JP7759629B2 - Behavioral support system, behavioral support method, and behavioral support program - Google Patents

Description

The present invention relates to an action support system, an action support method, and an action support program.

In recent years, efforts to utilize approaches based on theories of behavioral science, which is the scientific study of human behavior, in service development have been expanding in various fields, including public policy, healthcare, retail, and education. In addition, systems that technically realize supporting behavioral change and habit formation for subjects in these various fields are also being considered (for example, Patent Document 1).

The system described in Patent Document 1 analyzes behavioral data including various data measured on the behavior of multiple subjects, and based on the results of the analysis of the behavioral data, defines stage indicators that are indicators used as standards for multiple stages that gradually lead to a behavior that is a goal for habituation, and each of the multiple stages according to the stage indicators; identifies a gap between the two stages that make up each pair of adjacent stages; and identifies, for each stage pair, reasons/measures that are at least one of the reason for the existence of the identified gap and measures to cause a subject belonging to the lower stage to change their behavior and transition to the higher stage, from relationship information that defines the relationship between the gap and the reasons/measures.
It is described that processing related to the reason/measure identified for each stage pair is performed.

Japanese Patent Application Laid-Open No. 2020-140596

In order to promote behavioral change in subjects, it is thought to be effective to implement individualized interventions based on the subjects' behavioral characteristics using personal information such as their attributes, income, and residential area.
However, using subjects' personal information raises the following issues. First, obtaining personal information requires high levels of security and legal compliance, making it difficult for both the provider and recipient of personal information to implement the system. Second, the personal information needed to identify a subject's behavioral characteristics is often not known in advance. Therefore, even if personal information is obtained after overcoming various hurdles, it may not be useful in designing and implementing a behavior change program. Third, the personal information obtained for behavior change interventions is personal information from a single point in time in the subject's past and may not reflect the subject's state at the time of intervention. Therefore, interventions based on such past personal information may not produce the expected results.

Therefore, one of the objects of the present invention is to provide a behavior support system, a behavior support method, and a behavior support program that more effectively support a subject's behavioral change without using the subject's personal information or acquiring knowledge of effective techniques for the subject in advance.

A behavioral support system according to one embodiment of the present invention is a behavioral support system that supports the behavioral change of a subject, and includes: a memory unit that stores a plurality of message groups, each of which is linked to a plurality of behavioral change techniques that act on one or more of a plurality of behavioral change factors, and each message group includes one or more messages; a generation unit that extracts a message from each of the one or more message groups among the plurality of message groups and generates a message set consisting of the extracted messages for each subject; and a control unit that controls the messages in the message set to be output to the subject once or multiple times.

According to this aspect, a message set that acts on multiple behavioral change factors can be sent to the subject, making it possible to effectively encourage behavioral change in the subject without using the subject's personal information or obtaining knowledge in advance, such as providing information that is effective for the subject.

In the above aspect, the generation unit selects, for each of the plurality of behavior change factors, a behavior change technique that acts only on a specific behavior change factor from among the plurality of behavior change techniques, and generates one message from each of a group of messages linked to the selected behavior change technique.
The first message set may be generated by extracting one message from the plurality of behavioral change factors. According to this aspect, a comprehensive message set that acts on a plurality of behavioral change factors can be transmitted to the subject.

In the above aspect, the system may further include an acquisition unit that acquires response information from the subject in response to the output message, and an analysis unit that extracts, based on the response information, from the plurality of behavioral change factors, behavioral change factors to which the subject from whom the response information was acquired responded, wherein the generation unit may further select, for the subject from whom the response information was acquired, behavioral change techniques that act on the behavioral change factors to which the subject responded, and generate a second message set by extracting one or more messages from each of the message groups linked to the selected behavioral change techniques. According to this aspect, intervention can be performed for each subject that is adapted to the effects on the behavioral change factors due to the passage of time or changes in the environment, making it possible to more effectively encourage the subjects to change their behavior.

In the above aspect, the generation unit may randomly extract a message from each message group.

In the above aspect, the plurality of behavior change factors may include ability, opportunity, and motivation in the COM-B model.

In the above aspect, the plurality of behavioral change factors may include at least two of experiential attitudes, instrumental attitudes, indicative norms, descriptive norms, sense of behavioral control, self-efficacy, knowledge, skills, importance of behavior, environmental constraints, and habituation.

