CN114529086A

CN114529086A - Medicine data processing method, device and equipment

Info

Publication number: CN114529086A
Application number: CN202210155051.6A
Authority: CN
Inventors: 虞明星; 王文祥; 赵大平; 祝莎莎; 唐力伟; 周炜; 王琪; 黄智勇; 黄克华; 吴铭
Original assignee: Winning Health Technology Group Co Ltd
Current assignee: Winning Health Technology Group Co Ltd
Priority date: 2022-02-21
Filing date: 2022-02-21
Publication date: 2022-05-24

Abstract

The application provides a method, a device and equipment for processing medicine data, wherein the method comprises the following steps: acquiring historical period consumption data of a target drug in a historical period, predicting by adopting the historical period consumption data to obtain predicted period consumption data of the target drug in a preset future period, calculating predicted consumption data in each sub-period in the preset future period according to the predicted period consumption data, generating a purchase order for the target drug according to the predicted consumption data in each sub-period and a preset safety stock threshold value of the target drug if a purchase order generation service request input aiming at the target drug is detected, wherein the preset safety stock threshold value is used for indicating a safety stock corresponding to the target drug when the preset shelf life fluctuates, and the purchase order comprises: the amount of targeted drug procurement. And the medicine purchase cost is reduced by automatically generating the medicine purchase order.

Description

Medicine data processing method, device and equipment

Technical Field

The present application relates to the field of data processing technologies, and in particular, to a method, an apparatus, and a device for processing drug data.

Background

The medicine purchasing and inventory management is one of the important parts of daily management of hospitals, and the medical information system can help relevant departments of the hospitals to accurately record the medicine purchasing, ex-warehouse and consumption conditions in real time, so that the medicine inventory management level is effectively improved.

Currently, most hospital drug procurement plans are manually determined by relevant management personnel, in combination with recent consumption and procurement records, as to the list and quantity of drugs to be procured. However, the purchase scheme for listing thousands of medicines is time-consuming and labor-consuming, and more medicines are stocked, which increases the purchase cost of medicines.

Disclosure of Invention

An object of the present application is to provide a method, an apparatus and a device for processing drug data to automatically generate a drug purchase order, so as to reduce the drug purchase cost.

In order to achieve the above purpose, the technical solutions adopted in the embodiments of the present application are as follows:

in a first aspect, an embodiment of the present application provides a drug data processing method, including:

acquiring historical period consumption data of the target medicine in a historical period;

predicting by adopting the historical periodic consumption data to obtain predicted periodic consumption data of the target medicine in a preset future period;

calculating the predicted consumption data in each sub-period in the preset future period according to the predicted period consumption data;

if a purchase order generation service request input aiming at the target medicine is detected, generating a purchase order aiming at the target medicine according to the predicted consumption data in each sub-period and a preset safety stock threshold value of the target medicine, wherein the preset safety stock threshold value is used for indicating a safety stock corresponding to the target medicine when a preset arrival date fluctuates, and the purchase order comprises: a purchase amount of the target drug.

Optionally, the predicting by using the historical periodic consumption data to obtain the predicted periodic consumption data of the target drug in a preset future period includes:

predicting the historical periodic consumption data by adopting an exponential smoothing model to obtain first predicted consumption data of the target medicine in the preset future period;

if the time distribution of the historical periodic consumption data meets a first preset condition, predicting the historical periodic consumption data by adopting a mean value prediction model to obtain second predicted consumption data of the target drug in the preset future period;

and obtaining the predicted periodic consumption data according to the first predicted consumption data and the second predicted consumption data.

Optionally, the predicting by using the historical periodic consumption data to obtain the predicted periodic consumption data of the target drug in a preset future period further includes:

if the time distribution of the historical periodic consumption data meets a second preset condition, predicting the historical periodic consumption data by adopting a time series prediction model to obtain third predicted consumption data of the target medicine in the preset future period;

and obtaining the predicted periodic consumption data according to the first predicted consumption data and the third predicted consumption data.

Optionally, the obtaining the predicted cycle consumption data according to the first predicted consumption data and the second predicted consumption data includes:

carrying out model verification on the exponential smoothing model to obtain a first verification index;

carrying out model verification on the mean value prediction model to obtain a second verification index;

judging whether the first check index and the second check index are respectively in corresponding check ranges;

and if the first check index and the second check index are both in the corresponding check range, obtaining the prediction period consumption data according to the first prediction consumption data and the second prediction consumption data.

Optionally, the obtaining the predicted periodic consumption data according to the first predicted consumption data and the second predicted consumption data further includes:

and if the first check index and/or the second check index are not in the corresponding check range, determining that the second predicted consumption data are the predicted cycle consumption data.

Optionally, before generating a purchase order for the target drug according to the predicted consumption data in each sub-period and the preset safe inventory threshold of the target drug, the method further includes:

and calculating a preset safety stock threshold value of the target medicine according to the historical periodic consumption data, a preset arrival period corresponding to the target medicine and a preset safety factor.

Optionally, the method further comprises:

determining the number of available days of the target drug according to the predicted consumption data in each sub-period and the inventory data of the target drug;

and judging whether the target medicine is out of stock in a preset purchasing period and a arrival period according to the available days.

