Drug consumable SPD management system based on cloud

Abstract

The invention relates to the technical field of medicine consumable data processing, in particular to a cloud-based medicine consumable SPD management system, which comprises a cloud demand prediction module, a data processing module and a data processing module, and is characterized in that based on time period demand data and geographic area demand data of medicine consumables, the cloud demand prediction module aggregates similar demand modes in the time period demand data according to a time level, and sends the aggregated demand modes to a database; the method comprises the steps of obtaining demand aggregation data of a time period, aggregating areas in geographic area demand data into similar demand categories according to geographic position distances to obtain demand aggregation data of multiple areas, and extracting the demand aggregation data of the time period and key features of the demand aggregation data of the multiple areas. Comprising time period change features and inter-region demand similarity features, and generating a key demand feature set. According to the method, the key demand characteristics are extracted by layering and aggregating the time periods of the drug consumables and the geographic area demand data, so that the demand trends in different time periods and areas can be accurately predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of pharmaceutical consumables data processing, and in particular to a cloud-based pharmaceutical consumables SPD management system. Background Art

[0002] Drug and consumables data processing refers to the collection, storage, analysis and management of data related to drugs, drug consumables and other medical supplies in the fields of medical care, pharmaceuticals, supply chain, etc. Its technologies involve big data, Internet of Things, artificial intelligence and blockchain, etc., to achieve efficient traceability of drugs and drug consumables, inventory optimization, supply chain coordination, and clinical use monitoring, etc., and ensure timely and accurate allocation and use of medical resources by integrating data from hospitals, pharmaceutical companies, distribution companies and other parties.

[0003] Among them, the drug consumables SPD management system is specifically used for the supply and demand management of drugs and medical consumables in hospitals and medical institutions. Through automated data collection, inventory monitoring, order management, and distribution process optimization, it realizes the full life cycle management of drugs and consumables, thereby ensuring accurate supply and improving hospital management efficiency.

[0004] In the SPD management of pharmaceutical consumables, it is difficult for the existing technology to accurately grasp the changes in demand in different regions and time periods. For example, during the peak demand period, due to the inability to timely identify the differences in demand in different regions, there will be an uneven distribution of resource supply, resulting in insufficient supply in some areas during high-demand periods and inventory backlogs in low-demand areas, which not only affects the efficient use of resources, but also increases the pressure on supply chain management. In addition, the existing technology is difficult to timely identify and quantify the supply risk level of each region in terms of inventory risk control. It is impossible to set flexible inventory thresholds according to real-time demand changes, resulting in the inability to timely issue early warnings of supply risks when demand fluctuates. In the case of large demand fluctuations or supply delays, it is easy to affect the supply stability of pharmaceutical consumables. For the replenishment of pharmaceutical consumables, the system cannot dynamically adjust the replenishment frequency and quantity according to actual demand fluctuations, which makes it easy to over- or under-replenish in actual operations, resulting in unnecessary waste of resources or supply gaps. It limits the efficiency of resource allocation and inventory control, and affects the overall stability of hospital pharmaceutical consumables management and the response speed of the supply chain. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings in the prior art and propose a cloud-based pharmaceutical consumables SPD management system.

[0006] In order to achieve the above objectives, the present invention adopts the following technical solutions: The cloud-based drug consumables SPD management system includes:

[0007] The cloud demand forecasting module is based on the time period demand data and geographic area demand data of pharmaceutical consumables. It aggregates similar demand patterns in the time period demand data according to the time level to obtain the aggregated demand data of the time period. It aggregates the regions in the geographic area demand data into similar demand categories according to the geographical location distance to obtain the aggregated demand data of multiple regions. It extracts the key features of the aggregated demand data of the time period and the aggregated demand data of multiple regions, including the time period change features and the demand similarity features between regions, and generates a key demand feature set.

[0008] The cross-regional nonlinear allocation module identifies multiple demand periods in each region based on the key demand feature set, obtains a demand cycle feature map, and establishes a regional dynamic allocation parameter set for pharmaceutical consumables based on the demand similarity characteristics between regions and the currently acquired inventory data. Based on the regional dynamic allocation parameter set for pharmaceutical consumables, an optimal allocation plan for pharmaceutical consumables is defined to generate a cross-regional demand resource optimization result;

[0009] The risk control module, based on the cross-regional demand resource optimization result, calls the demand data of each type of pharmaceutical consumables in the demand cycle characteristic diagram, determines the area where insufficient inventory occurs in the specified demand cycle, obtains the supply risk area identification set, and classifies the supply risk of pharmaceutical consumables according to the supply risk area identification set to obtain the pharmaceutical consumables risk classification result;

[0010] The adaptive inventory replenishment module obtains the current inventory status and drug consumables consumption rate data, and according to the inventory threshold of the drug consumables in the drug consumables risk grading result, simulates the replenishment decision by setting the reward function, obtains the optimal replenishment plan, and calls the real-time inventory status according to the optimal replenishment plan to generate real-time inventory status update data.

[0011] As a further solution of the present invention, the step of obtaining the aggregated demand data for the time period is specifically as follows:

[0012] Based on the time period demand data of hospital drugs and consumables, the daily demand data is marked by timestamps and recorded daily, and weekly demand data is obtained by weekly aggregation, and then monthly demand data is accumulated in sequence to obtain hierarchical demand data for each time level;

[0013] Based on the stratified demand data at each time level, the demand trends of each hospital are compared, and by evaluating similar demand increases and decreases and demand fluctuations, hospitals with similar demand patterns are aggregated into similar demand categories to obtain aggregated demand data for the time period.

[0014] As a further solution of the present invention, the step of obtaining the multi-region demand aggregation data is specifically as follows:

[0015] Based on the drug and consumables usage and geographic location data of regional hospitals, the distance between hospitals is compared, and hospitals in adjacent locations are divided into similar demand categories to generate demand data aggregated in the current geographic area;

[0016] For the demand data aggregated for the current geographic area, the formula is used:

[0017]

[0018] Calculate the variance S of regional demand, quantify the degree of regional demand differences, and obtain the aggregated demand data of multiple regions;

[0019] Among them, D i is the actual demand of the ith region, is the average demand of all regions, and n is the number of regions.

