A medical instant retail insight system in o2o mode

Abstract

This invention discloses a medical real-time retail insight system under the O2O model, belonging to the field of intelligent supply chain management technology. It addresses the problem of information lag and inaccuracy caused by the system's inability to update inventory information in a timely manner when orders are generated or inventory changes. By combining real-time analysis of order impact metrics and dynamically adjusting the update frequency, it solves the problems of information lag and inaccuracy in traditional systems, ensuring that inventory information is synchronized with actual demand. Furthermore, considering the special characteristics of drug storage environments and drug types, it quantifies drug complexity and uses the TOPSIS algorithm to differentiate update methods, reducing the risk of drug expiration due to improper storage conditions or untimely updates, thus ensuring medical safety.

Description

Technical Field

[0001] This invention relates to the field of intelligent supply chain management technology, and more specifically, to a medical instant retail insight system under the O2O model. Background Technology

[0002] The O2O model is a business model that closely integrates online services with offline physical stores. It aims to guide consumers to physical stores through internet platforms, while converting offline customers into online customers, thus achieving a seamless connection between online and offline.

[0003] Medical on-demand retail is an emerging business model that combines the features of O2O (online-to-offline) and on-demand delivery. It aims to provide consumers with convenient medical products and services through online platforms and achieve rapid delivery through an efficient logistics system.

[0004] The existing technology has the following shortcomings:

[0005] Existing medical on-demand retail systems rely on traditional manual record-keeping or simple management systems, which cannot achieve real-time inventory monitoring. When an order is generated or inventory changes, the system cannot update inventory information in a timely manner, resulting in information lag and inaccuracy. Therefore, this application proposes an O2O medical on-demand retail system, aiming to dynamically adjust the frequency of inventory information updates and select appropriate inventory information update technologies, thereby improving service efficiency and cost control.

[0006] To address the above problems, this invention proposes a solution. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a medical instant retail insight system under the O2O model to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] A medical real-time retail insight system under the O2O model.

[0010] It includes a data acquisition module, a data classification module, a data matching module, and a feedback adjustment module, with signal connections between each module; among which:

[0011] The data acquisition module is used to collect inventory information update demand-related data and order impact measurement data, and to preprocess the data;

[0012] The data classification module is used to classify the demand for inventory information updates by level through clustering algorithms and to calculate a comprehensive index for measuring the impact of orders.

[0013] The data matching module is used to update the inventory information based on the demand and complexity of the drug, and selects the inventory information update method through the TOPSIS method.

[0014] The feedback adjustment module combines inventory information with demand level and order impact measurement to dynamically adjust the inventory information update frequency through fuzzy rules.

[0015] In a preferred embodiment, the inventory information update demand-related data collected by the data acquisition module includes drug supply arrival time, replenishment cycle, sales speed, return quantity, and storage environment; the order impact measurement-related data includes order quantity, order processing time, and order variability; and the preprocessing includes data standardization.

[0016] In a preferred embodiment, the replenishment cycle is the time interval from determining that replenishment is needed to actually completing the replenishment, including order generation time, supplier confirmation time, supplier processing time, transportation time, and warehousing time; the storage environment is classified into levels based on temperature parameters, including room temperature storage, cool and dry environment, constant temperature storage, refrigerated storage, and strict cold chain storage.

[0017] In a preferred embodiment, the data classification module uses the K-means clustering algorithm to classify the demand for inventory information updates into three levels: high demand, medium demand, and low demand. The clustering process includes:

[0018] The collected feature data are used to form a feature matrix, and the feature values ​​are standardized.

[0019] Set the number of clusters to 3, and iteratively update the cluster centers until convergence;

[0020] The tiers are defined based on the average characteristics of each cluster. The clusters with the shortest delivery time, shortest replenishment cycle, fastest sales speed, fewest returns, and most stable storage environment are classified as high-demand tiers.

