Clinical pharmacy intelligent management method and system and storage medium

Abstract

The invention relates to the technical field of pharmacy management, in particular to a clinical pharmacy intelligent management method and system and a storage medium. The method comprises the following steps: acquiring total element data of a drug treatment chain in real time, wherein the total element data comprises a prescription instruction stream, a drug preparation parameter, a drug administration execution record and a patient physiological feedback signal; constructing a drug treatment chain dynamic evaluation engine; establishing a digital twinning mapping relationship between the administration operation and the clinical terminal for the total element data of the drug treatment chain by utilizing a drug treatment chain dynamic evaluation engine, and generating a drug digital relationship chain; and performing drug treatment chain breakage risk point identification on the drug digital relation chain based on a multi-dimensional deviation detection algorithm to obtain drug treatment abnormal risk points. Through real-time data acquisition, digital twinborn mapping and intelligent risk identification and intervention, intelligent management and accurate regulation and control of the drug treatment chain are improved, and intelligence and accuracy of drug treatment chain management are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pharmaceutical management, and particularly to a method, system and storage medium for intelligent management of clinical pharmacy. Background Art

[0002] Initially, pharmaceutical management was limited to manually recording the procurement, inventory and distribution of drugs, relying on manual experience to judge drug compatibility and dosage. In the 1990s, the hospital information management system (HIS) gradually developed, and clinical pharmacy began to combine with computer technology, and drug management gradually became digital. Especially in drug inventory management and prescription review, the pharmacy information system (PIS) began to play a role. With the intelligent pharmaceutical management system using machine learning and natural language processing technologies, it is possible to automatically identify and warn of potential risks such as adverse drug reactions and drug interactions, improving the accuracy and safety of drug management. However, the current existing technologies usually analyze the drug treatment chain statically and singly, lacking data-based dynamic evaluation. At the same time, the risks of the drug treatment chain usually rely on manual judgment and experience, which are prone to delays or omissions, resulting in lower intelligence and accuracy in the management of the drug treatment chain. Summary of the Invention

[0003] Based on this, it is necessary to provide a method, system and storage medium for intelligent management of clinical pharmacy to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for intelligent management of clinical pharmacy, the method includes the following steps:

[0005] Step S1: Real-time obtain all-element data of the drug treatment chain, including prescription instruction flow, drug preparation parameters, administration execution records and patient physiological feedback signals;

[0006] Step S2: Construct a dynamic evaluation engine for the drug treatment chain; use the dynamic evaluation engine for the drug treatment chain to match the clinical medication mapping relationship of all-element data of the drug treatment chain, generate an individualized medication path map; extract the topological chain from the individualized medication path map through digital twin technology to generate a drug digital relationship chain;

[0007] Step S3: Identify risk points of drug treatment chain breakage for the drug digital relationship chain based on a multi-dimensional deviation detection algorithm to obtain abnormal drug treatment risk points; set drug intervention nodes for the drug digital relationship chain according to the abnormal drug treatment risk points to obtain drug intervention nodes;

[0008] Step S4: Judge the coverage of drug intervention nodes; trigger an intelligent compensation mechanism based on the coverage of drug intervention nodes, and perform intelligent regulation of drug dosage on all-element data of the drug treatment chain through the intelligent compensation mechanism to execute the intelligent management operation of clinical pharmacy.

[0009] By obtaining drug preparation, drug usage records, and physiological feedback data in real time, the present invention realizes the monitoring of the entire process of drug use. This method ensures that every link in the drug management process can be tracked and recorded in real time, thereby improving the management accuracy and response speed. By constructing a digital twin mapping of the drug treatment chain, a real-time digital model can be provided for each drug management link. This dynamic mapping relationship can provide real-time feedback to help relevant personnel optimize and adjust the drug management process at any time, thereby reducing human errors and improving the system operation efficiency. Using a multi-dimensional deviation detection algorithm to analyze the drug management process can detect potential risk points in advance. This algorithm can efficiently and accurately identify deviations in the process to help relevant personnel adjust the work plan in a timely manner, thereby reducing any defects or hidden dangers in the system. Once an abnormal risk is detected, the system will automatically adjust according to the set intervention nodes. This automated intervention measure ensures that the drug management process can respond according to individual needs, thereby improving the system's adaptive ability and ensuring the best management effect. When the system detects that there is a link that needs to be supplemented in the drug management process, the intelligent compensation mechanism will be automatically triggered to adjust the drug management strategy to ensure the coherence and efficiency of the process. This intelligent mechanism not only improves the drug management efficiency but also ensures the accuracy and timeliness of each link. By constructing a drug digital relationship chain, the data of each link can be clearly tracked. This not only provides a complete operation record for managers but also helps with post-event analysis and evaluation to ensure the transparency and operability of each link. With the help of a dynamic evaluation engine and data analysis tools, the system can provide real-time and accurate data support for managers, thereby optimizing the decision-making process. This enables managers to identify problems more efficiently, formulate countermeasures, and improve the operation efficiency of the entire management system. Therefore, the present invention improves the intelligent management and precise control of the drug treatment chain through real-time data acquisition, digital twin mapping, intelligent risk identification, and intervention, and enhances the intelligence and precision of drug treatment chain management.

[0010] Preferably, the construction of the dynamic evaluation engine of the drug treatment chain in step S2 includes:

[0011] Construct a dynamic pharmacy ontology framework, including decision rules in three dimensions: drug-gene interaction, metabolic pathway coupling, and drug delivery device adaptation;

[0012] Adopt path discretization modeling technology to decompose the full-element data of the drug treatment chain into medication decision units with time-dependent relationships;

[0013] Connect the medication decision units into a medication path; perform real-time deduction on the medication path based on a fuzzy Petri net, and dynamically update the activation probability of each medication decision unit through the results of the real-time deduction;

[0014] Establish a path credibility verification mechanism to trigger an artificial review process when the matching degree between the newly generated medication path and the preset historical successful case library is lower than 65%.

[0015] Through constructing a dynamic pharmaceutical ontology framework and combining multi-dimensional decision rules of drug-gene interaction, metabolic pathway coupling, and administration device adaptation, the present invention can comprehensively and precisely dynamically evaluate each link in the drug treatment chain. Such a framework can better integrate multiple influencing factors of drugs, ensure that decisions are based on comprehensive information, and improve the accuracy and scientific nature of drug use management. The path discretization modeling technology is adopted to convert the whole-element data of the drug treatment chain into medication decision-making units with time-dependent relationships, enabling each decision-making unit to make an optimal decision at a specific time point. Through this time-sequential management method, any temporal changes in the treatment process can be captured, avoiding missing key factors affecting the treatment effect. By using a fuzzy Petri net to deduce the medication path in real time, the activation probability of each medication decision-making unit can be dynamically updated, enabling the system to adapt to the physiological changes of the patient or external environmental changes in real time. This adaptive ability ensures that the drug use plan can be flexibly adjusted throughout the treatment process, thereby improving the adaptability and precision of the treatment. Through the path credibility verification mechanism, the system can compare the newly generated medication path with historical successful cases. When the matching degree is lower than 65%, an artificial review process is triggered, thus ensuring the reliability and safety of the medication path. This mechanism avoids the limitations of over-reliance on algorithms in automated decision-making, guarantees the accuracy of decisions through artificial review, and further enhances the credibility and safety of the system. Through real-time deduction and path verification mechanisms, potential risk points in the drug treatment chain can be quickly discovered and corresponding adjustment measures can be taken. The real-time deduction of the medication path by the system enables each medication decision to be corrected in a timely manner, thus avoiding the implementation of a drug treatment plan that is not suitable for the patient and greatly reducing the risk of adverse reactions during the drug treatment process.

[0016] Preferably, the construction method of the dynamic pharmaceutical ontology framework includes:

[0017] Establish a drug action quadruple model, where the quadruple includes chemical entities, target proteins, metabolic enzyme systems, and transport carriers, and each element is associated with at least 3,000 verified biomedical entities; adapt decision rules for the drug action quadruple model to generate drug-gene interaction decision rules;

[0018] Define a clinical path constraint rule library, including three types of constraint conditions: contraindication conflict detection rules, administration timing optimization rules, and device compatibility verification rules; adapt decision rules for the clinical path constraint rule library to generate metabolic pathway coupling decision rules;

[0019] Construct a drug property attenuation model, adapt the decision rules of the drug property attenuation model, and generate decision rules for dosing device adaptation; construct a dynamic pharmaceutical ontology framework based on drug-gene interaction decision rules, metabolic pathway coupling decision rules, and dosing device adaptation decision rules.