[0020] As a further solution of the present invention, the step of obtaining the key requirement feature set is specifically:

[0021] Based on the aggregated data of demand in the time period and the aggregated data of demand in multiple regions, the aggregated data are converted into vectors, and key feature information of each time period is extracted through standardization and data processing to obtain an information set of time period and regional demand;

[0022] Based on the information set of the time period and regional demand, by identifying the changing trends of demand between time levels and regions, comparing the demand patterns in each region, aggregating regions with similar demands, and establishing a key demand feature set.

[0023] As a further solution of the present invention, the steps for obtaining the regional dynamic allocation parameter set of the medicines and consumables are specifically as follows:

[0024] Based on the key requirement feature set, the formula is adopted:

[0025]

[0026] Calculate the demand cycle fluctuation range T p , get the demand cycle characteristic diagram;

[0027] Among them, d i is the average daily consumption in the ith region, f i is the historical demand fluctuation frequency of the ith region, s i is the inventory replenishment period of region i;

[0028] According to the demand cycle characteristic diagram, combined with the demand similarity characteristics between regions and the currently acquired inventory data, the similarity characteristics are analyzed, the areas with consistent demand patterns are identified, the inventory level and demand characteristic information are integrated, and a regional dynamic allocation parameter set for pharmaceutical consumables is established.

[0029] As a further solution of the present invention, the step of obtaining the cross-regional demand resource optimization result is specifically as follows:

[0030] According to the regional dynamic allocation parameter set of the pharmaceutical consumables, the formula is adopted:

[0031]

[0032] Calculate the resource allocation priority P of the i-th region under the current demand state i , get the resource allocation priority analysis result;

[0033] Among them, D i is the actual demand in the ith region, U i is the demand urgency coefficient of the i-th region, C i is the existing inventory in region i, R i is the estimated replenishment of the ith region;

[0034] Based on the resource allocation priority analysis results, the urgency of regional needs is compared, the resource allocation order is determined in turn, medicines and consumables in emergency areas are allocated first, and cross-regional demand resource optimization results are generated.

[0035] As a further solution of the present invention, the steps for obtaining the risk classification results of drugs and consumables are specifically as follows:

[0036] Based on the cross-regional demand resource optimization result, the drug and consumables demand data in the demand cycle characteristic diagram is called, the peak demand and low demand time periods between regions are analyzed, and the demand intensity and inventory status are compared to generate comparative data of the demand intensity and inventory status between regions;

[0037] Based on the comparative data of demand intensity and inventory status between the regions, the formula is adopted:

[0038]

[0039] Calculate the responsiveness of drug consumables demand to cycle changes E d , and obtain the demand analysis results between regions;

[0040] Among them, ΔQ d is the change in demand, ΔT d is the amount of change in the time period;

[0041] According to the demand analysis results between the regions, the areas with insufficient inventory are identified, and a supply risk area identification set is constructed. According to the supply risk area identification set, the supply risk of pharmaceutical consumables is graded. By combining the current inventory data with the supply risk level data, the inventory threshold is set to obtain the risk grading results of pharmaceutical consumables.

[0042] As a further solution of the present invention, the step of acquiring the real-time inventory status update data is specifically as follows:

[0043] Extract the current inventory status and consumption rate of consumables, combine the inventory threshold information in the risk grading results of the pharmaceutical consumables, define the inventory state space of reinforcement learning, and establish action selections for multiple replenishment plans to generate the state and action space of the pharmaceutical inventory;

[0044] According to the state and action space of the drug inventory, the replenishment frequency and quantity are defined, the reward function is set and multiple rounds of simulation tests are conducted, the relationship between the replenishment operation and the reward is recorded, and the various operations and rewards are gradually associated through data records to establish the associated data of the replenishment operation and the reward;

[0045] According to the associated data of the replenishment operation and the reward, the optimal replenishment strategy is screened out, and according to the real-time inventory status, the corresponding replenishment operation is judged and executed. After the replenishment is completed, the real-time inventory status update data is generated.

[0046] Compared with the prior art, the advantages and positive effects of the present invention are:

[0047] In the present invention, by layering and aggregating the time cycle and geographical area demand data of pharmaceutical consumables, key demand features are extracted, so that demand trends between different time cycles and regions can be accurately predicted. Based on the dynamic changes in the demand cycle and the similarity of demand between regions, resource priority allocation is carried out to accurately allocate resources during peak demand and low demand periods in different regions, ensuring the timely supply of materials in high-demand areas, reducing supply imbalances, and reducing the possibility of inventory backlogs and resource waste. By analyzing inventory and demand changes in real time, setting flexible inventory thresholds, and performing hierarchical management based on supply risk levels, potential areas of insufficient inventory can be identified and warned in advance, making inventory management more stable and flexible. By simulating and analyzing different replenishment strategies and dynamically adjusting the replenishment quantity and frequency, the timeliness and accuracy of resource supply can be ensured. The full process management of pharmaceutical consumables is optimized, making inventory and supply chain management more efficient and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a system flow chart of the present invention;

[0049] Figure 2 A flow chart for obtaining aggregated demand data for a time period for the present invention;

[0050] Figure 3 A flow chart for obtaining aggregated demand data for multiple regions according to the present invention;

[0051] Figure 4 A flow chart for obtaining a set of key required features for the present invention;

[0052] Figure 5 A flowchart for obtaining a regional dynamic allocation parameter set for pharmaceutical consumables according to the present invention;

[0053] Figure 6 A flow chart for obtaining cross-regional demand resource optimization results for the present invention;

[0054] Figure 7 A flow chart for obtaining the risk classification results of drugs and consumables for the present invention;

