Device and method for precise administration of traditional Chinese medicine preparation

Abstract

The invention discloses a device and method for precise administration of traditional Chinese medicine preparations. The device comprises a control module used for obtaining physical information and disease information of a to-be-administered user, performing multi-dimensional feature extraction to obtain multi-dimensional feature information, and generating a corresponding dynamic prescription and an individual administration scheme according to a first knowledge graph; obtaining a second knowledge graph; the traditional Chinese medicine efficacy association network constructs a synergistic effect relationship of different medicinal materials through a graph neural network; the medicine warehouse management module comprises a multi-temperature-zone medicine storage unit and a medicinal material weighing unit, and the multi-temperature-zone medicine storage unit divides independent storage spaces according to medicinal material attributes; the medicine feeding robot comprises a medicinal material extracting and packaging all-in-one machine; the control module controls the medicine library management module to dispense the medicinal materials according to the dynamic prescription, and controls the medicine delivery robot to perform directional extraction, concentration adjustment and time-phased packaging on the dispensed medicinal materials according to an individual taking scheme; the individualized administration regimen includes a pulsed administration timing calculated based on the metabolic cycle of the patient.

Description

Technical Field

[0001] The present application relates to the technical field of drug dispensing or application devices, and particularly relates to a device and method for accurately administering traditional Chinese medicine preparations. Background Art

[0002] At present, traditional Chinese medicine is taken in the form of preparations or patent traditional Chinese medicines. Among them, preparations require doctors to formulate prescriptions according to the physical constitution and condition of patients, decoct them into decoctions and take them in accordance with the dosage. Patent traditional Chinese medicines are prepared into pills according to ancient prescriptions and can be taken regardless of individual differences of patients.

[0003] Although preparations are more suitable for the physical constitution of patients, formulating prescriptions relies heavily on doctors' experience, the decocting time is long, and they can only be taken in accordance with the dosage during the taking process, which is not conducive to maximizing the efficacy. Although patent traditional Chinese medicines can be mass-produced, various physical constitutions need to be considered, so the efficacy is poor and long-term use is required. For cancer patients, a significant therapeutic and rehabilitation effect needs to be achieved in a gentle way. Therefore, the existing administration methods of patent traditional Chinese medicines cannot meet the rehabilitation needs of cancer patients.

[0004] Therefore, there is an urgent need to design a technical solution to solve at least one of the above technical problems. Summary of the Invention

[0005] The present application provides a device and method for accurately administering traditional Chinese medicine preparations, aiming to solve the problem that the existing administration methods of patent traditional Chinese medicines cannot meet the rehabilitation needs of cancer patients.

[0006] In a first aspect, the present application provides a device for accurately administering traditional Chinese medicine preparations, including:

[0007] A control module, which acquires the physical constitution information and disease information of the user to be administered medicine, extracts multi-dimensional feature information from the physical constitution information and disease information, and generates a dynamic prescription and an individualized taking plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph; the control module acquires a second knowledge graph; the second knowledge graph includes a traditional Chinese medicine efficacy association network, a disease-prescription matching model, and a patient feature analysis algorithm; the traditional Chinese medicine efficacy association network constructs the synergistic relationship of different medicinal materials through a graph neural network;

[0008] A medicine library management module, including a multi-temperature zone medicine storage unit and a medicinal material weighing unit, and the multi-temperature zone medicine storage unit divides independent storage spaces according to the attributes of medicinal materials;

[0009] A medicine delivery robot, including a medicinal material extraction and encapsulation integrated machine;

[0010] Among them, the control module controls the medicine library management module to dispense medicinal materials according to the dynamic prescription, and controls the medicine delivery robot to perform directional extraction, concentration adjustment and time-segmented packaging according to the individualized taking plan for the dispensed medicinal materials; the individualized taking plan also includes a pulsed drug delivery time sequence calculated based on the patient's metabolic cycle.

[0011] In some embodiments, the disease information includes traditional Chinese medicine four-diagnosis data and Western medicine biochemical indicators; the multi-dimensional feature information includes a fusion feature vector; the constitution information includes monitoring data collected by a preset wearable device; the obtaining of multi-dimensional feature information by performing multi-dimensional feature extraction on the constitution information and disease information includes: performing feature alignment on the traditional Chinese medicine four-diagnosis data, Western medicine biochemical indicators and monitoring data; extracting a tongue coating texture distribution matrix for the tongue image features corresponding to the traditional Chinese medicine four-diagnosis data, generating a three-dimensional feature vector of pulse position and pulse potential for the pulse condition data through a long short-term memory network, and extracting circadian rhythm features from the monitoring data through wavelet transform; and weighted fusing the tongue coating texture distribution matrix, the three-dimensional feature vector of pulse position and pulse potential, the circadian rhythm features and the Western medicine biochemical indicators according to an attention mechanism to generate the fusion feature vector and the traditional Chinese medicine syndrome classification label corresponding to the fusion feature vector.

[0012] In some embodiments, the generating of the dynamic prescription and the individualized taking plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph includes: mapping the multi-dimensional feature information to the disease-constitution feature sub-graph of the first knowledge graph, and matching a candidate prescription set corresponding to the multi-dimensional feature information in the disease-constitution feature sub-graph through a graph embedding algorithm; obtaining the efficacy-toxicity balance coefficient corresponding to each candidate prescription in the candidate prescription set through Monte Carlo tree search according to a preset efficacy synergy prediction model, and completing the optimization of the candidate prescription set; obtaining the real-time medicine library inventory status corresponding to the medicine library management module and the patient's previous medication records corresponding to the user to be administered, and generating a dynamic prescription topology graph according to the patient's previous medication records and the real-time medicine library inventory status; the dynamic prescription topology graph includes a medicinal material substitution plan; generating the dynamic prescription according to the dynamic prescription topology graph and the optimized candidate prescription set; generating a blood drug concentration curve corresponding to the dynamic prescription under different drug delivery time sequences according to a preset pharmacokinetic simulation model, and generating the individualized taking plan according to the blood drug concentration curve.

[0013] In some embodiments, the control of the medicine depot management module to dispense medicinal materials according to the dynamic prescription includes: parsing the dynamic prescription to obtain the topological relationship of the medicinal material components corresponding to the dynamic prescription; generating a medicinal material priority queue according to the topological relationship of the medicinal material components; in the medicinal material priority queue, the medicinal materials approaching the expiration date are preferentially matched with the prescriptions with high usage frequencies; controlling a preset medicine-taking robotic arm in the multi-temperature zone medicine storage unit to obtain target medicinal materials in the multi-temperature zone medicine storage unit according to the medicinal material priority queue; a flexible gripper and a near-infrared spectroscopy recognition probe are equipped at the end of the medicine-taking robotic arm; obtaining the weight of the target medicinal material measured by the medicinal material weighing unit, and eliminating the weight error caused by the change of the moisture content of the medicinal material in the medicinal material weight according to the dynamic weighing compensation algorithm; the dynamic weighing compensation algorithm corrects the medicinal material weight in real time based on the moisture absorption characteristic curve corresponding to the target medicinal material.

[0014] In some embodiments, the obtaining of the second knowledge graph includes: extracting entity-relationship triples from a preset digital library of traditional Chinese medicine ancient books, a modern pharmacology database, and a clinical trial database; mapping the data of the properties and meridians corresponding to the digital library of traditional Chinese medicine ancient books to the modern molecular pharmacology feature space of the modern pharmacology database according to a transfer learning framework; constructing an implicit association matrix of medicinal materials, diseases, and constitutions according to an adversarial generation network; generating the second knowledge graph according to the implicit association matrix, the modern molecular pharmacology feature space, and the entity-relationship triples.

[0015] In some embodiments, the medicinal material extraction and encapsulation integrated machine includes a supercritical carbon dioxide extraction device and a variable-volume packaging device.

[0016] Exemplarily, the control of the medicine delivery robot to perform directional extraction, concentration adjustment, and time-segmented encapsulation according to the individualized taking plan for the dispensed medicinal materials includes: dynamically adjusting the supercritical carbon dioxide extraction parameters corresponding to the supercritical carbon dioxide extraction device according to the medicinal material component characteristics corresponding to the dispensed medicinal materials; the supercritical carbon dioxide extraction parameters include a pressure gradient control curve and an entrainer ratio strategy; monitoring the concentration of the active ingredients of the extraction solution corresponding to the directional extraction in real time by a multi-spectral fusion detection device, and adjusting the solvent evaporation rate so that the concentration of the active ingredients of the extraction solution is the target concentration value; in the time-segmented encapsulation stage, performing a nano-coating treatment on the inner wall of the variable-volume packaging device, and the coating material selects a metal-organic framework material with specific adsorption characteristics according to the medicinal material component characteristics; controlling the variable-volume packaging device to complete the packaging of the extraction solution.

