A medicine purchasing plan generation method and system based on multi-dimensional classification

Abstract

The application discloses a kind of based on multi-dimensional classification's drug procurement plan generation method and system, belong to hospital drug inventory intelligent management technical field.The application fuses XYZ consumption stability classification, ABC consumption amount classification, VEN clinical importance classification three kinds of classic classification model, constructs three-dimensional drug classification system, by ABC-VEN dimension merging coupling XYZ fluctuation dimension, builds nine-part intelligent decision matrix, for different partition drug matching differentiation, custom procurement strategy, respectively configure order point procurement mode, safety stock replenishment mode, elastic control mode, simultaneously establish monthly dynamic iteration update mechanism, realize the whole process standardization, intelligentization, dynamic of drug classification standard, procurement logic, plan generation.The application effectively solves the industry pain point that traditional manual procurement mode extensive, supply and demand matching precision is low, greatly improves hospital drug inventory turnover efficiency, reduces inventory backlog cost and clinical shortage risk.

Description

Technical Field

[0001] This invention relates to the field of intelligent management technology for hospital drug inventory, specifically to a method and system for generating procurement plans by classifying drug consumption data in multiple dimensions and automatically matching differentiated procurement strategies. Background Technology

[0002] Pharmaceuticals are core supplies for clinical diagnosis, emergency treatment, and rehabilitation nursing in hospitals. The accuracy and scientific nature of pharmaceutical inventory management and procurement planning directly affect the continuity and safety of clinical care, as well as the hospital's operating costs and capital utilization. Currently, the vast majority of hospitals in China still rely on procurement managers to manually analyze and calculate pharmaceutical procurement plans based on historical consumption data and their personal experience. A standardized, intelligent, and systematic procurement decision-making mechanism has not yet been established, revealing many inherent technical deficiencies in practical application and severely hindering the improvement of hospital pharmaceutical inventory management.

[0003] Currently, hospital drug procurement plans are typically formulated manually by procurement personnel based on historical consumption data and personal experience, which presents the following technical problems: 1. Single classification dimension: Existing methods usually only classify drugs based on consumption amount (ABC classification) or clinical importance (VEN classification), without considering the stability of drug consumption (XYZ classification), resulting in classification results that cannot fully reflect the true consumption characteristics of drugs.

[0004] 2. Homogeneous procurement strategies: Using the same replenishment calculation method for drugs with different consumption characteristics leads to a situation where some drugs are stockpiled while others are out of stock, resulting in low inventory turnover efficiency.

[0005] 3. Static classification results: Once a drug classification is determined, it remains unchanged for a long time and cannot reflect the seasonal fluctuations in drug consumption and changes in clinical medication habits.

[0006] 4. Inconsistent decision-making standards: Different procurement personnel have different judgment standards for similar drugs, and the formulation of procurement plans relies on personal experience.

[0007] In summary, existing methods for generating drug procurement plans suffer from technical shortcomings such as one-sided classification, crude strategies, lack of standards, and poor dynamism. These shortcomings make it difficult to meet the development needs of modern hospitals for refined and intelligent inventory management. There is an urgent need for a multi-dimensional, differentiated, standardized, and dynamically iterative drug procurement plan generation solution. Summary of the Invention

[0008] The purpose of this invention is to overcome the technical defects of existing technologies, such as single classification dimensions of drug procurement, homogeneous procurement strategies, high dependence on manual labor, lagging dynamic response, low inventory turnover efficiency, and prominent contradictions between stockouts and backlogs, and to provide a drug procurement plan generation method based on multi-dimensional classification.

[0009] This invention provides a multi-dimensional classification method for generating drug procurement plans. By designing a decision path of "priority merging → stability classification → strategy matching → iterative correction", the standardization and dynamic generation of procurement plans can be achieved.

[0010] A method for generating drug procurement plans based on multidimensional classification includes the following steps: S1. Data Acquisition: Collect raw data on hospital drug sales within a preset statistical period. This raw data must include at least the generic name of the drug, drug specifications, manufacturer, monthly sales quantity, and monthly sales amount. The preset statistical period is 24 consecutive calendar months. Each month, the latest monthly data is included in the statistical window, while the earliest monthly data is removed, ensuring the data window always reflects the most recent 24 months. If the data completeness is less than 24 months, the shortest statistical period must be at least 12 consecutive months.

[0011] The data preprocessing stage determines and removes abnormal data according to the following quantitative rules: (1) Statistical outlier removal: Monthly data whose monthly consumption exceeds the range of ±3 times the average consumption of the drug over 24 months are marked as outliers and removed; (2) Elimination of sudden batch events: Monthly data whose monthly consumption exceeds 5 times the average monthly consumption of the drug and is marked as "emergency allocation" or "temporary batch purchase" by the system will be eliminated; (3) Business offset data removal: Inbound and outbound records marked as "return", "damage" or "inventory difference adjustment" by the hospital drug management system shall be removed; (4) Maximum removal ratio: The monthly data of a single drug that is removed shall not exceed 20% of the total number of months (i.e., a maximum of 4 months out of 24 months). If it exceeds this, the drug will not be included in the subsequent classification calculation and the data quality will be manually reviewed.