Optionally, the calculating predicted consumption data in each sub-period in the preset future period according to the predicted period consumption data includes:

acquiring the scale change rate of a use object corresponding to the target medicine;

and determining the predicted consumption data in each sub-period according to the predicted period consumption data and the scale change rate.

In a second aspect, another embodiment of the present application provides a drug data processing apparatus, including:

the acquisition module is used for acquiring historical period consumption data of the target medicine in a historical period;

the prediction module is used for predicting by adopting the historical periodic consumption data to obtain predicted periodic consumption data of the target medicine in a preset future period;

the calculation module is used for calculating the predicted consumption data in each sub-period in the preset future period according to the predicted period consumption data;

a generating module, configured to generate a purchase order for the target drug according to the predicted consumption data in each sub-period and a preset safety inventory threshold of the target drug if a purchase order generation service request input for the target drug is detected, where the purchase order includes: a purchase amount of the target drug.

In a third aspect, another embodiment of the present application provides an electronic device, including: a processor, a memory and a bus, the memory storing a computer program executable by the processor, the processor and the memory communicating via the bus when the electronic device is running, the processor executing the computer program to perform the method of any of the first aspect.

In a fourth aspect, another embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, where the computer program is executed by a processor to perform the method of any one of the first aspect.

The beneficial effect of this application is:

the application provides a medicine data processing method, a device and equipment, wherein the method comprises the following steps: acquiring historical period consumption data of a target drug in a historical period, predicting by adopting the historical period consumption data to obtain predicted period consumption data of the target drug in a preset future period, calculating predicted consumption data in each sub-period in the preset future period according to the predicted period consumption data, generating a purchase order for the target drug according to the predicted consumption data in each sub-period and a preset safety stock threshold value of the target drug if a purchase order generation service request input aiming at the target drug is detected, wherein the preset safety stock threshold value is used for indicating a safety stock corresponding to the target drug when the preset shelf life fluctuates, and the purchase order comprises: the amount of targeted drug procurement. By automatically generating the drug purchase order, the drug purchase cost is reduced.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.

Fig. 1 is a first schematic flow chart of a drug data processing method according to an embodiment of the present application;

fig. 2 is a schematic flow chart diagram of a drug data processing method according to an embodiment of the present application;

fig. 3 is a third schematic flowchart of a drug data processing method according to an embodiment of the present application;

fig. 4 is a fourth schematic flowchart of a drug data processing method according to an embodiment of the present application;

fig. 5 is a fifth flowchart illustrating a drug data processing method according to an embodiment of the present application;

fig. 6 is a schematic structural diagram of a drug data processing device according to an embodiment of the present application;

fig. 7 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it should be understood that the drawings in the present application are for illustrative and descriptive purposes only and are not used to limit the scope of protection of the present application. Additionally, it should be understood that the schematic drawings are not necessarily drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of the present application. It should be understood that the operations of the flow diagrams may be performed out of order, and steps without logical context may be performed in reverse order or simultaneously. One skilled in the art, under the guidance of this application, may add one or more other operations to, or remove one or more operations from, the flowchart.

In addition, the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.

It should be noted that in the embodiments of the present application, the term "comprising" is used to indicate the presence of the features stated hereinafter, but does not exclude the addition of further features.

Aiming at the problem that the high medicine purchasing cost is caused by manually determining the list and the quantity of the medicines to be purchased at present, the application provides a medicine data processing method, which is used for automatically generating the purchase order of the target medicines according to the historical periodic consumption data and the inventory condition of the target medicines, reduces the medicine purchasing cost and has higher analysis value and practical application economic value.

The following describes the drug data processing method of the present application in detail with reference to several specific embodiments.

Fig. 1 is a schematic flow chart of a drug data processing method according to an embodiment of the present application, where an execution main body of the embodiment may be an electronic device, such as a terminal device, a server, and the like. As shown in fig. 1, the method includes:

s101, acquiring historical period consumption data of the target medicine in the historical period.

The target medicine is any medicine to be purchased, and the historical period consumption data is consumption data of the target medicine summarized in the historical period, wherein the historical period can be divided according to weeks, and the historical consumption record of the target medicine comprises the historical consumption of the target medicine, so the historical consumption in the historical consumption record of the target medicine is summarized according to weeks, and the historical period consumption data of the target medicine can be obtained.

In some cases, for example, as the names of the drugs are updated, the target drugs may have different drug names in different time periods, the names of the target drugs need to be integrated to determine that the drugs corresponding to different drug names are the target drugs, and similarly, the specifications of the drugs and the manufacturers may be re-integrated to make the historical periodic consumption data more accurate.

It should be noted that the number of the historical periods may be multiple, and correspondingly, the number of the historical period consumption data may also be multiple, and after the historical period consumption data are collected by week, abnormal data and missing data in the historical period consumption data may also be processed to obtain week-time data of the target drug, where the week-time data includes multiple historical period consumption data, and for example, a year includes 53 weeks, and the week-time data may include 53 historical period consumption data.

The abnormal data may be data having a large difference from other historical period consumption data, for example, all the other historical period consumption data are about 100 (the deviation does not exceed a preset threshold), where one of the historical period consumption data is 1, which indicates that the historical period consumption data is the abnormal data, the historical period consumption data may be modified to 100, and the deviation of the modified historical period consumption data from the other historical period consumption data does not exceed the preset threshold. In addition, if the missing data is that the target drug is not used for one week, and thus the consumption amount of the target drug does not exist in the historical consumption record, the corresponding historical periodic consumption data may be filled with 0.