[0017] In some embodiments, the medicine repository management module further includes: a shelf life tracking unit. The multi-temperature zone medicine storage unit divides independent storage spaces according to the properties of traditional Chinese medicines, and the shelf life tracking unit updates the remaining shelf life of the traditional Chinese medicines in the independent storage spaces in real time; if the remaining shelf life is lower than a preset shelf life, the control module generates a warning signal.

[0018] In some embodiments, the medicine delivery robot further includes a medication monitoring unit. The medication monitoring unit monitors the RFID tag corresponding to the user to be medicated, associates the patient identity information corresponding to the user to be medicated, and records the medication timestamp.

[0019] In a second aspect, the present application provides a method for precise administration of traditional Chinese medicine preparations, which is applied to the control module of the device for precise administration of traditional Chinese medicine preparations provided in any embodiment of the present application; the method includes:

[0020] Obtain the physical constitution information and disease information of the user to be medicated, and perform multi-dimensional feature extraction on the physical constitution information and disease information to obtain multi-dimensional feature information;

[0021] Generate a dynamic prescription and an individualized medication plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph;

[0022] Obtain a second knowledge graph; the second knowledge graph includes a traditional Chinese medicine efficacy association network, a disease prescription matching model, and a patient feature analysis algorithm; the traditional Chinese medicine efficacy association network constructs the synergistic relationship between different traditional Chinese medicines through a graph neural network;

[0023] Control the medicine repository management module to dispense traditional Chinese medicines according to the dynamic prescription, and control the medicine delivery robot to perform directional extraction, concentration adjustment, and segmented packaging according to the medication plan on the dispensed traditional Chinese medicines; the individualized medication plan also includes a pulsed medication timing calculated based on the patient's metabolic cycle.

[0024] In a third aspect, the present application provides a device for precise administration of traditional Chinese medicine preparations, including:

[0025] An information acquisition unit, configured to obtain the physical constitution information and disease information of the user to be medicated, and perform multi-dimensional feature extraction on the physical constitution information and disease information to obtain multi-dimensional feature information;

[0026] A plan generation unit, configured to generate a dynamic prescription and an individualized medication plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph;

[0027] A relationship construction unit for obtaining a second knowledge graph; the second knowledge graph includes a traditional Chinese medicine efficacy association network, a disease prescription matching model, and a patient characteristic analysis algorithm; the traditional Chinese medicine efficacy association network constructs the synergistic relationship of different medicinal materials through a graph neural network;

[0028] A scheme control unit for controlling the medicine library management module to dispense medicinal materials according to the dynamic prescription, and controlling the delivery robot to perform directional extraction, concentration adjustment and time-segmented packaging according to the individualized taking scheme for the dispensed medicinal materials; the individualized taking scheme also includes a pulsed drug delivery time sequence calculated based on the patient's metabolic cycle.

[0029] In a fourth aspect, the present application provides a control module, which includes a memory and a processor; the memory is used for storing a computer program; the processor is used for executing the computer program and implementing the method provided in any embodiment of the present application when executing the computer program.

[0030] In a fifth aspect, the present application provides a computer-readable storage medium storing a computer program, and when the computer-readable instruction is executed by the processor, one or more processors are enabled to execute the method provided in any embodiment of the present application.

[0031] The present application provides a device and method for precise administration of traditional Chinese medicine preparations, aiming to improve the application effect of traditional Chinese medicine in the rehabilitation process of cancer patients through high-tech means. The following is a detailed description of the technical content and beneficial effects of the device:

[0032] The control module is responsible for collecting the physical constitution information and disease information of the user to be administered, and obtaining multi-dimensional feature information through multi-dimensional feature extraction technology. These information will be used to generate personalized prescriptions and taking schemes. The control module uses a preset first knowledge graph to generate a dynamic prescription and an individualized taking scheme, and at the same time obtains a second knowledge graph, which includes a traditional Chinese medicine efficacy association network, a disease prescription matching model, and a patient characteristic analysis algorithm. The traditional Chinese medicine efficacy association network is constructed by graph neural network technology to simulate the synergistic relationship between different medicinal materials.

[0033] The multi-temperature zone medicine storage unit divides independent storage spaces according to the attributes of the medicinal materials, such as sensitivity to temperature and humidity, to ensure the freshness and effectiveness of the medicinal materials.

[0034] The medicinal material weighing unit is used to accurately weigh the medicinal materials to ensure the accuracy of the prescription.

[0035] The medicine delivery robot is responsible for directionally extracting and adjusting the concentration of the dispensed medicinal materials according to the instructions of the control module, and performing segmented packaging according to the individualized dosing plan. Based on the calculation of the patient's metabolic cycle, pulsed drug delivery is achieved to optimize the distribution and action time of the drug in the body.

[0036] Through multi-dimensional feature extraction and individualized prescription generation, the device can provide a customized treatment plan for each patient, improving the treatment effect. The traditional Chinese medicine efficacy association network constructed by using graph neural networks can more accurately simulate and predict the synergistic effects between different medicinal materials, thereby optimizing the efficacy of the prescription. The automated operation of the medicine library management module and the medicine delivery robot reduces human errors and improves the accuracy and efficiency of drug delivery. The design of the multi-temperature zone medicine storage unit helps to maintain the best state of the medicinal materials and extend their shelf life. The design of the pulsed drug delivery timing enables patients to take medicine more conveniently according to the treatment plan, improving the compliance of treatment. By applying the disease prescription matching model and patient characteristic analysis algorithm in the second knowledge graph, the control module can make more scientific decisions based on a large amount of data. Through precise drug delivery and optimized treatment plans, the device helps tumor patients recover faster. The automated and intelligent drug delivery device reduces labor costs, and at the same time reduces drug waste through precise drug delivery, thereby reducing the overall medical cost. The application of the device can provide new data and methods for the research of traditional Chinese medicine, promoting the modernization development of traditional Chinese medicine. By providing more personalized and convenient treatment plans, the treatment experience of patients is improved, which helps to increase patients' satisfaction and trust in the treatment.

[0037] In summary, by combining modern technology with traditional Chinese medicine, the device provides a new precise drug delivery solution for tumor patients, with broad application prospects and profound social significance.

[0038] In summary, the device and method overcome the limitations of traditional methods in tumor patients through innovative magnetic labeling and detection technologies, achieving efficient and accurate BDNF detection, and providing a powerful tool for evaluating chemotherapy-induced depression.

[0039] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and do not limit this application. Brief Description of the Drawings

[0040] In order to more clearly illustrate the technical solutions of the embodiments of this application, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0041] Figure 1It is a schematic block diagram of a device for precise administration of traditional Chinese medicine provided by an embodiment of the present application;

[0042] Figure 2 It is a schematic flowchart of steps of a method for precise administration of traditional Chinese medicine provided by an embodiment of the present application;

[0043] Figure 3 It is a schematic block diagram of the structure of a control module provided by an embodiment of the present application.

[0044] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. Detailed implementation manners

[0045] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0046] The flowcharts shown in the accompanying drawings are only illustrative, not necessarily including all contents and operations / steps, nor necessarily executed in the described order. For example, some operations / steps can also be decomposed, combined or partially merged, so the actual execution order may be changed according to the actual situation.

[0047] It should be understood that in order to facilitate the clear description of the technical solutions in the embodiments of the present invention, in the embodiments of the present invention, terms such as "first" and "second" are used to distinguish the same items or similar items with basically the same functions and effects. Those skilled in the art can understand that the terms "first" and "second" do not limit the quantity and execution order, and the terms "first" and "second" do not necessarily mean different.

[0048] It should be understood that the terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification of the present application and the appended claims, unless otherwise clearly specified in the context, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0049] It should also be understood that the term " / and" used in the specification of the present application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0050] The following will describe in detail some embodiments of the present application with reference to the accompanying drawings. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0051] Currently, traditional Chinese medicine is taken in the form of preparations or proprietary Chinese medicines. Among them, for preparations, doctors need to formulate prescriptions according to the physical constitution and condition of patients, decoct them into decoctions and take them by the dose. Proprietary Chinese medicines, on the other hand, are prepared into pills according to ancient formulas and can be taken regardless of individual differences among patients.

[0052] Although preparations are more suitable for the physical constitution of patients, formulating prescriptions relies heavily on doctors' experience, the decocting time is long, and they can only be taken by the dose during the taking process, which is not conducive to maximizing the efficacy. Although proprietary Chinese medicines can be mass-produced, various physical constitutions need to be considered, so the efficacy is poor and long-term use is required. For cancer patients, a gentle way is needed to achieve significant treatment and rehabilitation effects. Therefore, the existing administration methods of proprietary Chinese medicines cannot meet the rehabilitation needs of cancer patients.

