A method and system for prescription review sampling

Abstract

The present application relates to medical management information technology field, disclose a kind of prescription review sampling method and system, including constructing prescription risk score model, output the comprehensive risk score of full amount of prescription of day;Full amount of prescription is sorted in descending order, generate the to-be-reviewed task list of day;The contribution rate of each department prescription quantity in to-be-reviewed task list is calculated, and corresponding contribution threshold is set;Multiple repetition, generate at least three consecutive days of to-be-reviewed task list, if the contribution rate of the prescription quantity of a department is higher than its corresponding contribution threshold for two consecutive days, then the prescription of the department in the to-be-reviewed task list of the two days is moved to the must-review prescription list, and prescription review is executed;Record structured result data;Calculate strategy effect feedback score, update risk score model when strategy effect feedback score is lower than expected threshold.The present application concentrates review resources on high-risk prescriptions through risk score model, improves the utilization rate of review resources and the discovery rate of problems.

Description

Technical Field

[0001] This invention relates to the field of medical management information technology, and in particular to a prescription review sampling method and system. Background Technology

[0002] Prescription review is a core component of hospital pharmacy management, used to identify irrational drug use and improve the quality control loop. However, current techniques generally employ a "fixed ratio + manual sampling" method, which has the following drawbacks: 1. Lack of strategic sampling: Random sampling based on proportions fails to prioritize high-risk scenarios, resulting in wasted review resources and a low problem detection rate; 2. Delayed strategy configuration: Specialized review tasks are set manually on a regular basis, which cannot quickly respond to sudden risks; 3. Data loop failure: The review results were not fed back to the sampling strategy, and the system lacked self-learning ability. Summary of the Invention

[0003] To solve the above problems, the technical solution adopted by the present invention is as follows: A prescription review sampling method includes the following steps: S1. Construct a prescription risk scoring model, taking multi-factor data from the drug dimension, patient dimension, medical order dimension, doctor dimension and treatment dimension of the day as input, and outputting a comprehensive risk score of all prescriptions of the day; S2. Based on the comprehensive risk score, sort all prescriptions in descending order, and combine the preset total number of review samples to perform high-risk mandatory sampling and medium-risk priority sampling to generate a list of review tasks for the day. S3. Classify the prescriptions in the task list to be reviewed for the day by department, calculate the contribution rate of the number of prescriptions in each department in the task list to be reviewed, and set a corresponding contribution threshold for the contribution rate of the number of prescriptions in each department. S4. Repeat steps S1-S3 twice to generate a list of tasks to be reviewed for at least three consecutive days, and monitor the contribution rate of prescriptions from each department in real time. If the contribution rate of prescriptions from a certain department is higher than its corresponding contribution threshold for two consecutive days, move the prescriptions from that department in the list of tasks to be reviewed for those two days to the list of prescriptions to be reviewed, and send the list of prescriptions to be reviewed to the pharmacist for prescription review. S5. After performing prescription review, record structured result data, which shall include at least the problem judgment result, problem type, severity level, intervention behavior, and influencing factors; S6. Calculate the strategy effect feedback score based on the structured result data. When the strategy effect feedback score is lower than the expected threshold, update the risk scoring model.

[0004] Furthermore, the prescription risk scoring model is constructed using linear weighting:

[0005] In the formula, For the first The comprehensive risk score of the prescription. For the first The values ​​of each risk factor in this prescription These are factor weights, derived from historical reviews or expert ratings, and support online incremental updates.

[0006] Furthermore, the high-risk mandatory sampling means that all prescriptions with a comprehensive risk score higher than the first threshold are included in the task list to be reviewed; the medium-risk priority sampling means that when the number of prescriptions in the task list to be reviewed is less than the preset total number of review samples, prescriptions with scores lower than the first threshold are added to the task list to be reviewed in descending order of scores until the number of prescriptions in the task list to be reviewed is equal to the preset total number of review samples.

[0007] Furthermore, in steps S3 and S4, the contribution threshold for each department is the percentage of the total number of prescriptions from the corresponding department in the total number of prescriptions.