[0020] By establishing a drug action quadruple model (including chemical entities, target proteins, metabolic enzyme systems, and transport carriers), the present invention can comprehensively capture the interactions between drugs and organisms. This quadruple model can not only accurately describe the mechanism of drug action but also ensure the comprehensiveness and accuracy of drug action by associating a large number of verified biomedical entities (at least 3000 entities are associated with each element), which helps to predict and optimize drug effects more deeply. The adaptation of decision rules for the drug action quadruple model can generate drug-gene interaction decision rules, which can help managers better understand the relationship between drugs and genes, thereby formulating more accurate drug usage plans and improving the personalization and effectiveness of drug treatment. The defined clinical pathway constraint rule library (including contraindication conflict detection rules, dosing timing optimization rules, and device compatibility verification rules) provides an important optimization framework for the drug treatment process. By adapting the decision rules of the clinical pathway constraint rule library, metabolic pathway coupling decision rules can be generated, which can effectively detect and avoid contraindication conflicts, optimize dosing timing, and ensure device compatibility. This pathway optimization helps to improve the safety and smoothness of drug treatment. Through the drug property attenuation model and its decision rule adaptation, dosing device adaptation decision rules are generated, which makes the matching between drug properties and dosing devices more accurate and ensures that the effectiveness of drugs is maximally utilized throughout the treatment process. The dynamic pharmaceutical ontology framework can automatically adjust decisions to adapt to different drug property changes and ensure that the treatment plan always matches the actual needs of patients. By combining the decision rules of drug-gene interaction, metabolic pathway coupling, and dosing device adaptation to construct a dynamic pharmaceutical ontology framework, the system can provide multi-dimensional decision support for drug management. This integrated decision framework can optimize drug use from multiple perspectives, avoiding the limitations brought by single rules, and thus improving the comprehensive efficiency of drug treatment management.

[0021] Preferably, the real-time deduction of the medication path is based on a fuzzy Petri net, and the activation probabilities of each medication decision unit are dynamically updated through the results of the real-time deduction, including:

[0022] Set a path node activation threshold function, where the formula of the path node activation threshold function is as follows:

[0023]

[0024] In the formula, P activateis the activation threshold of the path node, k is the drug sensitivity coefficient, and c actual is the real-time blood drug concentration monitoring value, and c standard is the standard reference value;

[0025] Calculate the activation probability of the unit node of the medication path through the path node activation threshold function to obtain the activation probability of the path unit node;

[0026] Analyze the difference in activation probability of adjacent nodes for the medication path according to the activation probability of the path unit node. When the difference in activation probability of adjacent nodes exceeds 40%, automatically insert a pharmaceutical care marker point;

[0027] Use the pharmaceutical care marker point to select different paths for the medication path, and deduce the path conflict of different paths through the fuzzy Petri net to obtain the result of real-time deduction;

[0028] Dynamically adjust the activation probability of the medication decision-making unit based on the result of real-time deduction.

[0029] By setting the path node activation threshold function, the present invention can calculate the activation probability of the path node according to the difference between the real-time blood drug concentration and the standard reference value. This mechanism enables the activation state of each medication decision-making unit in the drug treatment process to be closely associated with the actual drug response of the patient, thereby realizing precise medication path management. Based on the result of real-time deduction, the activation probability of each medication decision-making unit can be dynamically adjusted. When the reaction of the drug in the body changes, the system can automatically update the activation probability to ensure that the treatment plan matches the current physiological state of the patient. This dynamic adjustment keeps the drug treatment process in an optimal state all the time, thereby improving the treatment effect and safety. Through the analysis of the difference in activation probability of path nodes, when the difference in activation probability of adjacent nodes exceeds 40%, the system automatically inserts a pharmaceutical care marker point. This mechanism can timely detect abnormal fluctuations or potential problems in the drug treatment path, help medical staff identify risk points in the treatment process, and thus take timely intervention measures. Using the pharmaceutical care marker point, the system can make differential selections for different paths. By deducing the path conflict of different paths through the fuzzy Petri net, it can evaluate the conflicts or inconsistencies brought by different paths, provide a safer and more stable treatment path, and avoid the risks brought by path inconsistencies. The result of real-time deduction provides data-based decision support for the adjustment of the treatment path. By dynamically adjusting the activation probability of the medication decision-making unit, the treatment process can not only respond to the changes of the patient, but also continuously optimize according to real-time data, ensuring the scientificity and rationality of drug use in the treatment process.

[0030] Preferably, step S3 includes the following steps:

[0031] Step S31: Calculate the difference degree of the path form in the drug digital relationship chain based on the multi-dimensional deviation detection algorithm to obtain the difference degree of the path form of the relationship chain;

[0032] Step S32: Conduct path deviation interaction analysis on the drug digital relationship chain according to the difference degree of the path form of the relationship chain to generate path deviation interaction data;

[0033] Step S33: Use the path deviation interaction data to identify the path intersection points of the drug digital relationship chain, and mark the medication decision-making units corresponding to the identified path intersection points as drug treatment abnormal risk points;

[0034] Step S34: Set drug intervention nodes for the drug digital relationship chain according to the drug treatment abnormal risk points to obtain drug intervention nodes.

[0035] Through the multi-dimensional deviation detection algorithm, the present invention calculates the difference degree of the path form of the drug digital relationship chain, which can effectively identify abnormal deviations and inconsistencies in the drug treatment path. The introduction of this method enables any situation deviating from the preset path to be timely identified and processed during the drug treatment process, avoiding potential treatment problems. Through path deviation interaction analysis of the drug digital relationship chain, the system can obtain detailed path deviation interaction data, which can provide more accurate path analysis, help identify and analyze key nodes or factors leading to inconsistent drug treatment plans, and enhance the controllability of the drug treatment process. Through path intersection point identification, the abnormal risk points of drug treatment caused by path deviation can be effectively marked. Such an identification mechanism can ensure that the drug treatment path always conforms to the standard, timely discovers and marks potential risk points, and effectively improves the safety during the treatment process. The setting of drug intervention nodes can insert intervention measures in a timely manner in the drug treatment path to ensure that the drug treatment chain will not cause adverse reactions due to path deviation or abnormality. Through the identification of drug treatment abnormal risk points, the system can actively adjust the treatment plan and conduct necessary interventions to avoid instability during the treatment process. Through the calculation of the difference degree of the path form and the identification of path intersection points, the treatment path in the drug digital relationship chain can be optimized to make it more stable and reliable. This method reduces the occurrence probability of inconsistent or incorrect paths during the treatment process and improves the safety of the treatment plan. Through the real-time analysis of path deviation and the dynamic setting of intervention nodes, the system can implement intelligent pharmaceutical monitoring, automatically discover potential problems in the treatment path and conduct interventions. Such intelligent decision support not only improves the treatment efficiency but also ensures the personalization and precision of the drug treatment process.

[0036] Preferably, the multi-dimensional deviation detection algorithm in Step S31 is as follows:

[0037]

[0038] In the formula, D path is the difference degree of the relationship chain path form, N is the total number of path nodes, w geo is the geometric deviation weight, d geo (i) is the deviation degree of the i-th node in the geometric space, w topo is the topological deviation weight, d topo (i) is to calculate the topological deviation degree of the i-th node, w time is the time deviation weight, d time (i) is the deviation degree of the i-th node in the time dimension, w attr is the attribute deviation weight, d attr (i) is the deviation degree of the attribute information of the i-th node.