[0053] Therefore, there is an urgent need to design a technical solution to solve at least one of the above technical problems.

[0054] To solve the above problems, please refer to Figure 1 , the present application provides a device for accurately administering traditional Chinese medicine preparations, including: a control module, which obtains the physical constitution information and disease information of the user to be administered, extracts multi-dimensional feature information by performing multi-dimensional feature extraction on the physical constitution information and disease information, and generates a dynamic prescription and an individualized taking plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph; the control module obtains a second knowledge graph; the second knowledge graph includes a traditional Chinese medicine efficacy association network, a disease-prescription matching model, and a patient feature analysis algorithm; the traditional Chinese medicine efficacy association network constructs the synergistic relationship of different medicinal materials through a graph neural network; a medicine library management module, including a multi-temperature zone medicine storage unit and a medicinal material weighing unit, where the multi-temperature zone medicine storage unit divides independent storage spaces according to the attributes of medicinal materials; a medicine delivery robot, including a medicinal material extraction and encapsulation integrated machine; wherein, the control module controls the medicine library management module to dispense medicinal materials according to the dynamic prescription, and controls the medicine delivery robot to perform directional extraction, concentration adjustment and time-segmented encapsulation according to the taking plan on the dispensed medicinal materials; the individualized taking plan also includes a pulsed administration time sequence calculated based on the patient's metabolic cycle.

[0055] Specifically, the device for precise administration of traditional Chinese medicine of the present invention includes a control module, a medicine library management module, and a medicine delivery robot. The control module obtains the user's physical constitution information (such as physiological monitoring data collected by wearable devices) and disease information (traditional Chinese medicine four diagnostic data, western medicine biochemical indicators), extracts multi-dimensional features using deep learning algorithms, and generates a fused feature vector and traditional Chinese medicine syndrome classification labels. Based on a preset first knowledge graph (including the association relationship between diseases - physical constitution - prescriptions), the control module matches candidate prescriptions through a graph embedding algorithm, combines Monte Carlo tree search to optimize the efficacy - toxicity balance coefficient of the prescriptions, and finally generates a dynamic prescription. At the same time, an individualized dosing plan (such as a time - segmented pulsed dosing sequence) is generated according to the pharmacokinetic simulation model.

[0056] The multi - temperature - zone medicine storage unit of the medicine library management module divides independent storage spaces according to the properties of the medicinal materials (such as temperature and humidity sensitivity), and dynamically compensates for the moisture content error through the medicinal material weighing unit; the medicine delivery robot performs directional extraction of the medicinal materials through a supercritical carbon dioxide extraction device, and combines a variable - volume packaging device to package the medicine in segments according to the dosing plan. The control module coordinates the operations of each module in real - time to ensure the full - process automation from prescription generation to packaging.

[0057] The control module is the core of the device, responsible for processing and analyzing user information, and generating personalized prescriptions and dosing plans. Its specific functions include: Information acquisition and processing: Obtain the user's physical constitution information and disease information, perform multi - dimensional feature extraction, and obtain multi - dimensional feature information. Dynamic prescription generation: According to the preset first knowledge graph, combined with multi - dimensional feature information, generate corresponding dynamic prescriptions and individualized dosing plans. Knowledge graph application: Obtain a second knowledge graph, including a traditional Chinese medicine efficacy association network, a disease - prescription matching model, and a patient characteristic analysis algorithm, to achieve precise matching and efficacy analysis. Graph neural network construction: The traditional Chinese medicine efficacy association network is constructed through graph neural network technology to analyze the synergistic relationship between different medicinal materials.

[0058] The medicine library management module is responsible for the storage and weighing of medicinal materials to ensure the quality and precise proportioning of medicinal materials. Specifically, it includes: Multi - temperature - zone medicine storage unit: Divide independent storage spaces according to the properties of the medicinal materials to ensure that the medicinal materials are stored in a suitable environment. Medicinal material weighing unit: Accurately weigh the medicinal materials to ensure the accuracy of medicine dispensing.

[0059] The medicine delivery robot is responsible for the extraction, packaging, and distribution of medicinal materials. Its specific functions include: Medicinal material extraction and packaging integrated machine: According to the instructions of the control module, perform directional extraction, concentration adjustment of the dispensed medicinal materials, and package them in segments according to the dosing plan. Pulsed dosing sequence: The individualized dosing plan also includes a pulsed dosing sequence calculated based on the patient's metabolic cycle to achieve the best treatment effect.

[0060] If the user inputs physical constitution and disease information through the interface, the control module extracts key features through algorithms. The control module generates a personalized prescription based on the first knowledge graph and the user's characteristic information. The pharmacy management module takes out the corresponding medicinal materials from the multi-temperature zone medicine storage unit according to the dynamic prescription and weighs them. The medicine delivery robot receives the weighed medicinal materials, conducts extraction, concentration adjustment and encapsulation. According to the individualized taking plan, the medicine delivery robot delivers the encapsulated medicine to the user at different times.

[0061] Through multi-dimensional feature extraction and the application of knowledge graphs, the generation of personalized prescriptions is realized, improving the treatment effect. The precise weighing of the pharmacy management module and the design of the multi-temperature zone medicine storage unit ensure the accurate proportioning and quality of the medicinal materials. The automated operation of the medicine delivery robot reduces human errors and improves the accuracy and convenience of drug administration. The traditional Chinese medicine efficacy correlation network constructed by the graph neural network optimizes the synergistic effect of the medicinal materials and enhances the efficacy. Based on the pulsed drug delivery timing of the patient's metabolic cycle, the drug reaches the optimal concentration in the body, improving the curative effect.

[0062] In summary, this traditional Chinese medicine precise drug delivery device not only improves the treatment effect of traditional Chinese medicine through intelligent and personalized methods, but also optimizes the drug delivery process, providing a safer and more effective treatment plan for users.

[0063] Among them, multi-dimensional feature information is processed through the multi-source heterogeneous data processing center. A composite data structure is constructed, including a physical constitution feature library (including nine physical constitution classification standards) and a disease feature library (integrating ICD-11 and TCM diagnosis standards). Deep neural networks are used for feature fusion, specifically including: Convolutional feature extractor: processing image data such as tongue images and pulse conditions; LSTM time series analysis module: analyzing the time series data of the medical interview; Graph attention network: processing the symptom association network.

[0064] The node scale of the first knowledge graph reaches more than 500,000, covering: medicinal material entities (including genuine producing areas and processing technology attributes); chemical composition-target pathway mapping; classical prescription compatibility relationships. The TransR model is used for vector representation to achieve cross-modal semantic alignment.

[0065] Second knowledge graph: Medicinal material synergistic effect sub-graph: modeled based on the GAT-GCN hybrid network; Disease-syndrome mapping model: using a probabilistic graph model to quantify the correlation degree; Metabolic kinetics prediction module: integrating the PBPK model with the TCM meridian tropism theory.

[0066] The multi-temperature zone medicine storage unit adopts a three-level temperature control architecture: Ultra-low temperature zone (-20°C): storing animal-based medicinal materials; Constant humidity cold storage zone (4°C): storing medicinal materials with volatile components; Inert gas protection zone: storing easily oxidized medicinal materials; Equipped with a high-precision weighing unit (accuracy up to 0.01g).

[0067] The pressure range of the integrated machine for medicinal material extraction and encapsulation is 50 - 400 bar, and its membrane separation and concentration part adopts a 100 nm ceramic membrane module. The microencapsulation unit can achieve chronopharmacological administration timing.

[0068] In some embodiments, the disease information includes traditional Chinese medicine four diagnostic data and Western medicine biochemical indicators; the multi-dimensional feature information includes a fusion feature vector; the constitution information includes monitoring data collected by a preset wearable device; the obtaining of the multi-dimensional feature information by performing multi-dimensional feature extraction on the constitution information and the disease information includes: performing feature alignment on the traditional Chinese medicine four diagnostic data, Western medicine biochemical indicators, and monitoring data; extracting a tongue coating texture distribution matrix for the tongue image feature corresponding to the traditional Chinese medicine four diagnostic data according to a convolutional neural network, generating a three-dimensional feature vector of pulse position and pulse potential for the pulse condition data through a long short-term memory network, and extracting circadian rhythm features from the monitoring data through wavelet transform; and generating the fusion feature vector and the traditional Chinese medicine syndrome classification label corresponding to the fusion feature vector by weighted fusion of the tongue coating texture distribution matrix, the three-dimensional feature vector of pulse position and pulse potential, the circadian rhythm features, and the Western medicine biochemical indicators according to an attention mechanism.

