Intelligent opium drug management system and method

Abstract

The application belongs to the technical field of intelligent management of narcotic drugs, and discloses an intelligent narcotic drug management system and method, which comprises the following steps: processing operation context parameters and calculating a risk score; secondly, checking drug metering specifications by using a white box rule engine, and analyzing clinical path compliance and drug interactions; then, combining biological characteristics and geographic location information, real-time risk assessment and multi-level intervention measures are taken; next, a detailed visual audit report is generated according to the collected data; finally, the system performance is dynamically adjusted based on feedback to optimize the strategy; not only the safety and accuracy of medical operations are improved, but also the intelligent level and response efficiency of the system are significantly improved, and a highly safe operation verification process and precise prescription data checking capability are realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent management technology for narcotic and psychotropic drugs, and more specifically, to an intelligent management system and method for narcotic and psychotropic drugs. Background Technology

[0002] Narcotic and psychotropic drugs (narcotic drugs and psychotropic substances) play an important role in medical procedures due to their unique pharmacological effects, but they also pose a high risk of abuse. Therefore, the management of narcotic and psychotropic drugs requires special caution to ensure that their use meets clinical needs while preventing abuse. Traditional management systems for narcotic and psychotropic drugs typically rely on manual review and paper records, which are not only inefficient but also prone to human error or malicious tampering.

[0003] Existing operational verification methods typically rely on single authentication methods, such as passwords or simple biometrics (e.g., fingerprints). These methods fall short when faced with complex operational environments and changing behavioral patterns. The lack of comprehensive assessment of operational context parameters, including time granularity, geographical location, and device information, makes it impossible to accurately assess operational risks. Furthermore, existing prescription data verification technologies often overlook complex drug interactions and individualized patient health conditions (e.g., liver and kidney function), potentially leading to medication risks. Simultaneously, these systems fail to fully utilize the large amounts of data generated during operation for subsequent analysis and decision support, limiting the system's intelligence and response speed. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: an intelligent narcotic and psychotropic drug management system, comprising:

[0005] Operator authentication and dynamic generation of verification units: The operator performs verification based on the operation context parameters to obtain the operation risk score, and then constructs a knowledge-driven decision rule base to dynamically generate the operator's verification strategy. The operator's multimodal biometric identifier set is verified through multi-node collaborative verification to obtain the verification result and biometric hash.

[0006] Prescription data verification and early warning conflict intensity calculation unit: Based on the verification results, it captures the medication data of patients in all departments, verifies the drug dosage specifications and clinical pathway compliance level by level through the white box rule engine, identifies high-risk combinations through the pharmacokinetic interaction matrix, and obtains the early warning conflict intensity;

[0007] Real-time risk assessment and multi-level intervention blocking unit: Based on the operator's geographical location, combined with encrypted biometric hash, operation risk score, prescription data and early warning conflict intensity, a spatiotemporal-biological joint map is constructed to obtain a real-time risk score. Combining the real-time risk score and early warning conflict intensity, a comprehensive risk assessment value is generated. Based on the comprehensive risk assessment value, an interception instruction is generated to perform multi-level intervention blocking measures on high-risk behaviors and obtain an encrypted evidence chain package.

[0008] Visualized report generation unit: Based on operational risk scores, early warning conflict intensity, and encrypted evidence chain packages, it generates drug management trajectory files, constructs a visualized audit map, and generates audit reports and drug flow heat maps;

[0009] Dynamic adjustment and optimization unit: Based on audit reports and drug flow heat maps, dynamically adjust the operational risk threshold range and update the knowledge-driven decision rule base and verification strategy configuration scheme.

[0010] Furthermore, the construction method of the knowledge-driven decision rule base includes:

[0011] Operational context parameters include time granularity, pharmaceutics lineage and corresponding drug control level, historical behavior reputation value, geographic location and device information;

[0012] The weighted average method is used to calculate all influencing factors within the operational context parameters to obtain an operational risk score. An operational risk threshold range is then set, and the operational risk score is divided into different operational risk levels.

[0013] Set a geofence and a maximum deviation distance. If the actual path distance between the geographic location and the geofence range is greater than the maximum deviation distance, mark it as a location anomaly.

[0014] If the actual path distance between the geographic location and the geofence is less than or equal to the maximum deviation distance, it is marked as a normal location; normal location and abnormal location are used as the location status of the geographic location.

[0015] Different time granularities, location status, drug control levels, and operational risk levels are combined to form an operational scenario;

[0016] One or more verification strategies are randomly selected from the set of verification strategies, and a verification occurrence method is randomly selected from edge and cloud center verification. The selected verification strategies and verification occurrence methods are combined to form a verification strategy combination.

[0017] The verification strategy set includes vein recognition, fingerprint recognition, iris recognition and dynamic password. Edge and cloud center verification includes local node verification and cloud node verification. Local nodes are local servers and cloud nodes are remote data centers.

[0018] For each operation scenario, randomly select 'a' strategies from all strategy combinations as the initial verification strategy set for the operation scenario;

[0019] All operation scenarios and their corresponding initial verification strategy sets are integrated to construct an initial operation scenario-verification strategy mapping table.

[0020] The initial operation scenario-verification strategy mapping table is updated, and an expert rule engine is used to generate a knowledge-driven decision rule base based on the updated operation scenario-verification strategy mapping table.

[0021] Furthermore, the method for updating the initial operation scenario-verification strategy mapping table includes:

[0022] Based on the initial operation scenario-verification strategy mapping table, a verification strategy combination is randomly assigned to each operation scenario to obtain the initial operation scenario-verification strategy combination. Then, all initial operation scenario-verification strategy combinations are integrated to obtain a complete set of verification strategy configuration schemes. The assignment operation is repeated b times to obtain b different verification strategy configuration schemes, which in turn form the initial strategy configuration pool.

[0023] Based on the initial policy configuration pool, the effective value of each initial operation scenario-verification policy combination in each verification policy configuration scheme is calculated, and the average effective value of each verification policy configuration scheme is obtained by taking the average value.

[0024] For each operation scenario, extract the verification strategy combination with the highest effective value from the c verification strategy configuration schemes with the highest average effective value to obtain a new operation scenario-verification strategy combination, and then integrate it into a new verification strategy configuration scheme.

[0025] Replace the verification strategy configuration scheme with the lowest average effective value in the initial strategy configuration pool with the new verification strategy configuration scheme. Repeat the calculation, extraction and replacement of effective values ​​until the maximum number of iterations is reached or the effective value of all operation scenarios-verification strategy combinations no longer increases, and obtain the best verification strategy combination for all operation scenarios.

[0026] The operation scenario-verification strategy mapping table is updated using the optimal combination of verification strategies to obtain the updated operation scenario-verification strategy mapping table.

[0027] Furthermore, the verification results and biometric hashes are obtained in the following ways:

[0028] Based on the knowledge-driven decision rule base, a combination of verification strategies is matched for the operator. Based on the matched combination of verification strategies, the operator's multimodal biometric identifier is obtained through the device currently operated by the operator and integrated into a set of biometric identifiers.

