Drug safety management system based on multi-data fusion

Abstract

The invention relates to the technical field of medication safety management, in particular to a medication safety management system based on multi-data fusion. The method comprises the following steps: generating an initial prediction emergency degree sequence; primary auditing; determining an interception level; performing secondary auditing; determining an actual auditing completion sequence; and comparing the initial prediction emergency degree sequence with the actual auditing completion sequence, calculating the real-time matching degree, and correcting the abnormal type threshold value or the interception judgment parameter based on the real-time matching degree. According to the method, the received prescriptions are subjected to initial emergency degree prediction and sorting, the prescriptions of critical patients are ensured to preferentially enter an auditing process, the prescriptions are accurately distributed to the most suitable professional auditing personnel through a load balancing strategy by comparing the predicted emergency sequence of the prescriptions with the actual completion sequence and calculating the matching degree, and the auditing efficiency is improved. And targeted correction is automatically triggered, so that the dynamic intelligence of the prescription auditing process is realized, the auditing efficiency and accuracy are improved, and the medication safety of the patient is effectively guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medication safety management, and in particular to a medication safety management system based on multi-data fusion. BACKGROUND

[0002] Currently, medical institutions generally face problems such as data island, rigid audit process, and delayed risk response in medication safety management. Existing systems rely on single data sources (such as prescription information) for rule matching, lack of multi-dimensional data fusion analysis of patient history, disease state, and drug interactions, resulting in incomplete risk identification, high false interception rate, and delayed response for critical patients.

[0003] Chinese Patent Publication No. CN118866390A discloses a comprehensive detection method and system for prescription tampering behavior, including: obtaining medical insurance diagnosis and treatment data and extracting feature information, then calculating the number of prescriptions and drug dosage changes based on the feature information; descriptive features and trend features are obtained by statistical analysis of the number of prescriptions and drug dosage changes, and preliminary anomaly detection is performed based on these features; the descriptive features, trend features obtained by preliminary anomaly detection, and introduced behavior features constitute a data set, and the isolated forest algorithm is used for secondary anomaly detection on the data set to determine and mark the initial anomaly data points; a deep learning model is used to reconstruct the data based on the marked initial anomaly data points, and the final anomaly data points are determined by comparing the reconstruction error with the anomaly threshold in the final anomaly detection, and the prescription tampering behavior is determined based on the final anomaly data points. As can be seen, the existing prescription anomaly detection technology generally has the problems of delayed response to critical prescriptions caused by the static processing mode, reasonable medication misblocked caused by rigid interception rules, and low accuracy and efficiency of the system decision level due to the lack of self-optimization ability. SUMMARY

[0004] Therefore, the present application provides a medication safety management system based on multi-data fusion to overcome the problem of low audit efficiency caused by the lack of dynamic priority scheduling for prescription audit in the prior art.

[0005] To achieve the above-mentioned purpose, the present application provides a medication safety management system based on multi-data fusion, comprising: receiving a plurality of to-be-detected prescriptions issued by different departments of each medical institution in the target audit area; determining the initial predicted urgency of the to-be-detected prescriptions according to the emergency treatment demand intensity, and generating an initial predicted urgency order; sending the to-be-detected prescriptions to the preliminary detection end for one-time audit according to the initial predicted urgency order, and obtaining a one-time audit result; According to the results of the first review, the interception level of each prescription to be detected is determined, and whether to perform the interception operation is determined according to the interception proportion of the current batch of prescriptions; According to the interception level, the prescription to be reviewed is determined, and the prescription to be reviewed is dispatched to the corresponding target re-examination side detection end according to the professional category for secondary review to obtain the secondary review result; According to the interception proportion, whether to optimize the sorting and scheduling distribution strategy of the target re-examination side detection end is determined; According to the first review result, the interception result and the secondary review result, the actual review completion order of the prescription to be detected is determined; Compare the initial predicted urgency order with the actual review completion order, calculate the real-time matching degree, and modify the abnormal type threshold value or interception judgment parameter based on the real-time matching degree.

[0006] Further, according to the emergency treatment demand intensity, the initial predicted urgency order of the prescription to be detected is determined, which includes: Construct an emergency treatment demand intensity calculation model; Obtain the feature data of each prescription to be detected; Calculate the emergency treatment demand intensity value of each prescription to be detected by using the emergency treatment demand intensity calculation model; According to the order from high to low of the emergency treatment demand intensity value, the prescriptions to be detected are sorted to generate the initial predicted urgency order.

[0007] Further, the process of the first review includes: Match and compare the prescription to be detected with the drug prescription database: If the prescription to be detected and the drug in the patient's medication history data have the same matching item, determine whether there is a risk of repeated medication according to the drug storage amount, if there is, mark the prescription to be detected as the first type of abnormality; If there is a matching item between the prescription to be detected and the contraindication information in the drug prescription database, it is determined that there is a contraindication risk, and it is marked as the second type of abnormality; If there is a known interaction risk between multiple drugs in the prescription to be detected Determine the abnormality category according to the interaction risk level: If there is a contraindication interaction in the prescription to be detected, it is determined to be high risk, and it is marked as the fourth type of abnormality; If there is a serious interaction in the prescription to be detected, it is determined to be medium risk, and it is marked as the third type of abnormality; If there is a mild interaction in the prescription to be detected, the abnormality category of the prescription to be detected is determined according to the number of drug combinations; The abnormality category of the prescription to be tested is determined based on the number of drug combinations, including: Compare the number of drug combinations with the combination number threshold: If the number of drug combinations is less than or equal to the combination number threshold, it is determined that there is no abnormality and is marked as passed; If the number of drug combinations exceeds the combination number threshold, the prescription combination is deemed complex and marked as a fifth type of abnormality. The audit results include Category I anomalies, Category II anomalies, Category III anomalies, Category IV anomalies, and Category V anomalies.

