Information processing method, information processing device, and program

Abstract

To provide an information processing method, an information processing device, and a program capable of suitably supporting activity for a customer through a digital channel.SOLUTION: In a Medical Representative (MR) support system composed of a server 1 and a terminal 2, a server (information processing device) 1 executes processing of: acquiring medical worker information related to a medical worker; when inputting the medical worker information and use values for each of a plurality of channels, including a digital channel, which the MR uses when conducting an activity for the medical worker, by inputting the acquired medical worker information and use values for each channel of a plurality of patterns with different combinations of use values for each channel into a first model learned to calculate a result value of the MR activity, calculating result values for each pattern; and determining, on the basis of the calculated result values for each pattern, the use values of each channel that should be recommended to the MR.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an information processing method, an information processing device, and a program. [Background technology]

[0002] There are technologies that support salespeople who are engaged in activities such as product explanations to customers. For example, Patent Document 1 discloses an information processing system for creating sales plans for salespeople such as MRs (Medical Representatives), which constructs a prediction model for calculating predicted values ​​of sales evaluation indicators based on historical data of sales activities to customers and actual measurement data of performance evaluation indicators of products that are the object of the sales activities. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6676206 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the invention of Patent Document 1 merely constructs a prediction model by a simple regression analysis.

[0005] In one aspect, an object of the present invention is to provide an information processing method etc. that can favorably support activities for customers through digital channels. [Means for solving the problem]

[0006] In one aspect, an information processing method involves acquiring medical worker information related to a medical worker, and when the medical worker information and usage values ​​of a plurality of channels, including digital channels, that an MR uses when performing activities for the medical worker are input, a computer executes a process in which the acquired medical worker information and usage values ​​of each of a plurality of patterns, each having different combinations of usage values ​​of each channel, are input into a first model that has been trained to calculate an outcome value of the MR's activities, thereby calculating the outcome value corresponding to each pattern, and determining the usage value of each of the channels to be recommended to the MR based on the calculated outcome value for each pattern.

[0007] In one aspect of the information processing method, a degree of increase in prescriptions of a medicine by the medical professional is calculated as the outcome value.

[0008] In one aspect of the information processing method, the utilization value is the number of activities per unit period for the medical professional through the channel, and the number of activities for each channel is determined based on the outcome value.

[0009] In one aspect of the information processing method, the medical worker information includes attributes of the medical worker, attributes of the medical facility to which the medical worker belongs, characteristic values ​​representing the behavioral characteristics of the medical worker, or the degree of intimacy between the medical worker and the MR.

[0010] In one aspect of the information processing method, the characteristic value includes an aggregate value of viewing status of predetermined digital content related to medical care by the medical staff.

[0011] In one aspect, the information processing method estimates a utilization value of each channel by inputting the acquired medical professional information into a second model that has been trained to estimate the utilization value of each channel used by the medical professional when receiving activities from the MR when the medical professional information is input, generates a utilization value of each channel of the multiple patterns to be input into the first model based on the estimated utilization value of each channel, and calculates the outcome value corresponding to each pattern by inputting the utilization value of each generated pattern into the first model.

[0012] An information processing method according to one aspect estimates a utilization value of each of the channels using a plurality of the second models that have been trained to estimate utilization values ​​of the channels that are different from each other.

[0013] An information processing method according to one aspect calculates the contribution of each of a plurality of parameters included in the medical staff information to the utilization value based on the medical staff information input into the second model and the utilization value estimated by the second model, and outputs the calculated contribution of each parameter.

[0014] An information processing device according to one aspect includes an acquisition unit that acquires medical worker information related to a medical worker, a calculation unit that calculates the outcome value of the MR's activities by inputting the acquired medical worker information and the usage values ​​of each of a plurality of patterns in which the combinations of the usage values ​​of each channel are different into a first model that has been trained to calculate the outcome value of the MR's activities when the medical worker information and usage values ​​of each of a plurality of channels, including digital channels, that the MR uses when performing activities for the medical worker are input, and a determination unit that determines the usage values ​​of each of the channels to be recommended to the MR based on the calculated outcome value of each pattern.

[0015] In one aspect of the information processing method, a computer acquires medical professional information related to a medical professional, and when the medical professional information is input, a second model is trained to estimate the utilization value of each of a plurality of channels, including digital channels, that the medical professional uses when receiving activities from MR, by inputting the acquired medical professional information into the second model, which estimates the utilization value of each channel.

[0016] A program according to one aspect displays on a display unit a list of multiple medical professionals who are the targets of an MR's activities, and accepts a selection input to select one of the multiple medical professionals.When a selection input of a medical professional is accepted, a computer executes a process to display recommended usage values ​​for each of multiple channels, including digital channels, that the MR will use when carrying out activities with the medical professional, based on medical professional information related to the selected medical professional.