[0029] Among them, multimodal biometrics include vein biometrics, fingerprint biometrics, iris biometrics, and dynamic password biometrics;

[0030] The operator's biometric set is initially processed, including image preprocessing, biometric extraction, and multimodal biometric fusion;

[0031] If the matching verification strategy is local node verification, the pre-processed multimodal biometric identifier set will be sent to the local node.

[0032] If the matching verification strategy involves cloud node verification, the pre-processed multimodal biometric identifier set will be sent to the local node and d cloud nodes, where d is greater than 1.

[0033] For the device currently operated by the operator and each node participating in the verification, the received set of biometric identifiers is independently compared using its own database to obtain the corresponding verification results, including whether they match and the corresponding confidence score.

[0034] Based on all the verification results, calculate the difference between all the verification results. If the difference is less than the preset difference threshold, the verification result is determined to be correct and is taken as the final verification result. The corresponding verification result signal is generated. The types of verification result signals include verification success signal and verification failure signal.

[0035] If the difference value is greater than or equal to the preset difference value threshold, it is determined that there is abnormal behavior in the current verification, triggering the locking mechanism and generating a verification failure signal, while also providing feedback to the operator;

[0036] Based on the successful verification signal, the set of biometric identifiers is converted into biometric hash values ​​and encrypted to obtain the encrypted biometric hash.

[0037] Furthermore, the methods for providing early warning of conflict intensity include:

[0038] Based on the encrypted final verification result, the patient's prescription data is obtained, and then the corresponding prescription dosage is obtained;

[0039] By capturing and analyzing the prescription medication data of patients in all departments through the hospital's HIS system, the cumulative dosage of similar drugs within 24 hours can be calculated.

[0040] The white-box rule engine was used to retrieve the corresponding drug instructions in the hospital's HIS system and extract the maximum single dose and 24-hour cumulative dose thresholds.

[0041] The corrected cumulative dose threshold is obtained by adjusting the 24-hour cumulative dose threshold according to the patient's liver and kidney function status.

[0042] Based on the cumulative dose of similar drugs and the revised cumulative dose threshold, if the cumulative dose is less than or equal to the cumulative dose threshold, the drug dose is deemed to be in compliance with the standard.

[0043] If the cumulative dose exceeds the cumulative dose threshold, the drug dose is determined to be non-compliant, and a drug dose warning is generated.

[0044] We acquire patients’ disease history, allergy history, liver and kidney function indicators and treatment plans, and then construct a rule base for the association of drugs, diseases, test indicators and treatment plans. We then use the rule base to construct a clinical pathway knowledge graph.

[0045] The clinical pathway knowledge graph is used to perform subgraph matching on the current prescription data to calculate the clinical pathway compliance degree.

[0046] If the clinical pathway compliance rate is greater than or equal to the preset clinical pathway compliance rate threshold, the clinical pathway compliance rate is determined to be in compliance with the standard.

[0047] If the clinical pathway compliance rate is less than the preset clinical pathway compliance rate threshold, it indicates a deviation from the standard pathway, and the clinical pathway compliance rate is determined to be non-compliant, generating a clinical pathway compliance rate warning.

[0048] Pharmacokinetic interaction matrix was used to detect high-risk interaction combinations in prescription drug use data across all departments.

[0049] If a high-risk interaction combination is detected, the prescription is determined to have a high-risk interaction combination, and a high-risk warning is generated; if no high-risk combination is detected, the prescription is determined not to have a high-risk interaction combination.

[0050] The intensity of the warning conflict is determined based on the results of the determination of drug dosage specifications, clinical pathway compliance, and high-risk interaction combinations.

[0051] Furthermore, the corrected cumulative dose threshold is obtained in the following ways:

[0052] The patient's liver and kidney function indicators were obtained through the hospital's HIS system, and the kidney function correction factor was calculated using the Cockcroft-Gault formula based on the kidney function indicators.

[0053] Liver function correction factors were assessed using the Child-Pugh scoring system.

[0054] By combining renal function correction factors and liver function correction factors, a multi-organ function adjustment model based on the minimum rule is constructed, and the correction coefficient for the cumulative dose threshold is output.

[0055] The corrected cumulative dose threshold is obtained by multiplying the correction factor by the 24-hour cumulative dose threshold.

[0056] Furthermore, the pharmacokinetic interaction matrix is ​​constructed in the following ways:

[0057] Data on all narcotic and psychotropic drugs, as well as all other drugs that have known interactions with narcotic and psychotropic drugs, are collected and integrated into a drug category set, which includes the name, main ingredients, commonly used dosage range, route of administration, known interactions and mechanisms between different drugs, and the main pharmacokinetic parameters of each drug.

[0058] Based on the known interactions of drugs, drugs are classified into different categories according to their mechanisms. Based on the classified drugs, the interaction strength is assigned to each pair of interacting drugs.

[0059] Construct a two-dimensional matrix with one drug as the row and another drug as the column, where each cell in the matrix represents the interaction between a pair of drugs.

[0060] For each drug pair, the corresponding interaction information is filled into the corresponding cell in the matrix to obtain the completed pharmacokinetic interaction matrix.

[0061] Furthermore, the method for obtaining the encrypted evidence chain packet includes:

[0062] The operator's own identifier is used as the personal node, the operator's geographical location is used as the location node, the encrypted biometric hash is used as the time node, and the warning conflict intensity is used as the intensity node.

[0063] Define an edge connecting an individual node and a location node to represent the location of the device the operator is currently operating in during the current time period t; define an edge between an individual node and a time node to represent the biometric sample submitted by the operator during the current time period t; define an edge between an individual node and an intensity node to represent the risk assessment result of the current operation.

[0064] Construct a spatiotemporal-biological joint graph based on the defined nodes and edges;

[0065] The paths and connections in the spatiotemporal-biological joint map are analyzed using a predefined risk assessment model to calculate the real-time risk score for the current time period.

[0066] The comprehensive risk assessment value is obtained by weighting and summing the real-time risk score and the intensity of the early warning conflict.

[0067] Based on the comprehensive risk assessment value, an interception threshold range is set. If the comprehensive risk assessment value is less than the minimum value of the interception threshold range, the operator's current operation is determined to be risk-free and will not be intercepted.

[0068] If the comprehensive risk assessment value is greater than or equal to the minimum value of the interception threshold range and less than or equal to the maximum value of the interception threshold range, the operator's current operation is determined to be risky, and an interception instruction is generated.

[0069] If the comprehensive risk assessment value is greater than the maximum value of the interception threshold range, the operator's current operation is determined to be high-risk, an interception command is generated and physical isolation measures are implemented, and an alarm message is issued at the same time;

[0070] The operator's identifier, geographical location, encrypted biometric hash, and warning conflict intensity, along with the corresponding spatiotemporal-biological joint map, real-time risk score, comprehensive risk assessment value, and final interception decision, are integrated and encrypted using advanced encryption standards to form an encrypted evidence chain package.

[0071] Furthermore, the methods for generating the audit report and the drug flow heatmap include:

[0072] By combining operational risk scores, early warning conflict intensity, and encrypted evidence chains, a drug management trajectory profile is generated;

[0073] Based on knowledge graph technology, a visual audit graph is constructed according to the drug management trajectory archive. Based on the visual audit graph, audit reports and drug flow heat maps are generated.

