Apparatus and method for predicting demand for medicine in medical institution, and program thereof

Abstract

To provide an unprecedented method for predicting demand for medicines in a medical institution.SOLUTION: An apparatus 100 acquires prescription information corresponding to a plurality of prescriptions received by a pharmacy (S201). Then, the apparatus 100 distinguishes demand of patients coming to the pharmacy in the future again for treatment of the same diseases, from demand of first-time patients, and makes predictions separately (S202 and S203). The apparatus generates final time series of predicted demand quantities for every medicine based on each-day values of prediction results related to the respective medicines (S204).SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] This invention relates to an apparatus, method, and program for predicting the demand for pharmaceuticals in a medical institution. [Background technology]

[0002] Pharmacies and other medical institutions purchase pharmaceuticals from wholesalers and other distributors, and prepare and dispense one or more prescribed medications to patients who visit the pharmacy, as needed. Since medical institutions need to dispense medications at the patient's request, they strive to maintain sufficient stock in terms of both type and quantity.

[0003] Each drug is assigned a code to identify it, and even if the active ingredient is the same, different manufacturers may assign different codes to different products. In actual distribution, each product is sold in certain units, such as the minimum order unit (sales packaging unit) and the original packaging unit, which is a package containing multiple sales packaging units. Codes are also assigned to identify these packaging units, and a coding system called GS1 is used internationally. For example, a sales packaging unit of 10 PTP packaging sheets containing 10 tablets of "Loxonin Tablets 60mg" is assigned the code "14987081105400" according to the GS1 system. Medical institutions specify these codes and procure the necessary types and quantities of products. [Prior art documents] [Patent Documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 2020-166859 [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] However, currently, healthcare institutions are unable to predict the demand for each type of drug, the extent of the demand, and when it will occur. As a result, they are placing orders based on past experience, leading to inefficiencies such as insufficient stock and the need to urgently request deliveries, or conversely, being stuck with excessive inventory.

[0006] While Patent Document 1 describes a technology for medical institutions that learns the trends in the issuance of pharmaceuticals and predicts the demand for those pharmaceuticals based on the learning results, the inventors have come up with a new idea.

[0007] This invention has been made in view of these problems, and its purpose is to provide a novel device, method, and program for predicting the demand for pharmaceuticals in medical institutions. [Means for solving the problem]

[0008] To achieve this objective, a first aspect of the present invention is a method for predicting the demand for drugs in a medical institution, comprising: a first step of obtaining prescription information corresponding to a first prescription received by the medical institution; a second step of calculating the predicted visit date and predicted demand quantity of a patient associated with the prescription information for each drug included in the prescription information; a third step of repeating the first and second steps for a second prescription received by the medical institution; a fourth step of calculating the predicted visit date and predicted demand quantity of a first-time patient for each drug included in the prescription information corresponding to the first and second prescriptions; and a fifth step of generating a time series of predicted demand quantities for each drug based on the daily values ​​of the predicted demand quantities for each drug obtained in the third and fourth steps.

[0009] Furthermore, a second aspect of the present invention is the method of the first aspect, wherein the second step includes estimating one or more diseases in which one or more drugs included in the prescription information are used, and calculating at least one of the predicted visit date and predicted demand quantity using a disease-specific prediction model.

[0010] Furthermore, a third aspect of the present invention is the method of the first aspect, wherein the second step includes: estimating a single disease in which one or more drugs included in the prescription information are used; calculating the predicted number of times the patient will return for one or more drugs used for the disease; terminating the prediction process if the predicted number of times is 0; and calculating the predicted dates of subsequent visits and the predicted demand quantities of one or more drugs used for the disease if the predicted number of times is 1 or more.

