Intelligent closed-loop traceability management system in hemp essence medicine medical institution
By setting electronic tags on drugs and building a causal network, combined with patient data, intelligent management of the entire life cycle of drugs is achieved, solving the problems of cumbersome drug management and difficult data integration, and improving drug safety and treatment accuracy.
Patent Information
- Application Number
- CN202510751226.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-06
- Publication Date
- 2025-10-10
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing drug management and traceability systems in medical institutions rely on manual operations, resulting in cumbersome management, difficult data integration, real-time and privacy issues, making it difficult to achieve effective management and control of the entire life cycle of drugs.
Electronic tags are used for real-time tracking of drugs. Patient data and drug metabolism models are combined to build a causal network of drugs, patients and adverse reactions. Multimodal sensing technology and digital twins are used for full life cycle management to achieve real-time monitoring of drug status and accurate analysis of adverse reactions.
It achieves high-fidelity digital mapping of the entire life cycle of drugs, improves the efficiency of drug safety risk identification and the accuracy of handling, and forms a closed-loop management covering drug quality-patient individual characteristics-clinical operations, which can quickly locate the source of risk and carry out individualized treatment.
Smart Images
Figure CN120767009A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a field related to drug traceability, and in particular to an intelligent closed-loop traceability management system for narcotic and psychiatric drugs in medical institutions. Background Art
[0002] With the rapid development of the medical internet of things (IoT) and digital manufacturing technologies, drug management and monitoring systems are playing an increasingly important role in medical institutions and pharmaceutical companies. However, current drug safety traceability and management still rely primarily on traditional methods, which have numerous shortcomings. Traditional drug management methods often rely on manual operations, making the process of drug entry and exit from warehouses cumbersome and time-consuming, prone to human registration and management errors, and causing inconvenience to pharmaceutical warehouse management. Furthermore, existing drug traceability systems also have significant issues with data integration, real-time performance, and privacy, making it difficult to achieve effective control over the entire drug lifecycle. Summary of the Invention
[0003] The present invention overcomes the deficiencies of the prior art and provides an intelligent closed-loop traceability management system for narcotic drugs in medical institutions.
[0004] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is: an intelligent closed-loop traceability management system for narcotic drugs in medical institutions, comprising the following steps:
[0005] S1: Set electronic tags on drugs to track them in real time and record their real-time status;
[0006] S2: Extract the overall records of patients in the hospital system and record the real-time status of patients, align them in chronological order, and import the overall records and real-time status into the PD model to generate the patient's drug metabolism model to simulate the patient's drug metabolism status;
[0007] S3: Build digital twins of the drug and patient, import the real-time status of the drug in S1 and the overall records and real-time status of the patient in S2, as well as the drug metabolism model, into the digital twins. Align the two digital twins in chronological order to construct a causal relationship network between drug, patient, and adverse reaction.
[0008] S4: Use the relationship network built by S3 to check in real time whether the patient has adverse reactions, analyze the causes of adverse reactions, and dispose of the drugs.
[0009] In a preferred embodiment of the present invention, in S1, real-time tracking is specifically performed by scanning the electronic tag through the scanning terminal of the pharmacy and medical staff and recording the initial data and flow data of the drug in chronological order.
[0010] In a preferred embodiment of the present invention, the initial data includes the production time, ingredients, effective time and quality inspection report of the drug, and the circulation data includes the status of the drug's outbound, inbound, dispensing and use. The circulation data is arranged in order of scanning time.
[0011] In a preferred embodiment of the present invention, in S2, the overall record includes the patient's genotype, medical records, medication records and physical condition.
[0012] In a preferred embodiment of the present invention, in S2, the drug metabolism model specifically establishes a concentration-effect relationship for the drug action mechanism. For drugs with effect lag or complex signal cascades, time is introduced into the concentration-effect relationship established by the former mechanism, and the dynamic balance between the effect compartment concentration and the plasma concentration is characterized by a differential equation.
[0013] In a preferred embodiment of the present invention, the drug with delayed effect or complex signal cascade is a narcotic drug.