Another aspect of the behavioral support method of the present invention is a behavioral support method for supporting the behavioral change of a subject, comprising the steps of: extracting messages from one or more of a plurality of message groups, each of which is linked to a plurality of behavioral change techniques that act on one or more of a plurality of behavioral change factors, and generating a message set consisting of the extracted messages for each subject; and controlling the messages in the message set to be output to the subject once or multiple times.

Another aspect of the present invention is a behavioral support program executed by a computer to support the behavioral change of a subject, and the program executes the following steps: extracting messages from one or more of a plurality of message groups, each of which is linked to a plurality of behavioral change techniques that act on one or more of a plurality of behavioral change factors, and generating a message set consisting of the extracted messages for each subject; and controlling the messages in the message set to be output to the subject once or multiple times.

According to the present invention, it is possible to more effectively support the subject's behavioral change without using the subject's personal information or obtaining knowledge in advance, such as providing information that is effective for the subject.

1 is a table illustrating a matrix format of frames that associate behavior change factors with behavior change techniques. FIG. 10 is a diagram conceptually illustrating the data structure of a nudge message group. 1 is a diagram illustrating an example of a schematic configuration of a behavior support system according to an embodiment of the present invention. FIG. 1 is a diagram illustrating an example of the hardware configuration of devices in a behavior support system according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an example of the functional configuration of devices in the action support system according to the present embodiment. 10 is a flowchart illustrating an example of the operation of the action support system according to the present embodiment. FIG. 1 is a schematic diagram showing a specific example of an MECE-type nudge cocktail. FIG. 1 is a schematic diagram showing a specific example of an MECE-type nudge cocktail. FIG. 1 is a schematic diagram showing a specific example of an adaptive nudge cocktail.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described with reference to the accompanying drawings, in which the same reference numerals denote the same or similar components.

(Outline of behavioral support methods)
The behavioral support method according to this embodiment scientifically and objectively searches for important behavioral change factors for a subject of intervention, and encourages behavioral change by sending the subject nudge messages that effectively affect the subject's behavioral change factors. In this case, this embodiment uses a frame that associates multiple behavioral change factors with multiple behavioral change techniques that affect one or more of the multiple behavioral change factors.

<Frame>
Here, a behavioral change factor refers to a factor that causes a target person to take a specific action. Multiple behavioral change factors in a target field can be identified based on academic research or surveys such as attitude surveys, theories related to behavioral science, target persona settings, or behavioral process maps.

An example of a behavioral change factor is the "
Another example of a behavioral change factor is the Integrated Behavior Model.
The factors defined by the IBM (International Business Model) (also known as "IBM factors") are:
experiential attitude, instrumental attitude, injunctive norm, descriptive norm, and sense of behavioral control.
perceived control), self-efficacy, knowledge, and skills.
), salience of the behavior, environmental constraints
In this embodiment, a behavioral change factor including at least two of these factors may be identified.

It should be noted that the above-mentioned IBM factors are merely examples, and higher-level factors that include the above IBM factors may be specified as behavioral change factors. For example, the above-mentioned experiential attitude and instrumental attitude are included in "Attitude," and the above-mentioned indicative norms and descriptive norms are included in "Perceived Norms."
The sense of control and self-efficacy are included in the "personal norm" and
The knowledge and skills may be included in "Agency," the importance of the behavior may be included in "Importance," and environmental constraints may be included in "Friction." Alternatively, lower-level factors obtained by dividing the IBM factors may be identified as behavior change factors.

Of course, in this embodiment, any factors related to human behavioral intentions, such as factors defined by the COM-B model or behavioral models other than IBM, may be adopted as behavioral change factors.

The means or techniques that act on (influence) the above-mentioned factors for behavior change are called "Behavior Change Techniques (BCTs)." They are also sometimes called intervention functions. Hereinafter, these means or techniques will also be referred to as "BCTs."

BCT (or interventions) that affect behavioral change factors are theoretically known. For example,
Lou Atkins and Susan Michie (UCL Center for Behavior Change, University College
London), "Designing interventions to change eating behaviours"
A matrix showing the relationship between behavioral change factors and intervention effects in the COM-B model has been published ("Table 2. Matrix of links between COM-B model and intervention functions").
reference).