[0039] The present invention analyzes and integrates a formula of a multi-dimensional deviation detection algorithm. The principle of the formula is to calculate the difference degree of the relationship chain path form by comprehensively considering the deviation degrees of each node in the four dimensions of geometry, topology, time and attribute in the path. It is used for multi-dimensional deviation detection to help identify the deviations of each node on the path in different dimensions relative to the expected state, so as to evaluate the overall change or abnormality of the path form. This calculation method can more comprehensively reflect the potential abnormalities and changes on the path when dealing with complex spatial data. D path in the formula represents the difference degree of the path form, measuring the comprehensive result of the deviations of all nodes in the path in each dimension. The larger the value of D path , the more obvious the form change or abnormality of the path. N is the total number of path nodes, reflecting the length or complexity of the path. The more path nodes there are, the more complex the dimensions involved in the calculation of D path . w geo , w topo , w time and w attr these four parameters are weight coefficients, representing the importance of the geometry, topology, time and attribute dimensions in the overall difference degree respectively. By adjusting these weights, the contribution degree of each dimension to the total difference degree can be controlled. Different application scenarios require different weight settings and are adjusted according to the needs of specific tasks. d geo (i) represents the gap between the position of this node in the geometric space and the expected position. Geometric deviation reflects the change of the node in the spatial position, caused by the bending or movement of the path. d topo (i) describes the topological structure change of this node relative to other nodes. Topological deviation is usually used to detect whether the connection relationship between nodes has changed, such as the connection interruption in the network topology. d time(i) represents the difference between the timestamp of this node and the expected time point. The time deviation is used to analyze the timing changes in a path or data stream, and to detect whether data points are captured or processed within a predetermined time. d attr (i) represents the difference between the attribute data of this node and the expected value. Attribute deviation is usually used to analyze whether the characteristic information of a node (such as attribute value, classification label, etc.) has changed. When using the conventional multi-dimensional deviation detection algorithms in the art, the difference degree of the relationship chain path form can be obtained. By applying the multi-dimensional deviation detection algorithm provided by the present invention, the difference degree of the relationship chain path form can be calculated more precisely. By combining the deviations of geometry, topology, time, and attributes, the abnormality of the path can be evaluated more comprehensively. For example, a single geometric deviation is not sufficient to fully describe the path problem, but by considering topology, time, and attributes comprehensively, more detailed potential problems can be discovered. By weighted summarizing the deviations of each node in different dimensions, it can be located which specific dimensions or which nodes cause the abnormality of the overall path. In this way, not only can the abnormalities on the path be detected, but also the root cause of the problem can be revealed. This algorithm can automatically extract deviation information from data, has strong adaptability, and can be used to process different types of spatial, time, topological, and attribute data, making it highly versatile in practical applications.

[0040] Preferably, step S4 includes the following steps:

[0041] Step S41: Determine the coverage of the drug intervention node;

[0042] Step S42: Analyze the node intervention scope of the drug digital relationship chain based on the coverage of the drug intervention node, and generate the node intervention scope;

[0043] Step S43: Analyze the path propagation impact of the drug digital relationship chain through the node intervention scope, and generate path propagation impact data; use the path propagation impact data to perform intelligent regulation of the drug dosage of the drug digital relationship chain to execute the intelligent management operation of clinical pharmacy.

[0044] By judging the coverage of drug intervention nodes, the present invention ensures that drug intervention measures can cover key nodes, reduces the problems of omission or over-intervention, and thus achieves precise intervention. Combining with digital twin technology, analyzing the scope of action and path propagation impact of drug intervention nodes can monitor the dynamic changes in drug use in real time and conduct intelligent regulation. Such a regulation mechanism can dynamically adjust the drug dosage according to the individual needs of patients, treatment progress, and drug reactions, avoid overdose or underdose, and ensure the maximization of treatment effects. Through the generation and analysis of path propagation impact data, it can provide scientific decision-making support for clinical pharmacy, optimize drug use strategies, thereby improving medical efficiency, reducing unnecessary waste of medical resources, and reducing the occurrence of drug-related adverse reactions. Path propagation impact analysis provides a basis for drug use decisions and enhances the intelligence and scientific nature of clinical management.

[0045] Preferably, step S41 includes:

[0046] Set the coverage radius calculation formula, and the coverage radius calculation formula is as follows:

[0047] R cover =λ·D path ;

[0048] Wherein, R cover is the coverage radius of the drug intervention node, λ is the coverage radius adjustment factor, and D path is the difference degree of the relationship chain path form;

[0049] Calculate the coverage scope of the drug intervention node through the coverage radius calculation formula to obtain the coverage scope data;

[0050] Evaluate the coverage sufficiency of the drug intervention node to generate the coverage sufficiency evaluation data of the drug intervention node;

[0051] Judge the coverage of the drug intervention node. If the coverage sufficiency evaluation data is greater than or equal to the coverage scope data, no processing is performed; if the coverage sufficiency evaluation data is less than the coverage scope data, an intelligent compensation mechanism is triggered based on the coverage of the drug intervention node.

[0052] Through the set coverage radius calculation formula (R cover =λ·D path ), the present invention can accurately calculate the coverage scope of the drug intervention node, thereby ensuring that drug intervention can cover the required key nodes. This method combines the path form difference degree D pathAnd a regulation factor (λ) is used to make the coverage radius calculation more accurate, thereby improving the accuracy and efficiency of intervention. Through the coverage sufficiency assessment (based on the coverage radius and the coverage range of intervention nodes), the actual intervention effect of drug intervention nodes can be comprehensively evaluated. When the coverage sufficiency assessment data is greater than or equal to the coverage range data, no further intervention is required, reducing unnecessary intervention operations. If the coverage sufficiency assessment data is less than the coverage range data, the system can automatically trigger an intelligent compensation mechanism. The intelligent compensation mechanism performs supplementary intervention based on the coverage amount of drug intervention nodes, which can ensure that drug intervention is not omitted and improve the effectiveness of intervention measures. Through accurate coverage amount judgment and the intelligent compensation mechanism, the system can optimize the use of drug resources according to the actual situation, avoid ineffective or excessive intervention, which improves the efficiency of drug use and reduces potential resource waste at the same time.

[0053] In this specification, a clinical pharmacy intelligent management system is provided for implementing the above-mentioned clinical pharmacy intelligent management method. The clinical pharmacy intelligent management system includes:

[0054] A data acquisition module for real-time obtaining all-element data of the drug treatment chain, including prescription instruction flow, drug preparation parameters, administration execution records, and patient physiological feedback signals;

[0055] A relationship chain construction module for constructing a dynamic evaluation engine of the drug treatment chain; using the dynamic evaluation engine of the drug treatment chain to perform clinical medication mapping relationship matching on all-element data of the drug treatment chain, generating an individualized medication path map; extracting a topological chain from the individualized medication path map through digital twin technology to generate a drug digital relationship chain;

[0056] A risk point identification module for identifying risk points of drug treatment chain breakage in the drug digital relationship chain based on a multi-dimensional deviation detection algorithm to obtain abnormal drug treatment risk points; setting drug intervention nodes in the drug digital relationship chain according to the abnormal drug treatment risk points to obtain drug intervention nodes;

[0057] A drug dosage compensation module for judging the coverage amount of drug intervention nodes; triggering an intelligent compensation mechanism based on the coverage amount of drug intervention nodes, and performing intelligent regulation of drug dosage on all-element data of the drug treatment chain through the intelligent compensation mechanism to execute clinical pharmacy intelligent management operations.

[0058] The present invention also provides a storage medium on which a computer program is stored, and when the computer program is executed by a processor, it implements the above-mentioned clinical pharmacy intelligent management method.

[0059] The beneficial effects of the present invention are as follows: Through the data acquisition module, various data of the drug treatment chain can be obtained in real time, including prescription instructions, drug preparation parameters, administration records, and patient physiological feedback signals. This function ensures that each link in the drug treatment process is supported by accurate real-time data, providing a sufficient data basis for subsequent analysis and decision-making. The relationship chain construction module establishes a dynamic evaluation engine for the drug usage chain and uses digital twin technology to achieve an accurate mapping of all-element data of the drug usage chain to drug administration operations and clinical terminals. The digital twin relationship chain makes the drug usage process more transparent and traceable, improving the intelligent level of pharmacy management. The risk point identification module uses a multi-dimensional deviation detection algorithm to identify risk points in the drug digital relationship chain and promptly discovers potential abnormal risks. Identifying risk points in advance allows for timely adjustment of relevant operations, thereby reducing unnecessary risks. Through the setting of drug intervention nodes, the system can intelligently intervene in the drug usage chain and trigger an intelligent compensation mechanism based on the coverage of drug intervention nodes. The intelligent compensation mechanism can dynamically adjust the drug dosage, making personalized adjustments according to the actual situation and optimizing the operation effect. The drug dosage compensation module provides support for intelligent regulation, being able to adjust the drug usage amount in real time in a dynamic environment to ensure the accuracy and adaptability of drug usage. The system provides intelligent decision-making support to help managers make more accurate medication adjustments. Through this system, drug management can achieve more precise, personalized, and intelligent control, reducing manual intervention while improving work efficiency. The intelligent regulation of drugs and the timely identification of risk points help reduce operational risks and optimize the safety of the entire process. Therefore, the present invention improves the intelligent management and precise regulation of the drug treatment chain through real-time data acquisition, digital twin mapping, intelligent risk identification, and intervention, enhancing the intelligence and precision of drug treatment chain management. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic flow chart of the steps of a method for intelligent management of clinical pharmacy;

[0061] Figure 2 is Figure 1 a detailed implementation step flow chart of step S3 in

[0062] Figure 3 is Figure 1 a detailed implementation step flow chart of step S4 in

[0063] The realization, functional characteristics, and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0064] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative work belong to the scope of protection of the present invention.