[0014] In a preferred embodiment of the present invention, in S3, the causal relationship network of drug-patient-adverse reaction is constructed as follows:
[0015] A1: Build a triplet network of drugs, patient characteristics, and adverse reactions through the knowledge graph, connecting individual patient nodes, drug nodes, and adverse reaction nodes through relationship edges, including the "metabolic dependency" of drugs and enzymes, the "activity association" of patient genotypes and enzymes, and the "threshold trigger" of drug concentrations and adverse reactions.
[0016] A2: Based on the knowledge graph, a three-layer probabilistic dependency structure of the Bayesian network is defined. The root node is set as the patient's genetic polymorphism and concomitant medication, the intermediate node is the drug exposure, and the leaf node is the adverse reaction event and its severity level. The causal discovery algorithm is used to identify potential dependency paths, and latent variables are introduced to model unobserved confounding factors.
[0017] A3: By monitoring the patient's status in real time and updating the network weights according to the patient's status.
[0018] In a preferred embodiment of the present invention, the electronic tag is a multimodal sensing RFID chip that integrates a temperature sensor, an accelerometer and a Beidou positioning module to record the temperature, humidity, vibration intensity and geographic location coordinates of the environment in which the medicine is located in real time.
[0019] In a preferred embodiment of the present invention, an adaptive Kalman filter algorithm is introduced into the drug metabolism model to obtain the patient's blood drug concentration data in real time through wearable devices and dynamically correct the pharmacokinetic parameters. For narcotic drugs, the model is synchronously connected to the MIC susceptibility test results of the hospital's LIS system to adjust the bactericidal effect parameters in the differential equation.
[0020] In a preferred embodiment of the present invention, when the analysis result shows that the adverse reaction is caused by drug quality problems, the drug is recalled; if it is a normal adverse reaction for the patient, the patient's treatment plan is adjusted.
[0021] The present invention solves the defects existing in the background technology and has the following beneficial effects:
[0022] (1) The present invention provides an intelligent closed-loop traceability management system for narcotic drugs in medical institutions. It uses multimodal sensing technology to capture the physical state of drugs in real time, combines multidimensional physiological data such as patient genotype and metabolic indicators, and uses digital twin simulation technology to construct a causal chain of "drug quality decay-patient physiological response". When the drug's efficacy decreases due to environmental out-of-control, the system dynamically simulates the drug efficacy decay curve through a virtual twin, and conducts cross-dimensional correlation analysis with the patient's physiological indicators after taking the drug, so as to accurately locate the root cause of the problem - distinguishing between drug quality defects, patient metabolic abnormalities or medication operation errors. This achieves a leap from passive tracing to active intervention, significantly improves the efficiency of identifying and handling drug safety risks, and forms a closed-loop management covering the entire life cycle of "drug quality-patient individual characteristics-clinical operation".
[0023] (2) The present invention provides an intelligent closed-loop traceability management system for narcotic drugs in medical institutions, by constructing a high-fidelity digital mapping of the entire life cycle of drugs. Based on the real-time collection of drug environmental data (such as temperature, humidity, vibration records) and flow node information, the physical state and spatial position of the drug twin are dynamically updated, and combined with the three-dimensional spatial model of the virtual pharmacy, visual management of the drug storage environment and immersive monitoring of abnormal conditions are achieved. When the drug is exposed to risk conditions (such as cold chain breaks, expiration date approaching), the system triggers a risk warning through the twin state, and at the same time highlights the location coordinates of the problem drug and the deviation trajectory of the environmental parameters in the virtual scene, guiding management personnel to quickly locate the source of the risk. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments described in the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive efforts.
[0025] Figure 1 It is the process structure of the preferred embodiment of the present invention;
[0026] Figure 2 It is a module diagram of a preferred embodiment of the present invention. DETAILED DESCRIPTION
[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0028] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.