S102, predicting by adopting historical cycle consumption data to obtain predicted cycle consumption data of the target medicine in a preset future cycle.

S103, calculating the predicted consumption data in each sub-period in the preset future period according to the predicted period consumption data.

The preset period consumption data are consumption data collected by the target medicine in a preset future period, wherein a preset future period dividing principle corresponds to a historical period and can be divided according to weeks.

The method includes the steps of predicting to obtain predicted cycle consumption data of a target drug in a preset future cycle by using historical cycle consumption data, and then calculating preset consumption data in each sub-cycle in the preset future cycle according to the preset cycle consumption data, wherein the preset sub-cycle is a sub-cycle determined by periodically dividing the preset future cycle, for example, the preset future cycle is divided according to cycles, each sub-cycle in the preset future cycle is divided according to days, namely, each sub-cycle in the preset future cycle is divided by days, and the preset future cycle comprises 7 sub-cycles.

The predicted consumption data of the target medicine in each sub-period in the preset future period can be calculated according to the distribution proportion of the consumption data of the target medicine in the historical period and the predicted period consumption data of the target medicine in the preset future period, wherein the distribution of the consumption data of the target medicine can be determined according to the historical consumption record of the target medicine in the historical period, and the distribution proportion of the consumption data of the target medicine in one week can be determined according to the historical consumption record, for example, the consumption data percentage of Monday to Friday is 10%, and the consumption data percentage of Saturday to Friday is 7.5%. In the embodiment, the system operation speed is increased by summarizing data, and the accuracy of the predicted consumption data in each sub-period is ensured by considering data distribution.

And S104, if a purchase order generation service request input aiming at the target medicine is detected, generating a purchase order aiming at the target medicine according to the predicted consumption data in each sub-period and the preset safety stock threshold value of the target medicine.

The purchase order generation service request is used for requesting generation of a purchase order of a target medicine, if the purchase order of the target medicine needs to be generated, the purchase order generation request aiming at the target medicine can be input, correspondingly, if the purchase order generation service request aiming at the target medicine input is detected, the purchase order of the target medicine is generated according to predicted consumption data in each sub-period and a preset safety inventory threshold value of the target medicine, wherein the preset safety inventory threshold value is used for indicating a safety inventory corresponding to the target medicine when a preset arrival date fluctuates, and the purchase order comprises: and the purchase quantity of the target medicine is the sum of the predicted consumption data in each sub-period and the preset safety inventory threshold value of the target medicine.

The sum of the predicted consumption data of each sub-period and the preset safety inventory threshold is the purchase amount of the target medicine, that is, when the purchase amount of the target medicine is determined, the predicted consumption data of each sub-period is considered, and the preset safety inventory threshold set for avoiding shortage of goods is considered when the arrival period of the target medicine fluctuates, so that the purchase amount is more accurate, and the accuracy of the purchase order plan is ensured. For example, the arrival period of the target drug is tomorrow, but due to some reasons, the arrival period is delayed to the next day, and then in order to avoid the target drug being out of stock tomorrow, the preset safety stock threshold is set to ensure that the target drug has a certain stock tomorrow, that is, the target drug is not out of stock.

Optionally, in step S104, before generating a purchase order for the target drug according to the predicted consumption data in each sub-period and the preset safety inventory threshold of the target drug, the method further includes:

and calculating a preset safety stock threshold value of the target medicine according to the historical periodic consumption data, the preset arrival period corresponding to the target medicine and the preset safety factor.

The preset safety factor depends on the service level, the service level is used for indicating the satisfaction degree of the medicine quantity demand condition of the target medicine, the service level (%) -annual shortage times/annual ordering times, the higher the service level is, the fewer the shortage situations occur, and the values of the safety factor and the service level are shown in table 1.

Service level	Factor of safety z	Service level	Factor of safety z
				84.4	0.0	98.9	2.3
90.3	1.3	99.5	2.6
				94.5	1.6	99.9	3.0
97.7	2.0

TABLE 1

Referring to equation (1), the preset safety stock threshold SS may be expressed as:

wherein, σ is a standard deviation of week time series data of the target medicine, the week time series data includes a plurality of historical period consumption data, taking a year including 53 weeks as an example, the week time series data includes 53 historical period consumption data, then σ is a standard deviation of 53 historical period consumption data, L is a preset arrival period divided by 7 corresponding to the target medicine, and 7 represents that a week includes 7 days.

In the drug data processing method of this embodiment, historical period consumption data of a target drug in a historical period is obtained, the historical period consumption data is used for prediction, prediction period consumption data of the target drug in a preset future period is obtained, prediction consumption data in each sub-period in the preset future period is calculated according to the prediction period consumption data, if a purchase order generation service request input for the target drug is detected, a purchase order for the target drug is generated according to the prediction consumption data in each sub-period and a preset safety stock threshold of the target drug, the preset safety stock threshold is used for indicating a safety stock corresponding to the target drug when a shelf life fluctuates, and the purchase order includes: the amount of targeted drug procurement. And the medicine purchase cost is reduced by automatically generating the medicine purchase order.