[0008] Furthermore, the strategy effectiveness feedback scoring is achieved through:

[0009] The calculation yields F, which represents the strategy effectiveness feedback score, P, and Q, respectively. P is the problem detection rate, Q is the specific hit rate, and α and β are the strategy evaluation weights, with α + β = 1. When the strategy effectiveness feedback score is less than the set expected value, the first threshold for high-risk mandatory sampling is adjusted, or the factor weights are updated. At least one of them.

[0010] Furthermore, the problem detection rate is the ratio of problematic prescriptions to the list of prescriptions that must be reviewed, and the specific hit rate is the ratio of the type of problem that occurred to all problem types.

[0011] A prescription review sampling system includes a risk scoring module for collecting multi-dimensional data on drugs, patients, medical orders, doctors, and diagnosis and treatment, and outputting a comprehensive risk score for the prescription. The sampling task generation module, connected to the risk scoring module, is used to perform high-risk mandatory sampling and medium-risk priority sampling based on the comprehensive risk score, and generate a list of tasks to be reviewed. The classification module is used to classify prescriptions in the task list to be reviewed by department, calculate the contribution rate of each department in the task list to be reviewed, and set a corresponding contribution threshold for the contribution rate of each department's prescriptions. The review and judgment module is used to monitor the contribution rate of prescriptions in each department in real time. If the contribution rate of prescriptions in a certain department is higher than its corresponding contribution threshold for two consecutive days, the prescriptions of that department in the list of tasks to be reviewed for those two days will be moved to the list of prescriptions to be reviewed, and the list of prescriptions to be reviewed will be sent to the pharmacist's end to perform prescription review. The review execution module is used to perform prescription reviews and record structured result data. The structured result data includes at least the problem judgment result, problem type, severity level, intervention behavior, and influencing factors. The closed-loop optimization module is used to calculate the strategy effect feedback score based on the structured result data, and update the risk scoring model when the strategy effect feedback score is lower than the expected threshold.

[0012] The beneficial effects of this invention are: 1. By using a risk scoring model, review resources can be concentrated on high-risk prescriptions, thereby improving the utilization rate of review resources and the problem detection rate.

[0013] 2. By checking whether the contribution rate exceeds the contribution threshold, review resources can be concentrated on departments with consistently high rates of problems. In addition, since departmental problems fluctuate over time, such as a surge in respiratory prescriptions during flu season, monitoring for two consecutive days can filter out temporary fluctuations and occasional single-day events, allowing review resources to be accurately directed to problem departments. This avoids wasting review resources in low-risk departments and insufficient reviews in high-risk departments due to fixed allocation of review resources. Furthermore, by checking whether the contribution rate exceeds the contribution threshold for two consecutive days to implement special review tasks, it is possible to respond quickly to sudden risks.

[0014] 3. Use strategy effectiveness feedback scores to determine whether to update the risk scoring model, thereby optimizing the model by adjusting the weights of relevant factors. The entire set of prescriptions is re-ranked, and the factor weights are dynamically adjusted in real time based on the severity level, intervention behavior, and influencing factors in the structured outcome data. This allows for a more accurate and realistic comprehensive risk score for all prescriptions, thereby improving the problem detection rate. Attached Figure Description

[0015] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of the invention.

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1This is a schematic diagram of the steps of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0019] Unless otherwise defined, the technical or scientific terms used in this disclosure shall have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms "first," "second," and similar terms used in this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships, and these relative positional relationships may change accordingly when the absolute position of the described object changes.

[0020] like Figure 1 As shown, a prescription review sampling method includes the following steps: S1. Construct a prescription risk scoring model, taking multi-factor data from the drug dimension, patient dimension, medical order dimension, doctor dimension and treatment dimension of the day as input, and outputting a comprehensive risk score for all prescriptions of the day.