[0055] Figure 8 The present invention is a flow chart for obtaining real-time inventory status update data. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0057] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0058] See also Figure 1 , the cloud-based pharmaceutical consumables SPD management system includes:

[0059] The cloud demand forecasting module processes the data in layers based on the time period demand data and geographic area demand data of drugs and consumables. The time period demand data refers to the usage of drugs and drug consumables in daily demand data, weekly demand data, and monthly demand data. The geographic area demand data refers to the usage and demand data of drugs and consumables in hospitals in multiple regions. The same demand patterns in the time period demand data are aggregated according to the time level to obtain the aggregated demand data for the time period. The regions in the geographic area demand data are aggregated into the same demand categories according to the geographical location, and the quantitative index of the similarity of the demand in each region is calculated to obtain the aggregated demand data for multiple regions. The key features are extracted by performing dimensionality reduction processing on the aggregated demand data for the time period and the aggregated demand data for multiple regions. The key features include the time period change features and the demand similarity features between regions. The key features are used to summarize the main change trends in the aggregated demand data for the time period and the aggregated demand data for multiple regions to generate a key demand feature set.

[0060] The cross-regional nonlinear allocation module analyzes the current demand for pharmaceutical consumables based on the key demand feature set, identifies the peak demand time period and low demand time period of each region according to the average daily consumption of regional pharmaceutical consumables, inventory replenishment cycle, and historical demand fluctuation frequency indicators, and obtains the demand cycle feature map. On the basis of the demand cycle feature map, combined with the demand similarity characteristics between regions and the currently acquired inventory data, a regional dynamic allocation parameter set for pharmaceutical consumables is established. Based on the regional dynamic allocation parameter set for pharmaceutical consumables, the resource allocation priority of each region under the specified demand state is calculated, and the optimal allocation strategy of pharmaceutical consumables between different times and different regions is defined to generate cross-regional demand resource optimization results;

[0061] Based on the cross-regional demand resource optimization results, the risk control module calls the demand data of each type of pharmaceutical consumables in the demand cycle characteristic diagram, compares the demand intensity and inventory status of different regions according to the identified peak demand time period and low demand time period of each region, calculates the elasticity of demand for pharmaceutical consumables and the amplitude of demand change, determines the areas where insufficient inventory may occur within the specified demand cycle, obtains the supply risk area identification set, and divides the supply risk of pharmaceutical consumables into high, medium and low risk levels according to the supply risk area identification set. According to the current inventory data and the supply risk level data, the inventory thresholds of different categories of pharmaceutical consumables are set to obtain the risk grading results of pharmaceutical consumables;

[0062] The adaptive inventory replenishment module obtains the current inventory status and drug consumables consumption speed data, establishes the action space of reinforcement learning according to the inventory threshold of drug consumables in the drug consumables risk grading results, defines the replenishment frequency and replenishment quantity according to the supply risk level data of different drug consumables, simulates the replenishment decision by setting the reward function, conducts multiple rounds of replenishment plan experiments, records the relationship between each round of replenishment operations and rewards, obtains the optimal replenishment plan by analyzing the reward signal, calls the real-time inventory status according to the optimal replenishment plan, executes the replenishment operation, and generates real-time inventory status update data;

[0063] The key demand feature set includes the main change characteristics of the aggregated demand data over a time period, the similarity characteristics of geographical area demand, and the main trends of the aggregated demand data over a region. The cross-regional demand resource optimization results include the resource priority of inter-regional demand, time allocation priority, and dynamic allocation strategies for different time periods. The drug and consumables risk grading results include the supply risk area identification, risk level classification of inventory status, and inventory warning thresholds for each type of drug and consumables. Real-time inventory status update data include the current inventory level, drug and consumables consumption rate, and the replenishment frequency and quantity for different supply risk levels.

[0064] See also Figure 2 ,The specific steps for obtaining the aggregated data of demand for a time period are:

[0065] Based on the time period demand data of hospital drugs and consumables, the daily demand data is marked by timestamps and recorded daily, and weekly demand data is obtained by weekly aggregation, and then monthly demand data is accumulated in sequence to obtain hierarchical demand data for each time level;

[0066] First, the hospital's drug and consumable usage is broken down and recorded by time dimension to generate data files at different time levels. During the data collection process, the system automatically marks the daily demand data for drugs and consumables of each hospital by timestamp, and updates and summarizes it regularly every day in the cloud server. The daily demand data of multiple hospitals are sorted out to obtain the daily aggregated demand data in layers. Based on the daily demand data, the SPD system further summarizes the data on a weekly basis, selects seven days as a cycle to accumulate the demand data to obtain weekly demand data, and summarizes the monthly demand data in this way. The layered processing of daily, weekly, and monthly data helps to form a demand structure for multiple time periods, so that further analysis of the trends and changes in demand data can be carried out later.

[0067] Based on the stratified demand data of each time level, the demand trends of each hospital are compared. By evaluating similar demand increases and decreases and demand fluctuations, hospitals with similar demand patterns are aggregated into the same demand category to obtain aggregated demand data for the time period.

[0068] In order to analyze and aggregate similar demand patterns in time period demand data, the system evaluates the similarity of daily, weekly and monthly data and classifies similar demand patterns into the same category. For example, if within a certain time period, the antibiotic demand of Hospital A and Hospital B shows a similar growth trend, and there is consistency in the demand fluctuations between the two hospitals, the system will classify the antibiotic demand patterns of the two hospitals into the same category. The specific analysis process is as follows: First, the trend curves of the antibiotic demand data of Hospital A and Hospital B are compared according to daily records, and the increase and decrease changes are quantified in percentage form. The percentage data of similar demand trends are used as the basis for classification. Then, the system integrates the data of the two hospitals with similar trends according to the frequency, cycle and amplitude of demand changes. Through trend matching and similarity scores, hospitals with highly similar demand patterns are clustered into similar demands, and the demand is further aggregated, providing an effective basis for the supply and demand analysis of drugs among multiple hospitals.