[0017] In one aspect of the program, the recommended value is determined using a second model that has been trained to estimate the utilization value of each channel used by the medical professional when receiving activity from the MR when the medical professional information is input, and the program displays the contribution of each of a number of parameters included in the medical professional information to the utilization value by channel, calculated based on the medical professional information input into the second model and the utilization value of each channel estimated by the second model. Effect of the Invention

[0018] In one aspect, it is possible to effectively support activities with customers through digital channels. [Brief description of the drawings]

[0019] [Figure 1] FIG. 1 is an explanatory diagram showing a configuration example of an MR support system. [Diagram 2] FIG. 2 is a block diagram showing an example of the configuration of a server. [Diagram 3] FIG. 2 is a block diagram showing a configuration example of a terminal. [Figure 4] FIG. 2 is an explanatory diagram showing an example of the record layout of an MRDB, a doctor DB, a facility DB, and a sales DB. [Diagram 5] FIG. 1 is an explanatory diagram showing an overview of the present embodiment. [Figure 6] FIG. 1 is an explanatory diagram relating to a first model. [Figure 7] FIG. 11 is an explanatory diagram relating to a second model. [Figure 8] FIG. 2 is an explanatory diagram illustrating an example of a display screen of a terminal. [Figure 9] 11 is a flowchart showing a procedure for generating a first model. [Figure 10] 13 is a flowchart showing a procedure for generating a second model. [Figure 11] 13 is a flowchart showing a procedure for a process of determining a recommended value. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] The present invention will now be described in detail with reference to the drawings showing embodiments thereof. (Embodiment) Fig. 1 is an explanatory diagram showing a configuration example of an MR support system. In this embodiment, an MR support system is described that presents a recommended usage value of each channel (the number of activities with medical professionals through each channel) when an MR uses each digital channel such as a web interview, telephone, and email to conduct sales activities, etc., for medical professionals who are customers. The MR support system includes an information processing device 1, and terminals 2, 2, 2... Each device is connected to a network N such as the Internet.

[0021] In this specification, the term "digital channel" refers to a means of transmitting information via a communication line, and the specific means are not particularly limited. For example, digital channels may include web interviews (web conferences), telephone, e-mail, SNS (Social Networking Service), smartphone apps, etc.

[0022] The information processing device 1 is an information processing device capable of various information processing and transmitting and receiving information, such as a server computer, a personal computer, etc. In this embodiment, the information processing device 1 is assumed to be a server computer, and for the sake of brevity, will be referred to as a server 1 below. In this embodiment, the server 1 estimates the preferences of medical professionals for digital channels using a machine learning model, and determines a recommended value for the usage of each channel when an MR conducts sales activities, etc., with medical professionals based on the estimated preferences.

[0023] A medical professional is a customer for whom an MR is active, such as a doctor. Note that the medical professional is not limited to a doctor, and may be, for example, a pharmacist.

[0024] As described later, the server 1 uses the second model 52 that has been trained on predetermined training data to estimate the usage value of each channel, including digital channels, that is, the extent to which medical personnel will use each channel (see Figures 5, 7, etc.). In this way, the server 1 estimates the digital channel preferences of medical personnel. Then, based on the estimation results by the second model 52, the server 1 uses the first model 51 that has been trained on predetermined training data to determine the usage value of each channel to be recommended to MRs (hereinafter referred to as the "recommended value").

[0025] The usage value is the amount of use of each channel, such as web interviews, telephone calls, and emails, and in this embodiment, it refers to the number of activities with users through each channel per unit period (for example, one month). Note that the usage value is not limited to the number of activities, and may be the activity time (interview time), etc.

[0026] In addition, "channels including digital channels" may include information transmission means (digital channels) via communication lines such as web interviews, telephone, and e-mail, as well as information transmission means that do not involve communication lines such as face-to-face interviews and mailing of documents. In other words, in this embodiment, channels other than digital channels are also included as activity tools used by MRs.

[0027] MR activities refer to the activities that MRs perform toward medical professionals as part of their job. For example, MR activities include sales activities for pharmaceutical products provided by pharmaceutical companies. MR activities can also include not only sales activities, but also activities such as providing information about pharmaceutical products to medical professionals and collecting information from medical professionals.

[0028] The terminal 2 is an information processing terminal used by the MR, and is, for example, a personal computer, a smartphone, a tablet terminal, etc. As will be described later, the server 1 displays the recommended values ​​for each channel determined using the first model 51 on the terminal 2 and presents them to the MR.

[0029] 2 is a block diagram showing an example of the configuration of the server 1. The server 1 includes a control unit 11, a main storage unit 12, a communication unit 13, and an auxiliary storage unit . The control unit 11 has one or more processors such as a central processing unit (CPU), a micro-processing unit (MPU), a graphics processing unit (GPU), etc., and performs various information processing by reading and executing a program P1 stored in the auxiliary storage unit 14. The main storage unit 12 is a temporary storage area such as a static random access memory (SRAM), a dynamic random access memory (DRAM), a flash memory, etc., and temporarily stores data required for the control unit 11 to execute arithmetic processing. The communication unit 13 is a communication module for performing processing related to communication, and transmits and receives information to and from the outside.

[0030] The auxiliary storage unit 14 is a non-volatile storage area such as a large-capacity memory or a hard disk, and stores a program P1 and other data required for the control unit 11 to execute processing. The auxiliary storage unit 14 also stores a first model 51, a second model 52, an MRDB 141, a doctor DB 142, a facility DB 143, and a sales DB 144.

[0031] The first model 51 is a machine learning model that has learned predetermined training data, and is a model that calculates the outcome value of the MR's activity when medical professional information related to the medical professional and the usage value of each channel including the digital channel are input. The second model 52 is a machine learning model that has learned predetermined training data, and is a model that estimates the usage value of each channel of the medical professional when medical professional information is input. The first model 51 and the second model 52 are expected to be used as program modules that constitute part of artificial intelligence software.