[0074] Furthermore, an intelligent method for managing narcotic and psychotropic drugs includes:

[0075] S1: The operator performs verification based on the operation context parameters to obtain the operation risk score, and then constructs a knowledge-driven decision rule base to dynamically generate the operator's verification strategy. The operator's multimodal biometric identifier set is verified through multi-node collaborative verification to obtain the verification result and biometric hash.

[0076] S2: Based on the validation results, capture the medication data of patients in all departments, verify the drug dosage specifications and clinical pathway compliance through the white-box rule engine, identify high-risk combinations through the pharmacokinetic interaction matrix, and obtain the warning conflict intensity.

[0077] S3: Based on the operator's geographical location, combined with encrypted biometric hash, operational risk score, prescription data and warning conflict intensity, a spatiotemporal-biological joint map is constructed to obtain a real-time risk score. Combining the real-time risk score and warning conflict intensity, a comprehensive risk assessment value is generated. Based on the comprehensive risk assessment value, an interception command is generated to implement multi-level intervention and blocking measures for high-risk behaviors, and an encrypted evidence chain package is obtained.

[0078] S4: Based on operational risk scores, early warning conflict intensity, and encrypted evidence chain packages, generate drug management trajectory files, construct a visual audit map, and generate audit reports and drug flow heat maps;

[0079] S5: Based on the audit report and drug flow heat map, dynamically adjust the operational risk threshold range and update the knowledge-driven decision rule base and verification strategy configuration scheme.

[0080] The technical effects and advantages of the intelligent narcotic and psychotropic drug management system and method of the present invention are as follows:

[0081] This invention ensures high-precision operator authentication by processing operational context parameters and calculating operational risk scores, combined with a knowledge-driven decision rule base to dynamically generate personalized verification strategies. Next, a white-box rule engine is used to verify drug dosage standards and clinical pathway compliance at each level, and a pharmacokinetic interaction matrix is ​​used to identify potentially high-risk drug combinations, effectively preventing medication risks. Then, a spatiotemporal-biological joint map is constructed based on the operator's geographical location and other key information to assess risks in real time and implement multi-level interventional measures, achieving immediate response to high-risk behaviors. Subsequently, a detailed visual audit report is generated based on the operational risk score and encrypted evidence chain, enhancing the system's transparency and traceability. Finally, by dynamically adjusting the operational risk threshold range and optimizing the verification strategy configuration, the system's self-learning and optimization capabilities are continuously improved, reducing the need for manual intervention. Overall, this solution not only improves the safety and accuracy of medical operations but also significantly enhances the system's intelligence level and response efficiency. Attached Figure Description

[0082] Figure 1 This is a schematic diagram of a smart narcotic and psychotropic drug management system according to the present invention;

[0083] Figure 2 This is a schematic diagram of a unit flow of an intelligent narcotic and psychotropic drug management system according to the present invention;

[0084] Figure 3 This is a schematic diagram of an intelligent narcotic and psychotropic drug management method according to the present invention. Detailed Implementation

[0085] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0086] Example 1

[0087] Please see Figure 1 and Figure 2 As shown in this embodiment, an intelligent narcotic and psychotropic drug management system includes:

[0088] Operator authentication and dynamic generation of verification units: The operator performs verification based on the operation context parameters to obtain the operation risk score, and then constructs a knowledge-driven decision rule base to dynamically generate the operator's verification strategy. The operator's multimodal biometric identifier set is verified through multi-node collaborative verification to obtain the verification result and biometric hash.

[0089] Prescription data verification and early warning conflict intensity calculation unit: Based on the verification results, it captures the medication data of patients in all departments, verifies the drug dosage specifications and clinical pathway compliance level by level through the white box rule engine, identifies high-risk combinations through the pharmacokinetic interaction matrix, and obtains the early warning conflict intensity;

[0090] Real-time risk assessment and multi-level intervention blocking unit: Based on the operator's geographical location, combined with encrypted biometric hash, operation risk score, prescription data and early warning conflict intensity, a spatiotemporal-biological joint map is constructed to obtain a real-time risk score. Combining the real-time risk score and early warning conflict intensity, a comprehensive risk assessment value is generated. Based on the comprehensive risk assessment value, an interception instruction is generated to perform multi-level intervention blocking measures on high-risk behaviors and obtain an encrypted evidence chain package.

[0091] Visualized report generation unit: Based on operational risk scores, early warning conflict intensity, and encrypted evidence chain packages, it generates drug management trajectory files, constructs a visualized audit map, and generates audit reports and drug flow heat maps;

[0092] Dynamic adjustment and optimization unit: Based on audit reports and drug flow heat maps, dynamically adjust the operational risk threshold range and update the knowledge-driven decision rule base and verification strategy configuration scheme;

[0093] Operator authentication and dynamic generation verification unit → Prescription data verification and early warning conflict intensity calculation unit: Static verification will lead to low credibility of prescription review data, and incorrect verification results will cause subsequent misjudgments. Multi-node verification ensures that biometric hashes are true and reliable, providing reliable input for prescription verification.

[0094] Prescription data verification and early warning conflict intensity calculation unit → Real-time risk assessment and multi-level intervention blockade unit: Deviation in early warning conflict intensity calculation will affect risk assessment. Inaccurate early warning will lead to over / under-blockade. The accuracy of early warning is improved by pharmacokinetic tensor + organ correction (liver function + kidney function correction).

[0095] Real-time risk assessment and multi-level intervention blocking unit → Visualized report generation unit: The integrity of the encrypted evidence chain depends on the quality of real-time data. Evidence tampering will lead to audit failure. A tamper-proof evidence system is built by spatiotemporal-biological map + dual-chain evidence storage.

[0096] Visualized report generation unit → Dynamic adjustment and optimization unit: Manual analysis reports are difficult to drive system self-optimization, and strategy iteration lags behind risk changes. A closed loop of visualization heatmap → optimization algorithm → automatic parameter adjustment is used;

[0097] The construction methods for knowledge-driven decision-making rule bases include:

[0098] Operational context parameters include time granularity (the time when the operation occurred; since certain time periods (such as late at night) are outside working hours, operations during outside working hours can be considered higher risk; the division of different time periods can be set by relevant experts according to specific circumstances), pharmaceutics spectrum and corresponding drug control level (the type of drug being processed and its potential risk level; narcotic and psychotropic drugs are usually high-risk; by scanning the drug's RFID tag, the control level field is read, and the classification criteria are as follows: [Control level (Class I; Class II; Class III), risk label (high risk; medium risk; low risk), triggering conditions (morphine, fentanyl; tramadol, codeine; codeine-containing compound oral solutions)]), historical behavior reputation value (extracting the operator's historical operation records from the database to assess their behavioral patterns), geographical location (geographical information of the operation's occurrence), and equipment information (the type of equipment used and the equipment's safety status).

[0099] The weighted average method is used to calculate all influencing factors within the operational context parameters to obtain an operational risk score. An operational risk threshold range is set, and the operational risk score is divided into different operational risk levels, such as low risk, medium risk, and high risk.