[0065] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0066] It should be understood that although terms such as "first" and "second" may be used here to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit can be called the second unit, and similarly the second unit can be called the first unit. The term "and / or" used here includes any and all combinations of one or more of the listed associated items.

[0067] To achieve the above object, please refer to Figures 1 to 3 , a method for intelligent management of clinical pharmacy, the method comprising the following steps:

[0068] Step S1: Real-time obtain all-element data of the drug treatment chain, including prescription instruction flow, drug preparation parameters, administration execution records, and patient physiological feedback signals;

[0069] Step S2: Construct a dynamic evaluation engine for the drug treatment chain; use the dynamic evaluation engine for the drug treatment chain to match the clinical medication mapping relationship of all-element data of the drug treatment chain, generate an individualized medication path map; extract the topological chain of the individualized medication path map through digital twin technology to generate a drug digital relationship chain;

[0070] Step S3: Identify the risk points of drug treatment chain breakage for the drug digital relationship chain based on a multi-dimensional deviation detection algorithm to obtain drug treatment abnormal risk points; set drug intervention nodes for the drug digital relationship chain according to the drug treatment abnormal risk points to obtain drug intervention nodes;

[0071] Step S4: Determine the coverage of the drug intervention nodes; trigger the intelligent compensation mechanism based on the coverage of the drug intervention nodes, and perform intelligent regulation of the drug dosage for all elements data of the drug treatment chain through the intelligent compensation mechanism to execute the intelligent management operation of clinical pharmacy.

[0072] The present invention realizes the monitoring of the entire process of drug use by obtaining drug preparation, drug use records, and physiological feedback data in real time. This method ensures that every link in the drug management process can be tracked and recorded in real time, thereby improving the management accuracy and response speed. By constructing a digital twin mapping of the drug treatment chain, a real-time digital model can be provided for each drug management link. This dynamic mapping relationship can provide real-time feedback to help relevant personnel optimize and adjust the drug management process at any time, thereby reducing human errors and improving the system operation efficiency. Using a multi-dimensional deviation detection algorithm to analyze the drug management process can detect potential risk points in advance. This algorithm can efficiently and accurately identify the deviations in the process to help relevant personnel adjust the work plan in time, thereby reducing any defects or hidden dangers in the system. Once an abnormal risk is detected, the system will automatically adjust according to the set intervention nodes. This automated intervention measure ensures that the drug management process can respond according to individual needs, thereby improving the system's adaptability and ensuring the best management effect. When the system detects that there is a link that needs to be supplemented in the drug management process, the intelligent compensation mechanism will be automatically triggered to adjust the drug management strategy to ensure the coherence and efficiency of the process. This intelligent mechanism not only improves the drug management efficiency but also ensures the accuracy and timeliness of each link. By constructing a drug digital relationship chain, the data of each link can be clearly traced. This not only provides a complete operation record for the management personnel but also helps with post-event analysis and evaluation to ensure the transparency and operability of each link. With the help of a dynamic evaluation engine and data analysis tools, the system can provide real-time and accurate data support for the manager, thereby optimizing the decision-making process. This enables the management personnel to identify problems and formulate countermeasures more efficiently, improving the operation efficiency of the entire management system. Therefore, the present invention improves the intelligent management and precise regulation of the drug treatment chain through real-time data acquisition, digital twin mapping, intelligent risk identification, and intervention, and enhances the intelligence and precision of the drug treatment chain management.

[0073] In the embodiment of the present invention, with reference to Figure 1 As shown, it is a schematic diagram of the step flow of a method for intelligent management of clinical pharmacy according to the present invention. In this example, the method for intelligent management of clinical pharmacy includes the following steps:

[0074] Step S1: Obtain all elements data of the drug treatment chain in real time, including prescription instruction flow, drug preparation parameters, administration execution records, and patient physiological feedback signals;

[0075] In the embodiments of the present invention, 100 prescription data are acquired per second through the interface of the Hospital Information System (HIS), and the prescription content is parsed, including drug name, dosage, administration route, frequency, etc. An efficient data stream processing engine is adopted to control the data latency within ≤200 ms to ensure the real-time nature of the prescription instruction stream. Through the pharmacy automated dispensing system, 50 groups of drug preparation parameters (such as concentration, dissolution rate, mixing ratio, etc.) are collected per minute and standardized, with the error controlled within ±0.5%. An intelligent dispensing algorithm is adopted to optimize the preparation plan by combining historical data to improve the accuracy. Through intelligent infusion pumps, automatic drug delivery devices, etc., 20 times of drug administration execution data are recorded per second, including actual infusion volume, infusion rate, remaining dosage, etc., and the data integrity is verified (the missing rate ≤0.1%). A data synchronization mechanism is adopted to ensure a high degree of consistency between the drug administration execution record and the prescription instruction, and the data synchronization rate ≥99.9%. Through wearable devices and physiological monitoring systems, 10 groups of key physiological parameters (such as blood pressure, heart rate, blood oxygen, blood glucose, etc.) are collected per second and analyzed in real time. An intelligent anomaly detection model is adopted to ensure that the false alarm rate of physiological data ≤1% and the anomaly recognition accuracy rate ≥98%. A distributed data storage architecture is adopted to store more than 1TB of drug treatment data per hour and format it to improve the query efficiency. Through data tagging technology, efficient fusion of data from different sources is achieved, providing high-quality input for subsequent analysis, and finally, generating all-element data of the drug treatment chain.

[0076] Step S2: Construct a dynamic evaluation engine for the drug treatment chain; use the dynamic evaluation engine for the drug treatment chain to match the clinical medication mapping relationship of the all-element data of the drug treatment chain to generate an individualized medication path map; extract the topological chain from the individualized medication path map through digital twin technology to generate a drug digital relationship chain;

[0077] In the embodiments of the present invention, by adopting a deep reinforcement learning (DRL) framework, a dynamic evaluation model of the drug treatment chain is constructed. The model inputs include prescription instruction streams, drug preparation parameters, administration execution records, and patient physiological feedback signals. The reward function of reinforcement learning is set as: R = α·Ceff - β·Crisk - γ·Cdev; where: Ceff is the treatment effectiveness score (based on the improvement of physiological feedback, range 0 - 100), Crisk is the administration risk coefficient (based on allergic reactions, drug interactions, etc., range 0 - 1), Cdev is the operation deviation value (such as the deviation between actual administration and planned administration, range 0 - 1), and α, β, γ are regulation weights, with the initial values set as 1, 0.8, 0.5. The multi-modal data fusion technology is adopted to construct a mapping model between the administration operation and the clinical terminal, including: Structural mapping: the mapping relationship between drug dosage, route, frequency, and patient physiological state. Behavioral mapping: the time-synchronous association between the operations performed by nurses / doctors, patient feedback, and the administration process. Feedback mapping: matching the short-term trend of patient physiological signals with the drug effect, and calculating the drug effect feedback coefficient (range 0 - 1, and a value exceeding 0.8 is regarded as a significant effect). An efficient computing engine is adopted to control the mapping calculation time within ≤500 ms. The knowledge graph construction technology is adopted to generate an association graph of drugs, patients, prescriptions, administration processes, and physiological feedback: Nodes: drugs (name, dosage), patients (ID, physiological parameters), prescriptions (doctor, time), administration execution (device, time), physiological feedback (data points). Edges: time association, causal relationship, feedback action path. Weights: the influence weights between different factors (such as the influence coefficient of drug dosage on physiological parameters), generating a digital twin relationship chain database, storing ≥1 million association records, supporting real-time query, and the response time ≤1 s.