[0069] Data alignment is achieved by aligning the time stamps of traditional Chinese medicine four diagnostic data (tongue image, pulse condition), Western medicine biochemical indicators (such as blood routine), and wearable device monitoring data (such as heart rate, body temperature) to eliminate the deviation caused by the data acquisition interval. Tongue image processing extracts a tongue coating texture distribution matrix through a convolutional neural network (CNN) to identify features such as the color of the tongue body and cracks. Pulse condition processing uses a long short-term memory network (LSTM) to analyze the temporal changes of pulse position and pulse potential and generate a three-dimensional pulse condition feature vector. Wearable data processing extracts circadian rhythm features (such as body temperature fluctuation period) through wavelet transform. Through an attention mechanism, the above features are weighted and fused to generate a fusion feature vector representing the comprehensive health status of the user, and a traditional Chinese medicine syndrome classification label (such as "yin deficiency and excessive fire") is output.

[0070] By integrating traditional Chinese and Western medicine data and multi-modal deep learning, the objectivity and accuracy of constitution and disease analysis are improved, providing a reliable basis for dynamic prescription generation.

[0071] In some embodiments, generating the dynamic prescription and individualized administration plan corresponding to the multi-dimensional feature information according to the preset first knowledge graph includes: mapping the multi-dimensional feature information to the disease constitution feature sub-graph of the first knowledge graph, and matching the candidate prescription set corresponding to the multi-dimensional feature information in the disease constitution feature sub-graph through a graph embedding algorithm; obtaining the efficacy-toxicity balance coefficient corresponding to each candidate prescription in the candidate prescription set through Monte Carlo tree search according to the preset efficacy synergy degree prediction model, and completing the optimization of the candidate prescription set; obtaining the real-time drug library inventory status corresponding to the drug library management module and the patient's past medication records corresponding to the user to be administered, and generating a dynamic prescription topology graph according to the patient's past medication records and the real-time drug library inventory status; the dynamic prescription topology graph includes a medicinal material substitution plan; generating the dynamic prescription according to the dynamic prescription topology graph and the optimized candidate prescription set; generating the blood drug concentration curve corresponding to the dynamic prescription at different administration time sequences according to the preset pharmacokinetic simulation model, and generating the individualized administration plan according to the blood drug concentration curve.

[0072] By fusing the feature vector and mapping it to the disease constitution feature sub-graph of the first knowledge graph, matching the candidate prescription set through a graph embedding algorithm (such as Node2Vec). Using the efficacy synergy degree prediction model (based on graph neural network) to calculate the efficacy-toxicity balance coefficient of the candidate prescription, and screening the optimal prescription through Monte Carlo tree search (MCTS). Combining the real-time drug library inventory (such as the out-of-stock status of medicinal materials) and the user's medication records (such as allergy history), generating a dynamic prescription topology graph including substitute medicinal materials to ensure the executability of the prescription. Based on the pharmacokinetic simulation model (such as PK / PD model), simulating the blood drug concentration curve at different administration time sequences, and selecting the pulsatile administration time sequence (such as 3 peak concentration pulses per day) that meets the therapeutic window. The dynamic prescription takes into account efficacy, safety and actual inventory, and the individualized administration plan optimizes the administration effect through blood drug concentration simulation, reducing side effects.

[0073] In some embodiments, controlling the drug library management module to dispense medicinal materials according to the dynamic prescription includes: parsing the dynamic prescription to obtain the topological relationship of the medicinal material components corresponding to the dynamic prescription; generating a medicinal material priority queue according to the topological relationship of the medicinal material components; in the medicinal material priority queue, the medicinal materials approaching the expiration date are preferentially matched with the prescriptions with high usage frequencies; controlling the preset medicine-taking robotic arm in the multi-temperature zone drug storage unit to obtain the target medicinal materials in the multi-temperature zone drug storage unit according to the medicinal material priority queue; the end of the medicine-taking robotic arm is equipped with a flexible gripper and a near-infrared spectroscopy recognition probe; obtaining the medicinal material weight measured by the medicinal material weighing unit, and eliminating the weight error caused by the change in the moisture content of the medicinal materials in the medicinal material weight according to the dynamic weighing compensation algorithm; the dynamic weighing compensation algorithm corrects the medicinal material weight in real time based on the moisture absorption characteristic curve corresponding to the target medicinal materials.

[0074] Prescription parsing decomposes dynamic prescriptions into topological relationships of medicinal material components (such as the compatibility rules of monarch, minister, adjuvant and envoy). Priority queues generate queues based on the shelf life and frequency of use of medicinal materials, giving priority to medicinal materials that are about to expire to reduce waste. Medication collection is carried out by using a flexible gripper (adapting to different medicinal material forms) and a near-infrared spectral probe (to identify the authenticity of medicinal materials) to accurately grasp the target medicinal materials through the medicine collection robot arm. Weighing compensation dynamically corrects the weighing results according to the hygroscopic characteristic curve of the medicinal material (such as the humidity-weight relationship of astragalus) to eliminate the influence of environmental humidity. Intelligent medicine collection and dynamic weighing compensation improve the utilization rate and dispensing accuracy of medicinal materials, and flexible grippers and spectral recognition ensure the integrity and quality of medicinal materials.

[0075] In some embodiments, obtaining the second knowledge graph includes: extracting entity relationship triplets from a preset digital library of ancient Chinese medical books, a modern pharmacology database, and a clinical trial database; mapping the nature, flavor, and meridian data corresponding to the digital library of ancient Chinese medical books to the modern molecular pharmacology feature space of the modern pharmacology database according to a transfer learning framework; constructing an implicit association matrix of medicinal materials, diseases, and constitutions according to a generative adversarial network; and generating the second knowledge graph according to the implicit association matrix, the modern molecular pharmacology feature space, and the entity relationship triplets.

[0076] Entity relationship triplets (such as "Astragalus-enhancement-immunity") are extracted from the digital library of ancient Chinese medical books (such as the structured data of Compendium of Materia Medica), modern pharmacology databases (such as PubChem compound library) and clinical trial libraries. The nature, flavor and meridian data in ancient books (such as "warm in nature, belonging to the spleen meridian") are mapped to the molecular pharmacology feature space (such as target protein binding activity) through transfer learning. The generative adversarial network (GAN) is used to mine the implicit associations between medicinal materials, diseases and constitutions (such as "Coptis chinensis-damp-heat syndrome-high inflammatory factors") and generate an implicit association matrix. Integrate triplets, molecular features and implicit associations to construct a second knowledge graph that supports complex reasoning. The second knowledge graph integrates traditional experience and modern science to improve the comprehensiveness and explainability of prescription matching.

[0077] In some embodiments, the medicinal material extraction and packaging integrated machine comprises a supercritical carbon dioxide extraction device and a variable volume filler.

[0078] The supercritical carbon dioxide extraction device achieves efficient extraction of active ingredients of medicinal materials (such as flavonoids) by adjusting pressure (20-50MPa) and temperature (31-60℃); the variable volume filler adopts a telescopic cavity structure, and the capacity can be adjusted between 1-50mL to adapt to different dosage requirements. Supercritical extraction avoids high temperature damage to heat-sensitive components, and the variable filler realizes personalized dosage packaging.

[0079] Exemplarily, controlling the drug delivery robot to perform directional extraction, concentration adjustment, and segmented packaging according to the individualized dosing plan for the dispensed medicinal materials includes: dynamically adjusting the supercritical carbon dioxide extraction parameters corresponding to the supercritical carbon dioxide extraction device according to the characteristics of the medicinal material components corresponding to the dispensed medicinal materials; the supercritical carbon dioxide extraction parameters include a pressure gradient control curve and an entrainer ratio strategy; monitoring the concentration of the active ingredients in the extraction liquid corresponding to the directional extraction in real time by a multi-spectral fusion detection device, and adjusting the solvent evaporation rate so that the concentration of the active ingredients in the extraction liquid is the target concentration value; in the segmented packaging stage, performing nano-coating treatment on the inner wall of the variable-volume dispenser, and selecting a metal-organic framework material with specific adsorption characteristics as the coating material according to the characteristics of the medicinal material components; controlling the variable-volume dispenser to complete the dispensing of the extraction liquid.

[0080] By dynamically setting the pressure gradient control curve of supercritical CO2 (such as increasing pressure first and then decreasing pressure) and the ratio of entrainer (such as ethanol) according to the medicinal material components (such as the proportion of volatile components). Real-time monitoring of the extraction liquid concentration by multi-spectral fusion detection (ultraviolet-near infrared combination), and adjusting the evaporation rate to make the concentration meet the standard. Coating the inner wall of the dispenser with a metal-organic framework (MOF) coating (such as ZIF-8), selectively adsorbing easily oxidized components, and prolonging the stability of the medicine. Directional extraction maximizes the yield of active ingredients, and the MOF coating prevents component degradation during the dispensing process.