[0100] Set a geofence and a maximum deviation distance (e.g., a radius of 50m; for example, if the coordinates of a hospital pharmacy are 31.2304N, 121.4737E, and the set geofence radius is 50m, then the geofence area is within a 50m radius of the hospital pharmacy). If the actual path distance between the geographic location and the geofence area is greater than the maximum deviation distance, it is marked as an abnormal location; if the actual path distance between the geographic location and the geofence area is less than or equal to the maximum deviation distance, it is marked as a normal location. The normal and abnormal locations are then used as the corresponding location statuses.

[0101] Different time granularities, location status, drug control levels, and operational risk levels are combined to form an operational scenario;

[0102] One or more verification strategies are randomly selected from the set of verification strategies, and a verification occurrence method is randomly selected from edge and cloud center verification. The selected verification strategies and verification occurrence methods are combined to form a verification strategy combination.

[0103] The verification strategy set includes vein recognition, fingerprint recognition, iris recognition and dynamic password. Edge and cloud center verification includes local node verification and cloud node verification. Local nodes are local servers and cloud nodes are remote data centers.

[0104] For each operational scenario, 'a' strategies are randomly selected from all strategy combinations as the initial verification strategy set for the operational scenario. For example, an operational scenario is [time granularity = 10:30 AM (within the daytime working hours); location status (location anomaly); drug control level (Class III); operational risk level (medium risk)]. Two strategies are randomly selected, and the generated verification strategy combinations are [multimodal biometric identifier set (fingerprint + iris), local node verification] and [multimodal biometric identifier set (fingerprint + iris + vein recognition + dynamic password), cloud node verification].

[0105] All operation scenarios and their corresponding initial verification strategy sets are integrated to construct an initial operation scenario-verification strategy mapping table.

[0106] Based on the operation scenario-verification strategy mapping table, a verification strategy combination is randomly assigned to each operation scenario to obtain an initial operation scenario-verification strategy combination. Then, all initial operation scenario-verification strategy combinations are integrated to obtain a complete set of verification strategy configuration schemes (that is, all operation scenarios are randomly matched with a verification strategy combination in the corresponding initial verification strategy set, and then all operation scenarios are integrated with the matched verification strategy to obtain a set of verification strategy configuration schemes). The assignment operation is repeated b times to obtain b different verification strategy configuration schemes, which then constitute the initial strategy configuration pool.

[0107] Based on the initial policy configuration pool, the effective value of each initial operation scenario-verification policy combination in each verification policy configuration scheme is calculated, and the average effective value of each verification policy configuration scheme is obtained by taking the average value.

[0108] An example formula for calculating effective values ​​is as follows: ;

[0109] in, Indicates a valid value. , and This represents the preset scaling factor, and , This represents the success rate of validation, used to indicate the reliability of a strategy combination. It equals the ratio of the number of successful verifications to the total number of verifications. This represents the safety factor, used to indicate the strength of multi-factor validation. ; ( () indicates a security optimization item. It directly reflects the reliability of the strategy combination. Nonlinear enhancements are applied to multi-factor validation (e.g., two-factor aq=2, three-factor aq=3), through... The scaling factor reflects the principle of prioritizing safety (recommended). ≥0.6);

[0110] yzcl represents the number of verification factors, which is the number of verification methods in a combination of verification strategies. For example, if a combination of verification strategies uses (fingerprint + iris + vein recognition + dynamic password), then the number of verification factors is 3. This represents the average verification time, used to indicate user experience and system real-time performance. It is the average time for a single verification (a weighted average of the time for a single verification between local nodes and cloud nodes). This represents the maximum tolerance time, used to indicate business scenario constraints, i.e., the longest verification time threshold allowed by the system (e.g., typically set to 5 seconds in medical settings); The symbol () indicates an efficiency penalty; the penalty intensifies when the verification time exceeds the tolerance threshold (e.g., a 50% timeout results in a deduction of 0.5 points). ), to drive the system to optimize response speed under the premise of safety (emergency scenarios need to be ≥ );

[0111] This represents resource consumption, used to indicate hardware and communication costs, specifically device CPU utilization + network bandwidth + storage I / O. This indicates the upper limit of resources, used to represent the limitations of infrastructure, i.e., the upper limit of system resource capacity; The symbol () indicates a resource penalty; the closer the resource usage is to the limit, the higher the penalty percentage (e.g., 0.9 × [amount missing] when CPU usage is 90%). To avoid overloading edge devices (required in primary care hospital scenarios). ≥0.1);

[0112] Besides manual setting, the optimal weights for formula weights can be determined through training with historical data (e.g., grid search + cross-validation). Recommended weights for emergency scenarios: ( =0.5, , Recommended scenarios for pharmacy settings: ( =0.7, , );

[0113] In addition, effective values ​​can be customized according to specific circumstances, such as true positive rate, recall rate, true negative rate, precision rate, response time, computational complexity, and resource utilization.

[0114] For each operation scenario, extract the verification strategy combination with the highest effective value from the c verification strategy configuration schemes with the highest average effective value to obtain a new operation scenario-verification strategy combination, and then integrate it into a new verification strategy configuration scheme.

[0115] Replace the verification strategy configuration scheme with the lowest average effective value in the initial strategy configuration pool with the new verification strategy configuration scheme. Repeat the calculation, extraction and replacement of effective values ​​until the maximum number of iterations is reached or the effective value of all operation scenarios-verification strategy combinations no longer increases, and obtain the best verification strategy combination for all operation scenarios.

[0116] The operation scenario-verification strategy mapping table is updated using the best combination of verification strategies, and a knowledge-driven decision rule base is generated based on the updated operation scenario-verification strategy mapping table using an expert rule engine.

[0117] Based on the knowledge-driven decision rule base, a combination of verification strategies is matched for the operator. Based on the matched combination of verification strategies, the operator's multimodal biometric identifier is obtained through the device currently operated by the operator and integrated into a set of biometric identifiers.

[0118] Among them, multimodal biometrics include vein biometrics, fingerprint biometrics, iris biometrics, and dynamic password biometrics;

[0119] The operator's biometric set is initially processed, including image preprocessing, biometric extraction, and multimodal biometric fusion;

[0120] Image preprocessing involves noise reduction and contrast enhancement of the operator's vein, fingerprint, and iris images during vein recognition, fingerprint recognition, and iris recognition processes.

[0121] Biometric identifier extraction involves using feature extraction techniques to extract vein features, fingerprint features, iris features, and dynamic password features from preprocessed vein images, fingerprint images, and iris images, as well as dynamic passwords of the one-time verification code type.

[0122] Multimodal biometric fusion involves converting the corresponding features into a unified vector space based on a verification strategy, and then performing multimodal fusion using a weighted fusion method.

[0123] If the matching verification strategy is local node verification, the pre-processed multimodal biometric identifier set will be sent to the local node.

[0124] If the matching verification strategy involves cloud node verification, the pre-processed multimodal biometric identifier set will be sent to the local node and d cloud nodes (such as municipal medical certification centers and provincial medical certification centers), where d is greater than 1.

