Method for intelligent prediction and management of service consumable usage

Abstract

The application provides a service consumable usage intelligent prediction and management method, and relates to the technical field of data processing.The method comprises the following steps: processing multidimensional data, distinguishing consumables into patient one-to-one consumables and shared consumables according to consumable consumption characteristics, so as to obtain a consumable classification result; obtaining a first usage prediction value according to the patient one-to-one consumables in the consumable classification result and combining real-time outpatient registration data; and obtaining a second usage prediction value according to the shared consumables in the consumable classification result and combining operation habit parameters and consumable specification variables, wherein the habit parameters reflect the stable preferences of specific departments, medical groups and medical staff when using the shared consumables, and the consumable specification variables refer to physical attribute characteristics that affect the single consumption amount of the shared consumables.The application improves the prediction accuracy of consumable usage, reduces the risk of clinical supply shortage and waste of expired consumables.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for intelligent prediction and management of service consumable usage. Background Technology

[0002] Against the backdrop of the digital transformation of medical consumables management, the traditional model generally suffers from the problem of relying on manual experience or single historical data, failing to accurately classify consumables according to their consumption characteristics, and lacking integrated analysis of multi-dimensional dynamic data, resulting in frequent prediction errors and inventory imbalances.

[0003] For example, in the management of laparoscopic surgical consumables in the operating room of a tertiary hospital, a traditional model of historical monthly average consumption plus fixed replenishment points was adopted. This model included disposable trocars for patients (whose consumption fluctuates with the number of surgical patients) and instrument sterilization packs shared by multiple surgeries (whose consumption is strongly correlated with the doctor's operation time and equipment usage frequency) in the same prediction system. However, it failed to take into account key parameters such as real-time outpatient surgery appointments and the surgeon's operating habits, and only estimated the next month's demand based on the previous month's consumption. As a result, there were multiple instances where the inventory of disposable trocars was insufficient on peak surgical days, causing delays for two surgeries, while instrument sterilization packs experienced monthly expiration losses due to overstocking. The core technical flaw was that it did not design a classification prediction method for the different characteristics of one-to-one patient consumption and shared cyclical consumption of consumables. Demand judgment relied solely on static historical data, ignoring dynamic factors such as real-time surgery appointments and operating habits. Furthermore, the inventory replenishment strategy remained unchanged and could not adapt to the fluctuations in clinical consumption, ultimately resulting in both clinical risks and resource waste. Summary of the Invention

[0004] The technical problem to be solved by this invention is a method for intelligent prediction and management of service consumable usage, which improves the accuracy of consumable usage prediction and reduces the risk of clinical supply shortages and waste of expired consumables.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] Firstly, a method for intelligent prediction and management of service consumable usage, the method comprising:

[0007] Acquire multi-dimensional raw data related to consumable usage, including diagnosis and treatment, consumable attributes, consumption records, operation logs, time context, and inventory supply chain data; process the multi-dimensional data and classify consumables into patient-to-patient consumables and shared consumables based on their consumption characteristics to obtain consumable classification results;

[0008] Based on the patient-to-patient consumables in the consumables classification results, and combined with real-time outpatient registration data, a first usage prediction value is obtained; based on the shared consumables in the consumables classification results, and combined with operation habit parameters and consumable specification variables, a second usage prediction value is obtained. Among them, the habit parameters reflect the stable preferences of specific departments, medical groups, and medical staff when using shared consumables; the consumable specification variables refer to the physical attribute characteristics that affect the single consumption of shared consumables.

[0009] The first and second dosage prediction values ​​are converted into prediction data sets respectively; based on the distribution characteristics of the prediction data sets, a prediction analysis framework is constructed through the data boundaries.

[0010] In the predictive analysis framework, a benchmark reference range is established, and the benchmark reference range is adaptively partitioned according to the data fluctuation characteristics of the predictive data set to obtain multiple sub-ranges with different data characteristics.

[0011] Each predicted data set is assigned to a corresponding sub-range, and the predicted adjustment value is calculated based on the distribution density and fluctuation range of the data elements in each sub-range. The predicted usage value is then used to correct the obtained usage prediction value to obtain the optimized consumable usage prediction result.

[0012] A dynamic inventory management strategy is developed based on the optimized consumable usage forecast results; the consumable requisition operation data is monitored in real time, and the actual consumption data is compared with the optimized forecast results to obtain an updated inventory management strategy.

[0013] Furthermore, acquire multi-dimensional raw data related to consumable usage, including data on diagnosis and treatment, consumable attributes, consumption records, operation logs, time context, and inventory supply chain. Process this multi-dimensional data, classifying consumables into patient-specific consumables and shared consumables based on their consumption characteristics to obtain consumable classification results, including:

[0014] Acquire multi-dimensional raw data on consumable usage, including diagnosis and treatment, consumable attributes, consumption records, operation logs, time context, and inventory supply chain data; clean and standardize the multi-dimensional raw data to obtain a standardized consumable usage dataset;

[0015] Based on the consumable usage dataset, the consumption characteristics of each consumable specification are extracted, and a consumption characteristic vector is constructed based on the extracted consumption characteristics.

[0016] Based on the consumption characteristic vector, the consumption correlation index and sharing index of each consumable item are calculated; according to the comparison of the consumption correlation index and sharing index with the preset threshold, each consumable item is classified into patient-to-patient consumables or shared consumables to obtain the consumable classification results.

[0017] Furthermore, based on the patient-to-patient consumables in the consumables classification results, and combined with real-time outpatient registration data, a first usage prediction value is obtained; based on the shared consumables in the consumables classification results, and combined with operational habit parameters and consumable specification variables, a second usage prediction value is obtained. The habit parameters reflect the stable preferences of specific departments, medical teams, and medical staff when using shared consumables; the consumable specification variables refer to the physical attribute characteristics that affect the single consumption amount of shared consumables, including:

[0018] Based on the patient-to-patient consumables in the consumables classification results, obtain real-time outpatient registration data; for the real-time outpatient registration data, combine the correspondence between different patient types and consumable usage recorded in the historical consumable usage records to construct a mapping table reflecting the association between patient characteristics and consumable usage; based on the patient type distribution and number in the current real-time outpatient registration data, query the patient type and consumable usage mapping table, and obtain the first predicted usage value of the patient-to-patient consumables through weighted calculation;

[0019] Based on the shared consumables in the consumables classification results, operation habit parameters are extracted from the equipment operation logs and consumables specification variables are obtained from the consumables basic information database. Based on the extracted operation habit parameters and consumables specification variables, combined with the usage patterns recorded in the historical shared consumables consumption records, an association rule reflecting the relationship between operation characteristics, consumable attributes and usage is established, and the average consumption of shared consumables per unit time is calculated as the second usage prediction value.

[0020] Furthermore, the first and second usage forecasts are converted into forecast datasets; based on the distribution characteristics of the forecast datasets, a forecast analysis framework is constructed through data boundaries, including:

[0021] The first and second dosage prediction values ​​are aggregated according to a preset time period to obtain a prediction data set containing time series characteristics.

[0022] The distribution characteristics of the forecast dataset are analyzed, and statistical descriptive indicators of the forecast dataset, including mean, variance and extreme values, are calculated to identify the distribution patterns and fluctuation characteristics of the forecast data.

[0023] Based on the distribution patterns and fluctuation characteristics obtained from the analysis, the upper and lower boundary values ​​of the predicted data are determined by the percentile calculation method, and the data boundary range is established.

[0024] A predictive analytics framework is built based on data boundary ranges. The predictive analytics framework includes data validation rules and anomaly identification mechanisms.