[0081] In some embodiments, the medicine depot management module further includes: a shelf life tracking unit, the multi-temperature zone medicine storage unit divides independent storage spaces according to the properties of the medicinal materials, and the shelf life tracking unit updates the remaining shelf life of the medicinal materials in the independent storage spaces in real time; if the remaining shelf life is lower than the preset shelf life, the control module generates a warning signal.

[0082] The shelf life tracking unit records the storage time of the medicinal materials through RFID tags, and combines the temperature and humidity sensor data of the independent storage space to predict the remaining shelf life of the medicinal materials. When the remaining shelf life of a certain batch of medicinal materials is lower than the threshold (such as 30 days), the control module triggers a warning signal and preferentially schedules their use. Real-time shelf life monitoring reduces the waste of expired medicinal materials and ensures the safety of drug use.

[0083] In some embodiments, the drug delivery robot further includes a medication monitoring unit, and the medication monitoring unit monitors the RFID tag corresponding to the user to be administered, associates the patient identity information corresponding to the user to be administered, and records the medication timestamp.

[0084] The medication monitoring unit reads the RFID wristband tag worn by the user, automatically associates the identity information, and records the medication timestamp. If the medication signal is not detected within the set time window (such as the dispenser not being opened), a reminder is pushed through the APP. To ensure patient compliance, the medication record provides data support for efficacy evaluation.

[0085] Please refer to Figure 2 , Figure 2 which is a schematic flowchart of a method for precise administration of traditional Chinese medicine provided by an embodiment of the present application. The execution device of the method is the control module of the device for precise administration of traditional Chinese medicine provided by any embodiment of the present application.

[0086] As Figure 2 shown, the provided method includes steps S101 to S104. Among them, the control module can be a handheld terminal, a laptop, a wearable device, or a robot, etc. It is used to implement steps S101 to S104 and their corresponding embodiments.

[0087] Step S101. Obtain the physical constitution information and disease information of the user to be administered medicine, and perform multi-dimensional feature extraction on the physical constitution information and disease information to obtain multi-dimensional feature information.

[0088] Specifically, the purpose of this step is to convert the user's physical constitution information (such as physiological indicators collected by wearable devices) and disease information (traditional Chinese medicine four diagnostic data, western medicine biochemical indicators) into unified multi-dimensional feature information through multi-source data collection and deep learning technology, providing a data basis for subsequent dynamic prescription generation.

[0089] Real-time collect data such as heart rate variability (HRV), body temperature fluctuations, and circadian rhythms through intelligent wearable devices (such as smart bracelets and smart clothes).

[0090] Traditional Chinese medicine four diagnostic data: tongue image (captured by a high-resolution camera), pulse condition (pulse position and pulse potential waveform collected by a pressure sensor array), and medical history record (structured electronic questionnaire). Western medicine biochemical indicators: blood routine, liver and kidney function test data (imported through the hospital LIS system). Synchronize multi-source data based on timestamps, for example, align the heart rate data of the wearable device with the time series of the pulse waveform to eliminate the deviation caused by the sampling frequency difference.

[0091] Use a pre-trained convolutional neural network (CNN) (such as ResNet-50) to extract tongue coating texture features (such as cracks and thickness) and the tongue color distribution matrix (RGB channel decomposition). The output is a 128-dimensional tongue image feature vector.

[0092] The time series waveforms of pulse position (superficial and deep) and pulse trend (slippery and unsmooth) are modeled using a bidirectional long short-term memory network (Bi-LSTM) to generate a three-dimensional pulse feature vector (such as pulse position intensity, pulse trend frequency, waveform complexity).

[0093] The circadian rhythm features of physiological signals are extracted through wavelet transform (such as Daubechies wavelet basis), such as the peak phase of body temperature cycle and the low-frequency / high-frequency power ratio of heart rate variability. The multi-head attention mechanism is used to perform weighted fusion on tongue images, pulse conditions, wearable data, and biochemical indicators to generate a 256-dimensional fusion feature vector and output traditional Chinese medicine syndrome classification labels (such as "phlegm-dampness accumulation" and "qi stagnation and blood stasis").

[0094] Through multi-modal data alignment and attention mechanism weighting, the heterogeneity problem between traditional Chinese medicine subjective experience and Western medicine objective indicators is solved, and the comprehensiveness of feature representation is improved. Deep learning algorithms replace traditional manual syndrome differentiation to reduce the impact of individual physician experience differences on the diagnosis results.

[0095] Step S102. Generate a dynamic prescription and an individualized dosing plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph.

[0096] Specifically, in this step, a preset first knowledge graph (including the association relationship between diseases, constitutions, and prescriptions) is used, combined with an optimization algorithm and dynamic constraint conditions, to generate an optimal prescription and a dosing plan adapted to the user's characteristics.

[0097] Map the fusion feature vector generated in step S101 to the disease constitution feature sub-graph of the first knowledge graph (the sub-graph nodes include syndromes such as "yin deficiency" and "damp-heat blood stasis", and the edges represent the association weights between syndromes and prescriptions). Through a graph embedding algorithm (such as GraphSAGE), the user feature vector and the knowledge graph nodes are embedded into the same low-dimensional space, and the cosine similarity is calculated to match the candidate prescription set (such as "modified Liuwei Dihuang Decoction" and "Xuefu Zhuyu Decoction").

[0098] Based on a graph neural network (GNN), a pharmacodynamic synergy model is constructed to predict the balance coefficient of the efficacy (such as anti-inflammatory activity) and toxicity (such as liver toxicity) of different combinations of medicinal materials in the candidate prescriptions. In the candidate prescription set, the Monte Carlo tree search (MCTS) is used to simulate the impact of increasing or decreasing medicinal materials on efficacy and toxicity, and the prescription with the highest efficacy-to-toxicity ratio (ETR) is selected.

[0099] Obtain the real-time inventory status of the pharmacy management module (such as insufficient inventory of astragalus membranaceus), and generate a dynamic prescription topology graph containing substitution paths according to the medicinal material substitution rules (such as "astragalus membranaceus → codonopsis pilosula"). Combining the user's previous medication records (such as being allergic to a certain medicinal material), the taboo nodes in the topology graph are removed.

[0100] Based on a pharmacokinetic simulation model (such as a two-compartment model), simulate the blood drug concentration curves under different dosing schedules (such as three times a day, pulsed dosing), and select a pulsed dosing schedule that meets the target therapeutic window (such as peak concentration ≤ toxicity threshold).

[0101] Adjust the prescription by combining real-time inventory and medication records to ensure the executability of the prescription. Optimize the dosing schedule through blood drug concentration simulation to avoid cumulative drug toxicity.

[0102] Step S103. Obtain a second knowledge graph; the second knowledge graph includes a traditional Chinese medicine efficacy association network, a disease-prescription matching model, and a patient characteristic analysis algorithm; the traditional Chinese medicine efficacy association network constructs the synergistic relationship of different medicinal materials through a graph neural network.

[0103] Specifically, the second knowledge graph integrates the experience of traditional Chinese medicine ancient books, modern molecular pharmacology data, and implicit associations to construct a medicinal material-disease-constitution association network that supports complex reasoning.

[0104] : Extract entity relationships such as the nature, flavor, and meridian tropism of medicinal materials (such as "Coptis chinensis: bitter, cold, belonging to the heart meridian") and the composition of prescriptions from the structured databases of ancient books such as Treatise on Febrile and Miscellaneous Diseases and Systematic Differentiation of Warm Diseases. Obtain the molecular targets (such as baicalin inhibiting COX-2) and pharmacokinetic parameters of the active ingredients of medicinal materials from modern pharmacological databases such as PubChem and ChEMBL. Extract the medicinal material-efficacy association (such as "Salvia miltiorrhiza → improving myocardial ischemia") from clinical electronic medical records. Use a pre-trained BERT model to map the description of "nature, flavor, and meridian tropism" in ancient books (such as "pungent-warm relieving exterior syndrome") to the modern molecular feature space (such as volatile oil components activating the TRPV1 channel). Mine the potential associations between medicinal materials-syndromes-biochemical indicators (such as "Coptis chinensis → damp-heat syndrome → downregulation of IL-6") through a generative adversarial network (GAN) to generate an implicit association matrix. Integrate the entity relationship triples (head entity-relationship-tail entity), molecular features, and implicit association matrix to construct a traditional Chinese medicine efficacy association network based on a graph neural network to support synergistic effect reasoning (such as "Astragalus membranaceus + Angelica sinensis" enhancing immune regulation).

[0105] By bridging traditional experience and modern science, improve the scientific basis for prescription matching. Continuously mine new associations through GAN to support the iterative optimization of the knowledge graph.