[0078] Step S3: Based on the multi-dimensional deviation detection algorithm, identify the risk points of drug treatment chain breakage in the drug digital relationship chain to obtain the abnormal risk points of drug treatment; set the drug intervention nodes in the drug digital relationship chain according to the abnormal risk points of drug treatment to obtain the drug intervention nodes;

[0079] In the embodiments of the present invention, by adopting a multi-dimensional deviation detection algorithm that combines time series anomaly detection (TAD), high-dimensional data deviation analysis (HDA), and Bayesian risk assessment (BRA), identify the abnormal points in the drug digital relationship chain. Set the deviation detection formula to obtain the difference degree D of the relationship chain path form path , if D pathIf it is ≥0.7 (range 0 - 1), it is determined as a risk point for the breakage of the drug treatment chain. Combining historical data backtracking and real-time data stream analysis, the following strategies are adopted to detect anomalies: there are 3 consecutive deviation records within the most recent 10 minutes, the anomaly probability >5% within the past 24 hours, and the DBSCAN clustering algorithm is used to find high-density anomaly regions. Identify the risk points for drug treatment anomalies and generate a dataset of risk points for drug treatment anomalies. Adopt an adaptive dynamic intervention strategy, and set drug intervention nodes near the risk points: if the deviation of the risk point for drug treatment anomalies is ≥0.5, adjust the administration time; if the deviation of the risk point for drug treatment anomalies is ≥0.4, adjust the dosage. If the deviation of the risk point for drug treatment anomalies is ≥0.6, it is recommended to replace the drug or adjust the combination, generate a dataset of drug intervention nodes, and update it to the drug digital relationship chain.

[0080] Step S4: Judge the coverage of the drug intervention nodes; trigger the intelligent compensation mechanism based on the coverage of the drug intervention nodes, and perform intelligent regulation of the drug dosage for all elements of the drug treatment chain through the intelligent compensation mechanism to execute the intelligent management operation of clinical pharmacy.

[0081] In the embodiment of the present invention, the coverage of the drug intervention nodes in the drug treatment chain is calculated by using the area coverage rate model (ACR), and the coverage Ccover of the drug intervention nodes is obtained (range 0 - 1). Set the threshold: if Ccover≥0.85, it is considered that the coverage of the drug intervention nodes is sufficient and no additional compensation is required. If Ccover<0.85, the intelligent compensation mechanism is triggered. If the coverage Ccover<0.7, add new drug intervention nodes to optimize the administration time and plan. If the coverage 0.7≤Ccover<0.85, adjust the dosage parameters of the existing drug intervention nodes. If the intervention nodes are sufficient but the patient reports an anomaly, optimize the feedback processing process to improve the response accuracy of the physiological signals. Use the intelligent dosage prediction algorithm (IDA), combine the patient's individual data for dosage optimization calculation, set the intelligent regulation range, and the maximum dosage adjustment amplitude ≤±20% to avoid the risk of overdose. Update it to the digital twin relationship chain in real time to ensure that the drug dosage after intelligent compensation meets the individual treatment requirements. Set a 5-minute feedback interval, monitor the patient's physiological signals in real time, and perform secondary adjustment. Use the multi-objective optimization (MOO) method to balance the efficacy, risk, and dosage stability. Integrate the intelligent compensation mechanism into the hospital pharmacy system (HPS) to achieve automated intelligent pharmacy management.

[0082] Preferably, the construction of the dynamic evaluation engine for the drug treatment chain in step S2 includes:

[0083] Construct a dynamic pharmaceutical ontology framework, including decision rules in three dimensions: drug-gene interaction, metabolic pathway coupling, and administration device adaptation;

[0084] Adopt path discretization modeling technology to decompose the full-element data of the drug treatment chain into medication decision-making units with time-dependent relationships;

[0085] Connect the medication decision-making units into a medication path; perform real-time deduction on the medication path based on the fuzzy Petri net, and dynamically update the activation probability of each medication decision-making unit through the results of the real-time deduction;

[0086] Establish a path credibility verification mechanism to trigger an artificial review process when the matching degree between the newly generated medication path and the preset historical successful case library is lower than 65%.

[0087] In the embodiments of the present invention, data on known drug-gene interactions (DGIs) are collected, such as FDA drug labels, PharmGKB databases, etc. An ontology mapping method is used to associate concepts such as drugs, genes, and proteins to a unified knowledge graph. Medication decision rules based on DGIs are constructed, for example: the situation where gene mutations affect drug metabolism (such as CYP2C19 affecting the efficacy of clopidogrel). In combination with metabolic databases such as KEGG and Reactome, the interactions of drugs in different metabolic pathways are analyzed. A metabolic pathway topology analysis method is used to evaluate the influence degree of drugs at different pathway nodes, generating dynamic metabolic pathway coupling rules for guiding individualized medication. Adaptability rules are established for different drug dosage forms (oral, injection, implantation) and drug delivery devices (micropumps, sustained-release systems, etc.). A physicochemical modeling method is used to evaluate parameters such as drug solubility, drug release rate, and tissue permeability. The optimal dosing method is recommended according to the individual characteristics of the patient (age, liver and kidney function, etc.). The drug treatment chain is disassembled into multiple medication decision units (MDUs), including: the initial medication plan (based on patient characteristics), medication adjustment (adjusting the dose or drug according to treatment feedback), cross-medication management (avoiding drug interactions), and discontinuation or drug change decisions. The Markov decision process (MDP) is used to describe the time dependence of MDUs. State transition probabilities are set and dynamically updated according to real-time feedback data of the patient. A knowledge graph of the drug treatment chain is constructed using a graph database (such as Neo4j), with nodes being MDUs and edges representing decision transfer relationships. Combining individual patient data, an optimal medication path is dynamically generated. A fuzzy Petri net (FPN) is used to deduce the medication path and calculate the probabilities of different medication plans. The activation probabilities of each MDU are calculated through fuzzy rules, the path weights are updated, and an individualized medication recommendation plan is generated. A historical successful case database is used to calculate the matching degree between the new path and historical cases. A matching degree threshold of 65% is set, and when the value is lower than this, manual review is triggered. An expert system is called for in-depth review to evaluate the rationality of the new path. If it is confirmed as a valid path, it is included in the successful case database to improve the decision-making ability of the system.

[0088] Preferably, the method for constructing the dynamic pharmaceutical ontology framework includes:

[0089] Establish a quadruple model of drug action, where the quadruple includes chemical entities, target proteins, metabolic enzyme systems, and transport carriers, and each element is associated with at least 3000 verified biomedical entities; adapt decision rules to the quadruple model of drug action to generate drug-gene interaction decision rules;

[0090] Define a clinical pathway constraint rule library, which includes three types of constraint conditions: contraindication conflict detection rules, optimized administration timing rules, and device compatibility verification rules; adapt the decision rules of the clinical pathway constraint rule library to generate metabolic pathway coupling decision rules;

[0091] Construct a drug property attenuation model, adapt the decision rules of the drug property attenuation model to generate drug delivery device adaptation decision rules; construct a dynamic pharmaceutical ontology framework based on drug-gene interaction decision rules, metabolic pathway coupling decision rules, and drug delivery device adaptation decision rules.