Optionally, in step S104, generating a purchase order for the target drug according to the predicted consumption data in each sub-period and the preset safety inventory threshold of the target drug, where the method includes:

generating a purchase order aiming at the target medicine according to the predicted consumption data, the preset safety inventory threshold value of the target medicine, the existing inventory, the in-transit inventory, the preset purchase period and the arrival period in each sub-period, wherein the purchase order comprises: the amount of targeted drug procurement.

Wherein the existing inventory is the inventory which is currently warehoused, the in-transit inventory is the inventory which is placed and currently in transportation, the preset purchasing period is the purchasing period of the target medicine, the arrival period is the time length from the placing to the receiving of the target medicine, the sum of the preset purchasing period of the target medicine and the predicted consumption data in each sub-period before the arrival period is calculated, then, the sum is used to subtract the existing inventory and the in-transit inventory of the target drug to obtain the purchase amount of the target drug, and a adoption list is generated according to the purchase amount of the target drug, that is, when the purchase quantity of the target medicine is calculated, the sum of the preset purchase period and the predicted consumption data before the arrival period and the preset safety inventory threshold value are considered, the existing inventory and the in-transit inventory are brought into the category of the total inventory, the inventory data are more accurate, and the purchase quantity of the target medicine is more accurate.

Taking the sub-periods in days as an example, referring to equation (2), the purchase amount Q of the target drug can be expressed as:

wherein the stock safety threshold value is SS, the purchasing period is T, the arrival period is L, the existing stock is ES, the in-transit stock is TS, D_iIs the predicted consumption on day i, with the range of i from 1 to T + L.

The method is adopted to calculate the purchase quantity of all the medicines to be purchased, a purchase list is formed and stored in the system, a purchaser can adjust or not adjust the purchase list to obtain an actual execution purchase list and store the actual execution purchase list in the system, and then post feedback and analysis are carried out on the follow-up purchase list and the actual execution purchase list stored in the system, so that a closed loop is formed.

One possible implementation of predicting the historical cycle consumption data to obtain the predicted cycle consumption data of the target drug within the preset future cycle in step S102 is described below with reference to fig. 2.

Fig. 2 is a schematic flow chart diagram of a drug data processing method according to an embodiment of the present application, and as shown in fig. 2, predicting by using historical cycle consumption data to obtain predicted cycle consumption data of a target drug in a preset future cycle includes:

s201, predicting historical periodic consumption data by adopting an exponential smoothing model to obtain first predicted consumption data of the target medicine in a preset future period.

S202, if the time distribution of the historical period consumption data meets a first preset condition, predicting the historical period consumption data by adopting a mean value prediction model to obtain second predicted consumption data of the target drug in the preset future period.

And S203, obtaining predicted periodic consumption data according to the first predicted consumption data and the second predicted consumption data.

The historical periodic consumption data is used as input of an exponential smoothing model, the exponential smoothing model is adopted to predict the historical periodic consumption data, and first preset consumption data of the target drug in a preset future period are obtained, wherein the first preset condition can include that the historical use time is less than k weeks, or the historical use deficiency rate is greater than m, a consumption record exists in a latest preset time period, the preset time period can be the latest 1 month, the historical use deficiency rate is the deficiency rate of the target drug in the historical use process, the historical use deficiency rate can be the ratio of the use days of the target drug to the preset total days in a certain time period, and k and m can be respectively set to be 33% and 25%.

That is, if the time distribution of the historical period prediction data meets a first preset condition, the historical period consumption data is used as an input of a mean value prediction model, the historical period prediction data is predicted by the mean value prediction model to obtain predicted period consumption data, and then the predicted period consumption data is obtained according to the first predicted consumption data and the second predicted consumption data, wherein the preset period consumption data may be a weighted average value of the first predicted consumption data and the second predicted consumption data, and the weight of the exponential smoothing model and the mean value prediction model may be 0.5.

Optionally, the predicting by using the historical periodic consumption data to obtain the predicted periodic consumption data of the target drug in a preset future period, further includes:

and S204, if the time distribution of the historical period consumption data meets a second preset condition, predicting the historical period consumption data by adopting a time sequence prediction model to obtain third predicted consumption data of the target medicine in a preset future period.

And S205, obtaining predicted periodic consumption data according to the first predicted consumption data and the third predicted consumption data.

The second preset condition may include that the historical usage time is greater than or equal to k weeks, or the historical usage loss rate is less than m, and a consumption record exists in the latest preset time period, where the preset time period may be the latest 1 month.

If the time distribution of the historical period consumption data meets a second preset condition, taking the historical period consumption data as the input of a time series prediction model, predicting the historical period consumption data by adopting the time series prediction model to obtain third predicted consumption data of the target medicine in a preset future period, and then obtaining predicted period consumption data according to the first predicted consumption data and the third predicted consumption data, wherein the preset period consumption data can be a weighted average value of the first predicted consumption data and the third predicted consumption data, and the weight of the exponential smoothing model and the time series prediction model can be 0.5.

In the medicine data processing method of the embodiment, the multi-model integrated prediction is introduced, so that the prediction of all medicine consumption is met, and the method has wide applicability.

The exponential smoothing model can be a Holt-Winters exponential smoothing model, the time series prediction model can be a Prophet model, and the multi-model integrated prediction is shown in Table 2.