[0106] Step S104. Control the pharmacy management module to dispense medicinal materials according to the dynamic prescription, and control the delivery robot to perform directional extraction, concentration adjustment, and timed packaging according to the individualized dosing plan for the dispensed medicinal materials; the individualized dosing plan also includes a pulsed dosing schedule calculated based on the patient's metabolic cycle.

[0107] Specifically, this step realizes precise dispensing of medicinal materials, directional extraction of active ingredients, and time-segmented encapsulation through automated equipment to ensure the efficient execution of individualized medication regimens. According to the medicinal material components in the dynamic prescription topology diagram, a priority queue is generated (medicinal materials approaching expiration are used first), and the dispensing robotic arm is controlled to grab the target medicinal materials in the multi-temperature storage unit (such as the refrigerated area and the cool and shady area). The near-infrared probe (wavelength range 900 - 1700nm) at the end of the robotic arm scans the fingerprint spectrum of the medicinal materials in real time and compares it with the database to verify authenticity (such as differentiating Astragalus membranaceus from the counterfeit Althaea rosea root). According to the moisture absorption characteristic curve of the medicinal materials (such as the humidity-weight relationship of Poria cocos), real-time compensation is performed on the data of the medicinal material weighing unit. Optimization of supercritical CO2 extraction parameters: Dynamically adjust the pressure gradient (such as rising from 20MPa to 40MPa and then slowly decreasing) and the ratio of entrainer (such as ethanol) (5% - 15%) according to the characteristics of the medicinal material components (such as the proportion of volatile oil). The combination of ultraviolet-visible spectroscopy (detecting flavonoids) and near-infrared spectroscopy (detecting polysaccharides) is used to provide real-time feedback on the concentration of the extraction solution, and the solvent evaporation rate is adjusted through the PID control algorithm. According to the pulsed medication timing sequence (such as different doses in the morning, noon, and evening), control the telescopic cavity of the dispenser to adjust the volume (with an accuracy of 1mL). Spray a ZIF-8 type metal-organic framework (MOF) coating on the inner wall of the dispenser to selectively adsorb easily oxidized components (such as tanshinone) to prevent component degradation during the encapsulation process.

[0108] Priority scheduling and near-infrared identification reduce the risks of medicinal material waste and counterfeits. Supercritical extraction and MOF coating technology ensure the activity and stability of the active ingredients of the medicine. The variable-volume dispenser adapts to complex medication regimens and improves patient compliance.

[0109] In some embodiments, the disease information includes traditional Chinese medicine four-diagnosis data and Western medicine biochemical indicators; the multi-dimensional feature information includes a fusion feature vector; the constitution information includes monitoring data collected by a preset wearable device; the obtaining of the multi-dimensional feature information by performing multi-dimensional feature extraction on the constitution information and the disease information includes: aligning the features of the traditional Chinese medicine four-diagnosis data, Western medicine biochemical indicators, and monitoring data; extracting the tongue coating texture distribution matrix for the tongue image features corresponding to the traditional Chinese medicine four-diagnosis data according to a convolutional neural network, generating a three-dimensional feature vector of pulse position and pulse trend for the pulse condition data through a long short-term memory network, and extracting the circadian rhythm features from the monitoring data through wavelet transform; and weighted-fusing the tongue coating texture distribution matrix, the three-dimensional feature vector of pulse position and pulse trend, the circadian rhythm features, and the Western medicine biochemical indicators according to the attention mechanism to generate the fusion feature vector and the traditional Chinese medicine syndrome classification label corresponding to the fusion feature vector.

[0110] In some embodiments, generating the dynamic prescription and individualized administration plan corresponding to the multi-dimensional feature information according to the preset first knowledge graph includes: mapping the multi-dimensional feature information to the disease constitution feature sub-graph of the first knowledge graph, and matching the candidate prescription set corresponding to the multi-dimensional feature information in the disease constitution feature sub-graph through a graph embedding algorithm; obtaining the efficacy-toxicity balance coefficient corresponding to each candidate prescription in the candidate prescription set through Monte Carlo tree search according to the preset efficacy synergy prediction model to complete the optimization of the candidate prescription set; obtaining the real-time drug library inventory status corresponding to the drug library management module and the past medication records of the patient corresponding to the user to be administered, and generating a dynamic prescription topology graph according to the past medication records of the patient and the real-time drug library inventory status; the dynamic prescription topology graph includes a medicinal material substitution plan; generating the dynamic prescription according to the dynamic prescription topology graph and the optimized candidate prescription set; generating the blood drug concentration curve corresponding to the dynamic prescription at different administration time sequences according to the preset pharmacokinetic simulation model, and generating the individualized administration plan according to the blood drug concentration curve.

[0111] In some embodiments, controlling the drug library management module to dispense medicinal materials according to the dynamic prescription includes: parsing the dynamic prescription to obtain the topological relationship of the medicinal material components corresponding to the dynamic prescription; generating a medicinal material priority queue according to the topological relationship of the medicinal material components; in the medicinal material priority queue, the medicinal materials approaching the expiration date are preferentially matched with the prescriptions with high usage frequencies; controlling a preset medicine-taking robotic arm in the multi-temperature zone medicine storage unit to obtain the target medicinal materials in the multi-temperature zone medicine storage unit according to the medicinal material priority queue; the end of the medicine-taking robotic arm is equipped with a flexible gripper and a near-infrared spectroscopy identification probe; obtaining the medicinal material weight of the target medicinal materials measured by the medicinal material weighing unit, and eliminating the weight error caused by the change in the moisture content of the medicinal materials in the medicinal material weight according to the dynamic weighing compensation algorithm; the dynamic weighing compensation algorithm corrects the medicinal material weight in real time based on the moisture absorption characteristic curve corresponding to the target medicinal materials.

[0112] In some embodiments, obtaining the second knowledge graph includes: extracting entity-relationship triples from a preset digitalized traditional Chinese medicine ancient book library, a modern pharmacology database, and a clinical trial database; mapping the property-flavor and meridian tropism data corresponding to the digitalized traditional Chinese medicine ancient book library to the modern molecular pharmacology feature space of the modern pharmacology database according to a transfer learning framework; constructing an implicit association matrix of medicinal materials, diseases, and constitutions according to a generative adversarial network; generating the second knowledge graph according to the implicit association matrix, the modern molecular pharmacology feature space, and the entity-relationship triples.

[0113] In some embodiments, the medicinal material extraction and encapsulation integrated machine includes a supercritical carbon dioxide extraction device and a variable-volume packaging device.

[0114] Exemplarily, controlling the drug delivery robot to perform directional extraction, concentration adjustment, and time-segmented packaging according to the individualized medication plan for the dispensed medicinal materials includes: dynamically adjusting the supercritical carbon dioxide extraction parameters corresponding to the supercritical carbon dioxide extraction device according to the characteristics of the medicinal material components corresponding to the dispensed medicinal materials; the supercritical carbon dioxide extraction parameters include a pressure gradient control curve and an entrainer ratio strategy; monitoring the concentration of the active ingredients in the extraction solution corresponding to the directional extraction in real time by a multi-spectral fusion detection device, and adjusting the solvent evaporation rate so that the concentration of the active ingredients in the extraction solution is the target concentration value; in the time-segmented packaging stage, performing nano-coating treatment on the inner wall of the variable-volume dispenser, and selecting a metal-organic framework material with specific adsorption characteristics as the coating material according to the characteristics of the medicinal material components; controlling the variable-volume dispenser to complete the dispensing of the extraction solution.

[0115] In some embodiments, the medicine warehouse management module further includes: a shelf life tracking unit. The multi-temperature zone medicine storage unit divides independent storage spaces according to the attributes of the medicinal materials, and the shelf life tracking unit updates the remaining shelf life of the medicinal materials in the independent storage spaces in real time; if the remaining shelf life is lower than the preset shelf life, the control module generates an alarm signal.

[0116] In some embodiments, the drug delivery robot further includes a medication monitoring unit. The medication monitoring unit monitors the RFID tag corresponding to the user to be administered the drug, associates the patient identity information corresponding to the user to be administered the drug, and records the medication timestamp.

[0117] It should be noted that those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the methods and steps for precise administration of traditional Chinese medicine preparations described above can refer to the corresponding processes in the device embodiments for precise administration of traditional Chinese medicine preparations described in the above embodiments, and will not be elaborated here.

[0118] The embodiments of the present application also provide a device for precise administration of traditional Chinese medicine preparations. The device for precise administration of traditional Chinese medicine preparations is used to execute the steps of the methods for precise administration of traditional Chinese medicine preparations shown in the above embodiments. The device for precise administration of traditional Chinese medicine preparations can be a single server or a server cluster, or the device for precise administration of traditional Chinese medicine preparations can be a terminal, and the terminal can be a handheld terminal, a laptop computer, a wearable device, or a robot, etc.