The frame used in this embodiment may be one that applies the theories exemplified above, or may be one that has been developed based on theories, as well as surveys such as academic research or attitude surveys, persona settings of subjects, or behavioral process maps.

Figure 1 is a matrix showing an example of a framework that associates behavioral change factors with behavioral change techniques. In Figure 1, there are three behavioral change factors: "ability,""opportunity," and
The strength of the effects of the five BCTs on motivation—“motional instruction,” “social support,” “social norms,” “step-by-step task setting,” and “behavioral observation”—is expressed in vector form.
In the matrix shown in Figure 1, the sign "+" indicates that an effect was observed, and the number indicates the strength of the effect.

In Figure 1, for example, it is shown that the BCT's "action instruction" only affects the behavioral change factor "ability." Also, it is shown that "stepwise task setting" has a strong effect on the behavioral change factor "ability" (+1) and also affects "opportunity" (+1).
It has been shown that "behavioral observation" does not affect any of the three factors of behavioral change shown in Figure 1.

Hereinafter, the vector representing the effect of each BCT on a behavioral change factor will also be referred to as a "BCT vector." A specific BCT vector will also be referred to as an "action instruction vector,""social support vector," etc., using the name of the BCT. As shown in Figure 1, the action instruction vectors for the three behavioral change factors "ability,""opportunity," and "motivation" are <+1,0,0>.
It is expressed as:

<Nudge message>
In the behavioral support method according to this embodiment, a plurality of BCTs that act on one or more behavioral change factors among a plurality of behavioral change factors related to the target field of intervention are identified, and these BCTs
A message group is linked to each of the BCTs. Each message group includes one or more nudge messages (hereinafter simply referred to as messages) that embody the linked BCT. For example, the BCT "action instruction" may include instructions such as "Please do XX by XX month."
A message instructing an action is linked.

Figure 2 is a conceptual diagram illustrating the data structure of nudge message groups. In Figure 2, multiple messages are linked to each of the five types of BCTs shown in Figure 1. Each message may be assigned an identification code (e.g., a message ID). Furthermore, each message group linked to each behavioral change technique may be assigned a code (e.g., a group ID) to identify the message group to which it belongs.

When carrying out an intervention, one or more BCTs that act on behavior change factors identified according to the stage of intervention are selected from among multiple behavior change factors related to the target field, messages are extracted from each of the message groups linked to the selected BCTs, and a message set consisting of the extracted messages is generated.

Messages are selected from each message set, preferably randomly, for each subject. By incorporating randomness into the generation of message sets, intervention designers can avoid selection bias and achieve more stable intervention effects than traditional intervention designs, where intervention designers select nudges based on intuition and experience. This is because many intervention designers tend to use messages they have used in the past or that are well-known, rather than those based on scientific evidence.

However, a message group may contain a message (best message) that is thought to have an overwhelmingly greater impact on the target than other messages. Randomly extracting messages from such a message group by treating the best message and other messages equally may result in missed opportunities for intervention. Therefore, messages may be extracted from each message group using a reinforcement learning method such as the Epsilon-Greedy algorithm. The Epsilon-Greedy algorithm performs a search by randomly acting with probability ε and selects the action with the highest expected value with probability (1-ε). Specifically, the best message may be selected from a message group with probability (1-ε), and then randomly extracted from the other messages with probability ε. Alternatively, one message may be randomly extracted from excellent messages above a certain threshold with probability (1-ε), and the most recent message may be extracted with probability ε.

Note that there may be cases where only one message is associated with a certain BCT, and in such cases, that one message is extracted from the BCT.

<Nudge Cocktail>
As described above, when an intervention is performed, a message set is generated by extracting messages from each of the message groups associated with one or more BCTs. Hereinafter, such a message set will be referred to as a "nudge cocktail."

In this embodiment, two types of nudge cocktails are generated depending on the stage of the intervention. The first nudge cocktail (first message set) is generated for all subjects in the early stage of the intervention. Hereinafter, this first nudge cocktail will be referred to as the MECE (Mutually Exclusive and Collectively Exhaustive) nudge cocktail.

The MECE nudge cocktail is a combination of multiple BCs that act on behavioral change factors related to the target field.
It is composed of messages extracted from a group of messages linked to BCTs that act only on specific behavioral change factors in T. This is to make it easy to determine which behavioral change factors have acted when a BCT is accepted.