[0012] S2. Three-dimensional multidimensional classification: Based on the original data, each drug is classified into ABC consumption amount percentage and VEN clinical importance, respectively, to obtain the three-dimensional classification label of each drug. The classification results are merged into three priority categories: the first priority is category A or category V, the second priority is category B or category E, and the third priority is category C or category N. S3. Decision matrix construction: Calculate the coefficient of variation of the monthly consumption of each drug, and determine the XYZ category based on the coefficient of variation: X category is the variety with a coefficient of variation less than the first threshold, Y category is the variety with a coefficient of variation between the first and second thresholds, and Z category is the variety with a coefficient of variation greater than or equal to the second threshold. S4. Decision partition construction: Cross the priority category in step two with the XYZ classification in step three to form 9 decision partitions. Each partition corresponds to a procurement strategy and stockout tolerance. S5. Differentiated Procurement Strategy Execution: Based on the decision-making zone to which the drug belongs, match the corresponding reorder point mode, safety stock mode or flexible control mode, and calculate the core parameters of reorder point, safety stock, inventory upper and lower limits and replenishment quantity through the parameter algorithm of the corresponding mode. S6. Procurement Plan Output: Summarize all drug accounting and replenishment data, integrate basic drug information and procurement quantities, and automatically generate standardized drug procurement plan sheets.

[0013] Furthermore, the specific method for classifying the ABC consumption amount percentages in step S2 is as follows: The cumulative sales amount of all drugs within the statistical period is calculated, and the cumulative sales amount of each drug is calculated as a percentage of the total sales amount, sorted by sales amount from highest to lowest. The ABC classification thresholds can be adjusted according to the actual management needs of the hospital: the cutoff threshold for Class A drugs is selected within the range of 70% to 85%; the cutoff threshold for Class B drugs is adjusted accordingly within the range of 90% to 95%; after adjustment, the system automatically verifies the constraint that the proportion of Class A drugs to the total number of drugs in the hospital is ≤20% and the proportion of Class B drugs to the total number of drugs in the hospital is ≤30%.

[0014] Furthermore, the specific method for classifying the XYZ consumption stability in step S3 is as follows: The coefficient of variation (CV) is used to quantify the fluctuation of monthly drug consumption. The formula for calculating the coefficient of variation is: CV = STDEV.S (monthly consumption) / AVERAGE (monthly average consumption), where STDEV.S is the sample standard deviation function, AVERAGE is the mean function, the first threshold is 0.4, and the second threshold is 0.8.

[0015] Furthermore, the procurement strategies described in step S5 include: reorder point model, safety stock model, and flexible management model.

[0016] Furthermore, steps one through five are repeated monthly to update the classification and procurement strategies for each drug.

[0017] This invention also provides a drug procurement plan generation system based on multidimensional classification, the system comprising: The data acquisition module is used to collect raw data on hospital drug sales within a preset statistical period. The raw data includes at least the drug name, drug specifications, manufacturer, monthly sales quantity, and monthly sales amount. The data preprocessing module is used to preprocess the raw data according to the quantization rules described in S1 and remove abnormal data. The three-dimensional classification module is used to classify each drug by ABC consumption amount percentage and VEN clinical importance, and merge the classification results into three composite priority categories according to the union logic. The stability analysis module is used to calculate the coefficient of variation of the monthly consumption of each drug and to determine the XYZ classification based on the coefficient of variation. The decision partition matching module is used to cross the three composite priority categories with the XYZ classification to form 9 decision partitions, each partition corresponding to a procurement strategy and a security factor. The procurement strategy execution module is used to match the corresponding reorder point mode, safety stock mode or flexible control mode according to the decision-making zone to which the drug belongs, and to calculate the core parameters of reorder point, safety stock, upper and lower limits of inventory and replenishment quantity. The procurement plan output module is used to summarize the replenishment data of all medicines and generate a standardized medicine procurement plan.

[0018] The beneficial effects of this invention are as follows: 1. Achieve comprehensive and precise drug classification, eliminating blind spots in management. This invention integrates three core dimensions: stability, value, and clinical importance. It breaks through the one-sidedness of traditional single-dimensional classification, comprehensively and three-dimensionally depicting the inventory management characteristics of each drug. The classification results accurately match the actual supply and demand status of drugs, providing reliable data support for the formulation of differentiated procurement strategies and completely avoiding procurement errors caused by classification bias.