[0029] As shown in the figure, an intelligent closed-loop traceability management system for narcotic drugs in medical institutions includes the following steps:
[0030] It should be noted that this application builds a drug life cycle traceability system based on digital twin technology to track the dynamic data of the entire process of drugs from production, transportation, storage to patient use in real time. The system creates digital twins of drugs and patients in a virtual space. The drug twin records details such as the batch, temperature and humidity exposure, and expiration status of each drug. The patient twin integrates genetic data, vital signs, medication records, and dietary information to form a multidimensional dynamic archive. Through real-time data synchronization and causal inference models (such as Bayesian networks and knowledge graphs), the system automatically analyzes the root causes of adverse drug reactions - for example, distinguishing production batch quality problems, patient genetic metabolic defects, or omissions in the medication process, and triggering a hierarchical response strategy: automatically recalling and destroying problematic drugs, correcting medication plans for high-risk groups in real time, and optimizing operating specifications for process loopholes. Ultimately, a closed loop of "tracking-attribution-disposal-feedback" is formed, realizing the transformation from passive tracing to active prevention and control.
[0031] In this way, the digital twin system can be used to track drug details in real time, record the flow of drugs, and record the drug users and their status, diet, and reactions before and after taking the drugs. Different reactions are fed back to the system, a causal relationship judgment model is established, and then the drugs are judged and handled.
[0032] S1: Set electronic tags on drugs to track them in real time and record their real-time status;
[0033] In the present application, in S1, real-time tracking is specifically scanning the electronic tag by the code scanning terminal of the pharmacy and medical staff and recording the initial data and flow data of the drug in time sequence. The initial data includes the production time, ingredients, effective time and quality inspection report of the drug, and the flow data includes the state of the drug out of warehouse, into warehouse, dispensing and use, and the flow data is arranged in time sequence. The electronic tag is a multi-modal sensing RFID chip, integrating temperature sensor, accelerometer and Beidou positioning module, which can record the temperature, humidity, vibration intensity and geographic position coordinates of the environment where the drug is located in real time.
[0034] It should be noted that the present system uses a multi-modal sensing RFID chip as the core carrier of the electronic tag, which integrates a DS18B20 high-precision temperature sensor, an ADXL345 three-axis accelerometer and a Beidou / GPS dual-mode positioning module. The DS18B20 high-precision temperature sensor can monitor the environment temperature of the drug in real time with an error of ±0.5℃, with a sampling frequency of 1 time / minute, and the data is stored in the 4MB flash memory built-in the chip. The ADXL345 three-axis accelerometer is used to record the vibration intensity (range ±16g) during transportation, and automatically triggers an alarm mark when detecting an instantaneous impact exceeding 5g, avoiding excessive frequent shaking of the drug during transportation affecting the stability of the drug properties, such as liposome injection and freeze-dried powder injection. Severe vibration may cause particle aggregation or structural damage, and accelerometer data is used to evaluate whether the drug is exposed to a risky environment. The Beidou / GPS dual-mode positioning module obtains the geographic position coordinates through the NMEA-0183 protocol, and combines with the GIS geographic fence technology to alarm in real time when the drug moves abnormally out of the preset distribution area.
[0035] The RFID tag is embedded on the drug on the drug production line, recording the initial data (production time, ingredients, quality inspection report) to the blockchain. The RFID reader scans the state of each flow (into warehouse, out of warehouse, dispensing) and updates the digital twin state, and the key nodes (such as cold chain break) trigger the smart contract to record the exception. In the digital twin, the electronic tag displays the real-time position of the drug (such as a batch of insulin stored in A medicine shelf 3), the remaining validity period (countdown red warning), and the environmental risk (such as temperature exceeding the standard area flashing warning).
[0036] S2: Extract the overall record of the patient in the hospital system and record the real-time state of the patient, align and arrange in time sequence, and import the overall record and real-time state into the PD model to generate the drug metabolism model of the patient and simulate the drug metabolism state of the patient;
[0037] In the present application, in S2, the overall record includes the genotype, case, medication record and physical condition of the patient; the drug metabolism model is specifically a drug mechanism-based concentration-effect relationship, for drugs with effect lag or complex signal cascade, the time is introduced on the concentration-effect relationship established by the former mechanism, and the differential equation is used to describe the dynamic balance between the effect chamber concentration and the plasma concentration; the drugs with effect lag or complex signal cascade are specifically narcotic drugs. The drug metabolism model introduces an adaptive Kalman filter algorithm, acquires the blood drug concentration data of the patient in real time through a wearable device, and dynamically corrects the pharmacokinetic parameters; for narcotic drugs, the model synchronously accesses the MIC drug sensitivity test results of the hospital LIS system, and adjusts the bactericidal effect parameters in the differential equation.