[0032] The MRDB 141 is a database that stores information on MRs. The doctor DB 142 is a database that stores information on doctors (medical professionals) who are the targets of activities. The facility DB 143 is a database that stores information on medical facilities (hospitals, etc.) to which doctors belong (work). The sales DB 144 is a database that stores information on pharmaceutical sales (amounts prescribed by medical professionals).

[0033] The auxiliary storage unit 14 may be an external storage device connected to the server 1. The server 1 may be a multi-computer consisting of a plurality of computers, or may be a virtual machine virtually constructed by software.

[0034] In the present embodiment, the server 1 is not limited to the above configuration, and may include, for example, an input unit that accepts operation input, a display unit that displays images, etc. The server 1 may also include a reading unit that reads a non-transitory computer-readable recording medium 1a and reads the program P1 from the recording medium 1a. The program P1 may be executed on a single computer, or on multiple computers interconnected via a network N.

[0035] 3 is a block diagram showing an example of the configuration of the terminal 2. The terminal 2 includes a control unit 21, a main memory unit 22, a communication unit 23, a display unit 24, an input unit 25, and an auxiliary memory unit . The control unit 21 has one or more processors such as CPUs, and performs various information processing by reading and executing the program P2 stored in the auxiliary storage unit 26. The main storage unit 22 is a temporary storage area such as RAM, and temporarily stores data necessary for the control unit 21 to execute arithmetic processing. The communication unit 23 is a communication module for performing processing related to communication, and transmits and receives information to and from the outside. The display unit 24 is a display screen such as a liquid crystal display, and displays images. The input unit 25 is an operation interface such as a keyboard or mouse, and accepts operation input from the user. The auxiliary storage unit 26 is a non-volatile storage area such as a hard disk or large-capacity memory, and stores the program P2 and other data necessary for the control unit 21 to execute processing.

[0036] The terminal 2 may include a reading unit for reading the non-transitory computer-readable recording medium 2a and may read the program P2 from the recording medium 2a. The program P2 may be executed on a single computer or on multiple computers interconnected via a network N.

[0037] FIG. 4 is an explanatory diagram showing an example of the record layout of the MRDB 141, the doctor DB 142, the facility DB 143, and the sales DB 144. The MRDB141 includes an MRID column, an MR name column, a doctor in charge column, and an activity history column. The MRID column stores an MRID for identifying each MR. The MR name column, the doctor in charge column, and the activity history column store, in association with the MRID, the name of the MR, the doctor (medical worker) in charge of the MR, and the activity history of the MR with respect to the doctor. The activity history column stores, for example, the date and time of the activity, the medical worker who is the target of the activity, and the activity content (channel).

[0038] The doctor DB 142 includes a doctor ID column, a doctor name column, an affiliation column, a doctor attribute column, a behavioral characteristic column, and an intimacy column. The doctor ID column stores a doctor ID for identifying each doctor (medical worker). The doctor name column, affiliation column, doctor attribute column, behavioral characteristic column, and intimacy column store the doctor's name, the medical facility to which the doctor belongs, the doctor's attributes, behavioral characteristics, and intimacy with the MR in charge, respectively, in association with the doctor ID. The behavioral characteristic column stores characteristic values ​​that represent the doctor's behavioral characteristics, such as the number of times a website is viewed, the open rate of emails, and other aggregated values ​​that tally up the viewing status of specified digital content.

[0039] The facility DB143 includes a facility ID column, a facility name column, and a facility attribute column. The facility ID column stores a facility ID for identifying each medical facility. The facility name column and the facility attribute column store the name and attribute of the medical facility, respectively. The facility attribute column stores, for example, a classification of a medical facility uniquely defined by this system, a market share of pharmaceuticals, and a viewing rate of digital content within the medical facility.

[0040] The sales DB 144 includes a drug ID column, a drug name column, and a sales column. The drug ID column stores a drug ID for identifying each drug. The drug name column and sales column store the drug name and drug sales, respectively, in association with the drug ID. The sales column stores the prescription amount (sales quantity) and sales amount of the drug in association with, for example, the year and month, medical facility, and medical worker.

[0041] Fig. 5 is an explanatory diagram showing an outline of this embodiment. Fig. 5 illustrates a state in which a result value of an MR's activity (medical drug prescription growth rate) is estimated by inputting medical worker information (attributes of a medical worker (doctor), attributes of a medical facility, behavioral characteristics of a medical worker, and intimacy (relationship) with an MR) and a usage value (number of activities) of each of a plurality of channels into a first model 51. Also illustrated is a state in which a usage value of each channel of a medical worker is estimated by inputting medical worker information into a second model 52, and the estimated usage value is input into the first model 51. An outline of this embodiment will be described based on Fig. 5.

[0042] The first model 51 is a machine learning model that has learned predetermined training data, and is, for example, a neural network such as a CNN (Convolutional Neural Network). The first model 51 may be a neural network other than a CNN. The first model 51 may also be a model based on other learning algorithms, such as a decision tree, a random forest, or an SVM (Support Vector Machine).