[0019] 2. Refined and differentiated procurement strategies to fundamentally resolve the contradiction between shortages and overstocking. Customized quantitative procurement algorithms are applied to drugs with different volatility characteristics, clinical priorities, and value levels. This ensures a continuous supply of stable drugs, flexible management of fluctuating drugs, priority for essential drugs, and strict control of overstocking of auxiliary drugs. It completely abandons the traditional, homogeneous, and extensive procurement model, precisely balancing the two core needs of clinical supply security and inventory cost control.

[0020] 3. The procurement process is standardized and intelligent, eliminating reliance on human experience. All classification standards, parameter calculations, strategy matching, and plan generation in this invention are executed automatically and quantitatively, without any subjective human judgment. This forms a unified, replicable, and implementable standardized operating procedure, reducing human error rates, shortening the procurement plan development cycle, allowing new employees to quickly get started, and achieving the standardized implementation of the procurement control system.

[0021] 4. Possesses dynamic iteration capabilities and is highly adaptable to various scenarios. Through a monthly data iteration and update mechanism, it adapts in real time to the cyclical and sudden fluctuations in drug demand, ensuring that procurement strategies are synchronized with market demand and clinical changes. This significantly improves procurement response capabilities in complex scenarios and meets the dynamic and refined inventory management needs of modern hospitals.

[0022] 5. Significantly improves inventory turnover efficiency and reduces hospital operating costs. This invention can significantly optimize drug inventory structure, improve capital utilization, reduce drug shelf-life loss and inventory waste, and has both clinical application value and economic operation value. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a general flowchart of the overall method of the present invention; Figure 2 This is a flowchart of the nine-region differentiated procurement strategy matching process of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0026] This invention comprises six core steps, each logically connected and progressively advancing, achieving intelligent control over the entire process from data collection, intelligent classification, strategy matching, parameter calculation, plan generation to dynamic iteration: Step 1: Standardized Data Acquisition and Preprocessing This method first completes the accurate collection and cleaning preprocessing of basic data to avoid interference from abnormal data on classification results and procurement parameters. The preset statistical period is several consecutive natural months. During monthly updates, the latest monthly data is included in the statistical window, while the earliest monthly data is removed, maintaining a continuous preset length for the data window. The preset statistical period is preferably 24 months; when data integrity is insufficient, the shortest statistical period should not be less than 12 consecutive months.

[0027] The core data collected includes basic attribute data and dynamic consumption data of all drugs sold in the hospital pharmacy, specifically the generic name of the drug, drug specifications, manufacturer, total monthly outbound quantity, and total monthly sales amount.

[0028] In the data preprocessing stage, outlier data is removed according to the following quantitative rules, while routine consumption data generated from normal clinical diagnosis and treatment are retained to ensure the authenticity and accuracy of subsequent classification calculations and parameter verification: (1) Statistical outlier removal: Monthly data whose monthly consumption exceeds the average consumption of the drug within the statistical period by ±3 times the sample standard deviation are marked as outliers and removed; (2) Elimination of sudden batch events: Monthly data whose monthly consumption exceeds 5 times the average monthly consumption of the drug and is marked as "emergency allocation" or "temporary batch purchase" by the system will be eliminated; (3) Business offset data removal: Inbound and outbound records marked as "return", "damage" or "inventory difference adjustment" by the hospital drug management system shall be removed; (4) Maximum removal ratio: The monthly data of a single drug that is removed shall not exceed 20% of the total number of months (taking 24 months as an example, that is, a maximum of 4 months to be removed). If it exceeds this, the drug will not be included in the subsequent classification calculation, and the data quality will be manually reviewed.

[0029] Step 2: Three-dimensional independent classification to comprehensively depict the characteristics of drug attributes. Based on the preprocessed standardized consumption data, each drug is independently classified and judged from three independent dimensions: consumption stability, consumption value ratio, and clinical importance, forming a three-dimensional labeling system that comprehensively covers the dynamic consumption characteristics, economic value characteristics, and clinical application characteristics of drugs.

[0030] 1. XYZ Dimension – Consumption Stability Classification This dimension quantifies the fluctuation of monthly drug consumption and predicts the predictability of drug demand. Its core function is the coefficient of variation (CV), which is used for precise quantification. The formula is: CV = STDEV.S (monthly consumption) / AVERAGE (average monthly consumption). Here, STDEV.S is the sample standard deviation function, used to statistically analyze the dispersion of monthly consumption within the statistical period; AVERAGE is the average value calculation function, used to calculate the average monthly drug consumption. A smaller CV value indicates more stable drug consumption and stronger demand regularity; a larger CV value indicates more drastic consumption fluctuations and higher difficulty in demand prediction.

[0031] Specific classification criteria: Category X drugs (stable), coefficient of variation < 0.4, monthly consumption fluctuates little, demand is highly regular and can be accurately predicted; Category Y drugs (relatively stable), 0.4 ≤ coefficient of variation < 0.8, monthly consumption fluctuates slightly and regularly, overall demand trend is predictable; Category Z drugs (volatile), coefficient of variation ≥ 0.8, monthly consumption has no fixed pattern, is greatly affected by sudden medical needs, seasonal diseases, and public health events, demand prediction is more difficult.