TABLE 2

As an example, an analysis model used by the target drug is determined according to Table 2, and then the week time data of the target drug is input into the corresponding multi-model in Table 2 for prediction, so as to obtain the predicted week consumption data of the target drug in a preset future week. The respective models referred to in table 2 are explained below.

(1) Mean model

Intercepting the week-time data of a period of time, if the median of the week-time data is not zero, then taking the second predicted consumption data as the median, and the fluctuation interval of the second predicted consumption database is the quantile (such as 95%) designated by the week-time data, otherwise, taking the second predicted consumption data as the average value of the week-time data of the period of time, and the fluctuation interval of the second predicted consumption database is the upper and lower confidence level value (such as 90%) of the z-distribution of the average value.

(2) Prophet prediction model

The trend of the model to weekly time series data can be expressed by the following formula (2):

y(t)＝g(t)+s(t)+h(t)+ε (2)

where y (t) is third predicted consumption data, g (t) is a trend model with no periodic variation, s (t) represents a seasonal model of the period (e.g., week or year), h (t) is a model of holidays or events over one or more days, ε is an error term, and a normal distribution is assumed to be satisfied. The trend model, seasonal model, and holiday or event model are described below, one equation for each.

Referring to equation (3), the trend model is a linear model with transition points, and the specific equation is expressed as follows:

g(t)＝(k+α(t)^Tδ)t+(m+α(t)^Tγ) (3)

wherein t is the time point of research, k is the growth rate parameter, δ is the adjustment rate parameter of k, m is the compensation parameter, γ is the setting parameter, for different drug consumption transition point events, k, δ, m, α (t), γ correspond to different values respectively, and the occurrence time s of the event at the jth transition point_j，γ_jIs equal to-s_jδ_j，δ_jIs the adjustment rate parameter for the jth transition point event,

α_j(t) is α (t) for the jth transition point event.

The transition point event may refer to an event that causes a sharp increase or decrease in the consumption of the medicine due to a specific reason. A linear model of the transition point is used to evaluate drug consumption data that causes a sharp increase or decrease in drug consumption due to the transition point event of drug consumption.

Referring to equation (4), the seasonal model provides a flexible periodic equation model by fourier series, which is as follows:

where, for seasonal models of the year or week, N is in most cases equal to 10 or 3 works well, and P is equal to 365.25 or 7, a_nAnd b_nAll parameters are parameters to be estimated, and n is a fitting coefficient which can be selected according to actual conditions, such as 2, 3 and the like.

Seasonal models are used to evaluate drug consumption data that causes a sharp increase or decrease in drug consumption due to seasonal reasons.

Referring to equation (5), the equation for the holiday and event model is expressed as follows:

where κ is a parameter to be estimated, D is a holiday or event, 1 (t)∈D_i) Indicating the presence of holidays or events D at time t_iThen 1, otherwise 0, L represents the total number of holidays or events.

The holiday and event model is used to evaluate drug consumption data that causes a sharp increase or decrease in drug consumption due to holidays or other events.

Giving prior values for the parameters k, δ, m, a, b and κ by monte carlo random sampling, in combination with equations (2) - (5), yields the following expression (6):

y～normal((k+A*δ)*t)+(m+A*γ)+βX+κZ),ε) (6)

wherein y satisfies the positive-Taiwan normal, ε is the standard deviation of the positive-Taiwan,

Z＝[1(t∈D₁),…,1(t∈D_L)]。

in formula (6), A denotes α (t) in formula (3)^TAnd beta denotes a in the formula (4)_nAnd b_nAnd κ denotes in formula (5)

The parameters k, delta, m, a are given by Monte Carlo random sampling_n、b_nAnd kappa, fitting the formula with an estimated value of each parameter by using a quasi-Newton method (L-BFGS), predicting third predicted consumption data and a credible interval based on the estimated value of the parameter, concretely, substituting the fitted prior value of each parameter into a formula (4), predicting by using the formula (4) to obtain predicted cycle time sequence data of a historical cycle, then calculating the loss of the actual weekly time series data and the predicted weekly time series data of the historical period, adjusting each parameter, continuously predicting to obtain the predicted weekly time series data of the historical period, circulating the step until the loss of the actual weekly time series data and the predicted weekly time series data reaches the minimum, determining each parameter with the minimum loss as an estimated value of each parameter, substituting the estimated value of each parameter into formula (4), and predicting third predicted consumption data of a preset future period by using formula (4).

(3) Holt-Winters exponential smoothing model prediction

The model is used to predict time series data that are both trending and seasonal. To explain this model, the following four aspects of the time series, namely Level (Level), trend (tend), seasonality (seasonal) and noise (noise), are explained in advance.

Level: representing some level in the course of rising and falling in a time series;

trend is as follows: the levels of a time series vary in some pattern, called trending, and some common trends can be linear, squared, exponential, logarithmic, square root, reciprocal, and cubic or higher polynomials.

Seasonality: many time series show periodic up and down movements around the current level, which are called seasonality.

Noise: noise is only an aspect of the time series data that cannot (or is not intended to) be interpreted.

Levels, trends, seasonality and noise are considered to interact in an additive or multiplicative manner, producing the final values of the observed time series, as is often the case in real-world time series data with trends plus current levels multiplied by seasonal variations. The prediction formula for a time series starting from an arbitrary point i to k time steps in the future is as follows (7):

F_(i+k)＝(L_i+k*B_i)*S_(i+k-m) (7)

wherein, F_(i+k)Representing the first predicted consumption data, L, at time point i + k_i+k*B_iRepresenting the level estimate at time point i + k, k being the predicted number of weeks, S_i+k-mRepresenting a seasonal variation of time i + k and period length m, e.g. m equals 12 for annual changes and m equals 53 for weekly changes.