[0092] In the embodiments of the present invention, by setting a quadruple (C, T, M, Tt), where: C (Chemical Entity) refers to a drug molecule, such as aspirin. T (Target Protein) refers to the protein on which the drug acts, such as COX-1 / COX-2. M (Metabolic Enzyme System), such as the CYP450 enzyme system, affects the drug metabolism rate. Tt (Transporter) is responsible for drug transmembrane transport, such as P-gp (P-glycoprotein). Using natural language processing (NLP) technology, parse databases such as PubMed, DrugBank, and KEGG to extract quadruple information. Store the quadruple using a graph database (Neo4j), and each element is associated with at least 3,000 verified biomedical entities. Use the random walk algorithm to calculate the drug-protein-metabolism-transport association strength. Analyze the gene-level impact of the drug quadruple, such as the CYP2C19 gene affecting clopidogrel metabolism. Use the decision tree method (Decision Tree) to construct drug-gene interaction rules. Rule example: Rule 1 (Metabolic impact): CYP2D6*10 mutation -> Avoid using drugs that rely on CYP2D6 metabolism. Rule 2 (Target variation): EGFR mutation -> Select a suitable tyrosine kinase inhibitor (such as erlotinib). Establish a clinical pathway constraint rule base: For example, NSAIDs are contraindicated for patients with peptic ulcers, calcium channel blockers and beta blockers should be avoided from being used simultaneously, and insulin pumps are not suitable for some combined oral hypoglycemic drug treatment regimens. Use reasoning over knowledge graphs to match the medication plan with the contraindication database. Use metabolic network topology analysis to identify the risk of metabolic pathway conflicts. Use pharmacokinetic (PK) modeling to establish a drug concentration-time curve and calculate the half-life (T1 / 2) and bioavailability (F). Use a machine learning regression model (such as XGBoost) to predict the drug decay trend under different dosing methods. Adapt to different dosing devices (oral, injection, implantable devices, etc.). Rule example: Rule 1 (Controlled release adaptation): If T1 / 2 > 8h, a sustained-release preparation can be considered; if T1 / 2 < 2h, frequent dosing or continuous intravenous infusion is required. Rule 2 (Device adaptation): Oral hypoglycemic drugs are not suitable for patients with insulin pumps that rely on continuous blood glucose monitoring. Integrate the drug-gene interaction decision rule, the metabolic pathway coupling decision rule, and the dosing device adaptation decision rule to generate an ontology framework. Use an ontology language (OWL) and an inference engine (such as HermiT) for automated reasoning to generate an individualized medication plan.

[0093] Preferably, based on the fuzzy Petri net, the medication path is deduced in real time, and the activation probability of each medication decision unit is dynamically updated through the result of the real-time deduction, including:

[0094] Set the path node activation threshold function, and the formula of the path node activation threshold function is as follows:

[0095]

[0096] In the formula, P activate is the path node activation threshold, k is the drug sensitivity coefficient, c actual is the real-time blood drug concentration monitoring value, c standard is the standard reference value;

[0097] Calculate the activation probability of the unit node of the medication path through the path node activation threshold function to obtain the activation probability of the path unit node;

[0098] Analyze the difference in the activation probability of adjacent nodes of the medication path according to the activation probability of the path unit node. When the difference in the activation probability of adjacent nodes exceeds 40%, automatically insert a pharmaceutical care marker point;

[0099] Use the pharmaceutical care marker point to select different paths of the medication path, and deduce the path conflict of the different paths through the fuzzy Petri net to obtain the result of real-time deduction;

[0100] Dynamically adjust the activation probability of the medication decision-making unit based on the result of real-time deduction.

[0101] In the embodiment of the present invention, the path node activation threshold function is constructed by using the Logistic Function, and the formula is as follows: In the formula, P activate is the path node activation threshold, k is the drug sensitivity coefficient, c actual is the real-time blood drug concentration monitoring value, c standardis the standard reference value; the k value of the patient is calculated by using individualized pharmacokinetics (PK) modeling, and the parameters are updated by combining Bayesian estimation. By real-time monitoring the blood drug concentration of the patient, the real-time blood drug concentration monitoring value is dynamically adjusted. The activation probability of each path unit node is calculated by using the path node activation threshold function to obtain the path unit node activation probability set: Pnode = {P1, P2,..., Pn}; where the P value corresponding to each node is dynamically calculated by the real-time blood drug concentration. Calculate the difference in the activation probability of adjacent path unit nodes: ΔP = ∣Pi - Pi+1∣; set the adjacent node difference threshold of 40%, that is: ΔP > 0.4; if this condition is met, a pharmaceutical care marker point is automatically inserted into the path. Trigger the pharmacist review process (if the system determines that the risk is high). Enter the differential path selection module (if there is a potential path optimization plan). For the case where there is a care marker point on the path, the system selects the optimal path from the historical successful path library for comparison. The dynamic time warping (DTW) algorithm is used to calculate the matching degree between differential paths. Construct a fuzzy Petri net (FPN) model: state set (Places, P): represents different states of the medication path (such as "drug metabolism stage"). Transition set (Transitions, T): represents path switching events (such as "the drug is metabolized by CYP450"). Fuzzy weight (W): represents uncertainty (such as "individual metabolism rate"). Activation probability (FiringProbability): determined by the activation probability of the path unit node. Input the current medication path and candidate path, calculate the path conflict probability (based on fuzzy weight), identify conflict points (such as abnormal concentration caused by drug metabolism competition), and adjust the path switching strategy (such as adjusting the administration time or dose). If the conflict path probability is too high (such as > 60%), reduce the activation probability Pi of the corresponding node. If the new path has a high matching degree (such as > 80%), increase the activation probability Pi+1 of the new path. The reinforcement learning algorithm (such as Q-learning) is used to continuously optimize the decision-making unit, and combined with historical medication data, individualized path adjustment is performed to improve the prediction accuracy.

[0102] As an example of the present invention, refer to Figure 2 shown, in this example, step S3 includes:

[0103] Step S31: Calculate the difference degree of the path form in the drug digital relationship chain based on the multi-dimensional deviation detection algorithm to obtain the difference degree of the path form of the relationship chain;

[0104] Step S32: Perform path deviation interaction analysis on the drug digital relationship chain according to the difference degree of the path form of the relationship chain to generate path deviation interaction data;

[0105] Step S33: Identify the path intersection points of the drug digital relationship chain using the path deviation interaction data, and mark the drug administration decision units corresponding to the identified path intersection points as drug treatment abnormal risk points;

[0106] Step S34: Set drug intervention nodes for the drug digital relationship chain based on the drug treatment abnormal risk points to obtain drug intervention nodes.

[0107] In the embodiment of the present invention, by selecting three major dimensions of time series features, spatial features, and biocompatibility features, path shape deviation detection indicators are established: Time series features: drug administration time point, dynamic curve of blood drug concentration, metabolic half-life. Spatial features: drug action target, receptor distribution, metabolic pathway. Biocompatibility features: individual genotype, drug-protein binding ability, metabolic enzyme activity. The dynamic time warping (DTW) is used to calculate the shape matching degree between the standard path and the current drug administration path, and the relationship chain path shape difference degree Dshape is obtained. If Dshape≥0.3, it indicates that the path has a significant deviation, and the path deviation interaction analysis process is entered. Set the path deviation threshold: Mild deviation (0.1≤Dshape<0.3): Allow continued execution but monitor. Moderate deviation (0.3≤Dshape<0.6): Trigger a drug adjustment prompt. Severe deviation (Dshape≥0.6): Require manual intervention or path optimization. Calculate the local deviation degree of different path nodes to generate an interaction data matrix: where D m,n is the deviation degree of the nth node of the mth path on the drug digital relationship chain. In the path deviation interaction data matrix, find the high deviation points where multiple paths cross: If D m,n >0.5, it is determined as a path intersection point, ensuring that the path intersection point satisfies two conditions: Multiple paths overlap (at least 2 paths deviate at this point). There is a significant shape difference (deviation value>0.5). Mark the drug administration decision unit corresponding to the path intersection point as a drug treatment abnormal risk point. If the number of cross paths at this point>3, enter the intervention node setting process. For the risk points of metabolic conflict: Adjust the administration time (to avoid metabolic competition), replace drugs with different metabolic pathways. For the risk points of drug interaction: Reduce the dose or modify the drug administration plan, increase the monitoring frequency, and improve safety. Use a fuzzy inference system (FIS) to set the drug intervention plan: Input: risk point deviation degree, metabolic conflict level, patient's individual pharmacokinetic parameters. Output: adjusted drug intervention strategy.

[0108] Preferably, the multi-dimensional deviation detection algorithm in step S31 is as follows:

[0109]

[0110] In the formula, D path is the difference degree of the relationship chain path form, N is the total number of path nodes, w geo is the geometric deviation weight, d geo (i) is the deviation degree of the i-th node in the geometric space, w topo is the topological deviation weight, d topo (i) is the topological deviation degree of the i-th node, w time is the time deviation weight, d time (i) is the deviation degree of the i-th node in the time dimension, w attr is the attribute deviation weight, d attr (i) is the deviation degree of the attribute information of the i-th node.