[0032] 2. ABC Dimensions – Classification of Consumption Amount Value This dimension is used to classify the economic value level of drugs and identify core controlled drugs and auxiliary controlled drugs. All drugs are sorted from highest to lowest according to their cumulative sales amount within the statistical period, and the cumulative sales amount of each drug is calculated as a percentage of the total drug sales amount of the entire hospital, thus achieving value stratification.

[0033] Specifically, this invention sets default classification thresholds according to the Pareto principle (80 / 20 rule): Category A (core high-value drugs) accounts for 0% to 80% of the cumulative sales amount, with few varieties and high capital proportion, and is the core target of inventory control; Category B (mid-range stable drugs) accounts for 80% to 90% of the cumulative sales amount, with moderate value and stable consumption, and is of medium control priority; Category C (low-value auxiliary drugs) accounts for 90% to 100% of the cumulative sales amount, with many varieties and low capital proportion per variety, and is under less control pressure.

[0034] Preferably, the ABC classification threshold can be adjusted according to the actual management needs of the hospital. The adjustment rules are as follows: (1) The cutoff threshold for Class A drugs can be selected within the range of 70% to 85%. Lowering the threshold (e.g., taking 70%) means narrowing the scope of core control, which is suitable for hospitals that want to reduce management efforts; raising the threshold (e.g., taking 85%) means expanding the scope of core control, which is suitable for hospitals with higher requirements for inventory capital occupation control.

[0035] (2) The cutoff threshold for Category B drugs can be adjusted within the range of 90% to 95%, so that Category A and Category B drugs together cover 85% to 95% of the total cost of hospital drug consumption.

[0036] (3) Category C drugs are all remaining drugs.

[0037] (4) After adjustment, the system will automatically verify the following constraints: the number of Class A drugs accounts for ≤20% of the total number of drugs in the hospital, and the number of Class B drugs accounts for ≤30% of the total number of drugs in the hospital, so as to ensure that the classification results conform to the management logic of ABC classification and avoid classification failure due to improper threshold settings.

[0038] The above adjustments can be made by the system automatically recommending the optimal threshold combination based on the hospital's historical inventory turnover rate data, iterating through different threshold combinations with the goal of maximizing inventory turnover rate; alternatively, hospital administrators can manually set the thresholds in the system configuration interface, and the system will verify the constraints in real time and indicate whether the thresholds are valid.

[0039] 3. VEN Dimension – Classification of Clinical Importance This dimension is based on the priority of clinical diagnosis and treatment needs, avoiding the drawbacks of simply classifying data while ignoring actual clinical needs. To eliminate the subjectivity of inconsistent manual judgment standards, this invention adopts objective quantitative scoring rules to determine VEN classification. The scoring is based on the current version of the National Essential Medicines List issued by the National Health Commission, the current version of the WHO Standard List of Essential Medicines issued by the World Health Organization (WHO), the current version of the National Basic Medical Insurance Drug List issued by the National Healthcare Security Administration, the current version of the National Key Monitored Rational Drug Use List issued by the National Health Commission, and the latest versions of clinical practice guidelines, treatment guidelines, or expert consensus issued by national professional organizations such as the National Health Commission, the Chinese Medical Association, and the Chinese Society of Clinical Oncology (CSCO), or internationally recognized authoritative institutions such as NCCN, ESC, and WHO.

[0040] The scoring rules are as follows:

[0041] If the same drug is included in multiple guidelines with different recommendation levels, the score corresponding to the highest recommendation level is used, without duplication. The scores for each category are summed to obtain the drug's overall VEN score. The classification criteria are as follows: Category V (Clinically Essential and Irreplaceable Drugs): With a comprehensive score of ≥3 points, these drugs are extremely clinically essential, with no equivalent alternatives or very limited alternatives. Shortages will seriously affect clinical diagnosis and treatment and endanger patients' lives. Category E (Clinically Essential Drugs): Comprehensive score ≥1.5 and <3. These drugs are required for routine clinical diagnosis and treatment, have certain substitutability, and shortages will affect the efficiency of diagnosis and treatment but will not cause the stagnation of diagnosis and treatment. Category N (Clinical Non-Essential Auxiliary Drugs): Overall score <1.5 points. These drugs have a low degree of clinical necessity, abundant alternative resources, and shortages have no substantial impact on clinical diagnosis and treatment.

[0042] The scoring criteria are based on the latest published version of each catalog at the time of drug classification calculation. If any catalog version is updated, the latest version after the update date shall prevail. When the inclusion status of a drug changes in different catalogs, the VEN comprehensive score will be automatically recalculated in the monthly iterative update without manual intervention.