[0021] In a preferred embodiment, the process by which the data classification module calculates the comprehensive index of order impact measurement includes:

[0022] Calculate the impact coefficient of order quantity, the impact coefficient of order processing time, and the impact coefficient of order variability expressed as volatility;

[0023] Using the weighted summation formula ;in , , The weights are order quantity, order processing time, and order variability, respectively, with Qᵢ, Tᵢ, and Vᵢ being the corresponding influence coefficients.

[0024] In a preferred embodiment, the fuzzy rules of the feedback adjustment module include:

[0025] The input variables are the demand for inventory information updates, specifically divided into High, Medium, and Low, and the order impact measure, specifically divided into Big, Medium, and Small.

[0026] The output variable is the inventory information update frequency, specifically divided into Increase, Unchanged, and Decrease.

[0027] The rules include: if the demand for inventory updates is High and the order impact metric is Big, then the inventory update frequency is Increase; if the demand for inventory updates is Low and the order impact metric is Small, then the inventory update frequency is Decrease.

[0028] In a preferred embodiment, the drug complexity is calculated by a weighted sum of the storage environment level and the drug type level, as shown in the formula: In the formula For the storage environment of drugs, Drug type and These are the weighting coefficients for the storage environment and the drug type, respectively, and the sum of the weighting coefficients is 1; the drug type level is divided into 1-5 points according to its medical use.

[0029] In a preferred embodiment, the TOPSIS method employed by the data matching module includes:

[0030] The decision criteria were determined to be the need for inventory information updates and the complexity of the drugs, and the inventory information update need level was converted into a numerical value of 1-3;

[0031] The decision criteria data are normalized, and the ideal solution is set as the maximum value of inventory information update demand and the minimum value of drug complexity, while the negative ideal solution is set as the minimum value of inventory information update demand and the maximum value of drug complexity.

[0032] Calculate the Euclidean distance between each drug and the ideal solution and the negative ideal solution, and then use a comprehensive index. In the formula and are the distances between the i-th drug and the ideal solution and the negative ideal solution, respectively.

[0033] In a preferred embodiment, the data matching module selects an inventory information update method based on the composite index: automatic update is selected when the composite index is ≥ the 66th percentile; semi-automatic update is selected when the composite index is ≥ the 33rd percentile and < the 66th percentile; and manual update is selected when the composite index is ≤ the 33rd percentile.

[0034] In a preferred embodiment, the order volatility is calculated using volatility, including:

[0035] Calculate the logarithmic return on the number of orders in adjacent time periods Where yi is the number of orders in the i-th time period, and ln is the natural logarithm;

[0036] Where N is the total number of time periods, Logarithmic return It is the average of the logarithmic returns.

[0037] The technical effects and advantages of the medical instant retail insight system under the O2O model of this invention:

[0038] This invention uses a clustering algorithm to classify inventory update demand levels and combines real-time analysis of order impact metrics to dynamically adjust update frequency, solving the problems of information lag and inaccuracy in traditional systems and ensuring that inventory information is synchronized with actual demand. Addressing the specificities of drug storage environments and drug types, it quantifies drug complexity and uses the TOPSIS algorithm to differentiate update methods, reducing the risk of drug expiration due to improper storage conditions or untimely updates, thus ensuring medical safety. Fuzzy rules are used to respond to order variability, increasing the update frequency for drugs with high demand and significant order impact for rapid response to urgent orders; optimized update methods are used for low-demand, simple drugs, reducing unnecessary system computing power and manpower investment, balancing service efficiency and cost. Finally, it integrates supply chain data such as supply arrival time and replenishment cycle, using cluster analysis to clarify inventory update priorities, helping medical retail companies accurately control replenishment rhythm, reduce stockouts or backlogs, and improve supply chain stability. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the structure of a medical instant retail insight system under the O2O model of the present invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] This invention combines real-time analysis of order impact measurement with dynamic adjustment of update frequency, solving the problems of information lag and inaccuracy in traditional systems and ensuring that inventory information is synchronized with actual demand. Furthermore, considering the special characteristics of drug storage environment and drug type, it quantifies drug complexity and uses the TOPSIS algorithm to differentiate update methods, reducing the risk of drug failure due to improper storage conditions or untimely updates, and ensuring medical safety.