[0043] The first model 51 is a model that predicts the results of an MR's activities, that is, the extent of the results when the MR performs activities on medical professionals through each channel. Specifically, the first model 51 calculates the results of the MR's activities when medical professional information for a certain period (e.g., the most recent six months from the present time) and the usage values ​​of each channel for an estimated target period different from the certain period (e.g., one month from the present time) are input.

[0044] In this embodiment, a first model 51 for predicting results is used to determine the extent to which an MR should use each channel, and a recommended value for each channel. Specifically, the server 1 prepares multiple patterns of usage values ​​for each channel to be input to the first model 51, and inputs the usage values ​​of each channel in the multiple patterns to the first model 51 to calculate a performance value corresponding to each pattern. The server 1 determines a recommended value (usage value) for each channel based on the performance value of each pattern estimated by the first model 51.

[0045] Fig. 6 is an explanatory diagram of the first model 51. Fig. 6 conceptually illustrates input and output data of the first model 51 at the time of learning and at the time of estimation. The first model 51 will be described with reference to Fig. 6.

[0046] As described above, the first model 51 receives medical staff information and the usage values ​​of each channel of the MR as inputs. The medical staff information includes attributes of the medical staff, attributes of the medical facility to which the medical staff belongs, characteristic values ​​representing the behavioral characteristics of the medical staff, and the degree of intimacy with the MR.

[0047] The attributes of medical professionals include, for example, the medical department of the medical professional (doctor), the specialty (the academic society to which the doctor belongs, the number of papers submitted, etc.), age, whether the doctor is a target doctor who should be a focus of sales, etc. Note that these are merely examples, and other attributes (for example, the position in the hospital, etc.) may also be included.

[0048] The attributes of the medical facility include, for example, the classification of the medical facility unique to this system, the market share of pharmaceuticals held by the medical facility, the viewing rate of digital content within the medical facility, etc. Note that these are merely examples, and other attributes (for example, the area where the medical facility is located, etc.) may also be included.

[0049] The viewing rate of digital content refers to the ratio of medical staff viewing a given digital content. Digital content refers to medical-related content, such as a website that lists medical information, or e-mail sent or distributed to medical staff by medical representatives or pharmaceutical companies. The server 1 treats the viewing rate of digital content throughout the entire medical facility as an attribute of the medical facility.

[0050] The characteristic value representing the behavioral characteristics of medical workers refers to an index value measuring the digital or medical behavior or preferences of medical workers. In this embodiment, the characteristic value is a total value of the viewing status of the above-mentioned digital content, and whether or not the medical worker has selected the medicine, which is the outcome target, as the first-choice drug.

[0051] The aggregated values ​​of the viewing status of digital content include, for example, the number of times a website that lists medical information is viewed, the viewing time period, the open rate of e-mails sent or distributed to medical professionals by MRs or pharmaceutical companies, etc. The server 1 collects viewing logs of one or more websites, e-mail opening logs, etc., and aggregates the number of times the website is viewed, etc. The server 1 stores the aggregated values ​​in the doctor DB 142 as characteristic values ​​of medical professionals.

[0052] The first choice drug refers to a drug that a medical professional selects as the first drug to be administered for a certain disease. Whether or not a particular drug is the first choice drug may be determined automatically by the server 1 based on the medical professional's (doctor's) past drug prescription history, or may be manually set by a medical representative or the like.

[0053] The server 1 uses these characteristic values ​​as inputs to the first model 51. In particular, in this embodiment, by adding the aggregated value of the viewing status of digital content to the characteristic values, it is possible to appropriately determine the digital preferences of medical staff.

[0054] The above characteristic values ​​are merely examples, and for example, the server 1 may use the past usage values ​​(usage history) of each channel by medical staff as the characteristic values.

[0055] The intimacy with the MR is a numerical value indicating the intimacy between the medical staff and the MR, or a class that classifies the intimacy. As with the first-choice drug, the intimacy may be automatically determined by the server 1 from the MR's activity history (interview history with the medical staff), or may be manually set by the MR, etc.

[0056] When the medical professional information and the usage values ​​of each channel including the digital channel are input, the first model 51 calculates the performance value of the MR's activities. The performance value is an index value that measures the degree of achievement of the performance goal, and in this embodiment, the performance value is the prescription growth rate of the drug. The prescription growth rate is a numerical value obtained by dividing the prescription amount of the corresponding drug by the medical professional by a predetermined reference amount, and indicates the growth of drug sales.

[0057] In the above, the prescription growth rate is defined based on the prescription amount (sales volume) of the drug, but it may be defined based on the sales amount of the drug. Also, the outcome value is not limited to the prescription growth rate of the drug, and may be other measures such as the number of information provided by medical professionals to MRs about the drug (number of requests for information).

[0058] The server 1 generates a first model 51 by using training data in which correct answer values ​​of outcome values ​​are associated with an input data group consisting of medical worker information of each medical worker for a certain period and the utilization value of each channel when the MR performed activities for each medical worker in a target period following the certain period. For example, the server 1 uses the performance value of the MR's activities as training data. That is, the server 1 uses the medical worker information of each medical worker who was the target of the MR's activities, the number of web interviews, face-to-face interviews, phone calls, etc. with each medical worker, and the growth rate of drug prescriptions by each medical worker as training data.