[0043] Step 3: Multidimensional coupling to construct a nine-zone intelligent decision matrix To achieve effective integration of multi-dimensional data and avoid the limitations of single-dimensional classification, this invention adopts a matrix construction logic of "two-dimensional coupling + one-dimensional cross-referencing". First, the ABC value dimension and the VEN clinical dimension are merged and coupled to form three composite categories, matching the dual attributes of drug value and clinical application. Then, the composite categories are fully cross-referencing with the XYZ stability dimension, ultimately forming nine independent and clearly defined procurement decision partitions. Each partition corresponds to a specific procurement strategy, safety factor, and stockout tolerance.

[0044] ABC-VEN Merging Rules: Category A high-value drugs and Category V essential drugs are merged into the first composite category (high-priority essential drugs) using union logic. That is, drugs classified as Category A in ABC or Category V in VEN are included in the first composite category. Category B mid-range drugs and Category E basic essential drugs are merged into the second composite category (medium-priority routine drugs) using union logic. Category C low-value drugs and Category N non-essential drugs are merged into the third composite category (low-priority auxiliary drugs) using union logic. The classification is performed sequentially according to the first, second, and third priorities. Drugs already classified in a higher priority level are not considered in subsequent priority levels.

[0045] The nine-part decision matrix and its corresponding safety coefficient Z value constructed based on the above rules are as follows:

[0046] Table 1. Decision Matrix and Corresponding Safety Factor Z-value The stockout rates and service levels corresponding to each safety factor Z value are as follows:

[0047] The safety factor Z is obtained by using the inverse function Φ of the cumulative distribution function of the standard normal distribution. -1 (1-α) Calculation. Taking a stockout tolerance of 0.10% as an example: Service level = 1 - 0.10% = 99.90%, from the standard normal distribution table, Z≈3.09.

[0048] It is important to note that both Zone 2 and Zone 4 have a safety factor of 2.58, but their procurement strategies differ: Zone 2, for Class X stable drugs, uses a reorder point model, while Zone 4, for Class Y relatively stable drugs, uses a safety stock model. The safety factor is only an input value for parameter calculation and not the sole criterion for zone selection. The zone number is determined by a combination of XYZ stability and ABC-VEN priority.

[0049] Step 4: Regional adaptation and implementation of differentiated procurement strategies This invention uses a customized quantitative procurement algorithm to match the attribute characteristics of drugs in different regions, completely abandoning the traditional homogeneous procurement logic and achieving refined management of "one strategy per region and one standard per category". The specific execution logic and parameter calculation methods of the three procurement models are as follows: 1. Order point mode (suitable for zones 1-3, X-category stable consumable medicines) This model is suitable for drugs with highly stable monthly consumption and minimal fluctuations. It achieves precise, periodic replenishment through fixed reorder points and fixed order quantities, ensuring continuous clinical supply while minimizing inventory redundancy and reducing capital tied up. All core parameters are quantified and calculated using historical data, eliminating subjective human intervention and ensuring standardized, replicable processes.

[0050] Core parameter calculation formula: Average monthly amount = total consumption in all months within the statistical period ÷ number of valid statistical months; Standard deviation of demand = STDEV.S (monthly consumption). Safety stock = Safety factor of corresponding zone × Standard deviation of demand; Daily average quantity = Monthly average quantity ÷ 30; Procurement lead time L = the total number of days from the issuance of the purchase order to the acceptance and warehousing of the medicines, consisting of the following components: L = L1 + L2 + L3, where L1 is the order processing days (from the issuance of the purchase order to the supplier's confirmation of the order), L2 is the transit days (from the supplier's shipment to the hospital's pharmacy receiving the goods), and L3 is the warehousing and acceptance days (from the arrival of the medicines to the completion of quality inspection, warehousing, and shelf placement for sale).

[0051] Reorder point = average daily quantity × procurement lead time + safety stock; Fixed order quantity = average daily quantity × preset procurement cycle T; The procurement lead time is obtained by taking the average of the past 12 months and rounding it up, based on the actual delivery days of the same type of drugs from the same supplier in historical procurement records; if historical data is unavailable, L is set to 7 days by default.

[0052] The preset procurement cycle T is the number of days between two consecutive regular procurements, with a default value of 7 days.

[0053] Execution logic: The system monitors drug inventory data periodically. When the inventory value drops to the calculated reorder point, it automatically triggers a procurement warning and procurement instruction, generates procurement requirements according to a fixed order quantity, and does not require manual prediction. It is suitable for all routine replenishment scenarios of stable-consumption drugs.

[0054] 2. Safety Stock Mode (Adapted for 4-6 zones, for relatively stable consumption of Y-type drugs) This model targets Category Y drugs, which exhibit minor fluctuations but are generally predictable. It employs a "range-based control and replenishment to the upper limit" approach, setting upper and lower inventory limits to mitigate the risk of stockouts caused by minor demand fluctuations, while also avoiding overstocking issues resulting from excessive replenishment.