[0008] Furthermore, the interception level for each prescription to be tested is determined based on the results of a single review, including: The interception levels are divided into immediate interception, pending review, and approval. The prescriptions corresponding to the fourth type of abnormality are marked for immediate interception; The prescriptions corresponding to Category I, Category II, Category III, and Category V abnormalities are marked as prescriptions to be reviewed. The remaining prescriptions to be tested are marked as passed.

[0009] Furthermore, determining whether to perform an interception operation based on the interception rate of the current batch of prescriptions includes: The number of prescriptions marked as needing immediate interception after one review is counted to obtain the expected number of interceptions. The proportion of the expected number of interceptions to the total number of reviewed prescriptions is then calculated to obtain the interception percentage. The interception percentage is compared with a preset interception threshold; If the interception rate is greater than the preset interception threshold, prescriptions marked as needing immediate interception are marked as prescriptions awaiting review, and the sorting and scheduling allocation strategy is optimized; if the interception rate is less than or equal to the preset interception threshold, the prescriptions marked as needing immediate interception are intercepted by the system.

[0010] Furthermore, the process of dispatching prescriptions awaiting review to the corresponding target review testing terminals according to their professional categories includes: Extract the professional domain labels and real-time workload data from the testing terminals of each reviewing party; For any prescription awaiting review, select all online prescription testing platforms that match the corresponding professional category; From the matched review party detection terminals, select the one with the smallest real-time workload as the target review party detection terminal for secondary review.

[0011] Furthermore, the process of optimizing the sorting and scheduling allocation strategy is as follows: An analysis of the real-time processing capabilities of the review panel for any given professional category: If the real-time processing capacity is less than or equal to the first preset processing capacity, some prescriptions to be reviewed will be allocated to the target review end of the adjacent professional category according to the diversion ratio. If the real-time processing capacity is between the first preset processing capacity and the second preset processing capacity, the prescriptions to be reviewed will be scheduled to the corresponding target review prescription testing terminals according to their professional categories, in descending order of the emergency processing demand intensity value. If the real-time processing capacity is greater than or equal to the second preset processing capacity, the receiving range of the target review party detection terminal corresponding to the professional category is expanded according to the expansion ratio to receive prescriptions to be reviewed from adjacent professional categories.

[0012] Furthermore, based on the results of the first review, the interception, and the second review, the actual review completion order of the prescription to be tested is determined as follows: For each prescription to be tested, a timestamp of its key processing events is recorded, the timestamp including at least the initial review completion timetamp and the re-review completion timetamp; Based on the results of the initial review and the interception, the final disposal path of the prescription is determined; Based on the aforementioned final processing path, each prescription is assigned an actual review completion timestamp: If the final processing path of the prescription is direct interception or direct approval after the initial review, then the initial review completion timestamp is defined as the actual review completion timestamp; If the final disposal path of the prescription requires a second review and completion, then the timestamp of the second review completion is defined as the actual review completion timestamp; After the final status of all prescriptions is determined, all prescriptions are sorted in order of actual review completion timestamp from earliest to latest to generate the actual review completion order.

[0013] Furthermore, the process of calculating the real-time matching degree includes: Calculate the correlation coefficient between the initial predicted urgency sequence number set and the actual review completion sequence number set, as the real-time matching degree.

[0014] Furthermore, adjusting the anomaly type threshold or interception judgment parameter based on the real-time matching degree includes: Compare the real-time matching accuracy with the preset matching threshold: If the real-time matching degree is less than or equal to the preset matching threshold, the matching degree is determined to meet the standard. If the real-time matching degree is greater than the preset matching threshold, the matching degree is determined to be inconsistent with the standard. A matching queue is generated based on the comparison, and the number of batches in the matching queue determines whether to trigger root cause analysis. The root cause analysis process involves adjusting the preset interception threshold or the abnormal type threshold of a single review process based on the proportion determined.

[0015] Compared with the prior art, the beneficial effects of the present invention are that by constructing a quantitative model that integrates multiple factors such as the severity of the disease, drug characteristics, patient waiting status and drug inventory, the received prescriptions are initially predicted and sorted in terms of urgency, ensuring that prescriptions for critically ill patients are given priority in the review process. Subsequently, a tiered review is performed: the initial review is based on a rule base to conduct rapid risk screening, such as duplicate medication and contraindications, and a dynamic decision-making mechanism based on the batch interception ratio is introduced. When the expected interception ratio exceeds a preset threshold, high-risk prescriptions are automatically converted to re-review instead of being directly intercepted, effectively preventing excessive obstruction of reasonable medical processes. The review process intelligently schedules prescriptions based on their professional category and the real-time workload of reviewers, improving professional review efficiency. By comparing the predicted urgency order of prescriptions with the actual completion order, it calculates the matching degree and uses a load balancing strategy to accurately allocate prescriptions to the most suitable professional reviewers, improving review efficiency and automatically triggering targeted corrections, such as adjusting the interception threshold. This achieves dynamic and intelligent prescription review, which not only improves review efficiency and accuracy and effectively ensures patient medication safety, but also continuously adapts and learns through feedback loops, enabling the system's decision-making capabilities to continuously evolve, ultimately improving the optimal allocation of medical resources and the quality of medical care.