[0069] See also Figure 3 ,The specific steps for obtaining the aggregated demand data of multiple regions are as follows:

[0070] Based on the drug and consumables usage and geographic location data of regional hospitals, the distance between hospitals is compared, and hospitals in adjacent locations are divided into similar demand categories to generate demand data aggregated in the current geographic area;

[0071] First, collect the usage and demand data of drugs and consumables from hospitals in multiple regions, connect the system to the inventory and consumption records of each hospital, and update the demand of each hospital in real time. The data of each hospital is marked according to the geographical location and uploaded to the cloud database. The system automatically calculates the distance between hospitals to determine the distance of the geographical location. According to the set distance threshold (for example, setting 10 kilometers as the regional aggregation standard), the geographical locations of different hospitals are compared one by one. The system determines whether hospitals in adjacent locations meet the conditions of the same category and classifies qualified hospitals into the same area. After the regional division is completed, the data of the same area is summarized and stored.

[0072] For the demand data aggregated in the current geographic area, the formula is used:

[0073]

[0074] Calculate the variance S of regional demand, quantify the degree of regional demand differences, and obtain the aggregated demand data of multiple regions;

[0075] Among them, S is used to represent the degree of dispersion of demand between different regions. The larger the value, the greater the regional demand difference. Conversely, the closer the demand pattern is, the smaller the value is. iis the actual demand of the ith region, obtained through the demand records of hospitals in each region, and usually uploaded to the SPD system with the actual consumption of drugs and consumables in each hospital as the unit value. For example, the monthly demand of drugs and consumables in a hospital is the total consumption value directly recorded. The data is automatically uploaded and summarized through the inventory and consumption record module of the SPD system. It is the average value of all regional demands, which is used to compare the deviation of the demands in each region. The calculation formula is: i is used to represent the number of each area, and n is the number of areas, indicating the total number of areas that need to be analyzed. This value is generated based on the results of geographic location aggregation and is usually obtained through the regional division function in the SPD system.

[0076] For example, the drug demand in three areas is 120, 130 and 110 respectively, and the average demand is

[0077]

[0078] Calculate the square of the deviation of each region's demand from the average and add them up:

[0079] (120-120) 2 +(130-120) 2 +(110-120) 2 =0+100+100=200

[0080] Divide by n-1=2:

[0081]

[0082] The results show that the similarity index S = 100, assuming that the historical similarity index value is S = 50, the current index has increased compared with the historical value, indicating that the recent demand patterns in these three regions are more discrete or different than historical data. This change may indicate that the demand for drugs and consumables in hospitals in each region is deviating from the previous trend. For example, the demand in some regions may fluctuate due to seasonal factors or emergencies, resulting in a widening difference in demand.

[0083] See also Figure 4 , the specific steps for obtaining the key requirement feature set are:

[0084] Based on the aggregated data of time period demand and multi-region demand, the data is converted into vectors, and the key feature information of each time period is extracted through standardization and data processing to obtain the information set of time period and regional demand;

[0085] Based on the aggregated data of demand for time periods and the aggregated data of demand for multiple regions obtained in the previous order, the system performs dimensionality reduction processing on these data sets to reduce redundant information and extract core features. To this end, the principal component analysis (PCA) method is used to convert the daily, weekly and monthly demand data into vector form and extract the main features. First, the data of different time periods are made comparable through data standardization, and then the variance covariance matrix of each variable is obtained to extract the components with significant representativeness. The variance covariance matrix is ​​a matrix used to describe the relationship between multiple variables. It is constructed by the discreteness of each variable in the data set and the mutual changes between them. Assuming that there are drug demand data of multiple hospitals at different time points (such as daily demand, weekly demand and monthly demand of each hospital), each set of data represents the demand situation of a certain time period, and the variance covariance matrix can help observe the similarities and differences of demand data between different time periods. Specifically, the variance covariance matrix can be obtained by the following steps. First, the data is standardized so that each variable (such as daily, weekly and monthly demand) is comparable. Then, the fluctuation of demand in each time period is described by calculating the degree of dispersion (i.e., variance) of each variable itself. At the same time, the covariance between different variables is calculated, and the covariance describes the common change trend between two variables. For example, if the daily demand and the weekly demand show a trend of synchronous growth in most cases, the covariance between the two is positive; on the contrary, if the weekly demand decreases when the daily demand increases, the covariance is negative. Finally, by summarizing the variance of each variable itself and the covariance between different variables and arranging them into a matrix, the variance-covariance matrix is ​​obtained. This matrix not only reflects the independent changes in demand in each time period (variance), but also reveals the relationship between demands in different time periods (covariance), thereby providing useful information in dimensionality reduction or demand pattern analysis. Then, through eigenvalue decomposition, the system identifies and retains the key components that explain the changes in demand, gradually eliminates the components that contribute less to the overall demand pattern, and maps the data of each hospital and region into a low-dimensional feature space. During the dimensionality reduction process, the system extracts key features of the time period, including the periodic fluctuation characteristics of demand, the frequency of demand peaks and troughs, and seasonal variation characteristics. At the same time, after reducing the dimensionality of multi-regional data, the system obtains key features of demand similarity between different regions, such as the synchronization of demand between hospitals in adjacent regions under seasonal changes, specific disease cycles, etc. Through dimensionality reduction processing, the system ultimately generates a simplified feature data set that retains the main patterns of demand changes between time periods and regions, providing concise and representative feature information for subsequent demand forecasting and resource optimization allocation.