[0021] Specifically, the prescription risk scoring model is constructed using linear weighting:

[0022] In the formula, For the first The comprehensive risk score of the prescription. For the first The values ​​of each risk factor in this prescription These are factor weights, derived from historical reviews or expert ratings, and support online incremental updates.

[0023] In this invention, the risk factor values ​​in the drug dimension include at least whether it is a key monitored drug, incompatibilities, first-time use of the drug, drug dosage form suitability, and drug usage frequency. Key monitored drugs are determined by whether the drugs in the prescription are listed in the National Key Monitored Drugs List; if so, the risk factor value is 1, otherwise it is 0. Incompatibilities are detected using a drug interaction database, such as Micromedex. If a high-risk combination exists in the drug interaction database, the risk factor value is 0.7-1; if a medium-risk combination exists, the risk factor value is 0.3-7. For low-risk combinations in the database, the risk factor value is 0-0.3; for first-time drug use, the risk factor value is 1 if the patient is using the drug for the first time, and 0 otherwise; for drug dosage form suitability, the risk factor value is 0.7-1 if highly matched, 0.3-0.7 if partially matched, and 0 if completely mismatched, such as a child using an adult dosage form; for drug usage frequency, the risk factor value is the ratio of the usage frequency in the prescription to the usage frequency specified in the drug instructions. If the ratio is too low or too high, the risk factor value is 1; if it is 1 or a reasonable value, the risk factor value is 0.

[0024] The patient dimension includes at least age, liver function, kidney function, high-risk pregnancy, allergy history, and genetic testing results. Age is divided into several age groups: 0-1 years (neonatal group), 1-12 years (children), 12-65 years (adult group), and >65 years (elderly group). The risk factor value for the neonatal and elderly groups is 0.8-1, for children it is 0.4-0.8, and for adults it is 0-0.4. Liver function is assessed using the Child-Pugh score: 0 for grade A, 0.4-0.8 for grade B, and 1 for grade C. Kidney function is calculated using creatinine clearance (CrCl): [(140 - age) × weight × 0.85 (female)] / Serum creatinine (mL / min): If the creatinine clearance rate is greater than 90%, the risk factor value is 0; if the creatinine clearance rate is 60-89%, the risk factor value is 0.1-0.4; if the creatinine clearance rate is 30-59%, the risk factor value is 0.5-0.9; if the creatinine clearance rate is less than 30%, the risk factor value is 1. High-risk pregnancy: If it is a high-risk pregnancy, the risk factor value is 1; if it is not a high-risk pregnancy, the risk factor value is 0. History of allergies: If yes, the risk factor value is 1; if no, the risk factor value is 0. Genetic testing results: Polymorphisms of genes related to drug metabolism, such as CYP2C9 and CYP2D6, are counted as 1 if slow or fast metabolism is present, and 0 if normal.

[0025] The medical order dimensions include at least the dosage, the number of combined medications, and the route of administration. The dosage is the ratio of the actual dosage to the standard dosage (actual dosage / standard dosage) plus the ratio of the actual dosage to body weight (actual dosage / body weight). When the dosage is less than 0.8, it is considered insufficient dosage, or when the dosage is greater than 1.3, it is considered overdose, and the risk factor is set to 1. When the dosage is between 0.8 and 1.2, the dosage is considered reasonable, and the risk factor is set to 0. The number of combined medications is 1 when there are more than 5 types of combined medications and 0 when there are fewer than 5 types. The route of administration is whether the actual route of administration is the same as the recommended route of administration. If they are the same, the risk factor is set to 0; if they are different, the risk factor is set to 1.

[0026] The physician dimension includes at least the historical review problem rate, departmental problem density, and historical intervention status. The historical review problem rate is the percentage of problematic prescriptions for the physician in the past 3 months, i.e., the ratio of the number of problematic prescriptions to the total number of prescriptions. If the historical review problem rate is below 5%, the risk factor is 0; if it is between 5% and 10%, the risk factor is 0.5; and if it is above 10%, the risk factor is 1. The departmental problem density is the ratio of problematic prescriptions to the total number of prescriptions in that department. If the departmental problem density is below 5%, the risk factor is 0; if it is between 5% and 10%, the risk factor is 0.5; and if it is above 10%, the risk factor is 1. Historical intervention status refers to the number of rational drug use interventions received in the past 6 months. If the number of rational drug use interventions is greater than 5, the risk factor is 1; if it is less than 5, the risk factor is 0.