There is now a need to estimate the trend B_iHorizontal L_iAnd seasonal S_iReferring to equation (8), the estimated trend is expressed as:

B_i＝β*[L_i-L_(i-1)]+(1-β)*B_(i-1) (8)

wherein [ L ]_i-L_(i-1)]Representing the difference between two successive levels, representing the rate of change, B_(i-1)Is a recursive expression until B is reached₀Let B be₀The value of (b) is the initial condition.

Referring to equation (9), the estimated level is expressed as follows:

wherein, T_iRepresenting the sequence value at time i, i.e. the sum of the consumption data for week i, S_(i-m)Is a seasonal variation of period m at time i, L_(i-1)Represents the level estimate of the previous cycle, B_(i-1)Representing the amount of change, L, in the level estimates of i-1 to i_(i-1)+B_(i-1)Representing the level of i-1 plus the amount of change from i-1 to i.

Referring to equation (10), the estimated seasonality is expressed as follows:

note that equations (8) - (10) are recursive to L₀、B₀、S₀、T₀Wherein, T₀Is the initial data point in the training dataset. L is₀，B₀And S₀Respectively, the initial values of the level, trend and seasonal variation. The following equations (11) to (15) are used to calculate the initial value L₀、B₀、S₀、T₀：

L₀＝mean(T₀+T₁₂+T₂₄+···) (11)

S₀＝[(T₀-L₀),(T₁-L₀),(T₂-L₀),,···,(T_m-1-L₀)] (15)

Wherein L is₀、B₀Is a scalar quantity, S₀Are vectors, (12) and (14) correspond to multiplicative models, and (13) and (15) correspond to additive models.

Through formulas (7) - (15), with minimum Mean Square Error (MSE) as the optimal target loss minimum, parameters α, β and γ can be estimated, and then first prediction consumption data can be obtained, and the Prophet prediction model can be referred to for the relevant description about the target loss minimum.

Converting the time-of-week data into time-of-month data in units of years for the historical period, T being for equation (11)₀Consumption data for the first month of the first historical period, T₁₂Consumption data for the first month of the second historical period, T₂₄Consumption data for the first month of the third historical period.

For equation (12), T₁₂Consumption data for the first month of the second historical period, T₀Consumption data for the first month of the first historical period, T₁₃Consumption data for the second month of the second historical period, T₁Consumption data for the second month of the first history period, T₁₄Consumption data for the third month of the second historical period, T₂Consumption data for the first month of the first historical period, T₂₃Consumption data for the twelfth month of the third historical period, T₁₁Consumption data for the twelfth month of the second historical period.

For formula (14), T₀Consumption data for the first month of the first historical period, T₁For the second month of the first historical periodConsumption data, T_(m-1)Consumption data of the m month in the first history period.

For the parameters in equations (13) and (15), reference is made to the explanations in equations (12) to (14). In addition, both equations (14) and (15) are used to calculate S₀S can be calculated by the following equations (14) and (15)₀Is taken as the final S₀。

In the multi-model integrated prediction of the embodiment, factors such as holidays and the like are introduced when the consumption data are predicted, so that the accuracy of the consumption data prediction is improved.

Next, a possible embodiment of obtaining the predicted cycle consumption data from the first predicted consumption data and the second predicted consumption data in step S203 will be described with reference to fig. 3.

Fig. 3 is a third schematic flow chart of the drug data processing method according to the embodiment of the present application, and as shown in fig. 3, obtaining the predicted periodic consumption data according to the first predicted consumption data and the second predicted consumption data includes:

s301, carrying out model verification on the exponential smoothing model to obtain a first verification index.

And S302, carrying out model verification on the mean value prediction model to obtain a second verification index.

The first check index is used for evaluating the prediction performance of the index smoothing model, and the second check index is used for evaluating the prediction performance of the mean value prediction model. The first and second calibration indicators may respectively include a risk rate (risk _ rate), an error rate (error _ rate), a redundancy rate (rest _ rate), a Root Mean Square Error (RMSE), and a Mean Absolute Percentage Error (MAPE), where the risk rate, the error rate, and the redundancy rate are service indicators, and the root mean square error and the mean absolute percentage error are model indicators.

The risk rate refers to the proportion that the upper bound value of a 95% credible interval of a predicted value is smaller than the true value in n times of prediction; the error rate refers to the proportion of the predicted value plus the safety stock threshold value smaller than the true value in the n times of prediction; the redundancy refers to the mean value of (predicted value + preset safety stock threshold)/(real value + preset safety stock threshold) in n predictions; the root mean square error RMSE and the average absolute percentage error MAPE are shown in equations (16) - (17):

wherein,

y is the true value for the predicted value.

That is to say, an exponential smoothing model is used for predicting n historical period consumption data to obtain a first predicted value corresponding to each historical period consumption data, then model verification is performed on the exponential smoothing model according to the first predicted value and a true value (namely historical period consumption data) to obtain a first verification index, a mean value prediction model is used for predicting n historical period consumption data to obtain a second predicted value corresponding to each historical period consumption data, and then model verification is performed on the exponential smoothing model according to the second predicted value and the true value (namely historical period consumption data) to obtain a second verification index.