Such MECE-type nudge cocktails are generated by selecting BCTs that act only on specific behavior change factors for all behavior change factors related to the target field, and then extracting one message from each of the message groups linked to the selected BCTs. The reason for selecting BCTs that act only on specific behavior change factors for all behavior change factors related to the target field is to cover all behavior change factors to which the target person may respond.

The second nudge cocktail (second message set) is a nudge cocktail generated based on behavioral change factors adapted to the subject, and is generated for subjects who have responded in some way to the MECE-type nudge cocktail. Hereinafter, this second nudge cocktail will be referred to as:
This is referred to as an adaptive nudge cocktail.

The adaptive nudge cocktail is generated by identifying the behavioral change factors to which the subject responded from among multiple behavioral change factors related to the target field based on the subject's response, selecting behavioral change techniques that act on the identified behavioral change factors, and extracting one or more messages from each of a group of messages linked to the selected behavioral change techniques.

In both MECE and adaptive nudge cocktails, nudge messages are randomly selected from each message group for each target person, resulting in a different nudge cocktail for each target person.

(Details of behavioral support methods)
Next, a method for generating a nudge cocktail and a method for evaluating intervention in the behavior support method according to this embodiment will be described.

<Coordinate system>
In conventional behavioral support methods, the process and reasons for a subject's behavioral change are qualitatively assessed, but this method makes it difficult to design an individualized intervention for each subject.

In contrast to this, in this embodiment, a coordinate system is generated with multiple behavioral change factors as axes,
In this coordinate system, a target coordinate or target area (hereinafter collectively referred to as the target area) is set. Whether or not the subject has undergone behavioral change is then evaluated based on whether or not the subject's coordinates in the coordinate system have approached the target area. Hereinafter, the distance calculated in this coordinate system will be referred to as the "Behavioral Scientific Distance (BSD)."

In such a coordinate system, by setting the current coordinates and the target area to which intervention should be directed for each subject, the distance (BSD) from the current coordinates to the target area can be expressed by a multidimensional vector, making it possible to objectively and clearly visualize the BCT that should be adopted to fill this distance.

Below, we will explain the procedure for intervention based on behavioral distance using the frame shown in Figure 1 as an example.
(1) Selection of coordinate axes First, based on academic research or attitude surveys, or theories on behavioral science, persona settings of subjects, behavioral process maps, etc., multiple behavioral change factors in the target field are identified and defined as coordinate axes. Among the behavioral change factors shown in Figure 1, "ability"
The coordinate axis showing "opportunity" is called the C-axis, the coordinate axis showing "motivation" is called the O-axis, and the coordinate axis showing "motivation" is called the M-axis.

Here, the units of the multiple coordinate axes do not need to be the same for each other, nor do they need to be integers. For example, when designing an intervention to encourage behavioral change toward vaccination, the C-axis can be set to a 5-point scale, the O-axis to the number of accesses to the vaccination reservation site, and the M-axis to the number of seconds spent on the reservation site.
The dimensional space becomes a coordinate system that represents the process of behavioral change in the subject.

(2) Setting the target area (coordinates) Next, a target area is set in the coordinate system consisting of the selected coordinate axes.
The target area is set based on information about the state of the subject in the field of intervention, the type of intervention the subject received, and the behavior the subject took. The target area can be set as pinpoint coordinates, or it can be set as a broader area, such as "any value of the coordinates (c, o, m) is 3 or more." By setting the target area in this way, it is possible to clearly indicate the shortest path to behavioral change for each subject, taking into account the behavioral science characteristics of the field and the human characteristics of the subject.

The target area is set in common for all subjects. The influence of the passage of time, changes in the environment, and other factors on behavioral change factors can be easily expressed by changing the setting of the target area in the coordinate system. Examples of influences on behavioral change factors include changes in vaccination rates, improved convenience through improvements to the user interface (UI) and user experience (UX), and the approaching deadline for free medical checkups.

(3) Setting the subject's coordinates In the above coordinate system, the subject's initial coordinates are basically set to (c, o, m) = (0, 0, 0). However, for subjects whose information such as past medical history and attributes is available, the initial coordinates may be set individually based on certain coordinate transformation rules. For example, for subjects who have a medical history in the past, the initial coordinates are (c, o, m) because they understand how to receive medical treatment (i.e., they have capacity).
= (1,0,0).