[0027] The diagnostic and treatment dimensions include at least off-label use, necessity of medication, and rationality of drug selection. Off-label use refers to whether the use of the drug exceeds the type of information (dosage, indication, population, course of treatment) specified in the instructions. If the use exceeds the instructions, the risk factor is set to 1; otherwise, it is set to 0. Necessity of medication is assessed based on a necessity score (1-10) according to clinical guidelines. If the necessity score is greater than 8, the risk factor is set to 0; if it is less than 8, the risk factor is set to 1. Rationality of drug selection is assessed based on the degree of matching with the drugs recommended in clinical guidelines (1-10). If the degree of matching is less than 8, the risk factor is set to 1; if the degree of matching is greater than 8, the risk factor is set to 0.

[0028] S2. Based on the comprehensive risk score, sort all prescriptions in descending order, and combine the preset total number of review samples to perform high-risk mandatory sampling and medium-risk priority sampling to generate a list of tasks to be reviewed for the day.

[0029] Specifically, the high-risk mandatory sampling means that all prescriptions with a comprehensive risk score higher than the first threshold are included in the task list to be reviewed; the medium-risk priority sampling means that when the number of prescriptions in the task list to be reviewed is less than the total number of preset review samples, prescriptions with scores lower than the first threshold are added to the task list to be reviewed in descending order until the number of prescriptions in the task list to be reviewed is equal to the total number of preset review samples.

[0030] In this invention, a comprehensive risk score that can be ranked and compared is output for all prescriptions through a prescription risk scoring model. All prescriptions are sorted in descending order and a list of prescriptions to be reviewed is generated. This can concentrate high-risk prescriptions in the list of prescriptions to be reviewed, so that subsequent review resources can be prioritized over high-risk prescriptions, improve the detection rate of prescription problems, and avoid the waste of review resources.

[0031] In this invention, the total preset review sample size is set to 10%-20% of the total prescriptions issued on the same day. When the preset review sample size is less than 10%, the problem detection rate increases slowly; when the preset review sample size is greater than 20%, the marginal effect of improvement decreases. Therefore, a preset review sample size of 10%-20% is the most cost-effective, ensuring a high problem detection rate while avoiding waste of review resources.

[0032] S3. Classify the prescriptions in the task list to be reviewed for the day by department, calculate the contribution rate of the number of prescriptions in each department in the task list to be reviewed, and set a corresponding contribution threshold for the contribution rate of the number of prescriptions in each department.

[0033] Specifically, the contribution rate of each department's prescriptions is the ratio of the number of prescriptions in that department to the number of prescriptions on the task list to be reviewed. The contribution threshold for each department is the percentage of the total number of prescriptions in the corresponding department.

[0034] S4. Repeat steps S1-S3 twice to generate a list of tasks to be reviewed for at least three consecutive days, and monitor the contribution rate of prescriptions from each department in real time. If the contribution rate of prescriptions from a certain department is higher than its corresponding contribution threshold for two consecutive days, move the prescriptions from that department in the list of tasks to be reviewed for those two days to the list of prescriptions to be reviewed, and send the list of prescriptions to be reviewed to the pharmacist for prescription review.