[0038] It should be noted that the overall record of the patient is extracted from the hospital HIS system structured electronic medical record (including ICD-10 diagnosis code, operation record); the real-time state of the patient is obtained by wearing a wearable device (such as Philips Biosensor BX100) by the patient to acquire continuous vital sign data (heart rate, blood oxygen, respiratory rate), and the sampling rate is 10Hz.
[0039] For narcotic drugs, a two-compartment model is used, and the differential equation set is:
[0040] dCp / dt=-(k10+k12)Cp+k21Ct
[0041] dCt / dt=k12Cp-k21Ct
[0042] Wherein, Cp is the central chamber concentration, Ct is the peripheral chamber concentration, and k is the transfer rate constant.
[0043] For narcotic drugs, a two-compartment model is used to describe the dynamic distribution and elimination process of drugs in the human body. The model abstracts the body into two physiological compartments: central compartment (organs with rich blood flow, such as heart, liver) and peripheral compartment (tissues with slow blood flow, such as fat, muscle), and the mathematical expression of the model is based on the law of conservation of mass, including the following two differential equations:
[0044] Central chamber concentration change rate:
[0045] Wherein, C p is the central chamber drug concentration (ug / mL), k 10 is the first-order rate constant of drug elimination from the central chamber (h -1 ), k 12 is the rate constant of drug transfer from the central chamber to the peripheral chamber (h -1 ), k 21 is the rate constant of drug return from the peripheral chamber to the central chamber (h -1), the first term on the right side of the equation represents the loss of drugs from the central compartment due to elimination (such as renal excretion) and outward transport to the peripheral compartment, and the second term represents the reverse transport supplement of drugs from the peripheral compartment to the central compartment.
[0046] Peripheral room concentration change rate:
[0047] Among them, C t is the drug concentration in the peripheral compartment. The equation describes the dynamic equilibrium of drug concentration in the peripheral compartment, and its changes are driven only by bidirectional transport between the central compartment and the peripheral compartment.
[0048] A nonlinear mixed-effects model (NONMEM) was then fitted to the population data, collecting plasma drug concentration data at multiple time points (e.g., 0, 1, 4, and 12 hours) after intravenous administration. First-order conditional estimation (FOCE) was used to estimate the k10k10, k12k12, and k21k21 parameters, and the interindividual variation (IIV) and residual variation (RUV) were calculated. Trough concentrations were then predicted based on the estimated parameters to guide dose adjustments to maintain the therapeutic window (e.g., a target vancomycin trough concentration of 10-20 μg / mL).
[0049] The effect compartment model is introduced for targeted drugs (such as alfentanil), and the relationship between the effect compartment concentration Ce and Cp is defined as: dCe / dt=ke0(Cp-Ce)
[0050] Where ke0 is the equilibrium rate constant of the effect compartment, and the individualized parameters are estimated by a nonlinear mixed-effects model (NONMEM).
[0051] For targeted drugs such as alfentanil, it is necessary to correlate drug exposure with pharmacodynamic response. Therefore, the effect compartment model is introduced to characterize the delayed distribution process of drugs from plasma to the target site.
[0052] Mathematical expression of the effect compartment model:
[0053] Among them, C e Effect site drug concentration (ng / mL), representing the actual exposure level of the drug at the target site (such as μ opioid receptor); k e0 is the equilibrium rate constant of the effect compartment (h -1 ), determines the rate of drug diffusion from plasma to effect chamber; this equation reflects C e Tends to be similar to C p The hysteresis effect of the dynamic equilibrium process explains the delayed effect of drug observed in clinical practice (such as the analgesic effect lags behind the peak blood drug concentration).
[0054] Pharmacodynamics (PD), the effect compartment is further enhanced by S-type E max Models are associated with efficacy indicators (such as analgesia scores):
[0055] Among them, E max EC is the maximum effect intensity; 50 is the concentration that produces 50% of the maximum effect; γ is the Hill coefficient, which characterizes the steepness of the concentration-effect curve.