[0042] Example 1

[0043] Figure 1 This paper presents a medical real-time retail insight system under the O2O model, aiming to dynamically adjust the frequency of inventory information updates and select appropriate inventory information update technologies. The specific steps include:

[0044] Step 1: Collect relevant data needed to analyze the frequency of dynamically adjusting and updating inventory information, including inventory information update demand and order impact measurement, ensure the accuracy and consistency of the data, preprocess the data and store it in the database.

[0045] Step 2: Retrieve data from the database, use a clustering algorithm to calculate the demand for inventory information updates, and divide it into three levels.

[0046] Step 3: Collect order-related data, and calculate order variability, impact coefficient, and comprehensive indicators through standardized processing.

[0047] Step 4: Use fuzzy rules to define and comprehensively analyze the demand for inventory information updates and the impact of orders, so as to dynamically adjust the inventory information update frequency.

[0048] Step 5: Calculate the complexity of the drug based on its storage environment and type.

[0049] Step 6: Analyze drug inventory update methods. Use TOPSIS to comprehensively analyze the need for updating drug inventory information and the complexity of drugs in order to select the most suitable inventory update method.

[0050] Specifically, in step 1, dynamically adjusting and updating inventory information refers to the process of flexibly adjusting and updating inventory information based on real-time or near-real-time data. Inventory information update necessity refers to assessing the urgency or frequency of inventory information updates based on characteristic data. Characteristic data used for assessment includes, but is not limited to, drug supply arrival time, replenishment cycle, sales speed, return quantity, and storage environment. Supply arrival time refers to the time interval from the actual shipment of drugs in the supply chain to their arrival at the warehouse or distribution center; timestamps for each shipment and arrival can be recorded in the supply chain management system.

[0051] The replenishment cycle refers to the time interval from the point when replenishment is determined to be needed to the actual replenishment process during the sales process. Specifically, it includes order generation time, supplier confirmation time, supplier processing time, transportation time, and warehousing time. The replenishment cycle can be calculated by recording the start and end times of each process, subtracting the results from each other, and finally adding all the process times together. For example, recording the shipping and arrival times of pharmaceutical products can yield the transportation time, etc.

[0052] Sales speed refers to the quantity of drugs sold per unit of time, usually measured in days, weeks, or months; sales data (sales volume, etc.) can be recorded and collected in real time through POS systems or sales tables.

[0053] The quantity of returned goods refers to the number of medications returned from customers for various reasons; the quantity of returned goods can be recorded through return records and customer service feedback.

[0054] Storage environment refers to the environmental conditions in which a drug is stored and transported, such as temperature and humidity. Environmental monitoring equipment (such as temperature sensors and hygrometers) can be used to monitor the storage conditions of the drug in real time. For example, the temperature of the storage environment can be collected in real time. Excessively high or low temperatures may affect drug storage. Storage environments can be categorized as follows: ambient temperature storage (25°C to 30°C) is labeled as 1; cool and dry environment (15°C to 25°C) is labeled as 2; constant temperature storage (8°C to 15°C) is labeled as 3; refrigerated storage (2°C to 8°C) is labeled as 4; and strict cold chain storage (-20°C and below) is labeled as 5, etc. It should be noted that using temperature to assess the storage environment is only one example in this embodiment. In practice, humidity, light, etc., can also be used for assessment, which will not be elaborated on here.

[0055] Order impact measurement is the process of measuring the degree to which orders affect the frequency of inventory information updates, mainly including order quantity, order processing time, and order variability. Order processing time refers to the length of time from order generation to completion or delivery. Timestamps of order generation and completion can also be recorded to calculate order processing time. It should be noted that during preprocessing in the data acquisition module, all time-related raw data (such as order processing time) is uniformly converted to seconds as the basic unit to ensure consistency in subsequent calculations. Order variability refers to the degree of fluctuation and change in order quantity and type over a certain period, which can be calculated using the order quantity for each time period.