[0035] In this invention, there is significant heterogeneity in the incidence of problems among different departments. For example, the problem rate for antibiotics in the respiratory medicine department is typically 19.6%, with a relatively high contribution rate, while the rate in general internal medicine is only 8.2% during the same period, with a relatively low contribution rate. Compared to the traditional fixed allocation of review resources to departments, this invention can concentrate review resources on departments with consistently high problem rates by checking whether the contribution rate exceeds a contribution threshold. Furthermore, since departmental problems fluctuate over time, such as a surge in respiratory prescriptions during flu season, monitoring for two consecutive days can filter out temporary fluctuations and occasional single-day events, allowing review resources to be accurately directed to problem departments. This avoids wasting review resources in low-risk departments and insufficient review in high-risk departments due to fixed allocation of review resources. Moreover, by checking whether the contribution rate exceeds a contribution threshold for two consecutive days to implement specific review tasks, it is possible to quickly respond to sudden risks. For example, when a series of sudden events occur within a certain period, the contribution rate of the corresponding department will also increase continuously during that period, allowing for a rapid response and reasonable allocation of review resources.

[0036] S5. After performing prescription review, record structured result data, which includes at least the problem judgment result, problem type, severity level, intervention behavior, and influencing factors.

[0037] In this invention, after completing the prescriptions on the mandatory prescription review list, pharmacists need to record the problem judgment results of the prescriptions, i.e., whether the prescription has any problems, and standardize the problem codes; the problem type needs to be specified, and if the specific problem of the prescription is, such as overdose, over-treatment course, or incompatibility with indications, it can be marked with multiple tags; the severity level needs to be specified, indicating whether the problem of the prescription is mild, moderate, or severe; the intervention behavior needs to be specified, including only marking, feedback to the doctor, telephone intervention, and reporting to quality control; the influencing factors need to be specified, indicating which risk factors or factors led to the review result.

[0038] S6. Calculate the strategy effect feedback score based on the structured result data. When the strategy effect feedback score is lower than the expected threshold, update the risk scoring model.

[0039] Specifically, the strategy effectiveness feedback scoring is conducted through:

[0040] The calculation yields F, which represents the strategy effectiveness feedback score, P, and Q, respectively. P is the problem detection rate, Q is the specific hit rate, and α and β are the strategy evaluation weights, with α + β = 1. When the strategy effectiveness feedback score is less than the set expected value, the first threshold for high-risk mandatory sampling is adjusted, or the factor weights are updated. At least one of them.

[0041] Specifically, the problem detection rate is the ratio of problematic prescriptions to the list of prescriptions that must be reviewed, and the specific hit rate is the ratio of the type of problem that occurred to all types of problems.

[0042] In this invention, the risk scoring model is updated based on strategy effectiveness feedback scores to optimize the model. Generally, hospitals have limited review resources; therefore, when the first threshold is low or the factor weights are low... When the score is high, the number of prescriptions on the pending review list and the number of prescriptions on the review list will increase, thus diluting review resources. Therefore, when the strategy effectiveness feedback score is greater than the expected threshold, it indicates that the problem detection rate and the specific hit rate are both high, meaning that the proportion of problematic prescriptions in the total number of prescriptions is large. The first threshold or factor weight set at this point is then appropriate. Under these conditions, most of the issues can be categorized into the pending review list and the mandatory review prescription list, concentrating review resources on the issue prescriptions and avoiding waste of review resources; when the strategy effect feedback score is less than the expected threshold, it indicates that the current first threshold or factor weight is insufficient. Under these conditions, the detection rate and specific hit rate of prescription problems are both low, meaning that the proportion of problematic prescriptions in the total number of prescriptions is small. In this case, the first threshold or factor weight can be adjusted. This allows for adjustments to the sampling strategy. When high-risk mandatory sampling results in a number of prescriptions in the task list significantly exceeding the preset total sample size, the first threshold is increased to reduce the number of prescriptions sampled in the high-risk mandatory sampling method. This reduces the number of relatively low-risk prescriptions in both the task list and the mandatory review list, concentrating review resources more on high-risk prescriptions and improving the detection rate of prescription issues. Furthermore, the relevant factor weights are adjusted. The entire set of prescriptions is re-ranked, and the factor weights are dynamically adjusted in real time based on the severity level, intervention behavior, and influencing factors in the structured outcome data. This allows for a more accurate and realistic comprehensive risk score for all prescriptions, thereby improving the problem detection rate.