[0119] The device for precise administration of traditional Chinese medicine preparations includes:

[0120] An information acquisition unit, configured to acquire the physical constitution information and disease information of the user to be administered the drug, and perform multi-dimensional feature extraction on the physical constitution information and disease information to obtain multi-dimensional feature information;

[0121] A solution generation unit, configured to generate a dynamic prescription and an individualized dosing plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph;

[0122] A relationship construction unit, configured to obtain a second knowledge graph; the second knowledge graph includes a traditional Chinese medicine efficacy association network, a disease prescription matching model, and a patient feature analysis algorithm; the traditional Chinese medicine efficacy association network constructs the synergistic relationship of different medicinal materials through a graph neural network;

[0123] A solution control unit, configured to control the pharmacy management module to dispense medicinal materials according to the dynamic prescription, and control the delivery robot to perform directional extraction, concentration adjustment, and packaging in different time periods according to the individualized dosing plan for the dispensed medicinal materials; the individualized dosing plan further includes a pulsed dosing time sequence calculated based on the patient's metabolic cycle.

[0124] It should be noted that those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the above-described device for accurately administering traditional Chinese medicine preparations and each unit can refer to the corresponding processes in the method embodiments for accurately administering traditional Chinese medicine preparations described in the above embodiments, and will not be elaborated here.

[0125] The above method for accurately administering traditional Chinese medicine preparations is implemented in the form of a computer program, and this computer program can run on the above device.

[0126] Please refer to Figure 3 , Figure 3 , which is a schematic block diagram of the structure of the control module provided by an embodiment of the present application. The control module includes a processor, a memory, and a network interface connected through a device bus. Among them, the memory can include a storage medium and an internal memory.

[0127] The storage medium can store an operating device and a computer program. This computer program includes program instructions, and when the program instructions are executed, the processor can execute any embodiment of the method for accurately administering traditional Chinese medicine preparations.

[0128] The processor is used to provide computing and control capabilities to support the operation of the entire control module.

[0129] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any method for accurately administering traditional Chinese medicine preparations based on the device.

[0130] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art can understand that Figure 3The structure shown is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the terminal to which the solution of this application is applied. The specific control module may include more or fewer components than those shown in the figure, or combine some components, or have a different component arrangement.

[0131] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0132] Wherein, in one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0133] Obtain the physical constitution information and disease information of the user to be administered medicine, and perform multi-dimensional feature extraction on the physical constitution information and disease information to obtain multi-dimensional feature information;

[0134] Generate a dynamic prescription and an individualized dosing plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph;

[0135] Obtain a second knowledge graph; the second knowledge graph includes a traditional Chinese medicine efficacy association network, a disease prescription matching model, and a patient characteristic analysis algorithm; the traditional Chinese medicine efficacy association network constructs the synergistic relationship between different medicinal materials through a graph neural network;

[0136] Control the medicine library management module to dispense medicinal materials according to the dynamic prescription, and control the medicine delivery robot to perform directional extraction, concentration adjustment and timed packaging according to the dosing plan on the dispensed medicinal materials; the individualized dosing plan also includes a pulsed dosing time sequence calculated based on the patient's metabolic cycle.

[0137] In some embodiments, the disease information includes traditional Chinese medicine (TCM) four diagnostic data and Western medicine biochemical indicators; the multi-dimensional feature information includes a fusion feature vector; the constitution information includes monitoring data collected by a preset wearable device; and the obtaining of the multi-dimensional feature information by performing multi-dimensional feature extraction on the constitution information and the disease information includes: performing feature alignment on the TCM four diagnostic data, Western medicine biochemical indicators, and monitoring data; extracting a tongue coating texture distribution matrix for the tongue image features corresponding to the TCM four diagnostic data according to a convolutional neural network, generating a three-dimensional feature vector of pulse position and pulse trend for the pulse condition data through a long short-term memory network, and extracting circadian rhythm features from the monitoring data through wavelet transform; and generating the fusion feature vector and the TCM syndrome classification label corresponding to the fusion feature vector by weighted fusion of the tongue coating texture distribution matrix, the three-dimensional feature vector of pulse position and pulse trend, the circadian rhythm features, and the Western medicine biochemical indicators according to an attention mechanism.

[0138] In some embodiments, the generating of the dynamic prescription and individualized administration plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph includes: mapping the multi-dimensional feature information to a disease-constitution feature sub-graph of the first knowledge graph, and matching a candidate prescription set corresponding to the multi-dimensional feature information in the disease-constitution feature sub-graph through a graph embedding algorithm; obtaining the efficacy-toxicity balance coefficient corresponding to each candidate prescription in the candidate prescription set through Monte Carlo tree search according to a preset pharmacodynamic synergy prediction model to complete the optimization of the candidate prescription set; obtaining the real-time inventory status of the medicine library management module and the past medication records of the patient corresponding to the user to be administered, and generating a dynamic prescription topology graph according to the past medication records of the patient and the real-time inventory status of the medicine library; the dynamic prescription topology graph includes a medicinal material substitution plan; generating the dynamic prescription according to the dynamic prescription topology graph and the optimized candidate prescription set; generating a blood drug concentration curve corresponding to the dynamic prescription at different administration time sequences according to a preset pharmacokinetic simulation model, and generating the individualized administration plan according to the blood drug concentration curve.

[0139] In some embodiments, the control of the medicine depot management module to dispense medicinal materials according to the dynamic prescription includes: parsing the dynamic prescription to obtain the topological relationship of the medicinal material components corresponding to the dynamic prescription; generating a medicinal material priority queue according to the topological relationship of the medicinal material components; in the medicinal material priority queue, the medicinal materials approaching the expiration date are preferentially matched with the prescriptions with high usage frequencies; controlling a preset medicine-taking robotic arm in the multi-temperature zone medicine storage unit to obtain target medicinal materials in the multi-temperature zone medicine storage unit according to the medicinal material priority queue; a flexible gripper and a near-infrared spectroscopy identification probe are equipped at the end of the medicine-taking robotic arm; obtaining the medicinal material weight of the target medicinal materials measured by the medicinal material weighing unit, and eliminating the weight error caused by the change of the moisture content of the medicinal materials in the medicinal material weight according to the dynamic weighing compensation algorithm; the dynamic weighing compensation algorithm corrects the medicinal material weight in real time based on the moisture absorption characteristic curve corresponding to the target medicinal materials.

[0140] In some embodiments, the obtaining of the second knowledge graph includes: extracting entity-relationship triples from a preset digitalized traditional Chinese medicine ancient book database, a modern pharmacology database, and a clinical trial database; mapping the property and meridian tropism data corresponding to the digitalized traditional Chinese medicine ancient book database to the modern molecular pharmacology feature space of the modern pharmacology database according to a transfer learning framework; constructing an implicit association matrix of medicinal materials, diseases, and constitutions according to a generative adversarial network; generating the second knowledge graph according to the implicit association matrix, the modern molecular pharmacology feature space, and the entity-relationship triples.

[0141] In some embodiments, the medicinal material extraction and encapsulation integrated machine includes a supercritical carbon dioxide extraction device and a variable-volume dispenser.

[0142] Exemplarily, the control of the medicine delivery robot to perform directional extraction, concentration adjustment, and time-segmented encapsulation according to the individualized dosing plan for the dispensed medicinal materials includes: dynamically adjusting the supercritical carbon dioxide extraction parameters corresponding to the supercritical carbon dioxide extraction device according to the medicinal material component characteristics corresponding to the dispensed medicinal materials; the supercritical carbon dioxide extraction parameters include a pressure gradient control curve and an entrainer ratio strategy; monitoring the concentration of the active ingredients of the extraction solution corresponding to the directional extraction in real time according to a multi-spectral fusion detection device, and adjusting the solvent evaporation rate so that the concentration of the active ingredients of the extraction solution is the target concentration value; in the time-segmented encapsulation stage, performing a nano-coating treatment on the inner wall of the variable-volume dispenser, and the coating material selects a metal-organic framework material with specific adsorption characteristics according to the medicinal material component characteristics; controlling the variable-volume dispenser to complete the dispensing of the extraction solution.

[0143] In some embodiments, the medicine depot management module further includes: a shelf life tracking unit. The multi-temperature zone medicine storage unit divides independent storage spaces according to the properties of the medicinal materials, and the shelf life tracking unit updates the remaining shelf life of the medicinal materials in the independent storage spaces in real time; if the remaining shelf life is lower than the preset shelf life, the control module generates a warning signal.

[0144] In some embodiments, the medicine delivery robot further includes a medication monitoring unit. The medication monitoring unit monitors the RFID tag corresponding to the user to be administered medicine, associates the patient identity information corresponding to the user to be administered medicine, and records the medication timestamp.