[0025] Furthermore, a baseline reference range is established within the predictive analysis framework. Based on the data fluctuation characteristics of the predicted dataset, the baseline reference range is adaptively partitioned to obtain multiple sub-ranges with different data characteristics, including:

[0026] Based on the data boundary range established in the predictive analysis framework, and combined with the baseline level of historical consumable usage, a benchmark reference range for measuring the fluctuation range of predicted data is determined.

[0027] Within the benchmark reference range, analyze the data fluctuation characteristics of the forecast dataset, calculate the fluctuation amplitude and fluctuation frequency indices, identify the dynamic change patterns of the forecast data, and obtain the identified data fluctuation characteristics.

[0028] Based on the identified data fluctuation characteristics, the benchmark reference range is adaptively partitioned using a dynamic threshold segmentation method, dividing data with similar fluctuation characteristics into the same interval to obtain multiple sub-ranges with different data characteristics.

[0029] Furthermore, each predicted data set is assigned to a corresponding sub-range, and a prediction adjustment value is calculated based on the distribution density and fluctuation amplitude of data elements within each sub-range. The obtained usage prediction value is then corrected using this prediction adjustment value to obtain an optimized consumable usage prediction result, including:

[0030] Based on multiple sub-ranges with different data characteristics, each data element in the prediction dataset is assigned to the corresponding sub-range according to its feature attributes, and a mapping relationship between data elements and sub-ranges is established.

[0031] By analyzing the mapping relationship between data elements and sub-ranges, the distribution of data elements in each sub-range is analyzed, and the data distribution density and fluctuation amplitude index of each sub-range are calculated to quantify the data concentration and stability characteristics of each sub-range.

[0032] Based on the centralization and stability characteristics of each sub-range data, and combined with historical prediction deviation data, a weighted algorithm is used to generate prediction adjustment values ​​corresponding to each sub-range.

[0033] The predicted adjustment value is applied to the initial usage prediction value, and the first and second usage prediction values ​​are collaboratively corrected to obtain the optimized consumable usage prediction result.

[0034] Furthermore, a dynamic inventory management strategy is formulated based on the optimized consumable usage forecast results; consumable requisition operation data is monitored in real time, and the actual consumption data is compared with the optimized forecast results to obtain an updated inventory management strategy, including:

[0035] Based on the optimized consumable usage forecast, combined with inventory cost parameters and safety stock requirements, a dynamic inventory management strategy including replenishment trigger points and replenishment quantities is formulated.

[0036] Based on the dynamic inventory management strategy, monitor consumable requisition operation data in real time and collect actual consumption data;

[0037] The actual consumption data is compared and analyzed with the optimized consumable consumption prediction results to calculate the prediction deviation rate and consumption trend change index.

[0038] Based on the prediction deviation rate and consumption trend change indicators, the parameters in the dynamic inventory management strategy are adaptively adjusted to obtain the updated inventory management strategy and output it for execution.

[0039] Secondly, the intelligent prediction and management system for service consumables usage includes:

[0040] The acquisition module is used to acquire multi-dimensional raw data related to the use of consumables, including diagnosis and treatment, consumable attributes, consumption flow, operation logs, time context and inventory supply chain data; it processes the multi-dimensional data and classifies consumables into patient-to-patient consumables and shared consumables according to the consumption characteristics to obtain consumable classification results;

[0041] The calculation module is used to obtain a first usage prediction value based on the patient-to-patient consumables in the consumables classification results and real-time outpatient registration data; and a second usage prediction value based on the shared consumables in the consumables classification results and the operation habit parameters and consumable specification variables. The operation habit parameters reflect the stable preferences of specific departments, medical groups, and medical staff when using shared consumables; the consumable specification variables refer to the physical attribute characteristics that affect the single consumption of shared consumables. The first and second usage prediction values ​​are converted into prediction datasets respectively. Based on the distribution characteristics of the prediction datasets, a predictive analysis framework is constructed through data boundaries. A benchmark reference range is established within the predictive analysis framework, and the benchmark reference range is adaptively partitioned according to the data fluctuation characteristics of the prediction datasets to obtain multiple sub-ranges with different data characteristics.

[0042] The optimization module is used to allocate each predicted data set to the corresponding sub-range, calculate the prediction adjustment value based on the distribution density and fluctuation amplitude of the data elements in each sub-range, and correct the obtained usage prediction value through the prediction adjustment value to obtain the optimized consumable usage prediction result.

[0043] The processing module is used to formulate dynamic inventory management strategies based on the optimized consumable usage forecast results; monitor consumable requisition operation data in real time, compare the actual consumption data with the optimized forecast results, and obtain the updated inventory management strategy.

[0044] Thirdly, a computing device includes:

[0045] One or more processors;

[0046] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0047] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0048] The above-described solution of the present invention has at least the following beneficial effects:

[0049] Because it employs technical means such as classifying consumables into patient-specific consumables and shared consumables based on their consumption characteristics, conducting usage forecasting by type, constructing a predictive analysis framework based on the distribution characteristics of the predicted data set and adaptively partitioning the benchmark reference range, calculating adjustment values ​​to correct the prediction results by combining the distribution density and fluctuation amplitude of sub-range data elements, and dynamically formulating inventory management strategies and updating them through real-time consumption data comparison, it overcomes the technical problems of traditional consumable management, such as mixed prediction of consumables without classification, reliance on static data leading to large prediction deviations, and rigid inventory strategies that cannot adapt to dynamic clinical needs. As a result, it achieves the technical effects of improving the accuracy of consumable usage forecasting, reducing the risk of treatment delays and consumable expiration losses caused by clinical consumable shortages, reducing inventory holding costs, and increasing inventory turnover. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating the intelligent prediction and management method for service consumables usage provided in an embodiment of the present invention.

[0051] Figure 2 This is a schematic diagram of the intelligent prediction and management system for service consumables provided in an embodiment of the present invention. Detailed Implementation

[0052] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0053] like Figure 1 As shown, embodiments of the present invention propose a method for intelligent prediction and management of service consumable usage, the method comprising the following steps:

[0054] Step 1: Obtain multi-dimensional raw data related to consumable usage, including diagnosis and treatment, consumable attributes, consumption records, operation logs, time context, and inventory supply chain data; process the multi-dimensional data and classify consumables into patient-to-patient consumables and shared consumables according to their consumption characteristics to obtain consumable classification results.

[0055] Step 2: Based on the patient-to-patient consumables in the consumables classification results, and combined with real-time outpatient registration data, obtain the first usage prediction value; based on the shared consumables in the consumables classification results, and combined with operation habit parameters and consumable specification variables, obtain the second usage prediction value. Among them, the habit parameters reflect the stable preferences of specific departments, medical groups, and medical staff when using shared consumables; the consumable specification variables refer to the physical attribute characteristics that affect the single consumption of shared consumables.

[0056] Step 3: Convert the first and second usage prediction values ​​into prediction data sets respectively; based on the distribution characteristics of the prediction data sets, construct a prediction analysis framework through data boundaries;

[0057] Step 4: Establish a benchmark reference range in the predictive analysis framework, and adaptively partition the benchmark reference range according to the data fluctuation characteristics of the predictive data set to obtain multiple sub-ranges with different data characteristics.

[0058] Step 5: Allocate each predicted data set to the corresponding sub-range, calculate the prediction adjustment value based on the distribution density and fluctuation amplitude of the data elements in each sub-range, and correct the obtained usage prediction value using the prediction adjustment value to obtain the optimized consumable usage prediction result.