[0056] Specifically, in step 2, a clustering algorithm is used to calculate the demand for inventory information updates and divide it into three levels (clusters): high demand, medium demand, and low demand. Clustering is an unsupervised learning method used to divide a dataset into multiple groups, making data points within the same group more similar in some sense, and data points in different groups more different in some sense. The specific calculation process is as follows:

[0057] S1: Collect feature data, which refers to specific features used to assess the need for inventory information updates. This includes, but is not limited to, drug supply and delivery times, replenishment cycles, sales speed, return quantities, and storage conditions. The collected feature data is then structured into a feature matrix. Assuming a medical on-demand retail company has n drug products, and each drug product has m features, then the feature matrix X is... Let represent the j-th feature value of the i-th drug product. A sample is a vector containing all feature values. Each sample represents a drug product and contains all the features of that drug product. That is, each sample is an m-dimensional feature vector, where m is the number of features.

[0058] S2: Data Preprocessing. Standardize the eigenvalues ​​to ensure they have the same scale and units. A common method is to subtract the mean from each eigenvalue and then divide by its standard deviation to make the mean 0 and the standard deviation 1. Specific formulas are available. In the formula These are the standardized feature values, where x is the original feature value and α is the mean of the feature. It is the standard deviation of the feature.

[0059] S3: Choose an appropriate number of clusters. The number of clusters indicates how many clusters or groups the data should be divided into when using a clustering algorithm. Choosing an appropriate number of clusters is a crucial step in cluster analysis, directly impacting the quality and interpretability of the final clustering results. Based on this implementation case, the inventory information update demand can be divided into three levels: high, medium, and low, corresponding to high demand, medium demand, and low demand, respectively. Therefore, the number of clusters can be determined to be 3. By choosing 3 clusters, the desired three clusters can be obtained, each representing a level of inventory information update demand. Further adjustments can also be made using the elbow rule or silhouette coefficient, which will not be detailed here.

[0060] S4: Select and apply a clustering algorithm. Common clustering algorithms include K-means clustering, hierarchical clustering, and density clustering. Based on drug feature data, this solution uses K-means clustering as an example. The advantages of this algorithm are its simplicity, ease of understanding and implementation, and it is particularly suitable for handling large-scale datasets.

[0061] Randomly select k initial cluster centers and assign each sample to the cluster containing the nearest cluster center. Then calculate the new center of each cluster (i.e., the mean of the samples within the cluster) and use it as the new cluster center. Repeat the two steps of assigning samples and updating centers until the cluster centers no longer change or the maximum number of iterations is reached. Assigning samples means assigning each sample to the cluster containing the nearest cluster center, and updating centers means calculating the new center of each cluster (i.e., the mean of the samples within the cluster) and using it as the new cluster center.

[0062] We randomly select k initial cluster centers and label them μ1, μ2, ..., μk. For each sample, we calculate its distance to each cluster center, using the following formula: In the formula This represents the distance between sample xi and cluster center μj. It is assigned to the nearest cluster Cu (u=1, 2, 3), specifically... For each cluster Cu, calculate its new cluster center μj, which is the sample mean within the cluster, and have... In the formula, xi represents the i-th drug product, which is a feature vector containing all the features of the drug product.

[0063] S5: Interpret the clustering results. Shorter supply arrival times indicate the need for frequent inventory updates to ensure a smooth supply chain; shorter replenishment cycles suggest faster sales, requiring quicker replenishment and increased inventory update frequency; faster sales indicate high market demand, necessitating frequent inventory updates to maintain sufficient supply; fewer returns suggest better drug quality and higher customer satisfaction, requiring more frequent inventory updates; smaller (stable) storage environments mean the storage conditions are relatively easy to control, allowing for more frequent inventory updates. Therefore, the average values ​​of each feature for each cluster can be calculated to further define high-demand clusters, etc. For each cluster, the average value of its features is expressed as... Where u represents the cluster index and j represents the feature index. The specific formula is... In the formula It is the number of drugs in cluster Cu. This is the j-th feature value of the q-th drug in cluster Cu. By comparing the average value of each feature in a cluster, three clusters can be defined. For example, the cluster containing the fastest delivery time, shortest replenishment cycle, fastest sales speed, fewest returns, and moderate and stable storage environment can be defined as the high-demand cluster, and so on.