[0055] Core parameter calculation formula: Average monthly amount = total consumption in all months within the statistical period ÷ number of valid statistical months; Daily average quantity = Monthly average quantity ÷ 30; Standard deviation of demand = STDEV.S (monthly consumption). Safety stock = safety factor × standard deviation of demand; Upper limit of inventory = safety stock; Inventory minimum limit = Safety stock - Average daily quantity × Procurement lead time; The procurement lead time L = the total number of days from the issuance of the purchase order to the acceptance and warehousing of the medicines, which consists of the following components: L = L1 + L2 + L3, where L1 is the order processing days (from the issuance of the purchase order to the supplier's confirmation of the order), L2 is the transit transportation days (from the supplier's shipment to the arrival of the medicines in the hospital pharmacy), and L3 is the warehousing and acceptance days (from the arrival of the medicines to the completion of quality inspection, warehousing, and shelf placement for sale).

[0056] Actual replenishment quantity = upper limit of inventory - current real-time inventory; Preferably, considering the characteristic of small, regular fluctuations in the price of Class Y drugs, this invention also introduces a seasonal adjustment coefficient K_m to enhance the adaptability to seasonal demand. K_m is only applied to the safety stock mode (zones 4-6), and not to the reorder point mode (zones 1-3) or the flexible control mode (zones 7-9).

[0057] The calculation method for K_m is: K_m = monthly average consumption ÷ annual average monthly consumption; The monthly average consumption is calculated as follows: (sum of consumption for all months within the statistical period) ÷ (number of occurrences of that month); the annual monthly average consumption is calculated as: (total consumption within the statistical period) ÷ (total number of months).

[0058] When K_m>1, it means that the demand for the month is higher than the annual average, and the system automatically expands the upper and lower limits of inventory to cope with the peak demand; when K_m<1, it means that the demand for the month is lower than the annual average, and the system automatically reduces the inventory level to reduce backlog.

[0059] After introducing K_m, the parameters of the safety stock mode are adjusted as follows: Upper limit of inventory = safety stock × K_m; The lower limit of inventory is calculated as (safety stock - average daily quantity × procurement lead time) × K_m. When (safety stock - average daily quantity × procurement lead time) is negative, the lower limit is automatically truncated to the average daily quantity × 1 day and a supply shortage warning is triggered.

[0060] Actual replenishment quantity = upper limit of inventory - current real-time inventory; The value of K_m is limited to 0.5 to 2.0. When the calculated K_m exceeds this range, the system automatically truncates it to the boundary value (i.e., 0.5 when K_m < 0.5, and 2.0 when K_m > 2.0) and issues a data anomaly warning, prompting manual review of whether the seasonal fluctuation of the drug is reasonable.

[0061] When seasonal adjustment is not enabled, K_m = 1.0, and the above formula degenerates into the basic scheme.

[0062] Execution logic: The system periodically checks inventory data. When the real-time inventory of medicines falls below the lower limit, the replenishment process is immediately initiated. After the replenishment is completed, the inventory is restored to the upper limit level, effectively buffering small monthly demand fluctuations and balancing supply stability and inventory rationality.

[0063] 3. Flexible management mode (adapted to zones 7-9, Z-class unstable consumable drugs) This model targets Class Z drugs, whose consumption fluctuates wildly and has no fixed pattern. It abandons the fixed replenishment logic and splits the drug into two sub-models based on clinical priority, thus balancing the risks of stockouts and overstocking.

[0064] Sub-mode 1 (Interval control sub-mode): Adapted to Class Z drugs (zones 7-8) with large fluctuations but with phased patterns, adopting the replenishment logic of "interval control and replenishment to the upper limit", and adding a clinical demand confirmation step.

[0065] The parameter settings are as follows: Inventory ceiling = monthly average quantity; Inventory minimum limit = monthly average quantity × (1 - stockout tolerance); The default stockout tolerance is 40%, which means "stocking is allowed to be replenished only when inventory is depleted to 60% of its maximum limit".

[0066] Theoretical replenishment quantity = upper limit of inventory - current real-time inventory; Clinical confirmation process: After calculating the theoretical replenishment quantity, the system does not directly execute the purchase. Instead, it sends a replenishment confirmation notification to the clinical departments, including: drug name, current inventory, theoretical replenishment quantity, and calculation basis. After the clinical departments confirm their needs based on their recent treatment plans, the system uses the clinically confirmed needs as the final replenishment quantity to execute the purchase.

[0067] Execution logic: The system periodically checks inventory data. When the real-time drug inventory is lower than the lower limit, the replenishment process is initiated, the theoretical replenishment quantity is calculated, and clinical confirmation is initiated. After the clinical confirmation is completed, the replenishment is performed, and the inventory is restored to the upper limit level after the replenishment is completed.