[0059] Step 6: Develop a dynamic inventory management strategy based on the optimized consumable usage forecast results; monitor consumable requisition operation data in real time, compare the actual consumption data with the optimized forecast results, and obtain the updated inventory management strategy.

[0060] In this embodiment of the invention, by employing technical means such as distinguishing between patient-specific consumables and shared consumables based on their consumption characteristics, conducting usage prediction by type, constructing a predictive analysis framework based on the distribution characteristics of the predicted data set through data boundaries, adaptively partitioning the benchmark reference range according to data fluctuation characteristics, calculating prediction adjustment values ​​to correct prediction results by combining the distribution density and fluctuation amplitude of data elements in the sub-range, formulating dynamic inventory management strategies based on optimized prediction results, and updating through real-time consumption data comparison, the technical problems in traditional consumable management, such as unclassified and mixed prediction of consumables, large deviations caused by reliance on static data, and rigid inventory strategies that cannot adapt to dynamic clinical needs, are overcome. This achieves the technical effects of improving the accuracy of consumable usage prediction, reducing the risk of treatment delays and consumable expiration losses caused by clinical consumable shortages, reducing inventory holding costs, and increasing inventory turnover.

[0061] In a preferred embodiment of the present invention, step 1 above may include:

[0062] Step 1.1: Obtain multi-dimensional raw data on consumable usage, including data on diagnosis and treatment, consumable attributes, consumption records, operation logs, time context, and inventory supply chain. Clean and standardize this multi-dimensional raw data to obtain a standardized consumable usage dataset. Specifically, this includes: firstly, collecting multi-dimensional raw data on consumables related to laparoscopic surgery in the hospital operating room, including consumable usage data such as the daily quantity of disposable trocars, the department using them, the time of use, and the usage record of instrument sterilization packs, as well as the time of return and sterilization after use; patient diagnosis and treatment data such as the daily number of laparoscopic surgery appointments, the type of surgery for each patient, and the scheduled surgery time; and equipment operation data such as the daily number of times the laparoscopic equipment is started and the equipment used in a single surgery. Runtime; basic information on consumables, such as the specifications, production batch, rated number of uses, and expiration date of disposable trocars; the collected raw data is cleaned and processed, duplicate consumable requisition records are deleted, time recording errors caused by system delays in the equipment operation log are corrected, and abnormal data caused by manual input errors are removed, such as correcting the number of uses of the instrument sterilization pack from 10 to 5, which is in accordance with its rated number of uses; then standardization processing is performed, unifying the date format of all data to year, month, and day, and unifying the description of consumable specifications, such as unifying 5mm laparoscopic trocars and 5mm laparoscopic trocars as 5mm disposable laparoscopic trocars, and finally forming a standardized consumable usage data set with clear classification and unified format.

[0063] Step 1.2: Based on the consumable usage dataset, extract the consumption characteristics of each consumable specification, and construct a consumption characteristic vector based on the extracted consumption characteristics. Specifically, this includes: extracting the consumption characteristics of each consumable specification one by one based on the standardized consumable usage dataset. For disposable laparoscopic trocars, the extracted consumption characteristics include that each use corresponds to a single laparoscopic surgery for a single patient, it is used directly for the patient after issuance and is immediately discarded after use, and the number of uses is positively correlated with the number of laparoscopic surgery patients on that day. For laparoscopic instrument sterilization packs, the extracted consumption characteristics include that a single sterilization can be used for multiple laparoscopic surgeries, and the number of uses is influenced by the number of doctors. Surgical operation efficiency is affected; after use, the device must be recycled and disinfected before reuse; the number of uses is positively correlated with the total daily operating time of the laparoscopic equipment. Based on the extracted consumption characteristics, a consumption characteristic vector is constructed for each consumable specification. The vector dimensions cover the patient binding relationship, reusability, factors related to the number of uses, and post-use processing methods. The consumption characteristic vector of the disposable laparoscopic trocar is labeled as 1:1 binding with patients, non-reusable, the number of uses is related to the number of surgical patients, and it is discarded after use. The consumption characteristic vector of the laparoscopic instrument disinfection pack is labeled as many-to-many binding with patients, reusable, the number of uses is related to the equipment operating time, and it is disinfected after use.

[0064] Step 1.3: Based on the consumption characteristic vector, calculate the consumption correlation index and sharing index for each consumable specification; based on the comparison of the consumption correlation index and sharing index with preset thresholds, classify each consumable specification into patient-to-patient consumables or shared consumables to obtain consumable classification results. Specifically, this includes: based on the constructed consumption characteristic vector, calculating the consumption correlation index and sharing index for each consumable specification. When calculating the consumption correlation index, the number of patients corresponding to each use of the consumable within a certain period is statistically analyzed. For disposable laparoscopic trocars, it is statistically analyzed that each use in the past 30 days corresponds to 1 patient, and the calculated consumption correlation index is 98%; for laparoscopic instrument sterilization packs, it is statistically analyzed that each sterilization in the past 30 days is used for an average of 3 surgeries, corresponding to 3 patients, and the calculated consumption correlation index is 32%. When calculating the sharing index, the frequency of use of consumables by different operating rooms and different doctors within a certain period was statistically analyzed. For example, each disposable laparoscopic trocar was used only once by one doctor in one operating room within the past 30 days, with a sharing index of 5%. On average, each laparoscopic instrument sterilization pack was used by three doctors in two operating rooms within the past 30 days, with a sharing index of 85%. Preset thresholds were set, with a consumption correlation index threshold of 80% and a sharing index threshold of 20%. The indices of each consumable specification were compared with these thresholds. Disposable laparoscopic trocars, due to a consumption correlation of 98% (greater than or equal to 80%) and a sharing index of 5% (less than or equal to 20%), were classified as patient-to-patient consumables. Laparoscopic instrument sterilization packs, due to a consumption correlation of 32% (less than 80%) and a sharing index of 85% (greater than 20%), were classified as shared consumables, ultimately resulting in a clear consumable classification.

[0065] In this embodiment of the invention, because it employs a method of acquiring multi-dimensional raw data on consumable usage, including data on diagnosis and treatment, consumable attributes, consumption flow, operation logs, time context, and inventory supply chain, and then cleaning and standardizing the data, it uses the processed data to extract the consumption characteristics of each consumable specification and construct a consumption characteristic vector. Then, it uses this vector to calculate the consumption correlation index and sharing index of the consumable specification, and finally combines a preset threshold to classify the consumables into patient-specific consumables or shared consumables. Therefore, it overcomes the technical problems of traditional consumable classification, such as messy data, inaccurate extraction of consumption characteristics leading to ambiguous classification, and inability to clearly distinguish between patient-specific and shared consumables, thereby achieving the accuracy of consumable classification results.