[0125] For the device currently operated by the operator and each node participating in the verification, the received set of biometric identifiers is independently compared using its own database (such as Hamming distance calculation and cosine similarity) to obtain the corresponding verification results, including whether they match and the corresponding confidence score.

[0126] Based on all the verification results, calculate the difference between all the verification results. If the difference is less than the preset difference threshold, the verification result is determined to be correct and is taken as the final verification result. The corresponding verification result signal is generated. The types of verification result signals include verification success signal and verification failure signal.

[0127] If the difference value is greater than or equal to the preset difference value threshold, it is determined that there is abnormal behavior in the current verification, triggering the locking mechanism and generating a verification failure signal, while also providing feedback to the operator;

[0128] Based on the successful verification signal, the set of biometric identifiers is converted into biometric hash values, and the biometric hash values, the final verification result, and the operational risk score are encrypted.

[0129] It should be noted that existing systems often use fixed strategies for identity verification, which cannot dynamically adjust the verification strength according to real-time risks. This results in insufficient security protection in high-risk scenarios and low verification efficiency in low-risk scenarios. For example, traditional fingerprint verification is easily copied and misused when high-risk drugs are picked up at night, while dynamic passwords add unnecessary complexity to routine daytime operations.

[0130] Static verification strategies cannot perceive operational context, there is a contradiction between the accuracy and efficiency of multimodal biometric verification, and the lack of collaborative decision-making between local and cloud node verification leads to a high false positive rate.

[0131] Operator authentication and dynamic generation of verification units achieve a balance between security and efficiency by constructing a knowledge-driven decision rule base, dynamically generating verification strategy combinations based on operational risk scores, and through multi-node collaborative verification.

[0132] The methods for obtaining early warning conflict intensity include:

[0133] Based on the encrypted final verification result, obtain the patient's prescription data, including the drug name and dosage, and then obtain the corresponding prescription dosage;

[0134] By capturing and analyzing the prescription medication data of patients in all departments through the hospital's HIS system, the cumulative dosage of similar drugs within 24 hours can be calculated.

[0135] The white-box rule engine was used to retrieve the corresponding drug instructions in the hospital's HIS system and extract the maximum single dose and 24-hour cumulative dose thresholds.

[0136] The corrected cumulative dose threshold is obtained by adjusting the 24-hour cumulative dose threshold according to the patient's liver and kidney function status.

[0137] Based on the cumulative dose of similar drugs and the revised cumulative dose threshold, if the cumulative dose is less than or equal to the cumulative dose threshold, the drug dose is deemed to be in compliance with the standard.

[0138] If the cumulative dose exceeds the cumulative dose threshold, the drug dose is determined to be non-compliant, and a drug dose warning is generated.

[0139] We acquire patients’ disease history, allergy history, liver and kidney function indicators and treatment plans, and then construct a rule base for the association of drugs, diseases, test indicators and treatment plans. We then use the rule base to construct a clinical pathway knowledge graph.

[0140] The clinical pathway knowledge graph is used to perform subgraph matching on the current prescription data to calculate the clinical pathway compliance degree (i.e., by calculating the cosine similarity between the drugs in the prescription data and the clinical pathway, the cosine similarity value is used as the clinical pathway compliance degree).

[0141] If the clinical pathway compliance rate is greater than or equal to the preset clinical pathway compliance rate threshold, the clinical pathway compliance rate is determined to be in compliance with the standard.

[0142] If the clinical pathway compliance rate is less than the preset clinical pathway compliance rate threshold, it indicates a deviation from the standard pathway, and the clinical pathway compliance rate is determined to be non-compliant, generating a clinical pathway compliance rate warning.

[0143] Pharmacokinetic interaction matrix was used to detect high-risk interaction combinations in prescription drug use data across all departments.

[0144] If a high-risk interaction combination is detected, the prescription is determined to have a high-risk interaction combination, and a high-risk warning is generated; if no high-risk combination is detected, the prescription is determined not to have a high-risk interaction combination.

[0145] Based on the results of the determination of drug dosage specifications, clinical pathway compliance, and high-risk interaction combinations, the warning conflict intensity is obtained (one method is to record drug dosage compliance as 1 and non-compliance as 0, clinical pathway compliance as 1 and non-compliance as 0, the existence of high-risk interaction combinations as 0 and the absence of high-risk interaction combinations as 1, that is, drug dosage specification determination result = [0, 1], clinical pathway compliance determination result = [0, 1], and high-risk interaction combination determination result = [0, 1]. The three determination results are added together to obtain the warning conflict intensity. In one example, drug dosage specification is compliant, that is, the drug dosage specification determination result is 1; clinical pathway compliance is non-compliant, that is, the clinical pathway compliance determination result is 0; and there is no high-risk interaction combination, that is, the high-risk interaction combination determination result is 1. Therefore, the warning conflict intensity is 2).

[0146] It should be noted that traditional fingerprint verification has a high false recognition rate under fatigue operation. In one example, vein + iris verification can be dynamically enabled based on time granularity (night) and drug control level (Class I).

[0147] The revised cumulative dose threshold methods include:

[0148] The patient's liver and kidney function indicators were obtained through the hospital's HIS system, and the kidney function correction factor was calculated using the Cockcroft-Gault formula based on the kidney function indicators.

[0149] Liver function correction factors were assessed using the Child-Pugh scoring system.

[0150] By combining renal function correction factors and liver function correction factors, a multi-organ function adjustment model based on the minimum rule is constructed, and the correction coefficient for the cumulative dose threshold is output.

[0151] The multi-organ function adjustment model based on the minimum value rule is as follows: ;in, The correction factor representing the cumulative dose threshold. Indicates renal function correction factor, Indicates liver function correction factor;

[0152] The corrected cumulative dose threshold is obtained by multiplying the correction factor by the 24-hour cumulative dose threshold.

[0153] Methods for constructing pharmacokinetic interaction matrices include:

[0154] Data on all narcotic and psychotropic drugs, as well as all other drugs known to interact with them, were collected and integrated into a drug category set. This set included the name, main ingredient, usual dosage range, route of administration, known interactions between different drugs and their mechanisms (such as inhibition or induction of CYP450 isoenzymes), and key pharmacokinetic parameters, such as half-life, clearance, bioavailability, and volume of distribution. These parameters are crucial for assessing potential drug interactions.

[0155] Based on known drug interactions, drugs are classified into different categories according to their mechanisms, such as inhibiting or inducing CYP450 isoenzymes, competitively binding to plasma proteins, and altering gastrointestinal pH to affect absorption. Based on the classified drugs, each pair of interacting drugs is assigned an interaction strength (interaction strength is assigned based on expert experience and can be a grade, such as mild, moderate, or severe, or an intensity value, such as 0.3, 0.5, or 0.8).

[0156] Construct a two-dimensional matrix with one drug as the row and another drug as the column. If n drugs are considered, the matrix is ​​an n×n square matrix, and each cell in the matrix represents the interaction between a pair of drugs.

[0157] For each drug pair, fill in the corresponding interaction information in the corresponding cell of the matrix, including the interaction type (e.g., CYP450 inhibition), interaction strength (mild, moderate, severe), and any related pharmacokinetic parameter changes; for the diagonal cells of the matrix (i.e., the case of the same drug paired with itself), mark them as none, because the interaction between the same drug and itself is usually not considered; thus, the pharmacokinetic interaction matrix is ​​constructed.