[0086] Based on the information set of time period and regional demand, by identifying the changing trend of demand between time levels and regions, comparing the demand patterns within each region, aggregating regions with similar demand, and establishing a key demand feature set;

[0087] Based on the extracted key features of the time period and multi-region aggregated data, the system further identifies and summarizes the main demand change trends at the time level and between regions. Assume that in a certain province, the system analyzes the drug demand data of two hospitals A and B with similar geographical locations. The demand data of hospitals A and B show some similar change patterns, especially in the demand for antibiotics and common cold medicines, which have similar periodic fluctuations. The following is the comparison process of the demand characteristics of hospitals A and B: Short-term fluctuations in daily demand: The system first obtains the daily antibiotic demand of the two hospitals and finds that the daily demand fluctuations of hospitals A and B in the same season are similar. For example, in winter, the daily demand for antibiotics in both hospitals increases, indicating that seasonal factors have similar effects on antibiotic demand. This short-term demand fluctuation can reflect the similar demand characteristics of the two hospitals when responding to sudden illnesses (such as influenza peaks). Seasonal changes in weekly demand: By analyzing the total weekly demand for antibiotics in hospitals A and B, the system further identifies seasonal trends. Specifically, the weekly demand of hospitals A and B in winter increases week by week, reaches a peak, and then gradually decreases to the baseline level. Similar demand patterns reflect the synchronization of demand for antibiotics in the region, which helps the system determine whether the weekly demand change characteristics of the two hospitals during the peak winter demand period are consistent. Long-term trend of monthly demand: On a longer time scale, the system analyzes the monthly demand of Hospital A and Hospital B. By comparing the monthly antibiotic demand changes of the two hospitals, it was found that the monthly demand of Hospital A and Hospital B in winter each year was higher than that in summer, and the monthly demand changes showed a regular alternation of peaks (winter) and troughs (summer). This shows that the two hospitals have the same seasonal fluctuation characteristics in their demand for antibiotics throughout the year, and can be divided into categories with similar demand patterns.

[0088] See also Figure 5 , the specific steps for obtaining the regional dynamic allocation parameter set of medicines and consumables are as follows:

[0089] Based on the key requirement feature set, the formula is adopted:

[0090]

[0091] Calculate the demand cycle fluctuation range T p , get the demand cycle characteristic diagram;

[0092] Among them, d i is the average daily consumption in the ith region, representing the average daily demand for drugs or consumables in the region. This value is directly obtained through daily inventory and consumption data records.i is the historical demand fluctuation frequency of the ith region, which is used to indicate the frequency of changes in demand and reflect the degree of cyclical fluctuations in demand. This parameter is obtained by statistically analyzing the fluctuation frequency of historical data. For example, the frequency of growth or decline in demand in each region is calculated on a quarterly or monthly basis, and the frequency of occurrence is recorded. i is the inventory replenishment cycle of the ith region, which is used to reflect the replenishment frequency of medicines and consumables in the region. The data is obtained by analyzing the inventory replenishment records of the region. The replenishment cycle of each region may be different. For example, in the specific supply plan of the hospital, some routine medicines may be replenished every week, while the replenishment cycle of emergency medicines may be longer. The replenishment cycle is determined by analyzing the time intervals between each replenishment.

[0093] For example, there are three regions of data as follows:

[0094] Area 1: average daily consumption d1 = 200, fluctuation frequency f1 = 1.3, replenishment period s1 = 7;

[0095] Area 2: average daily consumption d2 = 250, fluctuation frequency f2 = 1.1, replenishment period s2 = 5;

[0096] Area 3: average daily consumption d3 = 150, fluctuation frequency f3 = 1.5, replenishment period s3 = 10;

[0097] Get the demand index for each region:

[0098] Region 1: Region 2: Region 3:

[0099] Extract the maximum and minimum values ​​and calculate the demand cycle fluctuation range:

[0100] T p =55.00-22.50=32.50

[0101] The results show that the demand cycle fluctuation range in area 1 is 37.14, the demand cycle fluctuation range in area 2 is 55, and the demand cycle fluctuation range in area 3 is 22.5.

[0102] In the result analysis, the calculated demand index results (37.14 for region 1, 55.00 for region 2, and 22.50 for region 3) are converted into demand cycle characteristic diagrams to intuitively identify the peak demand and low demand cycles of each region. First, the demand index data of each region is imported into the visualization tool, and the time axis is set as the horizontal axis and the demand index of each region is set as the vertical axis to ensure that the high and low demand indexes can directly show the differences between regions on the graph. Draw a line chart or bar chart with regions as units to display the demand index in each time period, and set different colors or icons to distinguish regions. For example, region 2 with the highest demand index is marked in red to show that it is in peak demand, while region 1 is marked in orange to indicate medium demand, and region 3 is marked in blue to indicate low demand. Add a marking line of the demand fluctuation range on the characteristic diagram, marking the critical points of peak demand (55.00) and low demand (22.50) to observe the overall amplitude of demand fluctuations and clarify the upper and lower limits of demand in each region. Through this visual demand cycle characteristic chart, managers can view the peak and trough cycles of demand in each region at a glance, providing data support and decision-making reference for subsequent resource allocation and optimization.

[0103] According to the demand cycle characteristic diagram, combined with the demand similarity characteristics between regions and the currently acquired inventory data, the similarity characteristics are analyzed, the regions with consistent demand patterns are identified, the inventory level and demand characteristic information are integrated, and the regional dynamic allocation parameter set of drugs and consumables is established;

[0104] Based on the demand cycle characteristic diagram, combined with the demand similarity characteristics between regions and the currently acquired inventory data, the system establishes a regional dynamic allocation parameter set for pharmaceutical consumables. The system first identifies the demand fluctuations in each region based on the demand cycle characteristic diagram, and retrieves the demand similarity characteristics between regions. These characteristics are extracted from the aggregated demand data of multiple regions, indicating the similarity of demand patterns in different regions. When analyzing the similarity characteristics one by one, the system compares the demand patterns of each group of similar regions to identify whether they are synchronized during peak and low demand periods. For example, the demand for antibiotics in regions A and B also increases significantly in winter, but tends to be stable in summer. Through this pattern consistency, the system identifies the similarity characteristics of demand fluctuations in regions A and B. After analyzing the similarity characteristics, the system integrates these characteristics with the inventory data, records the inventory levels of each region one by one, including the current inventory and consumption rate, and finally forms a regional dynamic allocation parameter set.