[0145] It should be noted that those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working process of the above-described processor can refer to the corresponding process in the method embodiments described in the above various embodiments, and will not be elaborated here.

[0146] An embodiment of the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and the computer program includes program instructions. The processor executes the program instructions to implement the steps of the method for precisely administering traditional Chinese medicine provided in the above various embodiments of the present application.

[0147] Among them, the computer-readable storage medium may be the internal storage unit of the control module described in the foregoing embodiments, such as the hard disk or memory of the control module. The computer-readable storage medium may also be an external storage device of the control module, such as a plug-in hard disk equipped on the control module, a smart media card (SMC), a secure digital (SD) card, a flash card, etc.

[0148] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or substitutions within the technical scope disclosed by the present application, and these modifications or substitutions should all be covered by the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A device for precise administration of traditional Chinese medicine preparations, characterized in that, Including: A control module, which acquires the physical constitution information and disease information of the user to be administered medicine, extracts multi-dimensional features from the physical constitution information and disease information to obtain multi-dimensional feature information, and generates a dynamic prescription and an individualized dosing plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph; the control module acquires a second knowledge graph; the second knowledge graph includes a traditional Chinese medicine efficacy association network, a disease-prescription matching model, and a patient characteristic analysis algorithm; the traditional Chinese medicine efficacy association network constructs the synergistic relationship of different medicinal materials through a graph neural network; A medicine library management module, including a multi-temperature zone medicine storage unit and a medicinal material weighing unit, where the multi-temperature zone medicine storage unit divides independent storage spaces according to the attributes of the medicinal materials; A medicine delivery robot, including a medicinal material extraction and packaging integrated machine; Among them, the control module controls the medicine library management module to dispense medicinal materials according to the dynamic prescription, and controls the medicine delivery robot to perform directional extraction, concentration adjustment and timed packaging according to the individualized dosing plan on the dispensed medicinal materials; the individualized dosing plan also includes a pulsed dosing time sequence calculated based on the patient's metabolic cycle.

2. The device according to claim 1, wherein The disease information includes traditional Chinese medicine four diagnostic data and Western medicine biochemical indicators; the multi-dimensional feature information includes a fusion feature vector; the physical constitution information includes monitoring data collected by a preset wearable device; The extracting multi-dimensional features from the physical constitution information and disease information to obtain multi-dimensional feature information includes: Aligning the features of the traditional Chinese medicine four diagnostic data, Western medicine biochemical indicators and monitoring data; Extracting the tongue coating texture distribution matrix according to the convolutional neural network for the tongue image features corresponding to the traditional Chinese medicine four diagnostic data, generating a three-dimensional feature vector of the pulse position and pulse potential through a long short-term memory network for the pulse condition data, and extracting the circadian rhythm features from the monitoring data through wavelet transform; Weightedly fusing the tongue coating texture distribution matrix, the three-dimensional feature vector of the pulse position and pulse potential, the circadian rhythm features and the Western medicine biochemical indicators according to the attention mechanism to generate the fusion feature vector and the traditional Chinese medicine syndrome classification label corresponding to the fusion feature vector.

3. The device according to claim 1, characterized in that, The generating the dynamic prescription and the individualized dosing plan corresponding to the multi-dimensional feature information according to the preset first knowledge graph includes: Mapping the multi-dimensional feature information to the disease-physical constitution feature sub-graph of the first knowledge graph, and matching the candidate prescription set corresponding to the multi-dimensional feature information in the disease-physical constitution feature sub-graph through a graph embedding algorithm; Obtaining the efficacy-toxicity balance coefficient corresponding to each candidate prescription in the candidate prescription set through Monte Carlo tree search according to a preset efficacy synergy prediction model, and completing the optimization of the candidate prescription set; Obtaining the real-time inventory status of the medicine library corresponding to the medicine library management module and the patient's previous medication records corresponding to the user to be administered medicine, and generating a dynamic prescription topology graph according to the patient's previous medication records and the real-time inventory status of the medicine library; the dynamic prescription topology graph includes a medicinal material substitution plan; Generating the dynamic prescription according to the dynamic prescription topology graph and the optimized candidate prescription set; Generate the blood drug concentration curves corresponding to the dynamic prescriptions under different drug administration timings according to a preset pharmacokinetic simulation model, and generate the individualized dosing plan according to the blood drug concentration curves.

4. The device according to claim 1, characterized in that Controlling the drug storage management module to dispense medicinal materials according to the dynamic prescription includes: Analyze the dynamic prescription to obtain the topological relationship of the medicinal material components corresponding to the dynamic prescription; Generate a medicinal material priority queue according to the topological relationship of the medicinal material components; in the medicinal material priority queue, the medicinal materials approaching the expiration date are preferentially matched with the prescriptions with high usage frequencies; Control a preset medicine-taking robotic arm in the multi-temperature zone medicine storage unit to obtain target medicinal materials in the multi-temperature zone medicine storage unit according to the medicinal material priority queue; a flexible gripper and a near-infrared spectrum identification probe are equipped at the end of the medicine-taking robotic arm; Obtain the medicinal material weight of the target medicinal material measured by the medicinal material weighing unit, and eliminate the weight error caused by the change in the moisture content of the medicinal material in the medicinal material weight according to a dynamic weighing compensation algorithm; the dynamic weighing compensation algorithm corrects the medicinal material weight in real time based on the moisture absorption characteristic curve corresponding to the target medicinal material.

5. The device according to claim 1, characterized in that Obtaining the second knowledge graph includes: Extract entity-relationship triples from a preset digital library of traditional Chinese medicine ancient books, modern pharmacology database, and clinical trial database; Map the property and meridian tropism data corresponding to the digital library of traditional Chinese medicine ancient books to the modern molecular pharmacology feature space of the modern pharmacology database according to a transfer learning framework; Construct an implicit association matrix of medicinal materials, diseases, and constitutions according to a generative adversarial network; Generate the second knowledge graph according to the implicit association matrix, modern molecular pharmacology feature space, and entity-relationship triples.

6. The device according to claim 1, characterized in that The medicinal material extraction and encapsulation integrated machine includes a supercritical carbon dioxide extraction device and a variable-volume packaging device.

7. The device according to claim 6, characterized in that, Controlling the medicine delivery robot to perform directional extraction, concentration adjustment, and time-segmented encapsulation of the dispensed medicinal materials according to the individualized dosing plan includes: Dynamically adjust the supercritical carbon dioxide extraction parameters corresponding to the supercritical carbon dioxide extraction device according to the medicinal material component characteristics corresponding to the dispensed medicinal materials; the supercritical carbon dioxide extraction parameters include a pressure gradient control curve and an entrainer ratio strategy; Monitor the concentration of the active ingredients in the extraction solution corresponding to the directional extraction in real time according to a multi-spectral fusion detection device, and adjust the solvent evaporation rate so that the concentration of the active ingredients in the extraction solution is the target concentration value; In the time-segmented encapsulation stage, perform nano-coating treatment on the inner wall of the variable-volume packaging device, and select a metal-organic framework material with specific adsorption characteristics as the coating material according to the medicinal material component characteristics; control the variable-volume packaging device to complete the packaging of the extraction solution.

8. The device according to claim 1, characterized in that The drug storage management module further includes: An expiration date tracking unit, the multi-temperature zone medicine storage unit divides independent storage spaces according to the medicinal material attributes, and the expiration date tracking unit updates the remaining expiration date of the medicinal materials in the independent storage spaces in real time; If the remaining expiration date is lower than the preset expiration date, the control module generates a warning signal.

9. The device according to claim 1, characterized in that The medicine delivery robot further includes a medication monitoring unit, which associates the patient identity information corresponding to the user to be medicated by monitoring the RFID tag corresponding to the user to be medicated and records the medication timestamp.

10. A method for precise administration of traditional Chinese medicine preparations, characterized in that, For the control module applied to the device according to any one of claims 1-9, the method includes: Obtaining the physical constitution information and disease information of the user to be medicated, and performing multi-dimensional feature extraction on the physical constitution information and disease information to obtain multi-dimensional feature information; Generating a dynamic prescription and an individualized dosage plan corresponding to the multi-dimensional feature information according to a preset first knowledge graph; Obtaining a second knowledge graph; the second knowledge graph includes a traditional Chinese medicine efficacy association network, a disease-prescription matching model, and a patient feature analysis algorithm; the traditional Chinese medicine efficacy association network constructs the synergistic relationship between different medicinal materials through a graph neural network; Controlling the medicine library management module to dispense medicinal materials according to the dynamic prescription, and controlling the medicine delivery robot to perform directional extraction, concentration adjustment, and time-segmented packaging according to the dosage plan on the dispensed medicinal materials; the individualized dosage plan further includes a pulsed drug delivery timing calculated based on the patient's metabolic cycle.

Images