[0011] Furthermore, a fourth aspect of the present invention is the method of the first aspect, wherein the second step includes estimating a plurality of diseases in which one or more drugs included in the prescription information are used; calculating the predicted number of times the patient will revisit for each of the plurality of diseases for one or more drugs used for that disease; terminating the prediction process if the predicted number of times is 0, and calculating the predicted visit dates for subsequent visits and the disease-specific predicted demand quantities of one or more drugs used for that disease if the predicted number of times is 1 or more, wherein the predicted demand quantities calculated by the second step are calculated by multiplying the predicted demand quantity for each disease by a value corresponding to the weight assigned to each of the plurality of diseases, and summing the results for the plurality of diseases.

[0012] Furthermore, a fifth aspect of the present invention is the method according to the third or fourth aspect, wherein in the second step, at least a portion of the patient's past prescription information is used to calculate the predicted number of times.

[0013] Furthermore, a sixth aspect of the present invention is a method according to any of the first to fifth aspects, wherein the second step is to calculate the predicted visit date using the visit history for each disease.

[0014] Furthermore, a seventh aspect of the present invention is a method according to any of the first to sixth aspects, wherein the second step is to calculate the predicted visit date based on the results of individually predicting the next visit week and the next day of the week.

[0015] Furthermore, an eighth aspect of the present invention is a method according to any of the first to seventh aspects, the fourth step comprising: generating a time series of drug-specific demand quantities by first-time patients in a region based on a plurality of prescriptions received by a plurality of pharmacies in the same region as the medical institution; classifying the plurality of drugs for which drug-specific time series have been generated into a plurality of clusters to generate a time series of cluster-specific demand quantities; and for each drug included in the prescription information corresponding to the first and second prescriptions, calculating the predicted visit date and predicted demand quantity of first-time patients using, in addition to prescription information received by the medical institution in a first period of one year, features obtained from the cluster-specific time series in a second period prior to the first period of the cluster to which the drug belongs.

[0016] Furthermore, a ninth aspect of the present invention is a program for causing a computer to perform a method for predicting the demand for drugs in a medical institution, the method comprising: a first step of obtaining prescription information corresponding to a first prescription received by the medical institution; a second step of calculating the predicted visit date and predicted demand quantity of a patient associated with the prescription information for each drug included in the prescription information; a third step of repeating the first and second steps for a second prescription received by the medical institution; a fourth step of calculating the predicted visit date and predicted demand quantity of a first-time patient for each drug included in the prescription information corresponding to the first and second prescriptions; and a fifth step of generating a time series of predicted demand quantities for each drug based on the daily values ​​of the predicted demand quantities for each drug obtained in the third and fourth steps.

[0017] Moreover, a tenth aspect of the present invention is an apparatus for predicting the demand for drugs in a medical institution, which acquires prescription information corresponding to a first prescription received by the medical institution, and for each drug included in the prescription information, calculates a predicted visit date and a predicted demand quantity for the patient's next visit associated with the prescription information, repeats the calculation of the predicted visit date and the predicted demand quantity for a second prescription received by the medical institution, calculates a predicted visit date and a predicted demand quantity for a first-time visiting patient for each drug included in the prescription information corresponding to the first and second prescriptions, and generates a time series of the predicted demand quantity for each drug based on the values of the predicted demand quantity for each day of each obtained drug.

Effects of the Invention

[0018] According to one aspect of the present invention, by acquiring prescription information corresponding to a plurality of prescriptions received by a pharmacy, distinguishing and separately predicting the demand by patients who will visit the pharmacy in the future and the demand by first-time visiting patients, it is possible to provide a novel method for predicting the demand for drugs in a medical institution that has never existed before.

Brief Description of the Drawings

[0019] [Figure 1] It is a diagram showing a demand prediction apparatus according to an embodiment of the present invention. [Figure 2] It is a diagram showing an outline of a demand prediction method according to an embodiment of the present invention. [Figure 3] It is a diagram showing the flow of a demand prediction method for repeat visitors according to an embodiment of the present invention.

