed to obtain the prediction results and the early warning information; The treatment plan is output by an architecture combining a convolutional neural network and a fully connected layer.

4. The peritoneal dialysis control method of claim 3, wherein, The CNN-Transformer-GRU-Attention combined model is used for multi-task prediction in peritoneal dialysis to obtain the prediction result and the warning information, including: Local features and time sequence information of the multi-dimensional data are extracted by a one-dimensional convolutional layer in the convolutional neural network to obtain important diagnostic information contained in the local features and dynamic change features in the time sequence information. The time sequence data is reconstructed by the Transformer model to extract hidden features of different sudden diseases, and the Transformer model input part is coupled with the GRU layer. An attention mechanism is added after the GRU layer, and different weight coefficients are given to the hidden features of different sudden diseases. The final output values of different tasks are obtained by mapping through a Softmax function, and the prediction result and the warning information are obtained according to the final output values, wherein the prediction result is used to reflect the occurrence probability of predicting different sudden diseases.

5. The peritoneal dialysis control method of claim 3, wherein the peritoneal dialysis control method is performed by the remote intelligent home peritoneal dialysis system of claim 1. The architecture combining the convolutional neural network and the fully connected layer outputs the treatment plan, including: The multi-dimensional data are sequentially subjected to one-dimensional convolution for feature extraction, batch normalization for data standardization, maximum pooling layer for down-sampling, and flattening layer for converting the features into one-dimensional vectors. Three treatment parameters, i.e., daily exchange frequency, dialysis cycle time, and dialysate concentration, are output by the fully connected layer. The treatment plan is generated according to the daily exchange frequency, the dialysis cycle time, and the dialysate concentration. 6.The peritoneal dialysis control method of the remote intelligent home peritoneal dialysis system according to claim 1, wherein, The method further includes: An intelligent consultation platform is integrated to formulate or adjust a prescription based on the treatment plan and set parameters through a remote tool.

7. A terminal, characterized by comprising: The terminal includes a memory, a processor, and a peritoneal dialysis control program of a remote intelligent home peritoneal dialysis system stored in the memory and executable on the processor, and the processor executes the peritoneal dialysis control program of the remote intelligent home peritoneal dialysis system to realize the steps of the peritoneal dialysis control method of the remote intelligent home peritoneal dialysis system according to any one of claims 2-6.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores the peritoneal dialysis control program of the remote intelligent home peritoneal dialysis system, and the peritoneal dialysis control program of the remote intelligent home peritoneal dialysis system realizes the steps of the peritoneal dialysis control method of the remote intelligent home peritoneal dialysis system according to any one of claims 2-6 on the computer-readable storage medium.

Images