[0056] Based on therapeutic drug monitoring (TDM) data and real-time feedback from wearable devices: integrating population pharmacokinetic data with individual patient genotypes (such as OPRM1 A118G polymorphism), and dynamically modifying k using a Bayesian feedback algorithm. e0 and EC 50 For patients on postoperative analgesia pump (PCA), update k every 30 minutes e0 The system estimates the value and adjusts the infusion rate through the PID controller to maintain the pain score within the target range (VAS 3-4 points). Through the coordinated application of the two types of models, the system can achieve closed-loop management from drug exposure prediction to dynamic regulation of drug efficacy, providing quantitative decision support for the personalized treatment of narcotic drugs.
[0057] S3: Build digital twins of the drug and patient, import the real-time status of the drug in S1 and the overall records and real-time status of the patient in S2, as well as the drug metabolism model, into the digital twins. Align the two digital twins in chronological order to construct a causal relationship network between drug, patient, and adverse reaction.
[0058] In a preferred embodiment of the present invention, in S3, the causal relationship network of drug-patient-adverse reaction is constructed as follows:
[0059] A1: Build a triplet network of drugs, patient characteristics, and adverse reactions through the knowledge graph, connecting individual patient nodes, drug nodes, and adverse reaction nodes through relationship edges, including the "metabolic dependency" of drugs and enzymes, the "activity association" of patient genotypes and enzymes, and the "threshold trigger" of drug concentrations and adverse reactions.
[0060] A2: Based on the knowledge graph, a three-layer probabilistic dependency structure of the Bayesian network is defined. The root node is set as the patient's genetic polymorphism and concomitant medication, the intermediate node is the drug exposure, and the leaf node is the adverse reaction event and its severity level. The causal discovery algorithm is used to identify potential dependency paths, and latent variables are introduced to model unobserved confounding factors.
[0061] A3: By monitoring the patient's status in real time and updating the network weights according to the patient's status.
[0062] It should be noted that the drug twin adopts Unity3D engine to build 3D visualization model, mapping the environmental exposure history of physical medicine (such as temperature-time integral value T1℃·h), and triggering the expiration correction algorithm when the cumulative exposure value exceeds the USP <1079> standard; the patient twin realizes real-time data fusion through TensorFlow Extended (TFX) pipeline, and updates the metabolic state prediction (such as the cerebrospinal fluid / plasma concentration ratio of fentanyl) every 5 minutes.
[0063] The structured mapping of medical knowledge is carried out in the knowledge graph layer, and the core of the knowledge graph layer is to convert the complex relationship between drugs, patient characteristics and adverse reactions into a computable semantic network. A semantic model conforming to the RDF standard is built through the Apache Jena framework to accurately describe the association between medical entities in the form of triples. For example, the triple: Patient_X: hasGenotype: CYP2D6*10 / *10 indicates that patient X carries a specific drug metabolizing enzyme genotype (CYP2D6*10 / *10). This genotype is clinically associated with significantly reduced enzyme activity, directly affecting drug metabolism efficiency. The definition of the drug metabolism path Oxy codone: metabolizedBy: CYP2D6 indicates that the main metabolism of oxy codone depends on the CYP2D6 enzyme. Such relationships are derived from pharmacology databases, providing a basis for subsequent analysis of the association between drug exposure and adverse reactions. In the genotype-enzyme activity quantification, CYP2D6*10 / *10: reducesActivityTo 0.3 numerically represents the impact of genotype variation on enzyme activity (e.g., activity reduced to 30% of normal level), which is used to calculate the change in drug metabolism rate. The construction of the knowledge graph not only realizes the digital expression of medical knowledge, but also supports dynamic expansion. For example, when a new drug and enzyme inhibition relationship is added, a clinical alert rule (such as "strong inhibitor combination requires dose adjustment") can be automatically triggered through semantic reasoning.