Modes for Carrying Out the Invention

[0020] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[0021] Figure 1 shows a demand forecasting device according to one embodiment of the present invention. The device 100 acquires prescription information corresponding to multiple prescriptions and generates a time series of predicted demand quantities for each drug included in the prescription information. Prescription information can be acquired by generating it by capturing a two-dimensional code written on a prescription received by a pharmacy using equipment with an image sensor installed in the pharmacy, and inputting the generated prescription information into the device 100 via wired or wireless connection, or by electronically receiving prescription information corresponding to prescriptions received by a pharmacy from equipment that can communicate via a computer network such as an IP network. The time series of predicted demand quantities can be a series of daily predicted demand quantities for each drug. Preferably, the device 100 is a server that provides demand forecasting services to multiple pharmacies.

[0022] The device 100 comprises a communication unit 101 such as a communication interface, a processing unit 102 such as a processor or CPU, and a storage unit 103 including a storage device or storage medium such as memory or a hard disk. The processing unit 102 executes a program for performing each process. The device 100 may include one or more devices, computers, or servers. The program may include one or more programs and can be recorded on a computer-readable storage medium to form a non-transient program product. The program can be stored in the storage unit 103 or in a storage device or storage medium such as a database 104 accessible via an IP network from the device 100, and executed in the processing unit 102. The data described below as being stored in the storage unit 103 may be stored in the database 104, and vice versa.

[0023] Figure 2 shows an overview of a demand forecasting method according to one embodiment of the present invention. The device 100 acquires prescription information corresponding to multiple prescriptions received by a pharmacy (S201). Next, the device 100 distinguishes between the demand from patients who will return to the pharmacy for treatment of the same disease and the demand from patients who are visiting for the first time, and forecasts them separately (S202, S203). Then, by calculating the sum of the daily values ​​of the forecast results for each drug, it generates a time series of the final forecasted demand quantity for each drug (S204). In Figure 2, the forecast for returning patients is shown followed by the forecast for first-time visiting patients, but the order does not matter. Also, in Figure 2, the order is shown in which prescription information is acquired first and then the forecasting process is performed, but prescription information may be acquired sequentially as needed for the forecasting process. Furthermore, although it has been explained that a time series is generated by calculating the sum for each day, more generally, the time series of forecasted demand quantity can be generated by calculating the final forecasted demand quantity for each day based on the daily values ​​of the forecast results for each drug.

[0024] Below, we will first explain the predictions for returning patients, and then the predictions for first-time patients. The reason for performing predictions for returning patients and first-time patients separately is that the inventors have found that these predictions are highly independent.

[0025] In this specification, it is preferable to consider probabilistic elements in all predictions or estimations and to calculate the expected value and the upper and lower limits of the confidence interval as a result. For example, a normal distribution can be assumed for the predicted demand quantity, so that the expected value of the predicted demand quantity can be obtained by summing the expected value predicted for returning patients and the expected value predicted for first-time patients. Furthermore, due to the additivity of variances in a normal distribution, the standard deviation can be obtained by taking the square root of the variance obtained for returning patients and the variance obtained for first-time patients, and the confidence interval can be determined using this standard deviation. However, it is not necessary to consider probabilistic elements in all predictions or estimations described in this specification.

[0026] Predictions regarding returning patients Figure 3 shows the flow of a demand forecasting method for returning patients according to one embodiment of the present invention. The device 100 performs demand forecasting for prescription information corresponding to each of the multiple prescriptions received by the pharmacy that is the target of demand forecasting, and these results form a time series as a series of predicted demand quantities. Figure 3 shows the processing performed on prescription information corresponding to each prescription.