[0016] Furthermore, through multi-source data fusion and intelligent scheduling, early identification, graded processing, and resource optimization of medication safety risks can be achieved, thereby improving review efficiency and accuracy, reducing false interception rates, ensuring patient medication safety, and possessing continuous self-optimization capabilities. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the medication safety management system based on multi-data fusion, as described in an embodiment of the present invention. Figure 2 A flowchart illustrating the process of determining the initial predicted urgency order of the prescriptions to be tested in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the process of dispatching prescriptions to be reviewed to the corresponding target review agent testing terminal according to professional categories in an embodiment of the present invention; Figure 4 This is a flowchart illustrating the process of determining the actual review completion order of the prescription to be tested in an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0019] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0020] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0021] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0022] Please see Figure 1 The diagram shown is a flowchart of a medication safety management system based on multi-data fusion according to an embodiment of the present invention. The present invention provides a medication safety management system based on multi-data fusion, comprising: Step S1: Receive multi-source medication data from various medical institutions within the target area, including several prescriptions to be tested issued by different departments, patient medication history, disease diagnosis, and drug knowledge base; Step S2: Based on multi-source data fusion, determine the initial predicted urgency of the prescription to be tested according to the intensity of urgent treatment needs, and generate the initial predicted urgency order. Step S3: According to the initial predicted urgency order, send the prescription to be tested to the initial reviewer's testing terminal for a review and obtain a review result; Step S4: Determine the interception level of each prescription to be tested based on the results of one review, and determine whether to perform an interception operation based on the interception percentage of the current batch of prescriptions. Step S5: Determine the prescription to be reviewed based on the interception level, and dispatch the prescription to be reviewed to the corresponding target review party detection terminal according to professional category for secondary review in order to obtain the secondary review result; Among them, the interception ratio is used to determine whether to optimize the sorting and scheduling allocation strategy for the target review party detection end; Step S6: Determine the actual review completion order of the prescription to be tested based on the first review result, the interception result, and the second review result; The initial predicted urgency order is compared with the actual review completion order to calculate the real-time matching degree, and the anomaly type threshold or interception judgment parameter is corrected based on the real-time matching degree.

[0023] By constructing a quantitative model that integrates multiple factors such as disease urgency, drug characteristics, patient waiting status, and drug inventory, the system predicts and prioritizes the urgency of received prescriptions. This ensures that prescriptions for critically ill patients are given priority in the review process. Subsequently, a tiered review process is implemented: the initial review stage performs rapid risk screening based on a rule base, such as for duplicate medications and contraindications, and introduces a dynamic decision-making mechanism based on batch interception ratios. When the expected interception ratio exceeds a preset threshold, high-risk prescriptions are automatically converted to secondary review instead of being directly intercepted, effectively preventing excessive disruption to the rational medical process. The secondary review stage intelligently schedules prescriptions based on their professional category and the real-time workload of reviewers, improving the efficiency of professional review. By comparing the predicted urgency order of prescriptions with the actual completion order, the matching degree is calculated, and a load balancing strategy accurately allocates prescriptions to the most suitable professional reviewers, improving review efficiency and automatically triggering targeted corrections, such as adjusting the interception threshold. This achieves dynamic and intelligent prescription review, which not only improves review efficiency and accuracy and effectively ensures patient medication safety, but also continuously adapts and learns through feedback loops, enabling the system's decision-making capabilities to continuously evolve, ultimately improving the optimal allocation of medical resources and the quality of medical care.

[0024] See Figure 2 As shown, it is a flowchart illustrating the process of determining the initial predicted urgency order of the prescription to be tested according to an embodiment of the present invention; Specifically, in step S2, determining the initial predicted urgency order of the prescriptions to be tested based on the intensity of urgent treatment needs includes: Step S2001: Construct an emergency response demand intensity calculation model; Step S2002: Obtain the feature data of each prescription to be tested; Step S2003: Calculate the emergency treatment demand intensity value of each prescription to be tested using the emergency treatment demand intensity calculation model; Step S2004: Sort the prescriptions to be tested according to the order of the emergency treatment demand intensity value from high to low to generate the initial predicted urgency order.