[0158] It should be noted that traditional prescription review mostly relies on human experience, making it difficult to detect drug interactions across departments in real time. In particular, it is slow to identify hidden threats such as metabolic pathway conflicts and dose accumulation risks. For example, doctors may overlook dose correction for patients with liver or kidney dysfunction, or fail to detect the toxicity accumulation caused by competitive inhibition of CYP3A4 enzyme.

[0159] In other words, single-point dose verification does not take into account dynamic changes in organ function, clinical pathway matching is based solely on diagnostic names, lacks semantic association of quality protocols, and the drug interaction library is updated in a lagging manner, failing to cover the risks of novel combinations.

[0160] The prescription data verification and early warning conflict intensity calculation unit introduces a white-box rule engine, which dynamically adjusts the dose threshold by combining organ function correction factors (including liver function and kidney function), and scans the risk of multi-drug combination in real time through pharmacokinetic interaction tensor (three-dimensional structure), thus realizing the quantitative output of early warning conflict intensity.

[0161] The methods for obtaining the encrypted evidence chain package include:

[0162] The operator's own identifier is used as the personal node, the operator's geographical location is used as the location node, the encrypted biometric hash is used as the time node, and the warning conflict intensity is used as the intensity node.

[0163] Define an edge connecting an individual node and a location node to represent the location of the device the operator is currently operating in during the current time period t; define an edge between an individual node and a time node to represent the biometric sample submitted by the operator during the current time period t; define an edge between an individual node and an intensity node to represent the risk assessment result of the current operation.

[0164] Construct a spatiotemporal-biological joint graph based on the defined nodes and edges;

[0165] A predefined risk assessment model (which can be built and trained using a linear regression model or a machine learning model, and is used to take the spatiotemporal-biological joint map as input and output a real-time risk score) is used to analyze the paths and connections in the spatiotemporal-biological joint map and calculate the real-time risk score for the current time period.

[0166] The comprehensive risk assessment value is obtained by weighting and summing the real-time risk score and the intensity of the early warning conflict.

[0167] Based on the comprehensive risk assessment value, an interception threshold range is set. If the comprehensive risk assessment value is less than the minimum value of the interception threshold range, the operator's current operation is determined to be risk-free and will not be intercepted.

[0168] If the comprehensive risk assessment value is greater than or equal to the minimum value of the interception threshold range and less than or equal to the maximum value of the interception threshold range, the operator's current operation is determined to be risky, and an interception instruction is generated (the interception instruction can refuse the operator further access, i.e., refuse the operator's current drug application).

[0169] If the comprehensive risk assessment value is greater than the maximum value of the interception threshold range, the operator's current operation is determined to be high-risk. An interception command is generated and physical isolation measures are implemented (physical isolation measures can be to forcibly lock the narcotic and psychotropic drug management cabinet for a certain period of time, such as 30 seconds or 30 minutes, during which time it cannot be opened by non-destructive means. Operation can only continue when the locking time ends). At the same time, an alarm message is issued.

[0170] The operator's identifier, geographical location, encrypted biometric hash, and warning conflict intensity, along with the corresponding spatiotemporal-biological joint map, real-time risk score, comprehensive risk assessment value, and final interception decision, are integrated and encrypted using advanced encryption standards to form an encrypted evidence chain package.

[0171] It should be noted that most existing risk interception methods rely on single-dimensional data (such as the number of prescriptions) and cannot identify spatiotemporal abnormal patterns (such as high-frequency cross-regional drug collection) and biometric fraud, resulting in a high false negative rate for risks such as "illegal operation with legitimate identity". In existing technologies, geographic location data and biometric verification results are often analyzed in isolation, the risk scoring model is static and cannot adapt to the evolution of abuse methods, the blocking measures are singular (such as only pop-up prompts) and lack a graded response mechanism.

[0172] The real-time risk assessment and multi-level intervention blocking unit constructs a spatiotemporal-biological joint map, integrates and analyzes operator trajectories, biometric hashes, and prescription data, and triggers graded blocking (such as monitoring / alarms / physical isolation) by constructing a comprehensive risk assessment value, and generates an encrypted evidence chain package to support judicial tracing;

[0173] The methods for generating audit reports and drug flow heatmaps include:

[0174] By combining operational risk scores, warning conflict intensity, and encrypted evidence chain packages, a drug management trajectory profile is generated. This profile includes the operator's basic information, the time sequence of the operation, geographical location, biometric verification results, warning conflict intensity, and intervention measures taken.

[0175] Using cross-chain evidence storage protocols (such as blockchain technology or other cross-chain evidence storage protocols), the generated drug management trajectory archives are stored on a distributed ledger to ensure their immutability and transparency, and a multi-party verification mechanism (such as a multi-node consensus algorithm) is adopted to enhance the credibility of the archives.

[0176] Based on knowledge graph technology, a visual audit graph is constructed according to the drug management trajectory archive. Based on the visual audit graph, audit reports and drug flow heat maps are generated. The visual audit graph displays key information in the drug management trajectory archive, including the operator's activity path, risk assessment results, warning conflict intensity, and drug flow information.

[0177] Optimization algorithms (such as genetic algorithms and ant colony algorithms) are used to dynamically update the operational risk threshold range based on audit reports. The knowledge-driven decision rule base and verification strategy configuration scheme are then updated according to the updated operational risk threshold range, reducing manual intervention and enhancing the system's self-learning and optimization capabilities.

[0178] Deploy updated rule base and verification strategy configuration schemes, continuously monitor system performance, and further adjust parameters based on the latest data to ensure continuous system optimization.

[0179] Example 2

[0180] Please see Figure 3 As shown, the parts not described in detail in this embodiment are described in Embodiment 1. A smart method for managing narcotic and psychotropic drugs is provided, including:

[0181] S1: The operator verifies based on the operation context parameters to obtain an operation risk score, and then constructs a knowledge-driven decision rule base to dynamically generate the operator's verification strategy. The operator's multimodal biometric identifier set is verified through multi-node collaborative verification to obtain the verification result and biometric hash.

[0182] S2: Based on the validation results, capture the medication data of patients in all departments, verify the drug dosage specifications and clinical pathway compliance through the white-box rule engine, identify high-risk combinations through the pharmacokinetic interaction matrix, and obtain the warning conflict intensity.

[0183] S3: Based on the operator's geographical location, combined with encrypted biometric hash, operational risk score, prescription data and warning conflict intensity, a spatiotemporal-biological joint map is constructed to obtain a real-time risk score. Combining the real-time risk score and warning conflict intensity, a comprehensive risk assessment value is generated. Based on the comprehensive risk assessment value, an interception command is generated to implement multi-level intervention and blocking measures for high-risk behaviors, and an encrypted evidence chain package is obtained.

[0184] S4: Based on operational risk scores, early warning conflict intensity, and encrypted evidence chain packages, generate drug management trajectory files, construct a visual audit map, and generate audit reports and drug flow heat maps;

[0185] S5: Based on the audit report and drug flow heat map, dynamically adjust the operational risk threshold range and update the knowledge-driven decision rule base and verification strategy configuration scheme.