[0068] When Submode 2 is preferred: When the consumption of Class Z drugs is extremely unstable and the monthly average amount is completely unreliable (such as zero consumption for several consecutive months or a sudden large-dose consumption in a single month), the interval control mode is not applicable, and Submode 2 (minimum guarantee mode) should be used instead.

[0069] Preferably, when the system triggers replenishment twice in a row and the clinically confirmed demand is unclear, the pharmacy department is automatically prompted to switch the drug to sub-mode two (minimum guarantee mode) to avoid continuous backlog caused by over-reliance on the average monthly quantity.

[0070] Sub-mode 2 (Minimum Guarantee Sub-mode): Suitable for Class Z drugs (including Zone 9) with highly unstable consumption and mainly driven by sudden demand. Since the monthly consumption of this type of drug has no fixed distribution pattern, the average monthly consumption does not have reliable demand prediction value, and it is not advisable to use a fixed formula based on historical data to calculate replenishment quantities.

[0071] The execution logic is as follows: (1) Triggering conditions: The system monitors drug inventory data in real time. When the inventory drops to zero or below the preset warning value (default is zero), the procurement demand confirmation process is triggered. (2) Confirmation of clinical needs: The system sends a procurement needs confirmation notice to the clinical department (or the designated personnel of the pharmacy department), and the clinical personnel shall provide feedback on the future demand quantity of the drug based on the recent treatment plan; (3) Supplier confirmation: The system simultaneously queries the corresponding supplier for the minimum order quantity and expected delivery cycle of the drug; (4) Determining the purchase quantity: The actual purchase quantity shall be the larger of the "clinical confirmation requirement" and the "supplier's minimum order quantity"; (5) Procurement execution: The system generates procurement instructions based on the confirmed procurement quantity.

[0072] If the clinic does not respond, the purchase will not be made this time, and the next trigger will still be "zero inventory". If there is no response for 3 consecutive times, the system will automatically suspend the confirmation of purchase requests for this drug and manually mark it as "suspended purchase" until the clinic actively submits a request.

[0073] Step 5: Data aggregation and generation of standardized procurement plan. After completing the regional matching, parameter calculation, and replenishment quantity calculation for all medicines, the system automatically summarizes the procurement needs for all categories, removes medicines with zero replenishment quantity or no need for procurement, and integrates key information such as medicine name, specifications, unit, current inventory, suggested procurement quantity, replenishment deadline, region, and procurement strategy type to generate a standardized and structured medicine procurement plan. The procurement plan can be directly exported as an Excel or PDF report file, supporting the entire process of procurement review, leadership approval, procurement execution, and ledger retention, with full data traceability and verification.

[0074] Step 6: Monthly iteration to achieve dynamic and intelligent updates To adapt to the dynamic changes in drug demand, this invention establishes a monthly routine iterative update mechanism. After completing the hospital-wide drug outbound settlement and data archiving each month, the system imports the latest monthly consumption data, using a preset statistical period as the data window (adding the latest monthly data while removing the oldest monthly data), and re-executes the entire process of data preprocessing, three-dimensional classification, matrix partitioning, parameter calculation, and plan generation, dynamically updating the classification label, decision partition affiliation, safety factor, inventory threshold, and procurement parameters for each drug.

[0075] Through the aforementioned monthly data iteration mechanism, the system can adapt to the following scenario changes in real time: (1) Fluctuations in drug consumption caused by seasonal epidemics; (2) Changes in drug types and dosages resulting from adjustments to the clinical department's diagnostic and treatment structure; (3) Changes in the drug catalog due to the introduction and elimination of drugs; (4) A surge in temporary demand caused by a public health emergency; (5) Fluctuations in procurement lead time due to changes in drug suppliers.

[0076] This mechanism ensures that procurement plans always align with the latest clinical needs, addressing the technical shortcomings of traditional classification methods, such as static and rigid results and delayed responses.

[0077] Implementation results: After adopting the method of the present invention, based on the historical consumption data of a preset statistical period (preferably 24 months) for classification and parameter calculation, after 6 consecutive months of rolling operation, the ending inventory turnover rate increased from 4.6% before implementation to 129.77%, and showed a gradual improvement trend; the turnover days decreased from 287.5 days to 70.46 days; and the proportion of products with a turnover days ≤30 days increased from 4.01% to 35.69%.