[0025] In this embodiment, the emergency treatment demand intensity calculation model is: Emergency Treatment Demand Intensity Value ; Where S: Disease Criticality Index, weight ; D: Drug characteristic coefficient, weighting ; W: Patient waiting status, weighted ; M: Coefficient of remaining drug quantity in the prescription, weighting ; The weights satisfy: ; The Disease Criticality Index (S) is scored as extremely high risk (10 points): emergency room, ICU, cardiopulmonary resuscitation, life-threatening condition requiring immediate medication; High risk (8 points): Emergency observation room, stroke center, chest pain center, critical care emergency, medication within the time window; Intermediate risk (6 points): Tumor radiotherapy and chemotherapy, postoperative analgesia, anti-infection treatment, and continuity of treatment are important, but not immediately life-threatening; Low risk (3 points): General outpatient services, chronic disease management, health check-ups, and routine treatments may be appropriately delayed; The scoring criteria for the drug characteristic coefficient D is 10 points for emergency use: adrenaline, atropine, nifedipine, essential for emergency treatment, with a response time of minutes; Time-window sensitive category (8 points): Thrombolytic drugs, antidotes, first-dose antibiotics, and drugs with a clear time window for use; Special storage category (7 points): Insulin, cold chain biological agents, require special conditions and should not be stored for a long time; High-risk control category (6 points): Anesthetic drugs, psychotropic drugs, drugs requiring special management but not for emergency treatment; Common medicines (2 points): Vitamins, calcium tablets, common chronic disease medications, not time-sensitive; The patient waiting status W score is bedside waiting (10 points): the patient is waiting for medication at the hospital bedside; Pharmacy waiting area (8 points): Patients wait at the pharmacy window; In-hospital off-site waiting (5 points): The patient is in the hospital but not at the pharmacy; Waiting for discharge (2 points): The prescription from the Internet hospital indicates that the patient has already left the hospital; The scoring standard for the drug supply coefficient M is as follows: drug discontinuation (10 points): 0 days of drug supply remaining, risk of immediate drug discontinuation; Less than one day's supply (8 points): Medication will be discontinued within the day; 1-2 day supply (6 points): Medication will be discontinued soon; 3-7 day supply (4 points): There is a buffer period; More than 7 days (1 point): Sufficient dosage.

[0026] The priority level mapping relationship is as follows: highest priority (E≥7.0): requires immediate review, such as emergency rescue and critically ill patients; high priority (5.0≤E<7.0): requires rapid review, such as tumor chemotherapy and drug discontinuation risk. Medium priority (3.0 ≤ E < 5.0): Can be routinely reviewed, such as for ordinary inpatients; Low priority (E < 3.0): Can be processed in batches, such as prescription renewals for chronic diseases in internet hospitals; For example, for an emergency patient, the prescription characteristics are: Disease severity index: acute myocardial infarction (S=10); Drug characteristics: nitroglycerin, emergency (D=8); Patient waiting status: bedside waiting (W=10); Prescription inventory: first prescription (M=1); The calculation process is as follows: The system generates an initial predicted urgency order based on priority levels, so that prescriptions for critically ill patients can be processed first.

[0027] Specifically, an audit process includes: The prescription to be tested is matched and compared with the drug prescription database: If the prescription to be tested has the same matching item as the drug in the patient's medication history data, the risk of duplicate medication is determined based on the amount of drug in stock. If so, the prescription to be tested is marked as a first-type abnormality. If the prescription to be tested matches the contraindication information in the drug prescription database, it is determined that there is a risk of contraindication and it is marked as a second type of abnormality. If there are known interactions between the various medications in the prescription to be tested. Anomaly category is determined based on interaction risk level: If there is a contraindication to the prescription being tested, it is considered high-risk and marked as a Category IV abnormality; If there are serious interactions in the prescription to be tested, it is classified as medium risk and marked as a Category III abnormality. If there is a slight interaction in the prescription to be tested, the abnormality category of the prescription to be tested is determined according to the number of drug combinations; The abnormality category of the prescription to be tested is determined based on the number of drug combinations, including: Compare the number of drug combinations with the combination number threshold: If the number of drug combinations is less than or equal to the combination number threshold, it is determined that there is no abnormality and is marked as passed; If the number of drug combinations exceeds the combination number threshold, the prescription combination is deemed complex and marked as a fifth type of abnormality. The audit results include Category I anomalies, Category II anomalies, Category III anomalies, Category IV anomalies, and Category V anomalies.

[0028] The drug prescription database in this embodiment contains a drug prescription database, with data sourced from authoritative pharmaceutical information sources, including but not limited to: drug instructions approved by the National Medical Products Administration (NMPA); recognized authoritative clinical pharmacy databases, such as Micromedex, Lexicomp, or the Martindale Drug Reference; domestic and international clinical practice guidelines and expert consensus; and abnormal categories determined during a single review process based on interaction risk levels, including: Extracting drug combinations: Analyzing the prescription to be tested and extracting the standardized codes (such as YPID) of all drugs; Knowledge base matching: The extracted drug code combinations are queried and matched with the "Drug Interaction Knowledge Base"; Risk level mapping: Based on predefined risk levels in the knowledge base, matched interactions are automatically mapped to "taboo," "serious," or "mild." If any "contraindication" level interaction exists in the prescription, then regardless of other interactions, the entire prescription is judged as having "contraindication interaction" and marked as a Type IV abnormality. If there is no "contraindication", but there is one or more "serious" level interactions, the whole thing is judged as "serious interaction exists" and marked as a Type III anomaly; If only a "mild" level interaction exists, the system will not directly make a final anomaly determination, but will trigger the subsequent logic of "determining the anomaly category based on the number of drug combinations". This is because a mild interaction is not sufficient to constitute a reason for interception or review, but the risk will increase when it is combined with complex multi-drug combinations.