[0064] Specifically, in step 3, the order-related data mainly includes order quantity, order processing time, and order variability. Order quantity and order processing time can be obtained through data collection or direct calculation. The calculation method for order variability is as follows:

[0065] First, data is collected, including the total number of orders for each time period (daily, weekly, monthly) and the timestamp of each order. Then, the standard deviation of the order count is calculated. Standard deviation is a common measure of variability, reflecting the average deviation of a dataset. The formula is as follows: In the formula Here, is the standard deviation, N is the number of orders, yi is the number of orders in each time period, and μ is the average number of orders. Finally, volatility is calculated to assess order variability. Volatility is a commonly used measure of volatility in finance and can also be used to calculate order variability. It typically uses time-series data and is measured by calculating the standard deviation of the logarithmic return of the number of orders over adjacent time periods (a common yield calculation method in finance used to measure the relative change in asset prices between two points in time). The logarithmic return of the number of orders over adjacent time periods has... Where yi is the number of orders in the i-th time period, and ln is the natural logarithm. The average logarithmic return is calculated as follows: Finally, the calculation formula is as follows: Where N is the total number of time periods, Logarithmic return It is the average of the logarithmic returns.

[0066] The order impact metric (OLL) is calculated using order quantity, order processing time, and order variability, as detailed below:

[0067] The specific formula for calculating the impact coefficient of order quantity is as follows: In the formula, Q represents the number of orders in the current time period. It is the average number of orders. This represents the standard deviation of the order quantity. Similarly, the order processing time impact coefficient Ti is obtained. This calculation formula uses the Z-Score standardization method commonly used in statistics. Its purpose is to convert the original data (such as order quantity and order processing time) into dimensionless impact coefficients, thereby eliminating the influence of different dimensions of the original data on the calculation of the comprehensive index Oll. The order variability impact coefficient is replaced by volatility and denoted as Vi.

[0068] The impact coefficients of order quantity, order processing time, and order variability are combined into a single index, OL, calculated using the following formula: ;in , , The weights for order quantity, order processing time, and order variability are determined based on relevant professional knowledge and specific needs, and will not be elaborated here.

[0069] A high Oll indicates a significant impact from orders, necessitating frequent inventory updates and thus requiring a higher update frequency. Conversely, a low Oll indicates a smaller impact from orders, resulting in lower inventory updates and allowing for a reduced update frequency. Specifically, this can be achieved by analyzing historical data... Statistical analysis was performed on the values, and two thresholds were set. and To divide The level. For example, it can be... The top 33% of the value distribution is set as 'Big' (higher), the middle 34% as 'Medium', and the bottom 33% as 'Small' (lower). These thresholds are set based on actual operational data and business needs, and fall within the scope that those skilled in the art can determine according to specific application scenarios.

[0070] Specifically, in step 4, fuzzy rules are used to comprehensively analyze the demand for inventory information updates and the impact of orders, thereby dynamically adjusting the inventory information update frequency. As follows:

[0071] The demand for inventory updates and the impact of orders are defined as input variables, and they are divided into different fuzzy sets, such as "Low", "Medium", and "High" for the demand for inventory updates, and "Small", "Medium", and "Big" for the impact of orders.

[0072] The inventory information update frequency is defined as an output variable, and it is also divided into different fuzzy sets, such as "Decrease", "Unchanged", and "Increase". For the inventory information update frequency, "Decrease" means to decrease the inventory information update frequency, "Unchanged" means to keep the inventory information update frequency unchanged, and "Increase" means to increase the inventory information update frequency.