[0059] The server 1 calculates a performance value by inputting training input data to the first model 51. The server 1 updates parameters such as weights between neurons so that the calculated performance value approximates a correct answer value. The server 1 sequentially provides each input data to the first model 51 to update the parameters, and finally generates the first model 51 with optimized parameters.

[0060] The server 1 uses the above-mentioned first model 51 to determine the recommended value of each channel when the MR performs an activity for the target medical worker. Specifically, the server 1 prepares a plurality of patterns of usage values ​​of each channel with different combinations of usage values ​​of each channel, and calculates a performance value corresponding to each pattern by inputting the usage values ​​of each channel of the plurality of patterns into the first model 51. Then, the server 1 determines the recommended value of each channel based on the calculated performance value.

[0061] The input and output data of the first model 51 when determining the recommended value is conceptually shown in the lower part of Fig. 6. Fig. 6 shows how the same medical professional information and different channel usage values ​​(shown in bold frames) are input for each pattern, and different prescription elongation degrees are output for each pattern.

[0062] The server 1 reads out medical staff information related to the target medical staff from each database (doctor DB 142, facility DB 143, etc.). The server 1 then generates multiple patterns of usage values ​​of each channel to be combined with the medical staff information. The usage values ​​of each channel in the multiple patterns are generated under constraints such as the usage values ​​of each channel of the medical staff estimated by the second model 52 described later, as well as the target number of activities (upper limit and / or lower limit) set for each department in the pharmaceutical company to which the MR belongs and the target number of activities (upper limit and / or lower limit) set for each MR.

[0063] The server 1 switches the patterns of the utilization values ​​input together with the medical staff information, and sequentially inputs the input data of each pattern to the first model 51. In this way, the server 1 calculates the outcome value corresponding to each pattern.

[0064] The server 1 determines the recommended value of each channel to be recommended to the MR based on the calculated performance value of each pattern. For example, the server 1 determines the usage value of each channel of the pattern that maximizes the performance value as the recommended value. In other words, the server 1 determines the number of activities through each of the multiple channels that maximizes the prescription extension rate.

[0065] The above method of determining the recommended value is an example, and the present embodiment is not limited to this. For example, the server 1 may compare the performance value of each pattern with a predetermined threshold value, and present the usage values ​​of each channel of one or more patterns whose performance value is equal to or greater than the threshold value to the MR in an order according to the performance value of each pattern.

[0066] In this way, the server 1 uses the first model 51 that predicts the results of MR activities to determine the recommended values ​​(number of activities) for each channel when the MR performs sales activities through each channel to medical professionals. This makes it possible to support the MR's activities so as to maximize the results.

[0067] 7 is an explanatory diagram of the second model 52. In the above, it has been described that the server 1 inputs a plurality of patterns of utilization values ​​of each channel to the first model 51. In this embodiment, the server 1 further generates the utilization values ​​of each channel to be input to the first model 51 by using a second model 52 different from the first model 51.

[0068] The second model 52 is a machine learning model that has learned predetermined training data, and is, for example, a neural network such as CNN like the first model 51. Note that the second model 52 may be a neural network other than CNN, or a model based on other learning algorithms such as a decision tree, a random forest, or an SVM.

[0069] The second model 52 is a model that estimates the preference of a medical professional, that is, the degree to which the medical professional prefers each channel including digital channels. Specifically, when medical professional information for a certain period in the past is input, the second model 52 estimates the usage value of each channel including digital channels by the medical professional in the estimation target period. For example, like the first model 51, when medical professional information for the past six months is input, the second model 52 estimates the extent to which the medical professional will use each channel, such as web interviews, face-to-face interviews, and telephone calls, in the next month, and the number of times each channel will be used.

[0070] The medical worker information input to the second model 52 is almost the same as the medical worker information input to the first model 51, and includes attributes of the medical worker, attributes of the medical facility to which the medical worker belongs, characteristic values ​​representing the behavioral characteristics of the medical worker, and the degree of intimacy with the MR. The second model 52 estimates the utilization value of each channel from various information.

[0071] The server 1 generates a second model 52 using training data in which a group of medical worker information for a certain period is associated with the correct value of the usage value of each channel in a target period following the certain period. The training data for the second model 52 is actual values ​​like the training data for the first model 51, and is data stored in each database. The server 1 uses medical worker information for a certain period (e.g., six months) as input data for training, and uses the usage values ​​of each channel of medical workers in the next target period (e.g., one month) as the correct value.

[0072] In this embodiment, the server 1 generates a plurality of second models 52, 52, 52... corresponding to each channel in order to estimate a utilization value for each channel. That is, the server 1 generates a plurality of second models 52, 52, 52... that estimate utilization values ​​of different channels, such as a second model 52 for estimating the number of Web consultations, a second model 52 for estimating the number of face-to-face consultations, a second model 52 for estimating the number of phone calls, etc.

[0073] In this embodiment, the second model 52 is generated for each channel, but a single second model 52 may be capable of estimating the utilization value of each of a plurality of channels.

[0074] The server 1 inputs medical staff information for training into each second model 52 and estimates the utilization value of each channel. The server 1 compares the estimated utilization value of each channel with a correct answer value and optimizes parameters such as weights between neurons so that the two are close to each other. In this way, the server 1 generates each second model 52.