[0029] Because each drug-drug interaction record in the knowledge base has a predefined interaction mechanism, clinical consequences, risk level, and treatment recommendations, the drug combination in the prescription to be tested is matched with the knowledge base, and the risk level is automatically determined according to the following predefined, quantified standards: The criteria for contraindication is that the combined use of two or more drugs will produce serious life-threatening or permanent damage consequences, and there is a lack of effective monitoring or intervention methods; such combined use is absolutely prohibited in clinical practice. For example, if the prescription contains both "terfenadine" and "clarithromycin," according to the knowledge base, both are metabolized by the liver enzyme CYP3A4, and combined use will cause a sharp increase in terfenadine blood concentration, triggering fatal torsades de pointes ventricular tachycardia, therefore it is determined to be "contraindicated"; the criteria for serious interaction is that the combined use of drugs may produce serious clinical consequences, but through... Close monitoring, dose adjustment, or the use of alternative medications can manage the risks. Such interactions require focused assessment and intervention by physicians and pharmacists. For example, if a prescription contains both warfarin and levofloxacin, and the knowledge base indicates that levofloxacin may enhance the anticoagulant effect of warfarin and significantly increase the risk of bleeding, the system will classify it as a "serious interaction," suggesting close monitoring of the patient's prothrombin time (PT / INR) and consideration of adjusting the warfarin dose. Mild interactions are defined as those where the combined use of drugs produces minor clinical consequences, usually manifested as a slight enhancement or weakening of drug efficacy, or a slight increase in the incidence of side effects. These generally do not affect continued treatment and do not require special monitoring or intervention. For example, if a prescription contains both omeprazole and clopidogrel, and the knowledge base indicates that omeprazole may slightly weaken the antiplatelet activity of clopidogrel by inhibiting CYP2C19. For most patients, this interaction is not clinically significant and is therefore classified as "mild interaction". The risk level of interaction between multiple drugs in the prescription to be tested is determined by querying and analyzing the drug interaction knowledge base in the drug prescription database.

[0030] Specifically, the interception level for each prescription to be tested is determined based on the results of a single review, including: The interception levels are divided into immediate interception, pending review, and approval. The prescriptions corresponding to the fourth type of abnormality are marked for immediate interception; The prescriptions corresponding to Category I, Category II, Category III, and Category V abnormalities are marked as prescriptions to be reviewed. The remaining prescriptions to be tested are marked as passed.

[0031] Specifically, determining whether to perform an interception operation based on the interception rate of the current batch of prescriptions includes: The number of prescriptions marked as needing immediate interception after one review is counted to obtain the expected number of interceptions. The proportion of the expected number of interceptions to the total number of reviewed prescriptions is then calculated to obtain the interception percentage. The interception percentage is compared with a preset interception threshold; If the interception rate is greater than the preset interception threshold, prescriptions marked as needing immediate interception will be marked as prescriptions awaiting review, and the sorting and scheduling allocation strategy will be optimized. If the interception rate is less than or equal to the preset interception threshold, then the prescription execution system marked as needing immediate interception will be intercepted.

[0032] In this embodiment, the preset interception threshold is a set ratio parameter used to determine whether to perform actual interception operation on prescriptions marked as immediately intercepted. When the interception ratio is greater than the preset interception threshold, there may be a risk of over-interception. In this case, prescriptions marked as immediately intercepted are changed to prescriptions awaiting review, and the optimized sorting and scheduling allocation strategy is triggered so that all suspicious prescriptions enter the secondary review process. At the same time, the global optimal allocation of review resources is achieved, realizing intelligent management.

[0033] Specifically, the process of optimizing sorting and scheduling allocation strategies is as follows: An analysis of the real-time processing capabilities of the review panel for any given professional category: If the real-time processing capacity is less than or equal to the first preset processing capacity, some prescriptions to be reviewed will be allocated to the target review end of the adjacent professional category according to the diversion ratio. If the real-time processing capacity is between the first preset processing capacity and the second preset processing capacity, the prescriptions to be reviewed will be scheduled to the corresponding target review prescription testing terminals according to their professional categories, in descending order of the emergency processing demand intensity value. If the real-time processing capacity is greater than or equal to the second preset processing capacity, the receiving range of the target review party detection terminal corresponding to the professional category is expanded according to the expansion ratio to receive prescriptions to be reviewed from adjacent professional categories.

[0034] In this embodiment, real-time processing capacity is determined based on pre-allocated intensity, which is used to measure the processing capacity and load balance of professional categories. The first preset processing capacity is set to 0.3, and the second preset processing capacity is set to 0.7. When the real-time processing capacity is less than or equal to the first preset processing capacity, it indicates a high load rate and strained processing capacity, representing low processing capacity. When the real-time processing capacity is between the first and second preset processing capacities, it indicates a moderate load and stable processing capacity. When the real-time processing capacity is greater than or equal to the second preset processing capacity, it indicates a low load rate and sufficient processing capacity. Therefore, by expanding the receiving range, the processing pressure on other inspection terminals is reduced, fully utilizing the processing capacity of this inspection terminal and improving review efficiency. (The text then mentions a traffic splitting ratio.) The diversion factor is 2.0, and the expansion ratio is... The expansion factor is 1.5; for example, when C=0.15, the split ratio is... This means that 30% of prescriptions awaiting review will be diverted to similar specialties; when C=0.85, the expansion ratio... This means that the system can receive 22.5% more prescriptions from neighboring specialties, effectively balancing the system load and improving overall processing efficiency.