[0066] In a preferred embodiment of the present invention, step 2 above may include:

[0067] Step 2.1: Based on the patient-to-patient consumables in the consumables classification results, obtain real-time outpatient registration data; for the real-time outpatient registration data, combine the correspondence between different patient types and consumable usage recorded in the historical consumable usage records to construct a mapping table reflecting the association between patient characteristics and consumable usage; based on the patient type distribution and number in the current real-time outpatient registration data, query the patient type and consumable usage mapping table, and obtain the first usage prediction value of the patient-to-patient consumables through weighted calculation, specifically including: firstly, for the patient-to-patient consumables in the consumables classification results, namely disposable laparoscopic trocars, obtain the hospital's real-time outpatient registration data for the day. The data must include information on all patients who have scheduled laparoscopic surgery, specifically covering the type of surgery scheduled by the patient, the number of patients scheduled for each type of surgery, and whether there are any special medical conditions in the remarks;

[0068] Next, the historical consumable usage records from the past six months were compiled to extract the correspondence between different patient types and the usage of disposable laparoscopic trocars. For example, statistics showed that patients undergoing laparoscopic cholecystectomy require one 5mm disposable laparoscopic trocar per surgery; patients undergoing laparoscopic appendectomy require one 5mm disposable laparoscopic trocar per surgery; and patients undergoing laparoscopic myomectomy require one 5mm disposable laparoscopic trocar and one 10mm disposable laparoscopic trocar per surgery due to the surgical procedure. Patients with special body types require an additional trocar of the appropriate size. Based on these correspondences, a mapping table reflecting the relationship between patient characteristics and consumable usage was constructed. The table clearly indicates the size of the disposable laparoscopic trocar and the number used per patient for each patient type.

[0069] Then, the distribution and number of patient types in the current real-time outpatient registration data are statistically analyzed. For example, there are 25 patients scheduled for laparoscopic cholecystectomy, 18 for laparoscopic appendectomy, and 12 for laparoscopic myomectomy on the same day, including 3 patients with special body types. Based on the data, the above mapping table is consulted, and the basic dosage is calculated according to patient type. Supplementary calculations are then performed for special cases. The first predicted dosage of disposable laparoscopic trocars for the day is obtained through weighted calculation. Specifically, the dosage for laparoscopic cholecystectomy is 25×1+1×1=26, including those for special body types; the dosage for laparoscopic appendectomy is 18×1=18; and the dosage for myomectomy is 12×(1+1)+2×1=26, including those for special body types. The total predicted dosage is 26+18+26=70.

[0070] Step 2.2: Based on the shared consumables in the consumables classification results, extract operation habit parameters from the equipment operation logs and obtain consumable specification variables from the consumables basic information database. Based on the extracted operation habit parameters and consumable specification variables, and combined with the usage patterns recorded in the historical shared consumables consumption records, establish association rules reflecting the relationship between operation characteristics, consumable attributes, and usage. Calculate the average consumption of shared consumables per unit time as the second usage prediction value. Specifically, this includes: First, for the shared consumables in the consumables classification results, namely the laparoscopic instrument sterilization pack, extract operation habit parameters from the laparoscopic equipment operation logs of the hospital equipment management. The parameters include the average single operation time of each surgeon performing laparoscopic surgery in the past 3 months, such as an average single operation of 45 minutes, Dr. Li's average single operation of 50 minutes, and Dr. Wang's average single operation of 40 minutes; the average number of times each laparoscopic device is started per day, such as a total of 30 times for 5 devices per day; the total daily operating time of the devices, such as an average of 500 minutes per day; and the proportion of additional time spent on equipment debugging and instrument replacement during surgical breaks.

[0071] At the same time, the specifications of the laparoscopic instrument sterilization packs are obtained from the hospital's basic information database of consumables. Specifically, these include the number of instrument sets contained in each sterilization pack, the number of complete laparoscopic operating instruments contained in one sterilization pack, the rated number of uses after a single sterilization, the number of consecutive laparoscopic surgeries that one sterilization pack can be used after sterilization, the duration of sterilization after each use (30 minutes per sterilization), and the effective usage period of the sterilization pack (valid for 7 days after sterilization).

[0072] Based on extracted operational habit parameters and consumable specification variables, combined with historical shared consumable consumption records from the past three months, usage patterns were analyzed and summarized. For example, statistics showed that for every 100 minutes of total daily operating time of laparoscopic equipment, two laparoscopic instrument sterilization packs were consumed; for every 5 minutes reduction in the average surgical time of the attending surgeon, each machine could complete one more surgery per day, corresponding to an additional 0.2 sterilization packs consumed. Based on these patterns, association rules reflecting operational characteristics, equipment operating time, average surgical time of doctors, consumable attributes, rated usage frequency, sterilization time, and usage were established. The rules clearly state that the total daily operating time of the equipment divided by 50 minutes equals the daily basic sterilization pack consumption, and for every 5 minutes change in the average surgical time of doctors compared to the historical average...

[0073] Finally, obtain the current operating habits parameters and consumable specifications variables. For example, if Dr. Zhang, Dr. Li, and Dr. Wang are scheduled to perform 8, 7, and 9 surgeries respectively on the same day, and the total estimated cumulative operating time of the equipment is 550 minutes, the average surgery time of the doctors is consistent with the historical average, and the consumable specifications variable remains unchanged, input the data into the above association rules to calculate the average consumption of laparoscopic instrument sterilization packs per unit time per day, i.e., 550 ÷ 50 = 11 packs. This value is used as the second usage prediction value.

[0074] In this embodiment of the invention, because a mapping table is constructed by combining real-time outpatient registration data with the historical correspondence between patient type and consumable usage for one-to-one consumables, and then the first usage prediction value is calculated by weighting the distribution and quantity of patient types in the current registration data; for shared consumables, operation habit parameters are extracted from the equipment operation log, consumable specification variables are obtained from the consumable basic information database, and association rules are established by combining historical shared consumable consumption patterns, and the average consumption per unit time is calculated as the second usage prediction value by inputting the current parameters, the technical problems of traditional consumable prediction, such as not predicting according to consumable type, relying on static historical data, and ignoring real-time outpatient registration dynamic data and equipment operation habit parameters, are overcome, thereby achieving accurate usage prediction for one-to-one consumables and shared consumables respectively, and reducing the prediction deviation of the two types of consumables.

[0075] In a preferred embodiment of the present invention, step 3 above may include:

[0076] Step 3.1: Aggregate the first and second usage prediction values ​​according to a preset time period to obtain a prediction data set containing time series characteristics. Specifically, this includes: clarifying the preset time period for the two types of consumables; combining the clinical operation rules of the operating room, aggregating the first usage prediction value of disposable trocars (patient-to-patient consumables) and the second usage prediction value of instrument sterilization packs on a daily basis as a unified time period; during aggregation, it is necessary to associate the daily surgical schedule information, binding the prediction value of disposable trocars with the number of patients scheduled for surgery that day, and binding the prediction value of instrument sterilization packs with the planned number of surgeries and equipment usage time periods for that day, ultimately forming a daily usage prediction value sequence for each type of consumable. The sequence contains the specific prediction values ​​for each day, which can intuitively reflect the time series characteristics of usage changes with date.

[0077] Step 3.2 involves performing distribution characteristic analysis on the predicted data set and calculating statistical descriptive indicators, including mean, variance, and extreme values, to identify the distribution patterns and fluctuation characteristics of the predicted data. Specifically, this includes: conducting distribution characteristic analysis on the predicted daily usage sequences of the two types of consumables; for the predicted value sequence of disposable puncture devices, calculating the mean, variance, and extreme values ​​of the daily predicted values ​​over the past three months to identify the distribution patterns of weekday peaks and weekend troughs, as well as the periodic fluctuation characteristics that change with the number of surgical appointments; for the predicted value sequence of instrument sterilization packs, similarly calculating the mean, variance, and extreme values, and combining this with the surgeon's operation time data to identify the intermittent fluctuation patterns that occur with specific surgeons' surgical days, such as a surgeon's fixed Tuesday and Thursday surgeries, and the sudden increase in usage due to increased operation time.