[0075] When determining the recommended value of each channel using the first model 51, the server 1 first estimates the utilization value of each channel of the medical worker by inputting the medical worker information into each second model 52. Then, the server 1 generates utilization values ​​of each channel of multiple patterns to be input to the first model 51 based on the estimated utilization value of each channel.

[0076] For example, the server 1 sets, for each channel, upper and lower limit values ​​that the usage value to be input to the first model 51 can take, by adding or subtracting a predetermined value from the usage value of each channel. The server 1 generates multiple patterns of usage values ​​for each channel within the numerical range from the upper limit to the lower limit. As described above, the server 1 generates multiple patterns of usage values ​​for each channel by referring to constraints such as the target number of activities for each department to which the MR belongs and the target number of activities for each MR.

[0077] Although FIG. 7 illustrates the utilization value output from the second model 52 as a single numerical value, the second model 52 may directly estimate the upper and lower limits, i.e., the numerical range, of the utilization value for each channel.

[0078] The server 1 combines the usage values ​​of each channel of the multiple patterns generated using the second model 52 with medical professional information related to the target medical professional, and inputs the combined values ​​to the first model 51. As described above, the server 1 determines the recommended values ​​of each channel to be recommended to the MR based on the outcome values ​​calculated by the first model 51.

[0079] Fig. 8 is an explanatory diagram showing an example of a display screen of the terminal 2. Fig. 8 shows a screen on which information on medical professionals (doctors) who are business targets can be viewed, the screen displaying the above-mentioned recommended values.

[0080] The screen includes a subject list 81, a usage value graph 82, and a contribution graph 83. The subject list 81 is a display field that displays a list of doctors (medical workers) registered in the doctor DB 142. The usage value graph 82 is a graph that displays the recommended value determined using the first model 51 and the actual value of the MR's activity by channel. The contribution graph 83 is a graph that displays the degree to which each parameter included in the medical worker information contributes to the channel usage, and the contribution of each parameter to the usage value by channel.

[0081] The terminal 2 displays, in a target person list 81, medical professionals who are the targets of the MR's activities. The terminal 2 then accepts a selection input to select one of the medical professionals from the target person list 81. The "items" in the target person list 81 indicate the medicines that the medical professionals often prescribe. The "scores" in the target person list 81 indicate the contribution of the medicine to sales per interview with the relevant medical professional. For example, the server 1 refers to the MR DB 141 and the sales DB 144, calculates a score for each medicine from the number of interviews (number of activities) between the MR and the medical professional and the amount of each medicine prescribed by the medical professional, and displays the medicine with the highest score and the score of the medicine in the target person list 81 in association with the medical professional.

[0082] Terminal 2 displays on the right side of the screen information on the medical worker selected in the subject list 81. Specifically, terminal 2 displays on the upper side of the screen the scores of each drug related to the selected medical worker, as well as a utilization value graph 82 and a contribution degree graph 83 related to the medical worker.

[0083] The utilization value graph 82 is a graph that displays the recommended value determined using the first model 51 and the actual value of the MR's activity. The terminal 2 displays the recommended value of each channel and the actual activity value by channel for a fixed period (e.g., one month) from the current time on the utilization value graph 82. This allows the MR to obtain an objective indicator of how much activity he or she should perform through each channel.

[0084] The contribution graph 83 is a graph showing the contribution of each parameter included in the medical worker information to the use of each channel by the medical worker. As described above, the server 1 estimates the utilization value of each channel of the medical worker by inputting the medical worker information into the second model 52. In this embodiment, the server 1 calculates the contribution of each parameter input into the second model 52 as medical worker information based on the input / output (medical worker information and utilization value) of the second model 52.

[0085] For example, the server 1 calculates a Shap value (Shapley Additive explanations) as the contribution degree. Shap is a visualization algorithm used to interpret a machine learning model, and is an algorithm that calculates the contribution degree of an input feature to the output of the model. Since Shap is a well-known algorithm, a detailed description is omitted. Note that the server 1 may calculate the contribution degree using an algorithm other than Shap.

[0086] As described above, the server 1 estimates the utilization value of each channel of the medical worker using the second models 52, 52, 52 ... corresponding to each channel. Based on the input and output of each second model 52, the server 1 calculates the contribution of each parameter of the medical worker information to the estimated utilization value for each channel.

[0087] For example, terminal 2 accepts, via a channel selection field 831, a selection input of a channel to be displayed in contribution graph 83. Terminal 2 displays the contribution of each parameter to the usage value of the selected channel in contribution graph 83. As shown in FIG. 8, the contribution is expressed as a positive or negative value, with a positive contribution indicating a positive impact on the usage of the channel and a negative contribution indicating a negative impact. The contribution graph 83 makes it possible to visualize important factors that contribute to the response (channel usage) of medical personnel.

[0088] In addition, the terminal 2 displays on the screen the attributes of medical staff (age, etc.), the attributes of medical facilities (category, digital viewer ratings, etc.), the viewing status of digital content, etc. (not shown in FIG. 8).

[0089] As described above, according to this embodiment, the first model 51 and the second model 52 are used to present the recommended values ​​for each channel, including the digital channel, as to how much to approach medical professionals through each channel. This makes it possible to appropriately support the activities of MRs.