[0027] First, the device 100 estimates the diseases in which one or more drugs included in the acquired prescription information are used (S301). Although the prescription r does not describe the disease d in which the patient is being treated, one or more drugs are prescribed to treat one or more diseases of the patient. Therefore, it is theoretically possible to estimate one or more diseases of the patient based on the one or more drugs included in the prescription information. In this embodiment, as an example, the correspondence between one or more drugs and the strength of the association between each drug and one or more diseases in which it can be used is stored in the storage unit 103, and the strength of the association with each disease can be determined by referring to this correspondence based on each drug included in the prescription information. Then, the strength of the association with each disease determined for all drugs included in the prescription information is summed up for each disease, and the top N diseases (N is an integer of 1 or more) can be estimated as the patient's diseases. The value of the strength of the association between each drug and one or more diseases in which it can be used can be determined, for example, according to the frequency with which each drug has been used for each disease in the past.

[0028] The strength of the association between drug m and disease d maintained by this correspondence is weighted ω by the value of that strength. m,d It can be expressed as follows: Mr is the set of one or more drugs included in prescription r, and ω is the sum of the values ​​for each disease. d This can be expressed by the following equation.

[0029]

number

[0030] The following table shows the drug m Ais associated with disease d1 with a strength of 2.0 and is associated with disease d2 with a strength of 1.5, and drug m B is shown to be associated with disease d1 with a strength of 3.0 and is associated with disease d3 with a strength of 1.0.

[0031]

Table 1

[0032] When summed for each disease, it becomes 5.0 for disease d1, 1.5 for disease d2, and 1.0 for disease d3. If the top 1 is used as a criterion, disease d1 is estimated, and if the top 2 are used as a criterion, diseases d1 and d2 are estimated. Hereinafter, an example of estimating the top 1 as a disease will be mainly described. In addition, when N (N is an integer of 2 or more) diseases are estimated, for each disease d n the predicted demand quantity for each disease is multiplied by the value a n corresponding to the weight ω dn assigned to that disease d n and the sum is calculated to obtain the predicted demand quantity of each drug based on one prescription r. As an example, for each disease d n the weight ω dn assigned to that disease d n and the value a

[0033]

Equation

[0034] Next, the device 100 inputs at least a portion of the prescription information into a disease-specific prediction model to predict the number of follow-up visits for patients associated with that prescription information (S202). The input data includes age, gender, and medical department. If the number is 0, the prediction process ends. If the number is n (where n is an integer greater than or equal to 1), the device predicts the date of the first visit (S303) and the quantity demanded (S304). If n is 2 or greater, the device predicts the dates of the second to nth visits and the quantities demanded as needed. For example, in the case of acute illnesses such as colds and influenza, there may be no follow-up visits, and treatment may be completed in the first visit. In the case of chronic diseases such as diabetes, several or more follow-up visits are expected.

[0035] A model for predicting the number of return visits can be generated using supervised learning, more specifically, using supervised learning with gradient boosting such as LightGBM or XGBoost. However, it can be generated using any machine learning method that is currently available or will be available in the future. Furthermore, the model for predicting the number of return visits is not limited to those generated by machine learning; it can also be any model that takes at least a portion of prescription information as input and outputs the number of return visits accordingly. It is preferable to include at least a portion of the patient's past prescription information associated with the prescription information as input data to obtain higher accuracy.

[0036] Predicting the date of a follow-up visit can be done using a predictive model generated by machine learning or any other currently available or future available method, but it can also be done using the visit interval distribution obtained from the past visit history of patients associated with the prescription information or patients with the associated disease. The visit interval distribution may be calculated in units of days, but it may also be calculated in units of weeks, and the visit day distribution may also be used to predict a specific visit date.

[0037] Based on the inventors' research, it is preferable to use the fact that the prediction of the interval between visits (in weeks) and the prediction of the day of the week for visits are highly independent when predicting the date of the next visit. Based on the results of predicting these separately, the prediction of the date of the next visit, which is the product of these two events, can be made.

[0038] Demand quantity forecasting can also be done using predictive models generated by machine learning or any other currently available or future available method, but it can also be done using, for example, the sales quantity distribution obtained from the historical sales history of patients with the disease to which the prescription information is associated.

[0039] Predictions regarding first-time patients The date of first-time patient visit and the quantity of medication requested can be predicted using a time-series forecasting algorithm, based on past prescription information associated with the pharmacy.