[0078] Step 3.3: Based on the distribution patterns and fluctuation characteristics obtained from the analysis, the upper and lower boundary values ​​of the predicted data are determined using the percentile calculation method to establish the data boundary range. Specifically, this includes: based on the identified distribution patterns and fluctuation characteristics of the two types of consumables, the upper and lower boundary values ​​of each are determined using the percentile calculation method; for disposable puncture devices, the predicted daily usage data for the past 6 months are collected, and the 95th percentile value of the dataset is calculated as the upper boundary value to ensure that 95% of the daily predicted usage will not exceed this boundary, meeting the demand on peak surgical days; the 10th percentile value of the dataset is calculated as the lower boundary value to avoid overstocking on low-usage days, such as weekends, due to the low boundary; for instrument sterilization packs, considering the more flexible fluctuation characteristics affected by operating habits, the predicted daily usage data for the past 6 months are collected, and the 90th percentile value is calculated as the upper boundary value to cover the usage demand during most peak surgical periods; the 15th percentile value is calculated as the lower boundary value to exclude interference from extreme low-usage situations such as equipment maintenance and doctors' leave, and finally, numerical ranges with clear upper and lower boundaries are established for the two types of consumables.

[0079] Step 3.4: Construct a predictive analysis framework based on the data boundary range. The predictive analysis framework includes data verification rules and an anomaly identification mechanism. Specifically, it includes: constructing a predictive analysis framework based on the established data boundary ranges of the two types of consumables. The data verification rules in the framework are set as follows: if the predicted daily usage of disposable puncture devices exceeds its upper and lower boundaries, a secondary verification of the outpatient surgery appointment data for the day is automatically triggered to check for errors in patient type statistics and surgical procedure verification; if the predicted daily usage of instrument sterilization packs exceeds its upper and lower boundaries, a secondary verification of equipment operation logs and consumable specification variables is automatically triggered; the anomaly identification mechanism in the framework is set as follows: when the predicted value of a certain type of consumable exceeds the upper boundary for 3 consecutive days, or is lower than the lower boundary for 5 consecutive days, it is automatically marked as an anomaly; at the same time, if the difference between a single predicted value and the predicted value of the same period in history exceeds 30% of the historical average, it is also marked as an anomaly.

[0080] In this embodiment of the invention, because the technical means of aggregating the first and second usage prediction values ​​according to a preset time period to obtain a prediction data set containing time series characteristics, analyzing the distribution characteristics of the set and calculating statistical descriptive indicators such as mean, variance, and extreme values, determining the upper and lower boundary values ​​of the prediction data through percentile calculation method to establish the data boundary range, and then constructing a prediction analysis framework containing data verification rules and anomaly identification mechanism based on this range, the technical problems of traditional consumable prediction, such as lack of time series integration of prediction data, lack of analysis of distribution patterns and fluctuation characteristics leading to blurred data boundaries, and difficulty in judging the rationality of prediction data due to the lack of verification and anomaly identification mechanisms, are overcome. Thus, the prediction data is made to have time dimension correlation, a clear reasonable fluctuation range of data, and the ability to effectively verify the validity of data and identify abnormal prediction values ​​are achieved.

[0081] In a preferred embodiment of the present invention, step 4 above may include:

[0082] Step 4.1: Based on the data boundary range established in the predictive analysis framework and combined with the baseline level of historical consumable usage, determine the benchmark reference range for measuring the fluctuation range of the predicted data. Specifically, this includes: determining the benchmark reference range for two types of consumables, disposable puncture devices and instrument sterilization packs. For disposable puncture devices, first retrieve the upper and lower boundary ranges established in the predictive analysis framework, then extract the actual usage data of the consumable over the past 12 months, calculate the average daily usage per month as the historical baseline level, combine the historical baseline level with the data boundary range, and take the interval within the boundary range that covers more than 90% of the actual usage of the historical baseline level as the benchmark reference range, ensuring that the range can reflect the fluctuation of routine usage and include the historical average consumption level. For instrument sterilization packs, similarly, combine their data boundary range with the historical average daily usage baseline over the past 12 months. Considering that they are more affected by operating habits, overlay the interval of 30% fluctuation above and below the historical baseline level with the data boundary range, and take the overlapping part as the benchmark reference range, so that the range not only conforms to the historical consumption pattern but also adapts to the fluctuations caused by operating habits.

[0083] Step 4.2: Within the baseline reference range, analyze the data fluctuation characteristics of the predicted data set, calculate the fluctuation amplitude and frequency indices, and identify the dynamic change patterns of the predicted data to obtain the identified data fluctuation characteristics. Specifically, within the baseline reference range of disposable puncture devices, extract the daily usage prediction data for the past 3 months, calculate the ratio of the difference between the daily predicted value and the mean of the range as the fluctuation amplitude, and count the number of days with a fluctuation amplitude exceeding 15% per week as the fluctuation frequency. Through analysis, it was found that the fluctuation amplitude is generally higher from Monday to Friday and lower on weekends, and the fluctuation frequency shows a pattern of dense fluctuations during the week and sparse fluctuations on weekends. Thus, the dynamic fluctuation characteristics of its weekly changes with the outpatient surgery appointment volume were identified. Within the baseline reference range of instrument sterilization packs, the daily usage prediction data for the past 3 months were also extracted, and the fluctuation amplitude and frequency were calculated. Combined with the analysis of the surgeon's surgical schedule, it was found that the fluctuation amplitude increased significantly when a specific doctor was on duty, and the fluctuation frequency was concentrated on the fixed surgical days of these doctors. Thus, the dynamic fluctuation characteristics concentrated with the operation time of specific doctors were identified.

[0084] Step 4.3: Based on the identified data fluctuation characteristics, the benchmark reference range is adaptively partitioned using a dynamic threshold segmentation method. Data with similar fluctuation characteristics are grouped into the same interval, resulting in multiple sub-ranges with different data characteristics. Specifically, based on the weekly fluctuation characteristics identified by the disposable puncture device, a dynamic threshold segmentation method is used to set segmentation thresholds on a weekly basis. The interval with fluctuation amplitude exceeding 20% ​​from Monday to Friday is divided into a high-fluctuation sub-range, the interval with fluctuation amplitude between 5% and 20% is divided into a medium-fluctuation sub-range, and the interval with fluctuation amplitude below 5% on weekends is divided into a low-fluctuation sub-range, so that each sub-range corresponds to the fluctuation characteristics of different time periods within the week. For the concentrated fluctuation characteristics of the instrument sterilization pack during operation periods, the operation day of the surgeon is used as the segmentation basis. The interval with fluctuation amplitude exceeding 25% on the operation day of a specific surgeon is divided into an operation-intensive sub-range, and the interval with fluctuation amplitude below 10% on other days is divided into an operation-smooth sub-range, so that each sub-range accurately matches the fluctuation characteristics under different operation intensities, achieving adaptive partitioning of the benchmark reference range.

[0085] In this embodiment of the invention, the technical means of determining the benchmark reference range by combining the data boundary range of the predictive analysis framework with the historical consumable usage benchmark level, identifying the dynamic change pattern by analyzing the fluctuation amplitude and frequency of the predictive data set, and finally using the dynamic threshold segmentation method to adaptively partition the benchmark reference range, overcomes the technical problems of traditional consumable prediction management that rely solely on fixed thresholds to divide the range, cannot adapt to the differentiated fluctuation patterns of different consumables, and ignores the dynamic changes in usage, resulting in a coarse range division. This achieves a benchmark reference range that is more in line with the actual consumable consumption characteristics, and each sub-range can accurately match the prediction data with different fluctuation characteristics, reducing prediction deviations caused by unreasonable range division.