[0035] Pre-allocated intensity in this embodiment ; in, ; Current load rate The current number of prescriptions pending processing is the number of prescriptions waiting for review in the queue. The maximum processing capacity is the maximum number of prescriptions that can be processed within a specified time, which is related to the number of pharmacists and their working hours. Processing efficiency coefficient = actual processing rate / baseline processing rate; the actual processing rate is the rate at which prescriptions are completed in the current monitoring cycle, and the baseline processing rate is the rate at which prescriptions are completed in the previous monitoring cycle. Professional matching degree = percentage of professionally relevant prescriptions × average professional proficiency; the percentage of professionally relevant prescriptions is the ratio of the number of professionally matched prescriptions to the total number of prescriptions processed, and the average professional proficiency is... ; For example, for the cardiovascular pharmacy specialty: if the current number of prescriptions to be processed is 45, the maximum processing capacity is 60, the actual processing rate is 12 prescriptions / hour, the baseline processing rate is 10 prescriptions / hour, the proportion of specialty-related prescriptions is 0.85, the average specialty proficiency is 0.9, and the weighting coefficient is... , The current load rate Processing efficiency coefficient Professional matching Pre-allocated intensity Then, in order of decreasing emergency processing demand intensity value, the prescriptions to be reviewed are sequentially dispatched to the corresponding target review prescription testing terminal according to professional category.

[0036] See Figure 3 As shown, it is a schematic diagram of the process of scheduling prescriptions to be reviewed to the corresponding target review party testing terminal according to professional categories in an embodiment of the present invention; Specifically, in step S5, the prescriptions to be reviewed are dispatched to the corresponding target review prescription testing terminal according to their professional categories, including: Step S5001: Extract the professional domain labels and real-time workload data of each review party's testing end; Step S5002: For any prescription to be reviewed, select all online prescription testing terminals that match the corresponding professional category; Step S5003: Select the one with the smallest real-time workload from the matched review party detection terminals as the target review party detection terminal for secondary review.

[0037] Specifically, determining whether there is a risk of duplicate medication based on the amount of medication in stock includes: Retrieve the patient's most recent valid medication record for the medication, including the date of prescription, total amount prescribed, and dosage. Calculate the patient's theoretical medication reserves based on medication records and the current date; Compare theoretical drug inventory with safety stock threshold: If the theoretical drug inventory is less than or equal to zero, it is determined that there is no risk of duplicate medication. If the theoretical drug inventory is greater than or equal to the safety stock threshold, it is determined to be a high-risk duplicate medication, and the prescription to be tested is marked as a fourth type of abnormality. If the theoretical drug inventory is between zero and the safety stock threshold, it is determined that there is a risk of duplicate medication. Theoretical drug inventory = Total prescription volume - (Current date - Prescription date) × Usage and dosage instructions; The safety stock threshold is the product of the number of days that medication can be prescribed ahead of schedule and the daily consumption.

[0038] See Figure 4 As shown, it is a flowchart illustrating the process of determining the actual review completion order of the prescription to be tested according to an embodiment of the present invention; Specifically, in step S6, determining the actual review completion order of the prescription to be tested based on the first review result, the interception result, and the second review result includes: Step S6001: Record the timestamps of key processing events for each prescription to be tested, wherein the timestamps include at least the initial review completion timestamp and the re-review completion timestamp; Step S6002: Determine the final disposal path of the prescription based on the first review result and the interception result; Step S6003: Based on the final processing path, assign an actual review completion timestamp to each prescription: If the final processing path of the prescription is direct interception or direct approval after the initial review, then the initial review completion timestamp is defined as the actual review completion timestamp; If the final disposal path of the prescription requires a second review and completion, then the timestamp of the second review completion is defined as the actual review completion timestamp; Step S6004: After the final status of all prescriptions is determined, sort all prescriptions according to the actual review completion timestamps from earliest to latest to generate the actual review completion order.

[0039] Specifically, the process of calculating the real-time matching degree includes: Calculate the correlation coefficient between the initial predicted urgency sequence number set and the actual review completion sequence number set, as the real-time matching degree.

[0040] In this embodiment, Spearman's rank correlation coefficient is used to calculate the real-time matching degree. The calculation formula is as follows: ; in, It is the first The difference in rank between two prescriptions This refers to the total number of prescriptions in this batch; real-time matching accuracy. The value range of is [-1, 1]; The closer to 1, the higher the match; the closer to 1, the better. The weaker the correlation, the lower the correlation.

[0041] The preset matching threshold is set to 0.7. If the value is lower than this, the matching degree is considered unsatisfactory. The preset consecutive batch is set to 3 or 5 to avoid overcorrection caused by abnormal data in a single batch. When the number of batches is greater than or equal to the preset consecutive batch, the root cause analysis process is triggered.