[0043] Specifically, the factor weights are adjusted through the following sub-steps: Step 1: Collect the number of times each risk factor listed in the problem prescription appears, and record the total number of times all risk factors appear; Step 2: Collect the severity level and intervention behavior corresponding to each occurrence of each risk factor specified in the problem prescription, and assign weights to the severity level and intervention behavior. The weights for severity level are: mild 0.2, moderate 0.6, and severe 1. The weights for intervention behavior are: marking only 0.3, feedback to doctor 0.6, telephone intervention 0.8, and reporting to quality control 1. Calculate the average severity level weight and average intervention behavior weight for each risk factor. Step 3: Calculate the percentage e of each risk factor in the total number of occurrences of all risk factors; Step 4: Update factor weights :

[0044] In the formula, The adjusted factor weights, For the current factor weights, Let be the learning rate, where =Average severity level weighting, =Average intervention behavior weighting, The learning constant is set to 0.002 by default, but can be adjusted according to the actual situation.

[0045] In this invention, risk factors that frequently appear in problematic prescriptions can be addressed through... Gradually increase its factor weight; in addition, for risk factors that do not appear in the influencing factors, their calculated... Since the factor weight is 0, its factor weight will gradually decrease. In summary, when calculating and ranking the comprehensive risk score, prescriptions corresponding to high-risk factors can be ranked first and prescriptions corresponding to low-risk factors can be ranked last. This makes it easier to collect prescriptions on high-risk prescriptions, realize the feedback of the review results to the sampling strategy, and improve the detection rate of prescription problems.

[0046] A prescription review sampling system includes a risk scoring module for collecting multi-dimensional data on drugs, patients, medical orders, doctors, and diagnosis and treatment, and outputting a comprehensive risk score for the prescription. The sampling task generation module, connected to the risk scoring module, is used to perform high-risk mandatory sampling and medium-risk priority sampling based on the comprehensive risk score, and generate a list of tasks to be reviewed. The classification module is used to classify prescriptions in the task list to be reviewed by department, calculate the contribution rate of each department in the task list to be reviewed, and set a corresponding contribution threshold for the contribution rate of each department's prescriptions. The review and judgment module is used to monitor the contribution rate of prescriptions in each department in real time. If the contribution rate of prescriptions in a certain department is higher than its corresponding contribution threshold for two consecutive days, the prescriptions of that department in the list of tasks to be reviewed for those two days will be moved to the list of prescriptions to be reviewed, and the list of prescriptions to be reviewed will be sent to the pharmacist's end to perform prescription review. The review execution module is used to perform prescription reviews and record structured result data. The structured result data includes at least the problem judgment result, problem type, severity level, intervention behavior, and influencing factors. The closed-loop optimization module is used to calculate the strategy effect feedback score based on the structured result data, and update the risk scoring model when the strategy effect feedback score is lower than the expected threshold.

[0047] (1) Unless otherwise defined, the same reference numerals in the embodiments and drawings of this disclosure have the same meaning.

[0048] (2) The accompanying drawings of the embodiments of this disclosure only involve the structures involved in the embodiments of this disclosure. Other structures can be referred to the general design.

[0049] (3) For clarity, components or areas are enlarged in the drawings used to describe embodiments of the present disclosure. It will be understood that when an element is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element, or there may be an intermediate element.

[0050] The above description is merely a specific embodiment of this disclosure, but the scope of protection of this disclosure is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this disclosure should be included within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims.