[0105] See also Figure 6 , the specific steps for obtaining the cross-region demand resource optimization results are:

[0106] According to the regional dynamic allocation parameter set of pharmaceutical consumables, the formula is used:

[0107]

[0108] Calculate the resource allocation priority P of the i-th region under the current demand state i , get the resource allocation priority analysis result;

[0109] Among them, D i is the actual demand in the ith region, U i is the demand urgency coefficient of the i-th region, indicating the urgency of the demand in the region. This coefficient is derived from the historical demand fluctuation characteristics of the region. By analyzing the increase and decrease rates in the historical demand data, the demand peak frequency in a specific period can be identified. For example, by evaluating the demand growth changes within a quarter, the system can derive the demand urgency of the region. i is the existing inventory in the i-th region, indicating whether the current inventory level can meet the recent demand. This parameter is directly obtained from the regional inventory management system and reflects the current inventory's ability to support demand. For example, the real-time update data of the inventory system can directly display the inventory status of each region. i is the estimated replenishment quantity of the ith area, indicating the possibility of future replenishment in the area. This data is obtained through the replenishment cycle record of the area and the estimated replenishment quantity. For example, based on the booking replenishment cycle and actual replenishment records of the area, the system can calculate the inventory quantity that is expected to be replenished in the near future.

[0110] For example, the parameters of the current regions A, B, and C are as follows:

[0111] Area A: Demand D A =100, urgency coefficient U A =1.5, inventory C A =50, estimated replenishment amount R A =20, Region B: Demand D B =80, urgency coefficient U B =1.2, inventory C B =60, estimated replenishment amount R B =15, Region C: Demand D C =120, urgency coefficient U C =1.3, inventory C C =100, estimated replenishment amount R C =30.

[0112] Calculate the priority index for each region:

[0113] Region A: Region B: Region C:

[0114] The results show that region A has a priority of 2.14, region B has a priority of 1.28, and region C has a priority of 1.2.

[0115] Based on the results of resource allocation priority analysis, the urgency of regional needs is compared, and the resource allocation order is determined in sequence. Drugs and consumables in emergency areas are allocated first, generating cross-regional demand resource optimization results.

[0116] According to the priority results of the three regions, the priority of region A is 2.14, region B is 1.28, and region C is 1.20. The priority values ​​indicate that the demand in region A is the most urgent, followed by region B, and the demand in region C is relatively low. According to the priority, the system first allocates resources to region A, where the demand is the most urgent, and determines the allocation amount based on the demand and urgency demand analysis. After the allocation of region A is completed, the system allocates resources to regions B and C in turn to ensure that the needs of each region are reasonably met. Considering the possible changes in demand in different time periods, the system also sets up dynamic monitoring for region A so that resources can be adjusted in real time according to changes in demand. The system finally summarizes the resource allocation plan and generates a cross-regional resource optimization result. This result includes the specific allocation amount and subsequent adjustment plan for each region, ensuring that high-demand areas are allocated drugs and consumables first, and are flexibly adjusted when demand changes, to ensure the optimal resource allocation of drugs and consumables in different regions and different time periods.

[0117] See also Figure 7 The specific steps for obtaining the risk grading results of drugs and consumables are as follows:

[0118] Based on the cross-regional demand resource optimization results, call the drug and consumables demand data in the demand cycle characteristic diagram, analyze the peak and low demand time periods between regions, and compare the demand intensity and inventory status to generate comparative data on demand intensity and inventory status between regions;

[0119] The demand data of various types of drugs and consumables in the demand cycle characteristic diagram are called up to analyze the peak and low demand periods in different regions in detail, and compare them with the existing inventory status region by region. Assuming that regions A, B, and C are three different hospital networks, the system finds that the demand for antibiotics in region A increases significantly during the winter from December to February each year, while the demand is low in the summer. This peak demand data is displayed through the demand cycle characteristic diagram. At the same time, the system retrieves the current inventory data of region A, showing that the antibiotic inventory in this region is close to the minimum safety inventory line. The demand for disinfectants in region B peaks around the flu season each year, and the demand cycle characteristic diagram shows that this type of demand fluctuates in March and October each year. The system compares the inventory status of region B with the demand data and finds that its inventory level can basically cover the peak demand during this cycle. For region C, due to relatively stable demand and sufficient inventory, the system marks its demand intensity and inventory status as being within the low risk range.

[0120] Based on the comparative data of demand intensity and inventory status between regions, the formula is adopted:

[0121]

[0122] Calculate the responsiveness of drug consumables demand to cycle changes E d , and obtain the demand analysis results between regions;

[0123] Among them, E d It is used to measure the volatility of demand so as to set the corresponding safety stock or replenishment frequency in inventory management. d It is the change in demand, indicating the increase or decrease in the demand for pharmaceutical consumables within a specific time period. This parameter is obtained by comparing the demand for the same type of drugs in different time periods in the demand cycle characteristic diagram. d It is the change in the time cycle, which represents the interval between two selected time periods. This parameter is usually defined in the monthly, quarterly or annual demand comparison analysis. The specific time span can be selected from the demand cycle characteristic diagram.

[0124] For example, the demand for antibiotics in region A increases in winter. The demand for antibiotics in the region was 400 units in the previous quarter (3 months), while the demand in the current quarter (also 3 months) is 600 units. At this time, the demand change ΔQ d is 200 units, the time period change ΔT d For 3 months.

[0125] Calculate the responsiveness of drug consumables demand to cycle changes:

[0126]

[0127] The results showed that the antibiotic response degree to periodic changes in region A was 66.67.