[0186] Example 3

[0187] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the operation mode of the intelligent narcotic and psychotropic drug management method described above.

[0188] Since the electronic device described in this embodiment is used to implement the intelligent narcotic and psychotropic drug management method described in this application embodiment, those skilled in the art can understand the specific implementation methods and various variations of the electronic device in this embodiment based on the intelligent narcotic and psychotropic drug management method described in this application embodiment. Therefore, how the electronic device implements the method in this application embodiment will not be described in detail here. Any electronic device used by those skilled in the art to implement the intelligent narcotic and psychotropic drug management method described in this application embodiment falls within the scope of protection of this application.

[0189] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0190] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An intelligent management system for narcotic and psychotropic drugs, characterized in that, include: Operator authentication and dynamic generation of verification units: The operator performs verification based on the operation context parameters to obtain the operation risk score, and then constructs a knowledge-driven decision rule base to dynamically generate the operator's verification strategy. The operator's multimodal biometric identifier set is verified through multi-node collaborative verification to obtain the verification result and biometric hash. The construction methods of the knowledge-driven decision rule base include: Operational context parameters include time granularity, pharmaceutics lineage and corresponding drug control level, historical behavior reputation value, geographic location and device information; The weighted average method is used to calculate all influencing factors within the operational context parameters to obtain an operational risk score. An operational risk threshold range is then set, and the operational risk score is divided into different operational risk levels. Set a geofence and a maximum deviation distance. If the actual path distance between the geographic location and the geofence range is greater than the maximum deviation distance, mark it as a location anomaly. If the actual path distance between the geographic location and the geofence is less than or equal to the maximum deviation distance, it is marked as a normal location; normal location and abnormal location are used as the location status of the geographic location. Different time granularities, location status, drug control levels, and operational risk levels are combined to form an operational scenario; One or more verification strategies are randomly selected from the set of verification strategies, and a verification occurrence method is randomly selected from edge and cloud center verification. The selected verification strategies and verification occurrence methods are combined to form a verification strategy combination. The verification strategy set includes vein recognition, fingerprint recognition, iris recognition and dynamic password. Edge and cloud center verification includes local node verification and cloud node verification. Local nodes are local servers and cloud nodes are remote data centers. For each operation scenario, randomly select 'a' strategies from all strategy combinations as the initial verification strategy set for the operation scenario; All operation scenarios and their corresponding initial verification strategy sets are integrated to construct an initial operation scenario-verification strategy mapping table. The initial operation scenario-verification strategy mapping table is updated, and an expert rule engine is used to generate a knowledge-driven decision rule base based on the updated operation scenario-verification strategy mapping table. The methods for updating the initial operation scenario-verification strategy mapping table include: Based on the initial operation scenario-verification strategy mapping table, a verification strategy combination is randomly assigned to each operation scenario to obtain the initial operation scenario-verification strategy combination. Then, all initial operation scenario-verification strategy combinations are integrated to obtain a complete set of verification strategy configuration schemes. The assignment operation is repeated b times to obtain b different verification strategy configuration schemes, which in turn form the initial strategy configuration pool. Based on the initial policy configuration pool, the effective value of each initial operation scenario-verification policy combination in each verification policy configuration scheme is calculated, and the average effective value of each verification policy configuration scheme is obtained by taking the average value. For each operation scenario, extract the verification strategy combination with the highest effective value from the c verification strategy configuration schemes with the highest average effective value to obtain a new operation scenario-verification strategy combination, and then integrate it into a new verification strategy configuration scheme. Replace the verification strategy configuration scheme with the lowest average effective value in the initial strategy configuration pool with the new verification strategy configuration scheme. Repeat the calculation, extraction and replacement of effective values ​​until the maximum number of iterations is reached or the effective value of all operation scenarios-verification strategy combinations no longer increases, and obtain the best verification strategy combination for all operation scenarios. The operation scenario-verification strategy mapping table is updated using the best combination of verification strategies to obtain the updated operation scenario-verification strategy mapping table. Prescription data verification and early warning conflict intensity calculation unit: Based on the verification results, it captures the medication data of patients in all departments, verifies the drug dosage specifications and clinical pathway compliance level by level through the white box rule engine, identifies high-risk combinations through the pharmacokinetic interaction matrix, and obtains the early warning conflict intensity; Real-time risk assessment and multi-level intervention blocking unit: Based on the operator's geographical location, combined with encrypted biometric hash, operation risk score, prescription data and early warning conflict intensity, a spatiotemporal-biological joint map is constructed to obtain a real-time risk score. Combining the real-time risk score and early warning conflict intensity, a comprehensive risk assessment value is generated. Based on the comprehensive risk assessment value, an interception instruction is generated to perform multi-level intervention blocking measures on high-risk behaviors and obtain an encrypted evidence chain package. Visualized report generation unit: Based on operational risk scores, warning conflict intensity, and encrypted evidence chain packages, it generates drug management trajectory files, constructs a visualized audit map, and generates audit reports and drug flow heat maps; Dynamic adjustment and optimization unit: Based on audit reports and drug flow heat maps, dynamically adjust the operational risk threshold range and update the knowledge-driven decision rule base and verification strategy configuration scheme.

2. The intelligent narcotic and psychotropic drug management system according to claim 1, characterized in that, The verification results and biometric hashes are obtained in the following ways: Based on the knowledge-driven decision rule base, a combination of verification strategies is matched for the operator. Based on the matched combination of verification strategies, the operator's multimodal biometric identifier is obtained through the device currently operated by the operator and integrated into a set of biometric identifiers. Among them, multimodal biometrics include vein biometrics, fingerprint biometrics, iris biometrics, and dynamic password biometrics; The operator's biometric set is initially processed, including image preprocessing, biometric extraction, and multimodal biometric fusion; If the matching verification strategy is local node verification, the pre-processed multimodal biometric identifier set will be sent to the local node. If the matching verification strategy involves cloud node verification, the pre-processed multimodal biometric identifier set will be sent to the local node and d cloud nodes, where d is greater than 1. For the device currently operated by the operator and each node participating in the verification, the received set of biometric identifiers is independently compared using its own database to obtain the corresponding verification results, including whether they match and the corresponding confidence score. Based on all the verification results, calculate the difference between all the verification results. If the difference is less than the preset difference threshold, the verification result is determined to be correct and is taken as the final verification result. The corresponding verification result signal is generated. The types of verification result signals include verification success signal and verification failure signal. If the difference value is greater than or equal to the preset difference value threshold, it is determined that there is abnormal behavior in the current verification, triggering the locking mechanism and generating a verification failure signal, while also providing feedback to the operator; Based on the successful verification signal, the set of biometric identifiers is converted into biometric hash values ​​and encrypted to obtain the encrypted biometric hash.