[0090] Fig. 9 is a flowchart showing the procedure of the process of generating the first model 51. The process of generating the first model 51 by machine learning will be described with reference to Fig. 9. The control unit 11 of the server 1 acquires training data for generating the first model 51 (step S11). The training data is data in which a correct answer value of the outcome value of the MR's activities is associated with an input data group consisting of medical worker information for a certain period and the usage value of each channel for a target period different from the certain period. The medical worker information includes the attributes of the medical worker, the attributes of the medical facility to which the medical worker belongs, a characteristic value representing the behavioral characteristics of the medical worker, and the degree of intimacy with the MR. The outcome value is, for example, the degree of increase in prescriptions of medicines by the medical worker.

[0091] The control unit 11 generates a first model 51 that calculates a result value of the MR's activity when medical staff information and the usage value of each channel including the digital channel are input based on the training data (step S12). For example, the control unit 11 generates a neural network as the first model 51. The control unit 11 calculates the result value by inputting training input data to the first model 51. The control unit 11 generates the first model 51 by optimizing parameters such as the weight between neurons so that the calculated result value approximates the correct value.

[0092] Fig. 10 is a flowchart showing the procedure of the process of generating the second model 52. Fig. 10 illustrates the process of generating the second model 52 by machine learning. The control unit 11 of the server 1 acquires training data for generating the second model 52 (step S31). The training data is data in which a group of medical worker information for a certain period is associated with the correct value of the usage value of each channel of the medical worker for a target period different from the certain period.

[0093] When medical worker information is input, the control unit 11 generates a second model 52 that estimates the utilization value of each channel of the medical worker in the estimation target period based on the training data (step S32). For example, the control unit 11 generates multiple neural networks as second models 52, 52, 52 ... for each channel, such as web interviews, face-to-face interviews, and telephone calls. The control unit 11 estimates the utilization value of the channel by the medical worker by inputting medical worker information for training into the second model 52. The control unit 11 generates the second model 52 by optimizing parameters such as the weights between neurons so that the estimated utilization value approximates the correct value.

[0094] 11 is a flowchart showing the procedure of the process of determining the recommended value. The process of determining the recommended value for each channel will be described with reference to FIG. The control unit 11 of the server 1 refers to the doctor DB 142 and causes the terminal 2 to display a list of multiple medical workers (doctors) who are the targets of the MR's activities (step S51). The control unit 11 accepts a selection input to select one of the multiple displayed medical workers (step S52).

[0095] The control unit 11 acquires medical worker information related to the selected medical worker from each database (step S53). Specifically, the control unit 11 acquires the attributes of the selected medical worker, characteristic values ​​representing behavioral characteristics, and intimacy from the doctor DB 142. In addition, the control unit 11 acquires the attributes of the medical facility to which the medical worker belongs from the facility DB 143.

[0096] The control unit 11 estimates the utilization value of each channel of the medical staff by inputting the acquired medical staff information into the second model 52 (step S54). Specifically, the control unit 11 estimates the utilization value of each channel in the estimation target period by inputting medical staff information for a certain period in the past into the second models 52, 52, 52 ... corresponding to each channel. The control unit 11 generates a plurality of patterns of utilization values ​​of each channel based on the utilization value of each channel estimated by the second model 52 (step S55).

[0097] The control unit 11 inputs the medical worker information and the utilization values ​​of each channel of the multiple patterns generated in step S55 to the first model 51, thereby calculating the outcome value of the MR activity corresponding to each pattern (step S56). That is, the control unit 11 switches the patterns of utilization values ​​of each channel to be combined with the medical worker information and inputs them to the first model 51, thereby sequentially estimating the outcome value corresponding to each pattern.

[0098] The control unit 11 determines the recommended value (use value) of each channel to be recommended to the MR based on the estimated performance value of each pattern (step S57). For example, the control unit 11 determines the usage value of each channel of the pattern that maximizes the performance value (prescription elongation) as the recommended value.

[0099] Furthermore, the control unit 11 calculates the contribution of each parameter included in the medical staff information to the utilization value estimated by the second model 52 for each channel (step S58). For example, the control unit 11 calculates the SHAP value as the contribution.

[0100] The control unit 11 causes the terminal 2 to display the recommended values ​​for each channel determined in step S57 and the contribution degree calculated in step S58 (step S59). For example, the control unit 11 causes the terminal 2 to display a usage value graph 82 showing the recommended values ​​and actual values ​​by channel. The control unit 11 also causes the terminal 2 to display a contribution degree graph 83 showing the contribution degree of each parameter of the medical worker information by channel. The control unit 11 ends the series of processes.

[0101] In the above description, the device that learns the first model 51 and the second model 52 and the device that performs inference using the first model 51 and the second model 52 are the same computer (server 1), but each device may be a separate computer. For example, the server 1 may learn each model, and the data of each model generated by the server 1 may be installed in the terminal 2, so that the terminal 2 may perform inference using each model.

[0102] In addition, although the above describes a form in which the present system is applied to the sales activities (sales of pharmaceuticals) of MRs, the present embodiment is not limited to this. For example, the present system may be applied to other fields such as sales of educational content, management consulting, counseling, etc. It is conceivable that the present system is applied when approaching customers through multiple channels in these fields.