[0073] Develop fuzzy rules to describe the impact of inventory update demand and order impact metrics on inventory update frequency. The rules can be defined based on industry expertise or through data analysis and experimentation. For example, if inventory update demand is labeled A, order impact metrics are labeled oll, and inventory update frequency is represented by R, then the following rules can be defined:

[0074] Rule1:IF(AisHigh)AND(ollisBig)THEN(RisIncrease)

[0075] Rule2:IF(AisLow)AND(ollisSmall)THEN(RisDecrease) ......

[0077] Fuzzy reasoning is performed based on fuzzy rules to dynamically adjust the update frequency of inventory information.

[0078] It should be noted that the division of fuzzy sets can be adjusted according to the actual situation. For example, this embodiment uses three fuzzy sets as an example. In reality, the inventory information update demand, order impact measurement, and inventory information update frequency can be divided into more than three sets, so as to more conveniently dynamically adjust the inventory information update frequency.

[0079] Further judgments regarding the low, medium, and high demand for inventory information updates can be made directly based on the clustering results of inventory information update demand, which will not be repeated here. Judgments regarding the small, medium, and large order impact metrics can be made by setting order impact metric thresholds for comparison. For example, when oll is greater than the first order impact metric threshold, it can be marked as "Big", etc., which will not be elaborated on here.

[0080] Specifically, in step 5, drug complexity refers to the stringency and management difficulty of the conditions required for drug storage, transportation, and inventory information updates. Drug complexity can be calculated using two main factors: storage environment and drug type. The storage environment can be assessed using only temperature, while the drug type can be converted into numerical data (data with specific values) based on its medical purpose or importance. The specific calculation is as follows:

[0081] Data on drug chemical composition, biological characteristics, main effects, and side effects are collected from drug instructions and supplier information. Safety records and efficacy evaluations are also obtained from medical institutions, drug manufacturers, and drug regulatory agencies. Based on this data, drugs are categorized and scored. For example, common disease treatments (such as cold medicine and painkillers) are scored 1 point; chronic disease management drugs (such as hypertension and diabetes medications) are scored 2 points; drugs for serious but treatable diseases (such as antibiotics and antiviral drugs) are scored 3 points; life-dependent drugs (such as emergency heart medications and certain cancer drugs) are scored 4 points; and critical emergency medications (such as resuscitation drugs and organ transplant rejection drugs) are scored 5 points, etc.

[0082] The storage environment for the drugs is the same as the collection method in step 1, and will not be repeated here.

[0083] Finally, a weighted summation is used to calculate a combined score for the drug based on both storage environment and drug type factors to assess drug complexity. The calculation formula is as follows: In the formula For the storage environment of drugs, Drug type and These are the weighting coefficients for storage environment and drug type, respectively, with a sum of 1. Their settings can be determined based on relevant domain expertise and specific needs. For example, if a drug is highly sensitive to temperature, the weighting coefficient for storage environment should be increased. No specific restrictions are imposed here. The resulting overall score ranges from 1 to 5. A higher overall score indicates a more complex drug, while a lower score indicates a less complex drug.

[0084] Specifically, in step 6, inventory information update methods can be divided into three types: automatic update, manual update, and semi-automatic update (a combination of manual and automatic). The choice between automatic, manual, or semi-automatic updates depends primarily on the need for updating drug inventory information and the complexity of the drug. Choosing manual or semi-automatic updates, especially for highly complex drugs (such as those requiring strict cold chain storage or critical emergency medications), is based on the specific requirements of medical point-of-care retail for drug storage safety and inventory information accuracy. Even if the system has automatic update capabilities, for highly complex drugs, manual verification or intervention (i.e., semi-automatic or manual) is still necessary to ensure the accuracy of the update, prevent drug expiration risks due to system errors or changes in storage conditions, and thus ensure medical safety. For example, for drugs requiring strict cold chain storage, semi-automatic updates may involve the system automatically recording data, but manual verification of temperature records and storage environment is required before final confirmation of the inventory information update; manual updates may involve manual inventory checks and data entry to ensure the absolute accuracy of inventory information for critical drugs. TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) is a multi-criteria decision analysis method used to evaluate the merits of candidate solutions relative to a set of specific attributes. This embodiment uses TOPSIS to perform a comprehensive analysis of both criteria, as detailed below:

[0085] P1: Determine decision criteria. Establish decision criteria for selecting an appropriate inventory update method, including the need for updating drug inventory information and drug complexity. Sample data is... That is, the need to update the inventory information of each drug and the complexity of the drug, with the sample size being the number of drugs that need to be updated.