[0128] According to the results of the demand analysis between regions, the areas with insufficient inventory are identified, and a supply risk area identification set is constructed. Based on the supply risk area identification set, the supply risk of drugs and consumables is graded. Combining the current inventory data with the supply risk level data, the inventory threshold is set to obtain the risk grading results of drugs and consumables;

[0129] Based on the obtained demand elasticity E d , the system identifies areas with high demand elasticity and marks areas at risk of insufficient inventory, and establishes a supply risk area identification set. For example, the antibiotic demand elasticity E in area A is d The calculated result is 66.67, while the system's demand elasticity benchmark value is 50. This indicates that the demand in region A fluctuates greatly, is sensitive to inventory changes, and has a potential risk of insufficient inventory. Next, the system sets inventory thresholds based on different supply risk levels. For high-risk areas (such as areas where the demand elasticity value exceeds the benchmark value), the system sets the inventory threshold at a higher level to ensure sufficient inventory during peak demand. For example, if the inventory threshold for antibiotics in region A is 300 units under low-risk conditions, then under high-risk identification, the antibiotic inventory threshold for the region will be adjusted to 400 units to maintain sufficient inventory during periods with higher demand elasticity.

[0130] See also Figure 8 , the specific steps for obtaining real-time inventory status update data are:

[0131] Extract the current inventory status and consumption rate of consumables, combine the inventory threshold information in the risk grading results of pharmaceutical consumables, define the inventory state space of reinforcement learning, and establish action selections for multiple replenishment plans to generate the state and action space of pharmaceutical inventory;

[0132] Define the inventory state space of the reinforcement learning agent. The system obtains the current inventory state and consumption rate data of pharmaceutical consumables from the real-time inventory management database of the pharmaceutical consumables SPD management module. First, the inventory level of each type of medicine and the average consumption rate of consumables are extracted, and on this basis, the inventory threshold information in the risk grading result is obtained. For example, the system identifies that the average daily consumption rate of antibiotics is 50 units and the inventory is 300 units, while the safety inventory threshold set in the risk grading is 200 units. Based on these data, the system sets the action space of reinforcement learning, in which each inventory state is associated with multiple possible operation options (such as different numbers of replenishment operations), forming a combination space of "current inventory and replenishment strategy". Taking antibiotics as an example, the system creates different combinations of replenishment frequencies and replenishment quantities, including "small frequency and small replenishment" and "large frequency and large replenishment". Specifically, the "small frequency and small replenishment" strategy refers to replenishing a small amount (such as 50-100 units) at a lower frequency (such as once a month) to maintain inventory levels when demand is stable or risk is low; the "large frequency and large replenishment" strategy is suitable for demand peaks or high-risk situations, and ensures inventory safety through replenishment at a higher frequency (such as once a week) and a larger quantity (such as 200-300 units). The system sets specific parameters for these replenishment strategies for each type of drug consumables based on the supply risk level of different drugs, and gradually tests the effects of these strategies in the reinforcement learning model, and finally selects the best strategy for dynamic adjustment of replenishment plans in real-time inventory management.

[0133] Based on the status and action space of drug inventory, the replenishment frequency and quantity are defined, the reward function is set and multiple rounds of simulation tests are conducted to record the relationship between replenishment operations and rewards. Through data records, multiple operations are gradually associated with rewards to establish the associated data of replenishment operations and rewards.

[0134] According to the supply risk level data of different pharmaceutical consumables, the system defines the specific replenishment frequency and replenishment quantity in the replenishment action space, and sets the reward function for the reinforcement learning model. In each round of simulation experiments, the system applies the replenishment operation to different consumables, and scores the effectiveness of the replenishment behavior according to pre-set standards. For example, when replenishment can maintain inventory close to the safety threshold, the system assigns a higher reward value to the operation. Assume that in the experiment, the operation with a replenishment frequency of "once a week" and a replenishment quantity of "150 units" can effectively meet the inventory demand of antibiotics, and the system sets its reward value to a high score. After each round of experiments, the system records the data comparison of the replenishment operation and the actual inventory changes, and gradually forms the correlation information between the replenishment operation and the reward to guide the subsequent operation optimization.

[0135] Filter out the optimal replenishment strategy based on the associated data of replenishment operations and rewards, judge and execute the corresponding replenishment operations based on the real-time inventory status, and generate real-time inventory status update data after the replenishment is completed;

[0136] By analyzing the reward signals generated during the experiment, the system automatically selects the replenishment plans that meet the high reward conditions under each demand scenario as the optimal replenishment strategy. Assuming that in the experiment, the best replenishment strategy for antibiotics is "replenish 150 units per week", the system sets this operation as the standard operation. Subsequently, in the formal inventory management process, the system automatically determines whether to trigger the optimal replenishment operation based on the current inventory status in the real-time inventory monitoring module, and executes the corresponding replenishment quantity and frequency. After the replenishment operation is completed, the system generates updated data on the inventory status, including the current inventory level, the next estimated replenishment time, the consumption status of consumables, etc., to provide the latest data support for real-time inventory management.

[0137] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. Cloud-based pharmaceutical consumables SPD management system, characterized by: The system comprises: The cloud demand forecasting module is based on the time period demand data and geographic area demand data of pharmaceutical consumables. It aggregates similar demand patterns in the time period demand data according to the time level to obtain the aggregated demand data of the time period. It aggregates the regions in the geographic area demand data into similar demand categories according to the geographical location distance to obtain the aggregated demand data of multiple regions. It extracts the key features of the aggregated demand data of the time period and the aggregated demand data of multiple regions, including the time period change features and the demand similarity features between regions, and generates a key demand feature set. The cross-regional nonlinear allocation module identifies multiple demand periods in each region based on the key demand feature set, obtains a demand cycle feature map, and establishes a regional dynamic allocation parameter set for pharmaceutical consumables based on the demand similarity characteristics between regions and the currently acquired inventory data. Based on the regional dynamic allocation parameter set for pharmaceutical consumables, an optimal allocation plan for pharmaceutical consumables is defined to generate a cross-regional demand resource optimization result; The risk control module, based on the cross-regional demand resource optimization result, calls the demand data of each type of pharmaceutical consumables in the demand cycle characteristic diagram, determines the area where insufficient inventory occurs in the specified demand cycle, obtains the supply risk area identification set, and classifies the supply risk of pharmaceutical consumables according to the supply risk area identification set to obtain the pharmaceutical consumables risk classification result; The adaptive inventory replenishment module obtains the current inventory status and drug consumables consumption rate data, and according to the inventory threshold of the drug consumables in the drug consumables risk grading result, simulates the replenishment decision by setting the reward function, obtains the optimal replenishment plan, and calls the real-time inventory status according to the optimal replenishment plan to generate real-time inventory status update data.