[0103] For example, in the sale of educational content, the target performance value can be defined by the grades of the customer (learner). In management consulting, the performance value can be defined by the business performance of the client company. In counseling, the performance value can be defined by the satisfaction of the person receiving counseling. The server 1 may determine the utilization value of each channel when the person in charge engages in activities with the customer, so as to maximize these performance values.

[0104] That is, the server 1 only needs to be able to calculate a performance value by inputting customer information related to a customer and the usage values ​​of each channel of multiple patterns into the first model 51, and to determine a recommended value for each channel based on the calculated performance value. Also, the server 1 only needs to be able to estimate a usage value of each channel of a customer by inputting customer information into the second model 52, and to generate a usage value of each channel of multiple patterns to be input into the first model 51 based on the estimated usage value. In this way, it is conceivable that the present system can be applied to various fields.

[0105] As described above, according to this embodiment, it is possible to favorably support activities for customers through digital channels.

[0106] The embodiments disclosed herein are illustrative in all respects and should not be considered as limiting. The scope of the present invention is defined by the claims, not by the above meaning, and is intended to include all modifications within the scope and meaning equivalent to the claims. [Explanation of symbols]

[0107] 1. Server (information processing device) 11 Control section 12 Main memory 13. Communications Department 14 Auxiliary storage P1 Program 51 1st Model 52 2nd Model 141 MRDB 142 Doctor DB 143 Facility DB 144 Sales DB 2. Terminal 21 Control section 22 Main memory 23 Communications Department 24 Display section 25 Input section 26 Auxiliary storage P2 Program

Claims

1. Obtain healthcare professional information related to healthcare professionals; a first model that has been trained to calculate an outcome value of an activity of a medical representative (MR) when the medical professional information and a usage value of each of a plurality of channels, including a digital channel, that the MR uses when carrying out activities with the medical professional are input, and the acquired medical professional information and a plurality of patterns of usage values ​​of each channel, each of which has a different combination of the usage values ​​of each channel, are input to the first model, which calculates the outcome value corresponding to each pattern; Based on the calculated outcome value of each pattern, a utilization value of each channel to be recommended to the MR is determined. An information processing method in which processing is performed by a computer.

2. The degree of prescription growth of the drug by the medical professional is calculated as the outcome value. The information processing method according to claim 1 .

3. the usage value being the number of activities per unit time period to the healthcare professional through the channel; The number of activities for each channel is determined based on the performance value.

3. The information processing method according to claim 1 or 2.

4. The medical worker information includes attributes of the medical worker, attributes of the medical facility to which the medical worker belongs, characteristic values ​​representing behavioral characteristics of the medical worker, or a degree of intimacy between the medical worker and the MR. The information processing method according to any one of claims 1 to 3.

5. The characteristic value includes a summary value of a viewing status of predetermined digital content related to medical care by the medical staff. The information processing method according to claim 4.

6. A utilization value of each of the channels is estimated by inputting the acquired medical worker information into a second model that has been trained to estimate a utilization value of each of the channels used by the medical worker when receiving an activity from the MR when the medical worker information is input; generating a utilization value of each of the plurality of patterns of channels to be input to the first model based on the estimated utilization value of each of the channels; The utilization value of each generated pattern is input to the first model to calculate the performance value corresponding to each pattern. The information processing method according to any one of claims 1 to 5.

7. Estimating a utilization value of each of the channels using a plurality of the second models that have been trained to estimate utilization values ​​of the different channels. The information processing method according to claim 6.

8. Calculating a contribution of each of a plurality of parameters included in the medical staff information to the utilization value based on the medical staff information input to the second model and the utilization value estimated by the second model; Output the calculated contribution of each parameter 8. The information processing method according to claim 6 or 7.

9. an acquisition unit for acquiring medical worker information related to a medical worker; a calculation unit that calculates the outcome value of the activity of the MR by inputting the acquired medical staff information and the utilization values ​​of each of a plurality of patterns of different combinations of the utilization values ​​of each channel into a first model that has been trained to calculate the outcome value of the activity of the MR when the medical staff information and utilization values ​​of each of a plurality of channels, including digital channels, that the MR uses when performing activities for the medical staff are input; and A determination unit that determines a utilization value of each of the channels to be recommended to the MR based on the calculated outcome value of each pattern; An information processing device comprising:

10. Obtain healthcare professional information related to healthcare professionals; The utilization value of each channel is estimated by inputting the acquired medical professional information into a second model that has been trained to estimate the utilization value of each of a plurality of channels, including a digital channel, that the medical professional uses when receiving an activity from an MR when the medical professional information is input. An information processing method in which processing is performed by a computer.

11. A list of multiple medical professionals who are the targets of the MR's activities is displayed on the display unit, accepting a selection input for selecting one of the plurality of medical professionals; When a selection input of the medical worker is received, a recommended value for each of a plurality of channels, including a digital channel, to be used by an MR when carrying out activities for the medical worker is displayed based on medical worker information related to the selected medical worker. A program that causes a computer to carry out processing.

12. The recommended value is a value determined using a second model that has been trained to estimate a utilization value of each channel that the medical worker will use when receiving an activity from the MR when the medical worker information is input, and Displaying the contribution of each of a plurality of parameters included in the medical staff information to the utilization value for each channel, the contribution being calculated based on the medical staff information input to the second model and the utilization value of each channel estimated by the second model, for each channel. The program according to claim 11.

Images