Claims

1. A sampling method for prescription review, characterized in that: Includes the following steps: S1. Construct a prescription risk scoring model, taking multi-factor data from the drug dimension, patient dimension, medical order dimension, doctor dimension and treatment dimension of the day as input, and outputting a comprehensive risk score of all prescriptions of the day; S2. Based on the comprehensive risk score, sort all prescriptions in descending order, and combine the preset total number of review samples to perform high-risk mandatory sampling and medium-risk priority sampling to generate a list of review tasks for the day. S3. Classify the prescriptions in the task list to be reviewed for the day by department, calculate the contribution rate of the number of prescriptions in each department in the task list to be reviewed, and set a corresponding contribution threshold for the contribution rate of the number of prescriptions in each department. S4. Repeat steps S1-S3 twice to generate a list of tasks to be reviewed for at least three consecutive days, and monitor the contribution rate of prescriptions from each department in real time. If the contribution rate of prescriptions from a certain department is higher than its corresponding contribution threshold for two consecutive days, move the prescriptions from that department in the list of tasks to be reviewed for those two days to the list of prescriptions to be reviewed, and send the list of prescriptions to be reviewed to the pharmacist for prescription review. S5. After performing prescription review, record structured result data, which shall include at least the problem judgment result, problem type, severity level, intervention behavior, and influencing factors; S6. Calculate the strategy effect feedback score based on the structured result data. When the strategy effect feedback score is lower than the expected threshold, update the risk scoring model.

2. The prescription review sampling method according to claim 1, characterized in that: The prescription risk scoring model is constructed using linear weighting: ； In the formula, For the first The comprehensive risk score of the prescription. For the first The values ​​of each risk factor in this prescription These are factor weights, derived from historical reviews or expert ratings, and support online incremental updates.

3. A prescription review sampling method according to claim 1 or 2, characterized in that: The high-risk mandatory sampling means that all prescriptions with a comprehensive risk score higher than the first threshold are included in the list of tasks to be reviewed. The medium-risk priority sampling means that when the number of prescriptions in the list of tasks to be reviewed is less than the total number of preset review samples, prescriptions with scores lower than the first threshold are added to the list of tasks to be reviewed in descending order until the number of prescriptions in the list of tasks to be reviewed is equal to the total number of preset review samples.

4. The prescription review sampling method according to claim 2, characterized in that: In steps S3 and S4, the contribution threshold for each department is the percentage of prescriptions from the corresponding department in the total number of prescriptions.

5. The prescription review sampling method according to claim 3, characterized in that: The strategy's effectiveness feedback score was obtained through: ； The calculation yields F, which represents the strategy effectiveness feedback score, P, and Q, respectively. P is the problem detection rate, Q is the specific hit rate, and α and β are the strategy evaluation weights, with α + β = 1. When the strategy effectiveness feedback score is less than the set expected value, the first threshold for high-risk mandatory sampling is adjusted, or the factor weights are updated. At least one of them.

6. The prescription review sampling method according to claim 5, characterized in that: The problem detection rate is the ratio of problematic prescriptions to the list of prescriptions that must be reviewed, and the specific hit rate is the ratio of the type of problem that occurred to all types of problems.

7. A prescription review sampling system, used in any one of the prescription review sampling methods as described in claims 1-6, characterized in that: This includes a risk scoring module, which collects multi-dimensional data on drugs, patients, medical orders, doctors, and diagnosis and treatment, and outputs a comprehensive risk score for prescriptions. The sampling task generation module, connected to the risk scoring module, is used to perform high-risk mandatory sampling and medium-risk priority sampling based on the comprehensive risk score, and generate a list of tasks to be reviewed. The classification module is used to classify prescriptions in the task list to be reviewed by department, calculate the contribution rate of each department in the task list to be reviewed, and set a corresponding contribution threshold for the contribution rate of each department's prescriptions. The review and judgment module is used to monitor the contribution rate of prescriptions in each department in real time. If the contribution rate of prescriptions in a certain department is higher than its corresponding contribution threshold for two consecutive days, the prescriptions of that department in the list of tasks to be reviewed for those two days will be moved to the list of prescriptions to be reviewed, and the list of prescriptions to be reviewed will be sent to the pharmacist's end to perform prescription review. The review execution module is used to perform prescription reviews and record structured result data. The structured result data includes at least the problem judgment result, problem type, severity level, intervention behavior, and influencing factors. The closed-loop optimization module is used to calculate the strategy effect feedback score based on the structured result data, and update the risk scoring model when the strategy effect feedback score is lower than the expected threshold.

Images