S303, judging whether the first check index and the second check index are respectively in the corresponding check ranges.

S304, if the first check index and the second check index are both in the corresponding check range, obtaining the predicted periodic consumption data according to the first predicted consumption data and the second predicted consumption data.

And judging whether the first check index and the second check index are in corresponding check ranges respectively, if the first check index and the second check index are in the corresponding check ranges, and the prediction performances of the exponential smoothing model and the mean value prediction model meet the requirements, obtaining predicted periodic consumption data according to the first predicted consumption data and the second predicted consumption data, wherein the preset periodic consumption data can be a weighted average value of the first predicted consumption data and the second predicted consumption data.

Optionally, obtaining predicted periodic consumption data according to the first predicted consumption data and the second predicted consumption data, further comprising:

s305, if the first check index and/or the second check index are not in the corresponding check range, determining that the second predicted consumption data are predicted period consumption data.

And if the first check index and/or the second check index are not in the corresponding check range, which indicates that the prediction performance of the model with the check index not in the corresponding check range does not meet the requirement, determining second predicted consumption data obtained by the prediction of the mean value prediction model as predicted periodic consumption data.

Similarly, obtaining predicted cycle consumption data from the first predicted consumption data and the third predicted consumption data comprises: and if the first check index and/or the third check index are not in the corresponding check range, determining the third predicted consumption data as the predicted periodic consumption data. The specific implementation process can be seen in steps S301-S306.

In the medicine data processing method of the embodiment, when multi-model prediction is introduced, model verification is also performed, so that the accuracy of predicting consumption data is further improved, and the accuracy of purchasing quantity is ensured.

Fig. 4 is a fourth schematic flowchart of a drug data processing method provided in the embodiment of the present application, and as shown in fig. 4, the method further includes:

s401, determining the available days of the target medicine according to the predicted consumption data and the inventory data of the target medicine in each sub-period.

The inventory data of the target medicines comprises the existing inventory and the in-transit inventory, wherein the existing inventory is the inventory which is currently put in storage, and the in-transit inventory is the inventory which is placed and currently in transportation.

The sum of the on-hand inventory and the in-transit inventory is determined as inventory data of the target drug, and then the number of days available for the target drug is determined based on the predicted consumption data for each sub-cycle and the inventory data of the target drug, wherein each sub-cycle in days in the future cycle is preset, that is, the number of days available for the target drug is determined based on the inventory data of the target drug and the predicted daily consumption data.

S402, judging whether the target medicine is out of stock in a preset purchasing period and a due period according to available days.

The preset purchasing period is the purchasing period of the target medicine, the arrival period is the time length from ordering to receiving of the target medicine, and the arrival period is set by considering logistics transportation.

And judging whether the available days are less than the days corresponding to the preset purchasing period and the arrival period or not, if the available days are less than the days corresponding to the preset purchasing period and the arrival period, determining that the target medicine is out of stock in the preset purchasing period and the arrival period, and if the available days are more than or equal to the sum of the days corresponding to the preset purchasing period and the arrival period, determining that the target medicine is not out of stock in the preset purchasing period and the arrival period. For example, the preset purchasing period is 10 days, the arrival period is 3 days, if the available days are 7 days, the target drug is out-of-stock in the preset purchasing period and the arrival period because the available days are less than the sum of the days corresponding to the preset purchasing period and the arrival period, and if the available days are 20 days, the target drug is not out-of-stock in the preset purchasing period and the arrival period because the available days are greater than the sum of the preset purchasing period and the arrival period.

If the target medicine is out of stock in the preset purchasing period and the arrival period, the out-of-stock early warning can be pushed, so that purchasing personnel can purchase the target medicine in time.

In one implementation, let the algorithm inputs be: daily predicted consumption D, existing inventory ES, in-transit inventory LS, purchase period T and arrival period L, and the algorithm output is as follows: available days omega, purchase period and whether the goods is out of stock in the arrival period are initialized to be 0, inventory data TS is calculated based on ES and LS, the forecast consumption D of the day of the last day is subtracted from the inventory data TS, whether the remaining inventory is larger than 0 is judged, if the remaining inventory is larger than 0, the number of available days ω is incremented by 1, the step is executed in a loop until the remaining inventory is less than 0, the number of available days ω is output, and judging whether the output available days omega is less than the sum of the days corresponding to the purchasing period T and the arrival period L, if the output available days omega are less than the sum of the days corresponding to the purchasing period T and the arrival period L, if the output available days omega are larger than the sum of the days corresponding to the purchasing period T and the arrival period L, the theta is 0, and the target medicine is not out-of-stock in the preset purchasing period and the arrival period.

In the medicine data processing method of the embodiment, in the aspect of medicine inventory calculation, the currently developed express industry is considered, in-transit inventory is brought into the category of total inventory, so that inventory data is more accurate, the purchase list plan is more accurate, and the early warning function of real-time medicine inventory is added while the purchase order service is provided, so that the whole service function is more comprehensive and robust.

Fig. 5 is a schematic flow chart diagram of a fifth method for processing drug data according to an embodiment of the present application, and as shown in fig. 5, step S103 is to calculate predicted consumption data in each sub-period in a preset future period according to the predicted period consumption data, and includes:

s501, acquiring the scale change rate of the use object corresponding to the target medicine.