[0040] Prescription information associated with a pharmacy may include not only prescriptions received by the pharmacy being forecasted, but also prescriptions received by other pharmacies associated with that pharmacy. For example, by using prescription information from one or more other pharmacies in the same area as the pharmacy in question, it is possible to determine the seasonality of drug demand in that area and make predictions about first-time patients based on that seasonality.

[0041] More specifically, the following example can be given. First, using prescription information based on multiple prescriptions received over a period of more than one year by multiple pharmacies in the same area as the pharmacy targeted for demand forecasting, a time series of drug-specific demand quantities by first-time patients in that area is generated. Each drug-specific time series reflects differences depending on the time of year, covering at least one year. Here, it is not necessary to generate a time series for some drugs, such as when there is insufficient data or when the need is not high. Also, it is not necessarily the case that less than one year is insufficient. Next, the multiple drugs for which time series have been generated are classified into multiple clusters using a clustering method such as the k-medoids method, and cluster-specific time series of demand quantities are generated. Each of the above time series can be a series of demand quantities on a daily basis, or a series of demand quantities on a predetermined unit such as a weekly basis. Then, from the multiple cluster-specific time series, the cluster-specific time series of the drug targeted for demand forecasting belongs is obtained, and the predicted visit date and predicted demand quantity for that drug by first-time patients are calculated using this cluster-specific time series. It should be noted that this prediction method for first-time patients, separate from predictions for returning patients, is in itself a novel method for forecasting drug demand in healthcare institutions.

[0042] Specifically, for a drug whose demand is being forecasted, the forecast date of first-time patients and the forecasted demand quantity can be calculated by using prescription information received by the pharmacy in the first period of the year, in addition to the demand quantity included in the cluster-specific time series for a second period prior to the first period, or a value obtained by applying the necessary calculations to that quantity. For example, if a pharmacy is only given prescription information based on prescriptions received between January and March for a certain drug, by using the demand quantity included in the cluster-specific time series for the cluster to which the drug belongs from April onwards, or a value obtained by applying the necessary calculations to that quantity, as a feature, it becomes possible to predict with high accuracy how the demand quantity will change from April onwards.

[0043] While the above explanation primarily uses pharmacies as an example, prescription information corresponding to prescriptions received by any medical institution can be used, and in that case, the part that says "visit to the pharmacy" should be understood as "visit to the pharmacy."

[0044] In the embodiments described above, unless the word "only" is used, such as "based only on XX," "according only to XX," or "in the case of XX only," it should be noted that this specification assumes that additional information may also be considered. Furthermore, as an example, the statement "if a, then b" does not necessarily mean "always b in the case of a" or "b immediately after a," unless explicitly stated otherwise. Also, the statement "each a constituting A" does not necessarily mean that A is composed of multiple components, but rather includes the possibility that a component is singular.

[0045] Furthermore, for the sake of clarity, even if there are aspects of operation in some method, program, terminal, device, server, or system (hereinafter referred to as "method, etc.") that differ from the operation described herein, each aspect of the present invention is intended to cover the same operation as any of the operations described herein, and the existence of operation different from the operation described herein does not mean that such method, etc. is outside the scope of each aspect of the present invention.

[0046] Furthermore, the "start" and "end" shown in Figures 2 and 3 are merely examples and do not necessarily mean that the method according to this embodiment will always start or end in the illustrated procedure. [Explanation of Symbols]

[0047] 100 devices 101 Communications Department 102 Processing Unit 103 Storage section 104 Databases

Claims

1. 1. A method for predicting drug demand in a medical institution, comprising: A first step in which a computer obtains prescription information corresponding to a first prescription; a second step in which the computer calculates, for each drug included in the prescription information, a predicted visit date and a predicted demand quantity for a patient associated with the prescription information; the computer repeating the first and second steps for a second prescription; the computer calculates a predicted visit date and a predicted demand quantity for each drug included in prescription information corresponding to the first and second prescriptions based on prescription information associated with a plurality of prescriptions received by a plurality of medical institutions in the same area as the medical institution; A method for predicting demand for a drug, including:

2. 10. The method of claim 1, The second step includes: A step of estimating one or more diseases for which one or more drugs included in the prescription information are used, At least one of the predicted visit date and the predicted demand quantity is calculated using a disease-specific prediction model.