3. The intelligent narcotic and psychotropic drug management system according to claim 2, characterized in that, The methods for issuing early warnings of conflict intensity include: Based on the encrypted final verification result, the patient's prescription data is obtained, and then the corresponding prescription dosage is obtained; By capturing and analyzing the prescription medication data of patients in all departments through the hospital's HIS system, the cumulative dosage of similar drugs within 24 hours can be calculated. The white-box rule engine was used to retrieve the corresponding drug instructions in the hospital's HIS system and extract the maximum single dose and 24-hour cumulative dose thresholds. The corrected cumulative dose threshold is obtained by adjusting the 24-hour cumulative dose threshold according to the patient's liver and kidney function status. Based on the cumulative dose of similar drugs and the revised cumulative dose threshold, if the cumulative dose is less than or equal to the cumulative dose threshold, the drug dose is deemed to be in compliance with the standard. If the cumulative dose exceeds the cumulative dose threshold, the drug dose is determined to be non-compliant, and a drug dose warning is generated. We acquire patients' disease history, allergy history, liver and kidney function indicators, and treatment plans, and then construct a rule base for the association of drugs, diseases, test indicators, and treatment plans. We then use this rule base to construct a clinical pathway knowledge graph. The clinical pathway knowledge graph is used to perform subgraph matching on the current prescription data to calculate the clinical pathway compliance degree. If the clinical pathway compliance rate is greater than or equal to the preset clinical pathway compliance rate threshold, the clinical pathway compliance rate is determined to be in compliance with the standard. If the clinical pathway compliance rate is less than the preset clinical pathway compliance rate threshold, it indicates a deviation from the standard pathway, and the clinical pathway compliance rate is determined to be non-compliant, generating a clinical pathway compliance rate warning. Pharmacokinetic interaction matrix was used to detect high-risk interaction combinations in prescription drug use data across all departments. If a high-risk interaction combination is detected, the prescription is determined to have a high-risk interaction combination, and a high-risk warning is generated; if no high-risk combination is detected, the prescription is determined not to have a high-risk interaction combination. The intensity of the warning conflict is determined based on the results of the determination of drug dosage specifications, clinical pathway compliance, and high-risk interaction combinations.

4. The intelligent narcotic and psychotropic drug management system according to claim 3, characterized in that, The corrected cumulative dose threshold is obtained in the following ways: The patient's liver and kidney function indicators were obtained through the hospital's HIS system, and the kidney function correction factor was calculated using the Cockcroft-Gault formula based on the kidney function indicators. Liver function correction factors were assessed using the Child-Pugh scoring system. By combining renal function correction factors and liver function correction factors, a multi-organ function adjustment model based on the minimum rule is constructed, and the correction coefficient for the cumulative dose threshold is output. The corrected cumulative dose threshold is obtained by multiplying the correction factor by the 24-hour cumulative dose threshold.

5. The intelligent narcotic and psychotropic drug management system according to claim 3, characterized in that, The pharmacokinetic interaction matrix is ​​constructed in the following ways: Data on all narcotic and psychotropic drugs, as well as all other drugs that have known interactions with narcotic and psychotropic drugs, are collected and integrated into a drug category set, which includes the name, main ingredients, commonly used dosage range, route of administration, known interactions and mechanisms between different drugs, and the main pharmacokinetic parameters of each drug. Based on the known interactions of drugs, drugs are classified into different categories according to their mechanisms. Based on the classified drugs, the interaction strength is assigned to each pair of interacting drugs. Construct a two-dimensional matrix with one drug as the row and another drug as the column, where each cell in the matrix represents the interaction between a pair of drugs. For each drug pair, the corresponding interaction information is filled into the corresponding cell in the matrix to obtain the completed pharmacokinetic interaction matrix.

6. The intelligent narcotic and psychotropic drug management system according to claim 5, characterized in that, The methods for obtaining the encrypted evidence chain packet include: The operator's own identifier is used as the personal node, the operator's geographical location is used as the location node, the encrypted biometric hash is used as the time node, and the warning conflict intensity is used as the intensity node. Define an edge connecting an individual node and a location node to represent the location of the device the operator is currently operating in during the current time period t; define an edge between an individual node and a time node to represent the biometric sample submitted by the operator during the current time period t; define an edge between an individual node and an intensity node to represent the risk assessment result of the current operation. Construct a spatiotemporal-biological joint graph based on the defined nodes and edges; The paths and connections in the spatiotemporal-biological joint map are analyzed using a predefined risk assessment model to calculate the real-time risk score for the current time period. The comprehensive risk assessment value is obtained by weighting and summing the real-time risk score and the intensity of the early warning conflict. Based on the comprehensive risk assessment value, an interception threshold range is set. If the comprehensive risk assessment value is less than the minimum value of the interception threshold range, the operator's current operation is determined to be risk-free and will not be intercepted. If the comprehensive risk assessment value is greater than or equal to the minimum value of the interception threshold range and less than or equal to the maximum value of the interception threshold range, the operator's current operation is determined to be risky, and an interception instruction is generated. If the comprehensive risk assessment value is greater than the maximum value of the interception threshold range, the operator's current operation is determined to be high-risk, an interception command is generated and physical isolation measures are implemented, and an alarm message is issued at the same time; The operator's identifier, geographical location, encrypted biometric hash, and warning conflict intensity, along with the corresponding spatiotemporal-biological joint map, real-time risk score, comprehensive risk assessment value, and final interception decision, are integrated and encrypted using advanced encryption standards to form an encrypted evidence chain package.

7. The intelligent narcotic and psychotropic drug management system according to claim 6, characterized in that, The methods for generating the audit report and drug flow heatmap include: By combining operational risk scores, early warning conflict intensity, and encrypted evidence chains, a drug management trajectory profile is generated; Based on knowledge graph technology, a visual audit graph is constructed according to the drug management trajectory archive. Based on the visual audit graph, audit reports and drug flow heat maps are generated.

8. A method for managing narcotic and psychotropic drugs, applied to the intelligent narcotic and psychotropic drug management system described in any one of claims 1 to 7, characterized in that, include: S1: The operator verifies based on the operation context parameters to obtain an operation risk score, and then constructs a knowledge-driven decision rule base to dynamically generate the operator's verification strategy. The operator's multimodal biometric identifier set is verified through multi-node collaborative verification to obtain the verification result and biometric hash. S2: Based on the validation results, capture the medication data of patients in all departments, verify the drug dosage specifications and clinical pathway compliance through the white-box rule engine, identify high-risk combinations through the pharmacokinetic interaction matrix, and obtain the warning conflict intensity. S3: Based on the operator's geographical location, combined with encrypted biometric hash, operational risk score, prescription data and warning conflict intensity, a spatiotemporal-biological joint map is constructed to obtain a real-time risk score. Combining the real-time risk score and warning conflict intensity, a comprehensive risk assessment value is generated. Based on the comprehensive risk assessment value, an interception command is generated to implement multi-level intervention and blocking measures for high-risk behaviors, and an encrypted evidence chain package is obtained. S4: Based on operational risk scores, early warning conflict intensity, and encrypted evidence chain packages, generate drug management trajectory files, construct a visual audit map, and generate audit reports and drug flow heat maps; S5: Based on the audit report and drug flow heat map, dynamically adjust the operational risk threshold range and update the knowledge-driven decision rule base and verification strategy configuration scheme.

Images