[0086] P2: Standardized data. A represents the inventory information update requirement of a drug. Clusters with low update requirement are labeled as 1, and all drugs within that cluster have an inventory information update requirement value of 1. This process is repeated, with clusters of medium update requirement labeled as 2, and clusters of high update requirement labeled as 3. The inventory information update requirement and drug complexity of the sample drugs are normalized using the following formula: In the formula, The value is the normalized value.

[0087] P3: Determine the ideal solution and the negative ideal solution. An ideal solution is the solution that achieves the best value for each decision criterion, while a negative ideal solution is the solution that achieves the worst value for each decision criterion. Generally, the higher the demand for inventory information updates, the greater the likelihood of using an automatic update method to quickly respond to changes in demand and maintain stable inventory levels; conversely, the lower the complexity (simpler) of the drug, the greater the likelihood that an automatic update method will be chosen, as it can improve update efficiency and flexibility while meeting the demand. Therefore, the ideal solution is the maximum value of the demand for inventory information updates and the minimum value of the drug complexity, i.e. Similarly, the negative ideal solution is the minimum demand for inventory information updates and the maximum drug complexity, i.e. .

[0088] P4: Calculate the similarity metric. Use the Euclidean distance formula to calculate the distance between each drug and the ideal solution. Specifically... In the formula It is the distance between the i-th drug and the ideal solution. It is the actual value of the i-th drug on the j-th decision criterion. The ideal solution is γ represents the number of decision criteria, which is 2, namely the need for updating drug inventory information and drug complexity. Similarly, by calculating the distance between each drug and the negative ideal solution, we obtain... .

[0089] P5: Calculate the composite index of the drugs. Calculate the composite index for each drug using the composite index formula, specifically: In the formula and are the distances between the i-th drug and the ideal solution and the negative ideal solution, respectively.

[0090] P6: Explain the results. Sort the drugs from largest to smallest based on their composite index. Drugs with a composite index greater than or equal to the 66th percentile (the value at the 66th percentile in a set of data) are updated manually; drugs with a composite index greater than or equal to the 33rd percentile but less than the 66th percentile are updated semi-automatically; drugs with a composite index less than or equal to the 33rd percentile are updated manually.

[0091] It is important to note that the determination of the ideal and negative ideal solutions is based on the analysis using the automatic update method. Therefore, the larger the final composite index, the more suitable the automatic update method is. If the ideal and negative ideal solutions are analyzed using the manual update method, then the larger the final composite index, the more suitable the manual update method is. Further details will not be elaborated upon here.

[0092] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0093] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0094] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0095] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0096] The above description is merely a specific embodiment of this application, but the scope of protection of this application 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 application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0097] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A medical instant retail insight system under the O2O model, characterized in that... ； It includes a data acquisition module, a data classification module, a data matching module, and a feedback adjustment module, with signal connections between each module; among which: The data acquisition module is used to collect inventory information update demand-related data and order impact measurement data, and to preprocess the data; The data classification module is used to classify the demand for inventory information updates by level through clustering algorithms and to calculate a comprehensive index for measuring the impact of orders. The data matching module is used to update the inventory information based on the demand and complexity of the drug, and selects the inventory information update method through the TOPSIS method. The feedback adjustment module combines inventory information with demand level and order impact measurement to dynamically adjust the inventory information update frequency through fuzzy rules.

2. The medical instant retail insight system under the O2O model according to claim 1, characterized in that: The data acquisition module collects inventory information update demand-related data including drug supply arrival time, replenishment cycle, sales speed, return quantity, and storage environment; the order impact measurement-related data includes order quantity, order processing time, and order variability; the preprocessing includes data standardization.