2. The cloud-based pharmaceutical consumables SPD management system according to claim 1, characterized in that: The specific steps for obtaining the aggregated demand data for the time period are: Based on the time period demand data of hospital drugs and consumables, the daily demand data is marked by timestamps and recorded daily, and weekly demand data is obtained by weekly aggregation, and then monthly demand data is accumulated in sequence to obtain hierarchical demand data for each time level; Based on the stratified demand data at each time level, the demand trends of each hospital are compared, and by evaluating similar demand increases and decreases and demand fluctuations, hospitals with similar demand patterns are aggregated into similar demand categories to obtain aggregated demand data for the time period.

3. The cloud-based pharmaceutical consumables SPD management system according to claim 2, characterized in that: The steps for obtaining the aggregated demand data of multiple regions are specifically as follows: Based on the drug and consumables usage and geographic location data of regional hospitals, the distance between hospitals is compared, and hospitals in adjacent locations are divided into similar demand categories to generate demand data aggregated in the current geographic area; For the demand data aggregated for the current geographic area, the formula is used: Calculate the variance S of regional demand, quantify the degree of regional demand differences, and obtain the aggregated demand data of multiple regions; Among them, D i is the actual demand of the ith region, is the average demand of all regions, and n is the number of regions.

4. The cloud-based pharmaceutical consumables SPD management system according to claim 3, characterized in that: The steps for obtaining the key requirement feature set are specifically as follows: Based on the aggregated data of demand in the time period and the aggregated data of demand in multiple regions, the aggregated data are converted into vectors, and key feature information of each time period is extracted through standardization and data processing to obtain an information set of time period and regional demand; Based on the information set of the time period and regional demand, by identifying the changing trends of demand between time levels and regions, comparing the demand patterns in each region, aggregating regions with similar demands, and establishing a key demand feature set.

5. The cloud-based pharmaceutical consumables SPD management system according to claim 4, characterized in that: The steps for obtaining the regional dynamic allocation parameter set of the medicines and consumables are specifically as follows: Based on the key requirement feature set, the formula is adopted: Calculate the demand cycle fluctuation range T p , get the demand cycle characteristic diagram; Among them, d i is the average daily consumption in the ith region, f i is the historical demand fluctuation frequency of the ith region, s i is the inventory replenishment period of region i; According to the demand cycle characteristic diagram, combined with the demand similarity characteristics between regions and the currently acquired inventory data, the similarity characteristics are analyzed, the areas with consistent demand patterns are identified, the inventory level and demand characteristic information are integrated, and a regional dynamic allocation parameter set for pharmaceutical consumables is established.

6. The cloud-based pharmaceutical consumables SPD management system according to claim 5, characterized in that: The steps for obtaining the cross-regional demand resource optimization result are specifically as follows: According to the regional dynamic allocation parameter set of the pharmaceutical consumables, the formula is adopted: Calculate the resource allocation priority P of the i-th region under the current demand state i , get the resource allocation priority analysis result; Among them, D i is the actual demand in the ith region, U i is the demand urgency coefficient of the i-th region, C i is the existing inventory in region i, R i is the estimated replenishment of the ith region; Based on the resource allocation priority analysis results, the urgency of regional needs is compared, the resource allocation order is determined in turn, medicines and consumables in emergency areas are allocated first, and cross-regional demand resource optimization results are generated.

7. The cloud-based pharmaceutical consumables SPD management system according to claim 6, characterized in that: The specific steps for obtaining the risk classification results of the drugs and consumables are as follows: Based on the cross-regional demand resource optimization result, the drug and consumables demand data in the demand cycle characteristic diagram is called, the peak demand and low demand time periods between regions are analyzed, and the demand intensity and inventory status are compared to generate comparative data of the demand intensity and inventory status between regions; Based on the comparative data of demand intensity and inventory status between the regions, the formula is adopted: Calculate the responsiveness of drug consumables demand to cycle changes E d , and obtain the demand analysis results between regions; Among them, ΔQ d is the change in demand, ΔT d is the amount of change in the time period; According to the demand analysis results between the regions, the areas with insufficient inventory are identified, and a supply risk area identification set is constructed. According to the supply risk area identification set, the supply risk of pharmaceutical consumables is graded. By combining the current inventory data with the supply risk level data, the inventory threshold is set to obtain the risk grading results of pharmaceutical consumables.

8. The cloud-based pharmaceutical consumables SPD management system according to claim 7, characterized in that: The steps for acquiring the real-time inventory status update data are specifically as follows: Extract the current inventory status and consumption rate of consumables, combine the inventory threshold information in the risk grading results of the pharmaceutical consumables, define the inventory state space of reinforcement learning, and establish action selections for multiple replenishment plans to generate the state and action space of the pharmaceutical inventory; According to the state and action space of the drug inventory, the replenishment frequency and quantity are defined, the reward function is set and multiple rounds of simulation tests are conducted, the relationship between the replenishment operation and the reward is recorded, and the various operations and rewards are gradually associated through data records to establish the associated data of the replenishment operation and the reward; According to the associated data of the replenishment operation and the reward, the optimal replenishment strategy is screened out, and according to the real-time inventory status, the corresponding replenishment operation is judged and executed. After the replenishment is completed, the real-time inventory status update data is generated.

Images