In this embodiment, intervention actions (such as sending a message) are determined and evaluated based on the coordinate system set in this way.

(4) Generation of MECE-type nudge cocktails BCTs that act only on specific behavioral change factors are represented by vectors in which only one coordinate value has a value (non-zero) and the other coordinate values are zero. MECE-type nudge cocktails are generated based on a collection of such BCT vectors.

Specifically, we extract a set whose sum of BCT vectors is <+1,+1,+1> and whose number of constituent BCT vectors is the minimum. If multiple sets of this kind are extracted, one of them is randomly selected for each subject. For example, see Figure 1.
In the case of the frame shown in Figure 1, the set of three BCT vectors (trivalent BCT vectors) is the action instruction vector <+1,0,0>, the social support vector <0,+1,0>, and the social norm vector <0,0,+1>.
Furthermore, by extracting one message from the message group linked to the BCT vector contained in this collection for each subject, a MECE-type nudge cocktail (trivalent nudge cocktail) consisting of three messages is generated. Note that the sum of the BCT vectors contained in the collection does not have to be <+1, +1, +1>. For example, in Figure 1
In the frame, the action instruction vector is <+1,0,0>, the social support vector is <0,+1,0>, and the social norm vector is <0,0,+1>. However, the action instruction vector is <+1,0,0>, the social support vector is <0,+2,0>, and the social norm vector is <0,0,+3>, and the action instruction vector, the social support vector, and the social norm vector whose sum is <+1,+2,+3> may be extracted as a collection. In this way, the collection includes a predetermined number of BCT vectors, and the sum of the predetermined number of BCT vectors may be a predetermined value. An n-valent nudge cocktail may be composed of messages corresponding to n (n is an integer > 0) types (i.e., behavioral change factors) of BCT vectors.

(5) Updating the Subject's Coordinates When a nudge cocktail intervention is implemented, the behavioral change factor to which the subject responded is identified based on the subject's response to the nudge cocktail, and the subject's coordinates in the coordinate system are updated. The behavioral change factor to which the subject responded can be identified from the BCT accepted by the subject. For example, if a subject accesses a reservation site guided by a message linked to a certain BCT, the BCT corresponding to that message is identified, and the behavioral change factor affected by this BCT can be identified using the frame illustrated in FIG. 1.

For example, if a subject accepts all the BCTs in the MECE-type nudge cocktail described above, the subject's state in the coordinate system is considered to have changed by the trivalent BCT vector (<+1,0,0>, <0,+1,0>, <0,0,+1>) from the initial coordinates (0,0,0). In other words, the total change due to the trivalent BCT vector is
<+1,0,0>+<0,+1,0>+<0,0,+1>=<+1,+1,+1>
Therefore, the coordinates of the subject are changed to (1,1,1) by adding this change to the initial coordinates (0,0,0).

In this way, by updating the subject's coordinates in the coordinate system, it is possible to objectively grasp the behavioral change factors to which the subject responded and the distance to the target area.

(6) Generation of Adaptive Nudge Cocktails The adaptive nudge cocktail is generated based on the current coordinates of each subject in the coordinate system. In particular, a vector (hereinafter also referred to as an intervention vector) representing the shortest route from the current coordinates of the subject to the nearest target area is first calculated. If there are multiple shortest route candidates, a route may be selected randomly for each subject. Furthermore, subjects for whom intervention should continue may be prioritized and selected based on the number of shortest routes. For example, subjects who are further away from the target area than a certain distance may be excluded from intervention.

Next, the intervention vector is decomposed to generate a set of BCT vectors whose sum is equal to the intervention vector. For example, if the intervention vector is <+2,+2,+2>, then
[<+1,0,0>×2, <0,+1,0>×2, <0,0,+1>×2], and
[<+2,+1,0>, <0,+1,0>, <0,0,+1> x 2]
These two types of aggregates are candidates.

If there are multiple candidates for a collection of BCT vectors equivalent to the intervention vector, a collection is randomly selected for each subject. Furthermore, if there are constraints on the intervention behavior and the number of BCT vectors is limited (for example, if there are constraints on the cost of sending messages and an upper limit on the number of messages that can be sent), a collection that meets the requirements is randomly selected from the candidates. The set of messages extracted from each of the message groups linked to the BCT vectors that make up the selected collection becomes the adaptive nudge cocktail for that subject.