[0042] Specifically, adjusting the anomaly type threshold or interception judgment parameter based on the real-time matching degree includes: Compare the real-time matching accuracy with the preset matching threshold: If the real-time matching degree is less than or equal to the preset matching threshold, the matching degree is determined to meet the standard. If the real-time matching degree is greater than the preset matching threshold, the matching degree is determined to be inconsistent with the standard. A matching queue is generated based on the comparison, and the number of batches in the matching queue determines whether to trigger root cause analysis. The root cause analysis process involves adjusting the preset interception threshold or the anomaly type threshold determined by the proportion of cases. Calculate the pass rate, which is the ratio of the number of prescriptions revised after review to the total number of prescriptions reviewed; If the approval rate exceeds the preset approval threshold, the exception type threshold for one review process will be adjusted according to the exception type. If the percentage of passage is less than or equal to the preset passage threshold, the preset blocking threshold is lowered until it is lowered to the corrected blocking threshold. The corrected interception threshold is the product of the preset interception threshold and the adjustment step size, where the adjustment step size is the difference between 1 and the adjustment coefficient; the adjustment coefficient... .

[0043] In this embodiment, the preset pass threshold is set to 70%. The process of adjusting the exception type threshold for a single review process based on the exception type is as follows: For the first type of exception, the safety stock threshold is increased to the corrected safety stock threshold. The safety stock threshold is 5 days' worth of medication, and the corrected safety stock threshold is 7 days' worth of medication. For the fifth type of exception, the combination number threshold is increased to the corrected number threshold. The combination number threshold is 3 types of drugs, and the corrected number threshold is 4 types of drugs. For the second type of exception, the contraindication matching sensitivity is reduced. For the third type of exception, the judgment criteria are adjusted, that is, the drug interaction of the current prescription is marked as having theoretical risk but clinically acceptable, and recorded in the drug prescription database so that when the same prescription is identified again, it is judged as having a mild interaction. For example, if the physician prescribes "proton pump inhibitors (such as oxaliplatin)", the drug interaction will be marked as having theoretical risk but clinically acceptable, and recorded in the drug prescription database. This will ensure that when the same prescription is identified again, it is judged as having a mild interaction. The prescription for "Meprazole" is used to prevent gastric damage caused by glucocorticoids, but the diagnosis field only shows "systemic lupus erythematosus." The system will block it because "the diagnosis does not match the drug's indication." However, it does not understand that this drug is used to prevent the side effects of another treatment drug. Therefore, the judgment criteria are adjusted to reduce misjudgments. For the second type of abnormality, the contraindications of the currently prescribed drugs are marked as theoretically risky but clinically common to reduce the sensitivity of contraindication matching and reduce misjudgments. This avoids setting some non-absolute contraindications as mandatory blocking rules. For example, a certain drug should be used with caution in patients with heart failure, but after evaluation by a physician, the patient's condition is stable and the benefits outweigh the risks. In this case, the system's blocking is a misjudgment. This avoids excessive blocking, ensures rational drug use, and reduces unnecessary review processes to improve review efficiency.

[0044] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A medication safety management system based on multi-data fusion, characterized in that, include: Receive several prescriptions to be tested from different departments of various medical institutions within the target review area; The initial predicted urgency of the prescription to be tested is determined based on the intensity of the urgent treatment demand, and an initial predicted urgency order is generated. According to the initial predicted urgency order, the prescription to be tested is sent to the initial reviewer's testing terminal for a first review, and a review result is obtained; The interception level of each prescription to be tested is determined based on the results of a single review, and whether to perform an interception operation is determined based on the interception rate of the current batch of prescriptions. Based on the interception level, the prescription to be reviewed is determined, and the prescription to be reviewed is dispatched to the corresponding target review party testing terminal for secondary review according to professional category to obtain the secondary review result; Based on the interception ratio, determine whether to implement an optimized sorting and scheduling allocation strategy for the target review party detection end; Based on the results of the first review, the interception, and the second review, the actual review completion order of the prescription to be tested is determined; The initial predicted urgency order is compared with the actual review completion order to calculate the real-time matching degree, and the anomaly type threshold or interception judgment parameter is corrected based on the real-time matching degree.

2. The medication safety management system based on multi-data fusion according to claim 1, characterized in that, The initial predicted urgency order of the prescriptions to be tested is determined based on the intensity of urgent treatment needs, including: Construct a model for calculating the intensity of emergency response needs; Acquire the characteristic data of each prescription to be tested; The emergency treatment demand intensity value of each prescription to be tested is calculated using the aforementioned emergency treatment demand intensity calculation model; The prescriptions to be tested are sorted in descending order of their emergency treatment demand intensity values ​​to generate the initial predicted urgency order.