3. 2. The method of claim 1, The second step includes: A step of estimating a single disease for which one or more drugs included in the prescription information are used; calculating an expected number of return visits for one or more medications for the condition; a step of terminating the prediction process when the predicted number of times is 0, and calculating a predicted visit date from the next visit onwards and a predicted demand quantity of one or more drugs to be used for the disease when the predicted number of times is 1 or more; Includes.

4. 2. The method of claim 1, The second step includes: A step of estimating a plurality of diseases for which one or more drugs included in the prescription information are used; For each of the plurality of diseases, calculating an expected number of return visits for one or more medications for the condition; a step of terminating the prediction process when the predicted number of times is 0, and calculating a predicted visit date from the next visit onwards and a disease-specific predicted demand quantity of one or more drugs to be used for the disease when the predicted number of times is 1 or more; Including, The predicted demand quantity calculated in the second step is calculated by multiplying the predicted demand quantity for each disease by a value corresponding to the weight assigned to each of the plurality of diseases and summing the results for the plurality of diseases.

5. 5. The method according to claim 3 or 4, In the second step, at least a portion of the patient's past prescription information is used to calculate the predicted number.

6. 6. A method according to any one of claims 1 to 5, comprising: The second step calculates the predicted visit date using visit history by disease.

7. 7. A method according to any one of claims 1 to 6, comprising: The second step calculates the predicted visit date based on the results of individually predicting the next and subsequent visit weeks and days of the week.

8. 8. A method according to any one of claims 1 to 7, comprising: The method further includes a step in which the computer generates a time series of drug-specific demand quantities by patients visiting for the first time based on multiple prescription information associated with multiple prescriptions received by multiple medical institutions in the same area as the medical institution.

9. 9. A method according to any one of claims 1 to 8, comprising: The method further includes a step in which the computer classifies the plurality of drugs into a plurality of clusters to generate a cluster-specific time series of demand quantities.

10. 10. A method according to any one of claims 1 to 9, comprising: The method further includes a step in which the computer calculates, for each drug included in prescription information corresponding to the first and second prescriptions, a predicted visit date and a predicted demand quantity for a first-time patient by using, in addition to the prescription information received by the medical institution during a first period of one year, features obtained from a cluster-specific time series for a cluster to which the drug belongs during a second period prior to the first period.

11. A program for causing a computer to execute a method for predicting drug demand in a medical institution, the method comprising: a first step of obtaining prescription information corresponding to a first prescription; a second step of calculating, for each drug included in the prescription information, a predicted visit date and a predicted demand quantity for a patient associated with the prescription information; a third step of repeating the first and second steps for a second prescription; calculating a predicted patient visit date and a predicted demand quantity for each drug included in prescription information corresponding to the first and second prescriptions based on prescription information associated with multiple prescriptions received by multiple medical institutions in the same area as the medical institution; Programs including.

12. A device for predicting drug demand in a medical institution, obtaining prescription information corresponding to the first prescription and calculating, for each drug included in the prescription information, a predicted return visit date and a predicted demand quantity for the patient associated with the prescription information; repeating the calculation of the predicted visit date and the predicted demand quantity for a second prescription; For each drug included in prescription information corresponding to the first and second prescriptions, calculate a predicted visit date and a predicted demand quantity for a first-time patient based on prescription information associated with multiple prescriptions received by multiple medical institutions in the same area as the medical institution; Based on the obtained values ​​of the predicted demand quantity for each drug on each day, a time series of the predicted demand quantity is generated for each drug.

Images