[0064] A dynamic probabilistic inference engine is used at the Bayesian network layer, which focuses on causal reasoning under quantitative uncertainty. The hierarchical probabilistic model is constructed using the GeNIe Modeler tool. The root node is designed to be based on the patient's genetic polymorphism (e.g., CYP3A4 rapid, intermediate, and slow metabolizer phenotypes) and concomitant medications (e.g., rifampicin). Prior probabilities for these nodes are set based on population epidemiological data. For example, the frequency of CYP3A4 slow metabolizers in Asian populations is approximately 5-10%. Modeling is performed at the intermediate nodes, with drug exposure (e.g., fentanyl's AUC0-24) as the intermediate node, and its conditional probability table is calculated using a pharmacokinetic model. For example, when a patient is an extensive CYP3A4 metabolizer and is taking a concomitant inducer, the fentanyl metabolism rate is accelerated, and the probability of a decrease in the AUC value increases significantly. Mapping is performed at the leaf nodes, with adverse events (e.g., the severity of respiratory depression) as the leaf nodes, and their probability distribution is dynamically linked to drug exposure and the patient's physiological state (e.g., respiratory rate). For example, when the fentanyl AUC exceeds the safety threshold (such as 60 ng·h / mL), the probability of severe respiratory depression increases exponentially.
[0065] The dynamic nature of Bayesian networks is reflected in real-time data-driven parameter updates. For example, using continuously monitored blood oxygen saturation data from patients, the probability distribution of drug exposure can be reversed to improve early warning accuracy.
[0066] Causal discovery algorithms identify potential causal pathways from real-world data, avoiding the limitations of prior knowledge by integrating multi-source data from the emergency department adverse reaction (ADR) case library, including patient genotypes, medication records, vital signs, etc. Statistical tests (such as partial correlation analysis) are used to identify independent relationships between variables, such as verifying whether respiratory depression is independent of patient age after controlling for drug concentration variables. The FCI algorithm is used to generate a maximum ancestral graph, clarifying three types of edge relationships, using direct causal arrows (→) to represent, for example, "CYP3A4 phenotype → fentanyl metabolic rate"; using potential confounding associations Indicates "untested intestinal flora Drug bioavailability"; spurious associations caused by selection bias are eliminated through data screening rules. When the causal network detects the combined event of "drug concentration exceeding the threshold" and "respiratory rate decrease", a graded alarm is automatically triggered (such as a pop-up prompt, an audible and visual alarm at the nurse station).
[0067] S4: Use the relationship network built by S3 to check in real time whether the patient has adverse reactions, analyze the causes of the adverse reactions, and dispose of the drugs; when the analysis results show that the adverse reactions are caused by drug quality problems, the drugs will be recalled. If it is a normal adverse reaction for the patient, the patient's treatment plan will be adjusted.
[0068] It should be noted that the dual-track processing logic of drug quality attribution and individualized treatment plan optimization is realized through the causal relationship network. When the system detects that a batch of drugs has ≥3 cases of similar ADR, the quality backtracking algorithm is activated: the spatiotemporal trajectory map of drug circulation is called to calculate the abnormal temperature and humidity exposure risk value (such as the cumulative time of temperature > 25°C accounts for more than 5% of the validity period); the hash value of the quality inspection report of the same batch of drugs is compared through the path backtracking algorithm. If the peak area difference between the original chromatogram (such as HPLC fingerprint) and the current sample is found to be >15%, it is judged as a production defect and the drug is recalled.
[0069] If the causal network shows a genotype-phenotype mismatch, such as a morphine metabolism rate of <1% in a patient with the CYP2D6*4 / *4 genotype taking codeine; or a drug-drug interaction, such as a >200% increase in the AUC when a strong CYP3A4 inhibitor (such as clarithromycin) is used simultaneously with fentanyl; or a physiological model prediction deviation, such as a >30% deviation between the predicted creatinine clearance (eGFR) of a patient's twin and the actual measured value, then the adverse reaction is determined to be caused by individual patient factors, and the patient's treatment plan is adjusted.
[0070] The above description is based on the ideal embodiment of the present invention. Based on the above description, relevant personnel can make various changes and modifications without departing from the technical scope of this invention. The technical scope of this invention is not limited to the content of the specification and must be determined according to the scope of the claims.