3. The medication safety management system based on multi-data fusion according to claim 1, characterized in that, An audit process includes: The prescription to be tested is matched and compared with the drug prescription database: If the prescription to be tested has the same matching item as the drug in the patient's medication history data, the risk of duplicate medication is determined based on the amount of drug in stock. If so, the prescription to be tested is marked as a first-type abnormality. If the prescription to be tested matches the contraindication information in the drug prescription database, it is determined that there is a risk of contraindication and it is marked as a second type of abnormality. If there are known interactions between multiple drugs in the prescription to be tested, the abnormality category shall be determined according to the interaction risk level: If there is a contraindication to the prescription being tested, it is considered high-risk and marked as a Category IV abnormality; If there are serious interactions in the prescription to be tested, it is classified as medium risk and marked as a Category III abnormality. If there is a slight interaction in the prescription to be tested, the abnormality category of the prescription to be tested is determined according to the number of drug combinations; The abnormality category of the prescription to be tested is determined based on the number of drug combinations, including: Compare the number of drug combinations with the combination number threshold: If the number of drug combinations is less than or equal to the combination number threshold, it is determined that there is no abnormality and is marked as passed; If the number of drug combinations exceeds the combination number threshold, the prescription combination is deemed complex and marked as a fifth type of abnormality. The audit results include Category I anomalies, Category II anomalies, Category III anomalies, Category IV anomalies, and Category V anomalies.

4. The medication safety management system based on multi-data fusion according to claim 1, characterized in that, The interception level for each prescription to be tested is determined based on the results of a single review, including: The interception levels are divided into immediate interception, pending review, and approval. The prescriptions corresponding to the fourth type of abnormality are marked for immediate interception; The prescriptions corresponding to Category I, Category II, Category III, and Category V abnormalities are marked as prescriptions to be reviewed. The remaining prescriptions to be tested are marked as passed.

5. The medication safety management system based on multi-data fusion according to claim 1, characterized in that, Determining whether to perform an interception operation based on the current batch prescription interception rate includes: The number of prescriptions marked as needing immediate interception after one review is counted to obtain the expected number of interceptions. The proportion of the expected number of interceptions to the total number of reviewed prescriptions is then calculated to obtain the interception percentage. The interception percentage is compared with a preset interception threshold; If the interception rate is greater than the preset interception threshold, prescriptions marked as needing immediate interception are marked as prescriptions awaiting review, and the sorting and scheduling allocation strategy is optimized; if the interception rate is less than or equal to the preset interception threshold, the prescriptions marked as needing immediate interception are intercepted by the system.

6. The medication safety management system based on multi-data fusion according to claim 1, characterized in that, The process of assigning prescriptions awaiting review to the corresponding target review testing terminal according to their professional categories includes: Extract the professional domain labels and real-time workload data from the testing terminals of each reviewing party; For any prescription awaiting review, select all online prescription testing platforms that match the corresponding professional category; From the matched review party detection terminals, select the one with the smallest real-time workload as the target review party detection terminal for secondary review.

7. The medication safety management system based on multi-data fusion according to claim 5, characterized in that, The process of optimizing sorting and scheduling allocation strategies is as follows: An analysis of the real-time processing capabilities of the review panel for any given professional category: If the real-time processing capacity is less than or equal to the first preset processing capacity, some prescriptions to be reviewed will be allocated to the target review end of the adjacent professional category according to the diversion ratio. If the real-time processing capacity is between the first preset processing capacity and the second preset processing capacity, the prescriptions to be reviewed will be scheduled to the corresponding target review prescription testing terminals according to their professional categories, in descending order of the emergency processing demand intensity value. If the real-time processing capacity is greater than or equal to the second preset processing capacity, the receiving range of the target review party detection terminal corresponding to the professional category is expanded according to the expansion ratio to receive prescriptions to be reviewed from adjacent professional categories.

8. The medication safety management system based on multi-data fusion according to claim 1, characterized in that, Based on the results of the first review, the interception, and the second review, the actual review completion order of the prescription to be tested is determined as follows: For each prescription to be tested, a timestamp of its key processing events is recorded, the timestamp including at least the initial review completion timetamp and the re-review completion timetamp; Based on the results of the initial review and the interception, the final disposal path of the prescription is determined; Based on the aforementioned final processing path, each prescription is assigned an actual review completion timestamp: If the final processing path of the prescription is direct interception or direct approval after the initial review, then the initial review completion timestamp is defined as the actual review completion timestamp; If the final disposal path of the prescription requires a second review and completion, then the timestamp of the second review completion is defined as the actual review completion timestamp; After the final status of all prescriptions is determined, all prescriptions are sorted in order of actual review completion timestamp from earliest to latest to generate the actual review completion order.

9. The medication safety management system based on multi-data fusion according to claim 1, characterized in that, The process of calculating the real-time matching degree includes: Calculate the correlation coefficient between the initial predicted urgency sequence number set and the actual review completion sequence number set, as the real-time matching degree.

10. The medication safety management system based on multi-data fusion according to claim 1, characterized in that, The correction of the anomaly type threshold or interception judgment parameter based on the real-time matching degree includes: Compare the real-time matching accuracy with the preset matching threshold: If the real-time matching degree is less than or equal to the preset matching threshold, the matching degree is determined to meet the standard. If the real-time matching degree is greater than the preset matching threshold, the matching degree is determined to be inconsistent with the standard. A matching queue is generated based on the comparison, and the number of batches in the matching queue determines whether to trigger root cause analysis. The root cause analysis process involves adjusting the preset interception threshold or the abnormal type threshold of a single review process based on the proportion determined.

Images