And S502, determining the predicted consumption data in each sub-period according to the predicted period consumption data and the scale change rate.

The target medicine corresponding to the use object can be a hospital, and the scale change rate is used for indicating the scale change condition of the hospital in a historical period and a preset future period.

The scale data of the target medicine in the historical period and the scale data in the preset future period are obtained, and then the scale change rate is calculated according to the scale data of the historical period and the scale data of the preset future period, wherein the scale data include but are not limited to the department number, the doctor number and the bed number of the hospital, and the scale data of the preset future period can be the scale data of the current time.

According to the predicted cycle consumption data of the target medicine and the distribution proportion of the target medicine consumption data in the historical cycle, the initial predicted consumption data of each sub-cycle in the preset future cycle can be calculated, and then the predicted consumption data of each sub-cycle in the preset future cycle is determined by multiplying the scale change rate by the initial predicted consumption data.

Referring to equation (18), the hospital scale rate of change R can be expressed as:

wherein, alpha, beta and gamma respectively represent the department number, the doctor number and the bed number, t is the current time, and t-1 is the historical time corresponding to the historical period.

Since hospitals are in different grades and types at present, and the types of medicines are extremely rich due to the diversity of actual diseases, in the medicine data processing method of the embodiment, the scale change rate of the hospital is quoted when the medicine consumption is predicted, so that the medicine data processing method can be used for pharmacies or drug depots of various types of hospitals, and has wide applicability.

Fig. 6 is a schematic structural diagram of a drug data processing apparatus according to an embodiment of the present application, where the apparatus may be integrated in an electronic device. As shown in fig. 6, the apparatus includes:

an obtaining module 601, configured to obtain historical period consumption data of a target drug in a historical period;

the prediction module 602 is configured to perform prediction by using the historical periodic consumption data to obtain predicted periodic consumption data of the target drug in a preset future period;

a calculating module 603, configured to calculate predicted consumption data in each sub-period in the preset future period according to the predicted period consumption data;

a generating module 604, configured to generate a purchase order for the target drug according to the predicted consumption data in each sub-period and a preset safety inventory threshold of the target drug if a purchase order generation service request input for the target drug is detected, where the purchase order includes: a purchase amount of the target drug.

Optionally, the prediction module 602 is specifically configured to:

Optionally, the calculating module 603 is further configured to:

Optionally, the method further comprises:

a determining module 605, configured to determine available days of the target drug according to the predicted consumption data in each sub-period and the inventory data of the target drug;

and the judging module 606 is configured to judge whether the target drug is out of stock in a preset purchasing period and a due date according to the available days.

Optionally, the calculating module 603 is specifically configured to:

The description of the processing flow of each module in the device and the interaction flow between the modules may refer to the related description in the above method embodiments, and will not be described in detail here.

Fig. 7 is a schematic structural diagram of an electronic device according to an embodiment of the present application, and as shown in fig. 7, the electronic device includes: a processor 701, a memory 702 and a bus 703, wherein the memory 702 stores a computer program executable by the processor 701, when the electronic device runs, the processor 701 communicates with the memory 702 through the bus 703, and the processor 701 executes the computer program to perform the above method embodiments.

Embodiments of the present application further provide a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to perform the above method embodiments.

It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the system and the apparatus described above may refer to corresponding processes in the method embodiments, and are not described in detail in this application. In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. The above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical division, and there may be other divisions in actual implementation, and for example, a plurality of modules or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or modules through some communication interfaces, and may be in an electrical, mechanical or other form.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The functions may be stored in a computer-readable storage medium if they are implemented in the form of software functional units and sold or used as separate products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application.

Claims

1. A method for processing drug data, comprising:

2. The method of claim 1, wherein the predicting using the historical cycle consumption data to obtain the predicted cycle consumption data of the target drug in a preset future cycle comprises:

3. The method of claim 2, wherein the predicting using the historical cycle consumption data to obtain the predicted cycle consumption data of the target drug in a preset future cycle further comprises:

if the time distribution of the historical periodic consumption data meets a second preset condition, predicting the historical periodic consumption data by adopting a time sequence prediction model to obtain third predicted consumption data of the target medicine in the preset future period;

4. The method of claim 2, wherein said deriving said predicted cycle consumption data from said first predicted consumption data and said second predicted consumption data comprises:

5. The method of claim 4, wherein said deriving said predicted cycle consumption data from said first predicted consumption data and said second predicted consumption data further comprises:

and if the first check index and/or the second check index are not in the corresponding check range, determining that the second predicted consumption data are the predicted period consumption data.

6. The method of claim 1, wherein before generating the purchase order for the target drug according to the predicted consumption data of the respective sub-periods and the preset safe inventory threshold of the target drug, the method further comprises:

7. The method of claim 1, further comprising:

8. The method according to claim 1, wherein said calculating predicted consumption data for each sub-period in the preset future period based on the predicted period consumption data comprises:

9. A drug data processing apparatus, comprising:

10. An electronic device, comprising: a processor, a memory and a bus, the memory storing a computer program executable by the processor, the processor and the memory communicating via the bus when the electronic device is running, the processor executing the computer program to perform the method of any of claims 1 to 8.