Claims
1. An intelligent closed-loop traceability management system for narcotic drugs in medical institutions, characterized by: The following steps are involved: S1: Set electronic tags on drugs to track them in real time and record their real-time status; S2: Extract the overall records of patients in the hospital system and record the real-time status of patients, align them in chronological order, and import the overall records and real-time status into the PD model to generate the patient's drug metabolism model to simulate the patient's drug metabolism status; S3: Build digital twins of the drug and patient, import the real-time status of the drug in S1 and the overall records and real-time status of the patient in S2, as well as the drug metabolism model, into the digital twins. Align the two digital twins in chronological order to construct a causal relationship network between drug, patient, and adverse reaction. S4: Use the relationship network built by S3 to check in real time whether the patient has adverse reactions, analyze the causes of adverse reactions, and dispose of the drugs.
2. The intelligent closed-loop traceability management system for narcotic drugs in medical institutions according to claim 1, characterized in that: In S1, real-time tracking is specifically to scan the electronic tag through the scanning terminal of the pharmacy and medical staff and record the initial data and circulation data of the drug in chronological order.
3. The intelligent closed-loop traceability management system for narcotic drugs in medical institutions according to claim 2, characterized in that: The initial data includes the production time, ingredients, effective time and quality inspection report of the drug, and the circulation data includes the status of the drug's outbound, inbound, dispensing and use. The circulation data is arranged in the order of scanning time.
4. The intelligent closed-loop traceability management system for narcotic drugs in medical institutions according to claim 1, characterized in that: In the S2, the overall record includes the patient's genotype, medical records, medication records and physical condition.
5. The intelligent closed-loop traceability management system for narcotic drugs in medical institutions according to claim 1, characterized in that: In the S2, the drug metabolism model specifically establishes a concentration-effect relationship for the drug action mechanism. For drugs with effect lag or complex signal cascades, time is introduced into the concentration-effect relationship established by the former mechanism, and the dynamic balance between the effect compartment concentration and the plasma concentration is characterized by differential equations.
6. The intelligent closed-loop traceability management system for narcotic drugs in medical institutions according to claim 5, characterized in that: The drug with delayed effect or complex signal cascade is specifically a narcotic drug.
7. The intelligent closed-loop traceability management system for narcotic drugs in medical institutions according to claim 1, characterized in that: In S3, the causal relationship network of drug-patient-adverse reaction is constructed as follows: A1: Build a triplet network of drugs, patient characteristics, and adverse reactions using a knowledge graph. Connect individual patient nodes, drug nodes, and adverse reaction nodes through edges, including the "metabolic dependency" of drugs and enzymes, the "activity association" of patient genotypes and enzymes, and the "threshold trigger" of drug concentrations and adverse reactions. A2: Based on the knowledge graph, a three-layer probabilistic dependency structure of the Bayesian network is defined. The root node is set as the patient's genetic polymorphism and concomitant medication, the intermediate node is the drug exposure, and the leaf node is the adverse reaction event and its severity level. The causal discovery algorithm is used to identify potential dependency paths, and latent variables are introduced to model unobserved confounding factors. A3: By monitoring the patient's status in real time and updating the network weights according to the patient's status.
8. The intelligent closed-loop traceability management system for narcotic drugs in medical institutions according to claim 1, characterized in that: The electronic tag is a multimodal sensing RFID chip that integrates a temperature sensor, an accelerometer and a Beidou positioning module to record the temperature, humidity, vibration intensity and geographic location coordinates of the environment in which the medicine is located in real time.
9. The intelligent closed-loop traceability management system for narcotic drugs in medical institutions according to claim 1, characterized in that: The drug metabolism model introduces an adaptive Kalman filter algorithm to obtain patient blood drug concentration data in real time through wearable devices and dynamically correct pharmacokinetic parameters. For narcotic drugs, the model synchronously accesses the MIC susceptibility test results of the hospital's LIS system to adjust the bactericidal effect parameters in the differential equation.
10. The intelligent closed-loop traceability management system for narcotic drugs in medical institutions according to claim 1, characterized in that: When the analysis results show that the adverse reaction is caused by drug quality problems, the drug will be recalled. If it is a normal adverse reaction for the patient, the patient's treatment plan will be adjusted.