[0111] The present invention analyzes and integrates the formula of a multi-dimensional deviation detection algorithm. The principle of the formula is to calculate the difference degree of the relationship chain path form by comprehensively considering the deviation degrees of each node in the four dimensions of geometry, topology, time, and attribute in the path. It is used for multi-dimensional deviation detection to help identify the deviations of each node on the path in different dimensions relative to the expected state, so as to evaluate the overall change or abnormality of the path form. This calculation method can more comprehensively reflect the potential abnormalities and changes on the path when dealing with complex spatial data. D in the formula path represents the difference degree of the path form, measuring the comprehensive result of the deviations of all nodes in the path in each dimension. D path The larger the value, the more obvious the change or abnormality of the path form. N is the total number of path nodes, reflecting the length or complexity of the path. The more path nodes, the more complex the dimensions involved in the calculation of D path . w geo , w topo , w time and w attr These four parameters are weight coefficients, representing the importance of the geometry, topology, time, and attribute dimensions in the overall difference degree respectively. By adjusting these weights, the contribution degree of each dimension to the total difference degree can be controlled. Different application scenarios require different weight settings, which are adjusted according to the needs of specific tasks. d geo (i) represents the gap between the position of the node in the geometric space and the expected position. Geometric deviation reflects the change of the node in the spatial position, caused by the bending or movement of the path. d topo (i) describes the topological structure change of the node relative to other nodes. Topological deviation is usually used to detect whether the connection relationship between nodes has changed, such as the connection interruption in the network topology. d time(i) represents the difference between the timestamp of this node and the expected time point. The time deviation is used to analyze the timing changes in a path or data stream, and to detect whether data points are captured or processed within a predetermined time. d attr (i) represents the difference between the attribute data of this node and the expected value. Attribute deviation is usually used to analyze whether the characteristic information (such as attribute values, classification labels, etc.) of a node has changed. When using the conventional multi-dimensional deviation detection algorithms in the art, the difference degree of the relationship chain path form can be obtained. By applying the multi-dimensional deviation detection algorithm provided by the present invention, the difference degree of the relationship chain path form can be calculated more precisely. By combining the deviations of geometry, topology, time, and attributes, the abnormality of the path can be evaluated more comprehensively. For example, a single geometric deviation is not sufficient to fully describe the path problem, but by considering topology, time, and attributes comprehensively, more detailed potential problems can be discovered. By weighted aggregation of the deviations of each node in different dimensions, it can be located which dimensions or which nodes cause the abnormality of the overall path. In this way, not only can the abnormalities on the path be detected, but also the root causes of the problems can be revealed. This algorithm can automatically extract deviation information from data, has strong adaptability, and can be used to process different types of spatial, time, topological, and attribute data, making it highly versatile in practical applications.

[0112] As an example of the present invention, refer to Figure 3 As shown, in this example, step S4 includes:

[0113] Step S41: Determine the coverage of the drug intervention node;

[0114] Step S42: Analyze the node intervention scope of the drug digital relationship chain based on the coverage of the drug intervention node, and generate the node intervention scope;

[0115] Step S43: Analyze the path propagation impact of the drug digital relationship chain through the node intervention scope, and generate path propagation impact data; use the path propagation impact data to perform intelligent regulation of the drug dosage of the drug digital relationship chain to execute the intelligent management operation of clinical pharmacy.

[0116] In the embodiments of the present invention, the coverage of the drug intervention node refers to the scope of the drug's role in the digital twin relationship chain, mainly referring to the degree of influence of the drug on different medication decision-making units. The coverage of each intervention node is calculated based on its degree of influence on other nodes in the path. By analyzing the interaction between the drug and adjacent nodes and the biological effects of the drug, the actual influence of the drug on each node in the path is determined. Nodes with a larger coverage have an impact on more decision-making units, while nodes with a smaller coverage only affect a local area. The coverage of each intervention node is calculated based on the physiological mechanism, pharmacodynamics, and pharmacokinetics models of the drug. The interaction strength between the drug and other relevant factors (such as genes, metabolic pathways, etc.) can be used as a measurement criterion. The coverage can be represented numerically and is usually classified into three degrees: low, medium, and high, to determine the strength of the intervention node. After obtaining the coverage of each intervention node, it is necessary to analyze its influence scope. Specifically, based on the influence of the intervention node, the scope of the drug intervention's effect on other nodes in the relationship chain is analyzed. This analysis considers the drug's mechanism of action, pharmacokinetics, pharmacodynamics, and clinical path constraints (such as dosing sequence, contraindication conflicts, etc.). Through this multi-dimensional analysis, the influence scope of the drug intervention node can be accurately determined, including direct and indirect influences. According to the analysis results, a map of the influence scope of the drug intervention node is generated. The other nodes directly or indirectly affected by the intervention node can be determined through path analysis algorithms, thereby defining the area or scope of the drug intervention. Using the scope of the node intervention effect obtained from the previous step, an impact analysis of the path propagation in the drug digital relationship chain is carried out. During the analysis, the influence intensity of the drug intervention on different nodes on the path is evaluated based on the coverage of the drug intervention node and its influence scope. Through the path propagation impact analysis, propagation impact data for each path are generated. These data reflect the propagation intensity of the drug intervention in the entire path and the influence of the drug intervention on different nodes. Based on these data, it can be evaluated whether the drug intervention meets the treatment requirements and whether adjustment or optimization is needed. Based on the path propagation impact data, intelligent regulation of the drug dosage is carried out. Specifically, by evaluating the influence of the drug intervention on each decision-making unit, the dosage and administration timing of the drug are intelligently adjusted to achieve personalized and precise treatment. This process relies on the clinical pharmacy intelligent management system, and the drug dosage is adjusted through algorithms to ensure the maximization of the drug effect and the reduction of side effect risks during the treatment process. For example, when the propagation impact of a certain path is too strong, the drug dosage needs to be reduced or the administration plan needs to be adjusted.

[0117] Preferably, step S41 includes:

[0118] Set the coverage radius calculation formula, where the coverage radius calculation formula is as follows:

[0119] R cover = λ·D path;

[0120] Among them, R cover is the coverage radius of the drug intervention node, λ is the coverage radius adjustment factor, and D path is the difference degree of the relationship chain path form;

[0121] Calculate the coverage range of the drug intervention node through the coverage radius calculation formula to obtain the coverage range data;

[0122] Evaluate the coverage sufficiency of the drug intervention node to generate the coverage sufficiency evaluation data of the drug intervention node;

[0123] Judge the coverage quantity of the drug intervention node. If the coverage sufficiency evaluation data is greater than or equal to the coverage range data, no processing is performed; if the coverage sufficiency evaluation data is less than the coverage range data, the intelligent compensation mechanism is triggered based on the coverage quantity of the drug intervention node.

[0124] In the embodiment of the present invention, by setting the coverage radius calculation formula, it is used to determine the influence range of the drug intervention node in the drug digital twin relationship chain. The formula is: R cover = λ·D path ; Among them, R cover is the coverage radius of the drug intervention node, λ is the coverage radius adjustment factor, and D pathIt is the difference degree of the relationship chain path form. Based on the coverage radius calculation formula, calculate the coverage range of each drug intervention node. The calculation result is the coverage radius (Rcover), which represents the path range that the drug intervention node can affect. According to this coverage radius, further analyze the number of nodes and paths it covers and their influence degree to form coverage range data. Based on the coverage range data, evaluate the coverage sufficiency of the drug intervention node. The coverage sufficiency evaluation determines whether its effect meets the treatment requirements by comparing the intersection of the coverage range of the intervention node and the actual treatment path. The evaluation criterion is: if the coverage sufficiency evaluation result is greater than or equal to the coverage range data, it means that the influence of the drug intervention is sufficient and the next step can be carried out; if the coverage sufficiency evaluation result is less than the coverage range data, supplementary measures need to be taken. If the difference between the coverage range data and the range covered by the actual intervention node is less than a preset threshold, it is considered that the node has sufficient coverage and no further processing is required. If the coverage sufficiency evaluation data is lower than the coverage range data, it means that the influence of the intervention node is insufficient and a compensation mechanism needs to be adopted. When the coverage amount of the drug intervention node is insufficient, trigger the intelligent compensation mechanism. This mechanism enhances the influence of the drug intervention on the path by dynamically adjusting the action range of the intervention node to ensure that all key paths are covered. By adjusting the adjustment factor (λ), the influence range of the intervention node is amplified, thereby enhancing the influence on the path nodes lacking coverage. According to the difference degree between the drug intervention node and other paths, appropriately modify the difference degree of the relationship chain path form to enhance the control of the path form. Add new intervention nodes in the areas or paths lacking coverage to ensure the integrity of the drug treatment chain.