[0078] 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 and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for generating drug procurement plans based on multidimensional classification, characterized in that, Includes the following steps: S1. Data Acquisition: Collect raw data on hospital drug sales within a preset statistical period. The raw data includes at least the generic name of the drug, drug specifications, manufacturer, monthly sales quantity, and monthly sales amount. The preset statistical period is 24 consecutive calendar months. When updating monthly, the latest monthly data is included in the statistical window, while the earliest monthly data is removed to keep the data window always showing the most recent 24 months. When the data completeness is less than 24 months, the shortest statistical period shall not be less than 12 consecutive months. The data preprocessing stage determines and removes abnormal data according to the following quantitative rules: (1) Statistical outlier removal: Monthly data whose monthly consumption exceeds the range of ±3 times the average consumption of the drug over 24 months are marked as outliers and removed; (2) Elimination of sudden bulk events: Monthly data whose monthly consumption exceeds 5 times the average monthly consumption of the drug and is marked as "emergency allocation" or "temporary bulk purchase" by the system will be eliminated; (3) Business offset data removal: Inbound and outbound records marked as "return", "damage" or "inventory difference adjustment" by the hospital drug management system shall be removed; (4) Maximum removal ratio: The monthly data of a single drug that is removed shall not exceed 20% of the total number of months. If it exceeds this, the drug shall not be included in the subsequent classification calculation and the data quality shall be manually reviewed. S2. Three-dimensional multidimensional classification: Based on the original data, each drug is classified into ABC consumption amount percentage and VEN clinical importance, respectively, to obtain the three-dimensional classification label of each drug. The classification results are merged into three priority categories: the first priority is category A or category V, the second priority is category B or category E, and the third priority is category C or category N. S3. Decision matrix construction: Calculate the coefficient of variation of the monthly consumption of each drug, and determine the XYZ category based on the coefficient of variation: X category is the variety with a coefficient of variation less than the first threshold, Y category is the variety with a coefficient of variation between the first and second thresholds, and Z category is the variety with a coefficient of variation greater than or equal to the second threshold. S4. Decision partition construction: Cross the priority category in step two with the XYZ classification in step three to form 9 decision partitions. Each partition corresponds to a procurement strategy and stockout tolerance. S5. Differentiated Procurement Strategy Execution: Based on the decision-making zone to which the drug belongs, match the corresponding reorder point mode, safety stock mode or flexible control mode, and calculate the core parameters of reorder point, safety stock, inventory upper and lower limits and replenishment quantity through the parameter algorithm of the corresponding mode. S6. Procurement Plan Output: Summarize all drug accounting and replenishment data, integrate basic drug information and procurement quantities, and automatically generate standardized drug procurement plan sheets.

2. The method for generating a drug procurement plan based on multidimensional classification according to claim 1, characterized in that, The specific method for classifying the ABC consumption amount percentages in step S2 is as follows: The cumulative sales amount of all drugs within the statistical period is calculated, and the cumulative sales amount of each drug is calculated as a percentage of the total sales amount, sorted by sales amount from highest to lowest. The ABC classification thresholds can be adjusted according to the actual management needs of the hospital: the cutoff threshold for Class A drugs is selected within the range of 70% to 85%; the cutoff threshold for Class B drugs is adjusted accordingly within the range of 90% to 95%; after adjustment, the system automatically verifies the constraint that the proportion of Class A drugs to the total number of drugs in the hospital is ≤20% and the proportion of Class B drugs to the total number of drugs in the hospital is ≤30%.

3. The method for generating a drug procurement plan based on multidimensional classification according to claim 1, characterized in that, The specific method for classifying the XYZ consumption stability in step S3 is as follows: The coefficient of variation (CV) is used to quantify the fluctuation of monthly drug consumption. The formula for calculating the coefficient of variation is: CV = STDEV.S (monthly consumption) / AVERAGE (monthly average consumption), where STDEV.S is the sample standard deviation function, AVERAGE is the mean function, the first threshold is 0.4, and the second threshold is 0.

8.

4. The method for generating a drug procurement plan based on multidimensional classification according to claim 1, characterized in that, The procurement strategies described in step S5 include: reorder point model, safety stock model, and flexible management model.

5. The method for generating a drug procurement plan based on multidimensional classification according to claim 1, characterized in that, Steps one through five are repeated monthly to update the classification and procurement strategies for each drug.

6. A drug procurement plan generation system based on multidimensional classification, the system comprising: The data acquisition module is used to collect raw data on hospital drug sales within a preset statistical period. The raw data includes at least the drug name, drug specifications, monthly sales quantity, and monthly sales amount. The data preprocessing module is used to preprocess the raw data according to the quantization rules described in S1 and remove abnormal data. The three-dimensional classification module is used to classify each drug by ABC consumption amount percentage and VEN clinical importance, and merge the classification results into three composite priority categories according to the union logic. The stability analysis module is used to calculate the coefficient of variation of the monthly consumption of each drug and to determine the XYZ classification based on the coefficient of variation. The decision partition matching module is used to cross the three composite priority categories with the XYZ classification to form 9 decision partitions, each partition corresponding to a procurement strategy and a security factor. The procurement strategy execution module is used to match the corresponding reorder point mode, safety stock mode or flexible control mode according to the decision-making zone to which the drug belongs, and to calculate the core parameters of reorder point, safety stock, upper and lower limits of inventory and replenishment quantity. The procurement plan output module is used to summarize the replenishment data of all medicines and generate a standardized medicine procurement plan.

Images