[0086] In a preferred embodiment of the present invention, step 5 above may include:

[0087] Step 5.1: Based on multiple sub-ranges with different data characteristics, assign each data element in the prediction dataset to the corresponding sub-range according to its characteristic attributes, and establish a mapping relationship between data elements and sub-ranges. Specifically, for the prediction datasets of disposable puncture devices and instrument sterilization packs, data is allocated according to the different data characteristic sub-ranges obtained. For disposable puncture devices, the daily predicted values ​​in their prediction datasets need to be matched to sub-ranges according to their characteristic attributes. If a day is Monday and the predicted value fluctuation exceeds 20%, the data element is assigned to the high fluctuation sub-range; if a day is Wednesday and the fluctuation is between 5% and 20%, it is assigned to the medium fluctuation sub-range; if a day is Sunday and the fluctuation is less than 5%, it is assigned to the low fluctuation sub-range. At the same time, the association between the data element and the corresponding sub-range is recorded to establish a clear mapping relationship. For instrument sterilization packs, if a day is the surgery day of a specific surgeon and the predicted value fluctuation exceeds 25%, the data element is assigned to the operation-intensive sub-range; for other days and the fluctuation is less than 10%, it is assigned to the operation-smooth sub-range. Similarly, a mapping relationship between data elements and sub-ranges is established.

[0088] Step 5.2: Analyze the distribution of data elements within each sub-range based on the mapping relationship between data elements and sub-ranges. Calculate the data distribution density and fluctuation amplitude indices for each sub-range to quantify the data concentration and stability characteristics of each sub-range. Specifically, this includes: Based on the established mapping relationship, analyze the distribution of data elements within each sub-range. For the high-fluctuation sub-range of disposable puncture devices, count the number of all data elements within the sub-range and divide it by the length of the sub-range's numerical interval to obtain the data distribution density. Higher density indicates more concentrated data within that interval. Calculate the percentage deviation between the data element and the sub-range mean to obtain the fluctuation amplitude index. A smaller index indicates stronger data stability. For fluctuating and low-fluctuation sub-ranges, calculate the distribution density and fluctuation amplitude in the same way to quantify the data concentration of each sub-range. For example, a high-fluctuation sub-range may have moderate density and poor stability. For the operation-intensive sub-range of instrument sterilization packs, calculate the data distribution density and fluctuation amplitude. The operation-smooth sub-range exhibits high density and low fluctuation amplitude characteristics.

[0089] Step 5.3: Based on the data concentration and stability characteristics of each sub-range, and combined with historical prediction deviation data, a weighted algorithm is used to generate the corresponding prediction adjustment value for each sub-range. Specifically, this includes: combining the data concentration, stability characteristics, and historical prediction deviation data of each sub-range, a weighted algorithm is used to generate the corresponding prediction adjustment value. For the high-fluctuation sub-range of disposable puncture devices, historical data shows that the predicted value of the sub-range is often lower than the average deviation of actual usage by 15%, and the data concentration is moderate with poor stability. Therefore, historical deviation is given a high weight, generating an upward adjustment value of 12%. Among them, the fluctuation sub-range has a smaller historical deviation of 5% on average, a high data concentration, and good stability, and is given a lower weight, generating an upward adjustment value of 3%. The low-fluctuation sub-range has a historical prediction slightly higher than the actual average deviation of -3%, and the data is concentrated and stable, generating a downward adjustment value of 2%. For the operation-intensive sub-range of instrument sterilization packs, historical predictions are often higher on average with a deviation of -8%, the data concentration is moderate with poor stability, generating a downward adjustment value of 6%. The operation-smooth sub-range has a smaller historical deviation of 2% on average, and the data is concentrated and stable, generating an upward adjustment value of 1%.

[0090] Step 5.4 involves applying the adjusted predicted values ​​to the initial usage forecast values, and collaboratively correcting the first and second usage forecast values ​​to obtain the optimized consumable usage forecast results. Specifically, this includes:

[0091] In this embodiment of the invention, the technical means of first allocating predicted data elements according to attributes and establishing mapping relationships based on different data feature sub-ranges, then analyzing the data distribution density and fluctuation amplitude of each sub-range to quantify the degree of concentration and stability, then combining historical prediction deviations to generate the prediction adjustment value of the corresponding sub-range through a weighted algorithm, and finally co-correcting the first usage prediction value of the patient's one-to-one consumables and the second usage prediction value of the shared consumables, overcomes the technical problems of traditional consumable prediction that do not distinguish data features, only uniformly correct the prediction value, resulting in insufficient targeting, ignoring the differences in the degree of concentration and stability of sub-range data, not combining historical deviations for dynamic adjustment, and being unable to adapt to the different consumption characteristics of the two types of consumables. Thus, the technical effects of more accurate correction of the predicted data of each sub-range, the ability of the usage prediction results of the two types of consumables to match their own consumption patterns, and the reduction of the overall prediction error are achieved.

[0092] In a preferred embodiment of the present invention, step 6 above may include:

[0093] Step 6.1: Based on the optimized consumable usage forecast, combined with inventory cost parameters and safety stock requirements, formulate a dynamic inventory management strategy that includes replenishment trigger points and replenishment quantities. Specifically, this includes: for two types of consumables, disposable trocars and instrument sterilization packs, formulate dynamic strategies based on the optimized usage forecast, combined with inventory cost parameters and safety stock requirements. For disposable trocars, the optimized forecast shows that their daily usage fluctuates between 15 and 30 units. The inventory cost parameters include a procurement and transportation cost of 200 yuan per batch, a storage cost of 0.5 yuan per unit per day, and an expiration loss cost of 50 yuan per unit. The safety stock requirement needs to cover 2 days of emergency surgical demand, calculated based on a maximum daily forecast of 30 units. Replenishment triggers are set accordingly. The replenishment trigger point is set at 50 units, with a safety stock of 80 units (60 units). Replenishment response time is reserved. The replenishment quantity is calculated by subtracting the current inventory from the total predicted usage for the next 3 days. For example, if the current inventory is 40 units and the predicted usage for the next 3 days is 70 units, then 30 units will be replenished. For medical device sterilization packs, the optimized daily predicted usage is 8 to 15 packs. Inventory cost parameters include a purchase cost of 30 yuan per pack, a storage cost of 0.2 yuan per pack per day, and an expiration cost of 20 yuan per pack. The safety stock is required to cover 3 days of demand, calculated at a maximum of 15 packs per day. The replenishment trigger point is set at 40 packs, with a safety stock of 85% of 45 packs. The replenishment quantity is calculated by subtracting the current inventory from the total predicted usage for the next 5 days. For example, if the current inventory is 35 packs and the predicted usage for the next 5 days is 60 packs, then 25 packs will be replenished.

[0094] Step 6.2: Based on the dynamic inventory management strategy, monitor the consumable requisition operation data in real time and collect actual consumption data. Specifically, this includes: updating the monitoring frequency set by the dynamic inventory management strategy every hour to track the requisition operation data of disposable trocars and instrument sterilization packs in real time; automatically recording the requisition time, quantity, corresponding surgery number, and patient information when a nurse requisitions a disposable trocar before surgery; simultaneously recording the requisitioning department, equipment number, and operating physician information when requisitioning instrument sterilization packs; summarizing the requisition data every hour, and generating actual consumption data for the two types of consumables after deducting unused consumables returned that day. For example, from 9:00 to 10:00 on a given day, 8 disposable trocars were requisitioned, none were returned, and 8 were actually consumed; 5 instrument sterilization packs were requisitioned, 1 unopened pack was returned, and 4 packs were actually consumed.