3. A medical instant retail insight system under the O2O model according to claim 2, characterized in that: The replenishment cycle is the time interval from determining that replenishment is needed to actually completing the replenishment, including order generation time, supplier confirmation time, supplier processing time, transportation time, and warehousing time; the storage environment is classified into levels based on temperature parameters, including room temperature storage, cool and dry environment, constant temperature storage, refrigerated storage, and strict cold chain storage.

4. A medical instant retail insight system under the O2O model according to claim 1, characterized in that... ； The data classification module uses the K-means clustering algorithm to divide the demand for inventory information updates into three levels: high demand, medium demand, and low demand. The clustering process includes: The collected feature data are used to form a feature matrix, and the feature values ​​are standardized. Set the number of clusters to 3, and iteratively update the cluster centers until convergence; The tiers are defined based on the average characteristics of each cluster. The clusters with the shortest delivery time, shortest replenishment cycle, fastest sales speed, fewest returns, and most stable storage environment are classified as high-demand tiers.

5. A medical real-time retail insight system under the O2O model according to claim 1, characterized in that: The process by which the data classification module calculates the comprehensive index of order impact measurement includes: Calculate the impact coefficient of order quantity, the impact coefficient of order processing time, and the impact coefficient of order variability expressed as volatility; Using the weighted summation formula ;in , , The weights are order quantity, order processing time, and order variability, respectively, with Qᵢ, Tᵢ, and Vᵢ being the corresponding influence coefficients.

6. A medical instant retail insight system under the O2O model according to claim 1, characterized in that: The fuzzy rules of the feedback adjustment module include: The input variables are the demand for inventory information updates, specifically divided into High, Medium, and Low, and the order impact measure, specifically divided into Big, Medium, and Small. The output variable is the inventory information update frequency, specifically divided into Increase, Unchanged, and Decrease. The rules include: if the demand for inventory updates is High and the order impact metric is Big, then the inventory update frequency is Increase; if the demand for inventory updates is Low and the order impact metric is Small, then the inventory update frequency is Decrease.

7. A medical instant retail insight system under the O2O model according to claim 1, characterized in that: The drug complexity is calculated by a weighted sum of the storage environment level and the drug type level, using the following formula: In the formula For the storage environment of drugs, Drug type and These are the weighting coefficients for the storage environment and the drug type, respectively, and the sum of the weighting coefficients is 1; the drug type level is divided into 1-5 points according to its medical use.

8. A medical real-time retail insight system under the O2O model according to claim 1, characterized in that: The TOPSIS method used by the data matching module includes: The decision criteria were determined to be the need for inventory information updates and the complexity of the drugs, and the inventory information update need level was converted into a numerical value of 1-3; The decision criteria data are normalized, and the ideal solution is set as the maximum value of inventory information update demand and the minimum value of drug complexity, while the negative ideal solution is set as the minimum value of inventory information update demand and the maximum value of drug complexity. Calculate the Euclidean distance between each drug and the ideal solution and the negative ideal solution, and then use a comprehensive index. In the formula and are the distances between the i-th drug and the ideal solution and the negative ideal solution, respectively.

9. A medical instant retail insight system under the O2O model according to claim 8, characterized in that: The data matching module selects the inventory information update method based on the composite index: automatic update is selected when the composite index is greater than or equal to the 66th percentile; semi-automatic update is selected when the composite index is greater than or equal to the 33rd percentile and less than the 66th percentile; and manual update is selected when the composite index is less than or equal to the 33rd percentile.

10. A medical instant retail insight system under the O2O model according to claim 1, characterized in that: The order volatility is calculated using volatility, including: Calculate the logarithmic return on the number of orders in adjacent time periods Where yi is the number of orders in the i-th time period, and ln is the natural logarithm; Where N is the total number of time periods, Logarithmic return It is the average of the logarithmic returns.

Images