[0125] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to embrace all changes within the meaning and scope of the equivalent elements of the application document in the present invention.

[0126] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for intelligent management of clinical pharmacy, characterized in that: The following steps are involved: Step S1: Real-time acquisition of all-element data of the drug treatment chain, including prescription instruction flow, drug preparation parameters, drug administration execution records, and patient physiological feedback signals; Step S2: Construct a dynamic evaluation engine for drug treatment chain; use the dynamic evaluation engine for drug treatment chain to match clinical drug mapping relationships for all element data of drug treatment chain to generate an individualized drug treatment pathway map; use digital twin technology to extract topological chains from individualized drug treatment pathway maps to generate a digital relationship chain for drugs; Step S3: Based on the multi-dimensional deviation detection algorithm, the drug digital relationship chain is used to identify the risk points of drug treatment chain breakage, and the abnormal risk points of drug treatment are obtained; According to the abnormal risk points of drug treatment, drug intervention nodes are set for the drug digital relationship chain to obtain drug intervention nodes; Step S4: Determine the coverage of drug intervention nodes; The intelligent compensation mechanism is triggered based on the coverage of the drug intervention node, and the drug dosage of all elements of the drug treatment chain is intelligently regulated through the intelligent compensation mechanism to perform clinical pharmacy smart management operations.

2. The method for intelligent management of clinical pharmacy according to claim 1, characterized in that: The step S2 of constructing a drug treatment chain dynamic evaluation engine includes: Construct a dynamic pharmaceutical ontology framework, including decision rules in three dimensions: drug-gene interaction, metabolic pathway coupling, and drug delivery device adaptation; Using path discretization modeling technology, the full-factor data of the drug treatment chain is decomposed into medication decision-making units with time dependencies; Connect medication decision units to medication paths; perform real-time deduction of medication paths based on fuzzy Petri nets, and dynamically update the activation probability of each medication decision unit based on the results of real-time deduction; A path credibility verification mechanism is established to trigger the manual review process when the newly generated medication path matches the preset historical successful case library by less than 65%.

3. The method for intelligent management of clinical pharmacy according to claim 2, characterized in that: The method for constructing the dynamic pharmaceutical ontology framework includes: Establish a drug action quadruple model, where the quadruple includes chemical entities, target proteins, metabolic enzymes and transporters, and each element is associated with at least 3,000 verified biomedical entities; adapt the decision rules of the drug action quadruple model to generate drug-gene interaction decision rules; Define the clinical pathway constraint rule base, including three types of constraints: contraindication conflict detection rules, medication timing optimization rules, and equipment compatibility verification rules; adapt the decision rules of the clinical pathway constraint rule base to generate metabolic pathway coupling decision rules; Construct a drug property attenuation model, adapt the decision rules of the drug property attenuation model, and generate drug delivery device adaptation decision rules; construct a dynamic pharmaceutical ontology framework based on drug-gene interaction decision rules, metabolic pathway coupling decision rules, and drug delivery device adaptation decision rules.

4. The method for intelligent management of clinical pharmacy according to claim 2, characterized in that: Based on the fuzzy Petri net, the medication path is deduced in real time, and the activation probability of each medication decision unit is dynamically updated according to the results of the real-time deduction, including: Set the path node activation threshold function, where the formula of the path node activation threshold function is as follows: Where P activate is the path node activation threshold, k is the drug sensitivity coefficient, c actual is the real-time blood drug concentration monitoring value, c standard is the standard reference value; The unit node activation probability of the medication pathway is calculated by using the pathway node activation threshold function to obtain the pathway unit node activation probability; According to the activation probability of the path unit node, the difference analysis of the activation probability of adjacent nodes in the medication path is performed. When the difference in the activation probability of adjacent nodes exceeds 40%, the pharmaceutical monitoring mark point is automatically inserted. Pharmaceutical monitoring mark points are used to select different paths for medication paths, and fuzzy Petri nets are used to deduce path conflicts for different paths to obtain real-time deduction results. The activation probability of the medication decision-making unit is dynamically adjusted based on the results of real-time deduction.

5. The method for intelligent management of clinical pharmacy according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: Calculate the path morphology difference of the drug digital relationship chain based on the multi-dimensional deviation detection algorithm to obtain the path morphology difference of the relationship chain; Step S32: performing path deviation interaction analysis on the drug digital relationship chain according to the relationship chain path morphology difference, and generating path deviation interaction data; Step S33: using the path deviation interaction data to identify the path intersections of the drug digital relationship chain, and marking the medication decision units corresponding to the identified path intersections as abnormal risk points of drug treatment; Step S34: setting drug intervention nodes for the drug digital relationship chain according to the abnormal risk points of drug treatment to obtain drug intervention nodes.

6. The method for intelligent management of clinical pharmacy according to claim 5, characterized in that: The multi-dimensional deviation detection algorithm in step S31 is as follows: Where D path is the difference degree of relationship chain path morphology, N is the total number of path nodes, w geo is the geometric deviation weight, d geo (i) is the deviation degree of the i-th node in the geometric space, w topo is the topological deviation weight, d topo (i) is to calculate the topological deviation of the i-th node, w time is the time deviation weight, d time (i) is the deviation of the i-th node in the time dimension, w attr is the attribute deviation weight, d attr (i) is the attribute information deviation of the i-th node.

7. The method for intelligent management of clinical pharmacy according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: determining the coverage of drug intervention nodes; Step S42: Analyze the node intervention scope of the drug digital relationship chain based on the coverage of the drug intervention node, and generate the node intervention scope; Step S43: Perform path propagation impact analysis on the drug digital relationship chain through the node intervention scope to generate path propagation impact data; use the path propagation impact data to intelligently regulate the drug dosage of the drug digital relationship chain to perform clinical pharmacy smart management operations.

8. The method for intelligent management of clinical pharmacy according to claim 7, characterized in that: Step S41 includes: Set the coverage radius calculation formula, where the coverage radius calculation formula is as follows: R cover =λ·D path ; Among them, R cover is the coverage radius of the drug intervention node, λ is the coverage radius adjustment factor, D path is the difference in the path morphology of the relationship chain; The coverage range of drug intervention nodes is calculated using the coverage radius calculation formula to obtain coverage range data; Conduct coverage adequacy assessment on drug intervention nodes and generate coverage adequacy assessment data for drug intervention nodes; Determine the coverage of the drug intervention node. If the coverage adequacy assessment data is greater than or equal to the coverage range data, no processing will be performed; if the coverage adequacy assessment data is less than the coverage range data, the intelligent compensation mechanism will be triggered based on the coverage of the drug intervention node.

9. A smart management system for clinical pharmacy, characterized in that: Used to execute the clinical pharmacy intelligent management method as claimed in claim 1, the clinical pharmacy intelligent management system comprises: Data acquisition module, used to obtain real-time data of all elements of the drug treatment chain, including prescription instruction flow, drug preparation parameters, drug administration execution records and patient physiological feedback signals; The relationship chain construction module is used to build a dynamic evaluation engine for the drug treatment chain. The dynamic evaluation engine for the drug treatment chain is used to match the clinical drug mapping relationship of all the data elements of the drug treatment chain to generate an individualized drug use pathway map. The digital twin technology is used to extract the topological chain of the individualized drug use pathway map to generate a digital relationship chain for drugs. The risk point identification module is used to identify the risk points of drug treatment chain breakage in the drug digital relationship chain based on the multi-dimensional deviation detection algorithm to obtain the abnormal risk points of drug treatment; according to the abnormal risk points of drug treatment, the drug intervention nodes are set for the drug digital relationship chain to obtain the drug intervention nodes; The drug dosage compensation module is used to determine the coverage of drug intervention nodes; based on the coverage of drug intervention nodes, the intelligent compensation mechanism is triggered, and the drug dosage of all elements of the drug treatment chain is intelligently regulated through the intelligent compensation mechanism to perform clinical pharmacy smart management operations.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for intelligent management of clinical pharmacy as described in any one of claims 1 to 7 is implemented.

Images