[0095] Step 6.3 involves comparing and analyzing the actual consumption data with the optimized consumable usage prediction results, calculating the prediction deviation rate and consumption trend change indicators. Specifically, this includes: at the end of each day, comparing the actual daily consumption data of disposable puncture devices and instrument sterilization packs with the optimized daily usage prediction results one by one. For disposable puncture devices, if the optimized daily usage prediction is 22 and the actual consumption is 25, the prediction deviation rate is calculated as follows: actual consumption minus optimized daily usage, divided by optimized daily usage, multiplied by 100%. Data from the past 7 days shows that the deviation rate between actual consumption and the predicted value gradually increased from 3% to 13.6%, indicating a continuous upward trend in consumption. For instrument sterilization packs, the optimized predicted usage was 12 packs, while the actual consumption was 10 packs, resulting in a deviation rate of -16.7%. Data from the past 7 days shows that the deviation rate decreased from -5% to -16.7%, and the daily actual consumption was lower than the predicted value, confirming a continuous downward trend in consumption. At the same time, the deviation rate for all days of the month is summarized at the end of each month to calculate the monthly average deviation rate, which serves as a basis for judging the long-term trend.

[0096] Step 6.4: Based on the prediction deviation rate and consumption trend change index, adaptively adjust the parameters in the dynamic inventory management strategy to obtain the updated inventory management strategy and output it for execution. Specifically, this includes: adaptively adjusting the dynamic inventory management strategy parameters for the two types of consumables according to the calculated prediction deviation rate and consumption trend change index; for disposable puncture devices, the monthly average deviation rate is 12% and the trend continues to rise, indicating that the original safety stock may be insufficient. The safety stock is increased from 60 to 70, and the corresponding replenishment trigger point is adjusted from 50 to 56, which is 80% of 70. At the same time, the calculation cycle for replenishment quantity is extended from 3 days to 4 days to avoid frequent replenishment and increased costs. For medical device sterilization packs, the monthly average deviation rate was -15% and the trend continued to decline, indicating that the original safety stock was too high, resulting in capital occupation. The safety stock was reduced from 45 packs to 35 packs, the replenishment trigger point was adjusted from 40 packs to 29.75 packs, and the replenishment quantity calculation cycle was shortened from 5 days to 4 days to reduce the risk of expiration caused by excessive inventory. The adjusted inventory management strategy was output to the purchasing department and inventory management in real time. The purchasing department initiated replenishment according to the new parameters, and the inventory level was monitored according to the new trigger point to ensure that the strategy was implemented.

[0097] In this embodiment of the invention, a dynamic inventory management strategy is adopted, which includes replenishment trigger points and replenishment quantities, based on optimized consumable usage prediction results combined with inventory cost parameters and safety stock requirements. This strategy is then used to monitor consumable requisition operation data in real time to collect actual consumption data. By comparing and analyzing the actual consumption data with the optimized prediction results, the prediction deviation rate and consumption trend change indicators are calculated. Finally, the strategy parameters are adaptively adjusted based on these indicators, and the execution is output. Therefore, this overcomes the technical problems of traditional consumable inventory management, such as fixed replenishment points and replenishment quantities failing to adapt to clinical consumption fluctuations, the accumulation of prediction deviations due to lack of real-time tracking of actual consumption, and rigid inventory strategies failing to cope with changes in surgical volume or operating habits. This achieves the technical effect of dynamically matching the actual consumption patterns of consumables, reducing surgical delays caused by stockouts and expired losses due to overstocking, and ensuring a stable supply of clinical consumables while reducing inventory holding costs.

[0098] like Figure 2 As shown, embodiments of the present invention also provide an intelligent prediction and management system for service consumables usage, including:

[0099] The acquisition module is used to acquire multi-dimensional raw data related to the use of consumables, including diagnosis and treatment, consumable attributes, consumption flow, operation logs, time context and inventory supply chain data; it processes the multi-dimensional data and classifies consumables into patient-to-patient consumables and shared consumables according to the consumption characteristics to obtain consumable classification results;

[0100] The calculation module is used to obtain a first usage prediction value based on the patient-to-patient consumables in the consumables classification results and real-time outpatient registration data; and a second usage prediction value based on the shared consumables in the consumables classification results and the operation habit parameters and consumable specification variables. The operation habit parameters reflect the stable preferences of specific departments, medical groups, and medical staff when using shared consumables; the consumable specification variables refer to the physical attribute characteristics that affect the single consumption of shared consumables. The first and second usage prediction values ​​are converted into prediction datasets respectively. Based on the distribution characteristics of the prediction datasets, a predictive analysis framework is constructed through data boundaries. A benchmark reference range is established within the predictive analysis framework, and the benchmark reference range is adaptively partitioned according to the data fluctuation characteristics of the prediction datasets to obtain multiple sub-ranges with different data characteristics.

[0101] The optimization module is used to allocate each predicted data set to the corresponding sub-range, calculate the prediction adjustment value based on the distribution density and fluctuation amplitude of the data elements in each sub-range, and correct the obtained usage prediction value through the prediction adjustment value to obtain the optimized consumable usage prediction result.

[0102] The processing module is used to formulate dynamic inventory management strategies based on the optimized consumable usage forecast results; monitor consumable requisition operation data in real time, compare the actual consumption data with the optimized forecast results, and obtain the updated inventory management strategy.

[0103] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligent prediction and management of service consumable usage, characterized in that, The method includes: Acquire multi-dimensional raw data related to consumable usage, including diagnosis and treatment, consumable attributes, consumption records, operation logs, time context, and inventory supply chain data; process the multi-dimensional data and classify consumables into patient-to-patient consumables and shared consumables based on their consumption characteristics to obtain consumable classification results; Based on the patient-to-patient consumables in the consumables classification results, and combined with real-time outpatient registration data, a first usage prediction value is obtained; based on the shared consumables in the consumables classification results, and combined with operation habit parameters and consumable specification variables, a second usage prediction value is obtained. Among them, the habit parameters reflect the stable preferences of specific departments, medical groups, and medical staff when using shared consumables; the consumable specification variables refer to the physical attribute characteristics that affect the single consumption of shared consumables. The first and second dosage prediction values ​​are converted into prediction data sets respectively; based on the distribution characteristics of the prediction data sets, a prediction analysis framework is constructed through the data boundaries. In the predictive analysis framework, a benchmark reference range is established, and the benchmark reference range is adaptively partitioned according to the data fluctuation characteristics of the predictive data set to obtain multiple sub-ranges with different data characteristics. Each predicted data set is assigned to a corresponding sub-range, and the predicted adjustment value is calculated based on the distribution density and fluctuation range of the data elements in each sub-range. The predicted usage value is then used to correct the obtained usage prediction value to obtain the optimized consumable usage prediction result. A dynamic inventory management strategy is developed based on the optimized consumable usage forecast results; the consumable requisition operation data is monitored in real time, and the actual consumption data is compared with the optimized forecast results to obtain an updated inventory management strategy.

2. The intelligent prediction and management method for service consumables usage according to claim 1, characterized in that, Acquire multi-dimensional raw data related to consumable usage, including diagnosis and treatment, consumable attributes, consumption records, operation logs, time context, and inventory supply chain data; Multi-dimensional data is processed, and consumables are categorized into patient-specific consumables and shared consumables based on their consumption characteristics to obtain consumable classification results, including: Acquire multi-dimensional raw data on consumable usage, including diagnosis and treatment, consumable attributes, consumption records, operation logs, time context, and inventory supply chain data; clean and standardize the multi-dimensional raw data to obtain a standardized consumable usage dataset; Based on the consumable usage dataset, the consumption characteristics of each consumable specification are extracted, and a consumption characteristic vector is constructed based on the extracted consumption characteristics. Based on the consumption characteristic vector, the consumption correlation index and sharing index of each consumable item are calculated; according to the comparison of the consumption correlation index and sharing index with the preset threshold, each consumable item is classified into patient-to-patient consumables or shared consumables to obtain the consumable classification results.

3. The intelligent prediction and management method for service consumables usage according to claim 2, characterized in that, Based on the patient-to-patient consumables in the consumables classification results, and combined with real-time outpatient registration data, a first usage prediction value is obtained; based on the shared consumables in the consumables classification results, and combined with operational habit parameters and consumable specification variables, a second usage prediction value is obtained. The operational habit parameters reflect the stable preferences of specific departments, medical teams, and medical staff when using shared consumables; the consumable specification variables refer to the physical attribute characteristics that affect the single consumption volume of shared consumables, including: Based on the patient-to-patient consumables in the consumables classification results, obtain real-time outpatient registration data; for the real-time outpatient registration data, combine the correspondence between different patient types and consumable usage recorded in the historical consumable usage records to construct a mapping table reflecting the association between patient characteristics and consumable usage; based on the patient type distribution and number in the current real-time outpatient registration data, query the patient type and consumable usage mapping table, and obtain the first predicted usage value of the patient-to-patient consumables through weighted calculation; Based on the shared consumables in the consumables classification results, operation habit parameters are extracted from the equipment operation logs and consumables specification variables are obtained from the consumables basic information database. Based on the extracted operation habit parameters and consumables specification variables, combined with the usage patterns recorded in the historical shared consumables consumption records, an association rule reflecting the relationship between operation characteristics, consumable attributes and usage is established, and the average consumption of shared consumables per unit time is calculated as the second usage prediction value.

4. The intelligent prediction and management method for service consumables usage according to claim 3, characterized in that, The first and second dosage forecasts are converted into forecast datasets respectively; based on the distribution characteristics of the forecast datasets, a forecast analysis framework is constructed through data boundaries, including: The first and second dosage prediction values ​​are aggregated according to a preset time period to obtain a prediction data set containing time series characteristics. The distribution characteristics of the forecast dataset are analyzed, and statistical descriptive indicators of the forecast dataset, including mean, variance and extreme values, are calculated to identify the distribution patterns and fluctuation characteristics of the forecast data. Based on the distribution patterns and fluctuation characteristics obtained from the analysis, the upper and lower boundary values ​​of the predicted data are determined by the percentile calculation method, and the data boundary range is established. A predictive analytics framework is built based on data boundary ranges. The predictive analytics framework includes data validation rules and anomaly identification mechanisms.

5. The intelligent prediction and management method for service consumables usage according to claim 4, characterized in that, In the predictive analytics framework, a baseline reference range is established. Based on the data fluctuation characteristics of the predictive dataset, this baseline reference range is adaptively partitioned to obtain multiple sub-ranges with different data characteristics, including: Based on the data boundary range established in the predictive analysis framework, and combined with the baseline level of historical consumable usage, a benchmark reference range for measuring the fluctuation range of predicted data is determined. Within the benchmark reference range, analyze the data fluctuation characteristics of the forecast dataset, calculate the fluctuation amplitude and fluctuation frequency indices, identify the dynamic change patterns of the forecast data, and obtain the identified data fluctuation characteristics. Based on the identified data fluctuation characteristics, the benchmark reference range is adaptively partitioned using a dynamic threshold segmentation method, dividing data with similar fluctuation characteristics into the same interval to obtain multiple sub-ranges with different data characteristics.

6. The intelligent prediction and management method for service consumables usage according to claim 5, characterized in that, Each predicted data set is assigned to a corresponding sub-range, and a prediction adjustment value is calculated based on the distribution density and fluctuation amplitude of data elements within each sub-range. The obtained usage prediction values ​​are then corrected using these adjustment values ​​to obtain optimized consumable usage prediction results, including: Based on multiple sub-ranges with different data characteristics, each data element in the prediction dataset is assigned to the corresponding sub-range according to its feature attributes, and a mapping relationship between data elements and sub-ranges is established. By analyzing the mapping relationship between data elements and sub-ranges, the distribution of data elements in each sub-range is analyzed, and the data distribution density and fluctuation amplitude index of each sub-range are calculated to quantify the data concentration and stability characteristics of each sub-range. Based on the centralization and stability characteristics of each sub-range data, and combined with historical prediction deviation data, a weighted algorithm is used to generate prediction adjustment values ​​corresponding to each sub-range. The predicted adjustment value is applied to the initial usage prediction value, and the first and second usage prediction values ​​are collaboratively corrected to obtain the optimized consumable usage prediction result.

7. The intelligent prediction and management method for service consumables usage according to claim 6, characterized in that, Develop dynamic inventory management strategies based on optimized consumable usage forecasts; Real-time monitoring of consumable requisition data, comparison of actual consumption data with optimized forecasts, and the resulting updated inventory management strategy, including: Based on the optimized consumable usage forecast, combined with inventory cost parameters and safety stock requirements, a dynamic inventory management strategy including replenishment trigger points and replenishment quantities is formulated. Based on the dynamic inventory management strategy, monitor consumable requisition operation data in real time and collect actual consumption data; The actual consumption data is compared and analyzed with the optimized consumable consumption prediction results to calculate the prediction deviation rate and consumption trend change index. Based on the prediction deviation rate and consumption trend change indicators, the parameters in the dynamic inventory management strategy are adaptively adjusted to obtain the updated inventory management strategy and output it for execution.

8. A service consumables usage intelligent prediction and management system, wherein the system implements the method as described in any one of claims 1 to 7, characterized in that, include: The acquisition module is used to acquire multi-dimensional raw data related to the use of consumables, including diagnosis and treatment, consumable attributes, consumption records, operation logs, time context, and inventory supply chain data. Multi-dimensional data is processed, and consumables are classified into patient-specific consumables and shared consumables based on their consumption characteristics to obtain consumable classification results; The calculation module is used to obtain the first usage prediction value based on the patient-to-patient consumables in the consumables classification results and real-time outpatient registration data; Based on the shared consumables in the consumables classification results, and combined with operational habit parameters and consumable specification variables, a second usage prediction value is obtained. The habit parameters reflect the stable preferences of specific departments, medical groups, and medical staff when using shared consumables; the consumable specification variables refer to the physical attribute characteristics that affect the single consumption amount of shared consumables. The first and second usage prediction values ​​are converted into prediction datasets respectively. Based on the distribution characteristics of the prediction datasets, a predictive analysis framework is constructed through data boundaries. A benchmark reference range is established within the predictive analysis framework, and the benchmark reference range is adaptively partitioned according to the data fluctuation characteristics of the prediction datasets to obtain multiple sub-ranges with different data characteristics. The optimization module is used to allocate each predicted data set to the corresponding sub-range, calculate the prediction adjustment value based on the distribution density and fluctuation amplitude of the data elements in each sub-range, and correct the obtained usage prediction value through the prediction adjustment value to obtain the optimized consumable usage prediction result. The processing module is used to formulate dynamic inventory management strategies based on the optimized consumable usage forecast results; monitor consumable requisition operation data in real time, compare the actual consumption data with the optimized forecast results, and obtain the updated inventory management strategy.

9. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images