When an intervention using an adaptive nudge cocktail is implemented, as described above, the behavioral change factors to which the subject responded are identified based on the subject's response to the nudge cocktail, and the subject's coordinates in the coordinate system are updated. If the subject's coordinates reach the target area, the intervention behavior is stopped. On the other hand, if the subject's coordinates have not yet reached the target area, a further adaptive nudge cocktail is generated based on the updated coordinates.

(Effect of nudge cocktail)
The reason for creating two types of nudge cocktails, MECE and adaptive, is as follows: The factors that change a subject's behavior are thought to be dynamic and change daily. Therefore, a nudge cocktail created based on static personal information that indicates the subject's state at a given point in time may have difficulty encouraging the subject to change their behavior. In this regard, the MECE nudge cocktail is a comprehensive, complete combination of BCTs that act on multiple behavior change factors identified based on objective research and theory, making it possible to take a multifaceted approach to all behavior change factors related to the intervention.

On the other hand, adaptive nudge cocktails consist of BCTs that act on the behavioral change factors to which the subject responded after receiving the MECE-type nudge cocktail. Therefore, by using adaptive nudge cocktails, it is possible to dynamically grasp the behavioral change factors of the subject based on their responses and take an approach that encourages behavioral change in an intensive manner.

By combining such MECE and adaptive nudge cocktails, even if the factors that cause a subject's behavioral change are unknown at first, it is possible to understand the factors that cause a subject's behavioral change through the sequential implementation of interventions using these nudge cocktails, and to effectively encourage behavioral change.

The number of messages constituting the nudge cocktail, in other words, the number of BCT vectors constituting the aggregate, can be determined based on intervention design conditions such as the target level of behavioral change rate for the subjects, the budget and time (duration) available for the intervention, and ethical constraints. Ethical constraints refer to ethical considerations, such as the possibility that too many interventions could lead to complaints from subjects or give the subject the impression of being overbearing. For example, if a trivalent nudge cocktail is expected to be effective for approximately 60% of all subjects, while a pentavalent nudge cocktail is expected to be effective for approximately 90% of all subjects, the choice of trivalent or pentavalent nudge cocktail can be determined based on the intervention design conditions described above.

When generating a nudge cocktail for the second or subsequent times, messages are extracted from the remaining message group excluding messages previously sent to the target person.

(Configuration of behavioral support system)
Next, the configuration of the behavior support system according to this embodiment will be described. In the following, a specific intervention action is assumed to be sending messages constituting a nudge cocktail to a subject via email or short message (SM).

Fig. 3 is a diagram showing an example of the schematic configuration of a behavior support system according to this embodiment. As shown in Fig. 3, the behavior support system 1 includes a behavior support device 10 and a terminal 20. The behavior support device 10 is an information processing device that provides the terminal 20 with a message that uses a nudge based on a behavior change factor to prompt a subject to change their behavior.

The terminal 20 is a terminal used by a person who is a target of behavioral modification, and may be, for example, any of various information devices such as a smartphone, a mobile phone, a tablet, or a personal computer.
The mobile phone is assumed to have a function for receiving messages such as short messages and e-mails, and to be able to use various social networking services (SNS) such as LINE (registered trademark).

FIG. 4 is a diagram illustrating an example of the hardware configuration of devices in the action support system 1.
As shown in FIG. 1, each device in the action support system 1 (for example, each of the action support devices 10 and the terminals 20) includes a processor 11 such as a CPU (Central Processing Unit) that corresponds to a computing device.
The computer 10 includes a storage device 12, a communication device 13, and an input/output device 14.
They are connected via a bus so that they can send and receive data to each other.

The processor 11 is, for example, a CPU (Central Processing Unit), and
The processor 11 is a control unit that controls the execution of programs stored in the memory device 12 and performs calculations and processing of data. The processor 11 receives various input data from the input/output device 14 and/or the communication device 13, and outputs (e.g., displays) the calculation results of the input data to the input/output device 14, stores them in the memory device 12, or transmits them via the communication device 13.

The storage device 12 includes a memory, a hard disk drive (HDD), and a solid state drive (SSD).
The storage device 12 of the action support device 10 may store the action support program executed by the processor 11. The storage device 12 may be called a "storage unit" or the like.

The communication device 13 is a device that communicates via a wired and/or wireless network, and may include, for example, a network card, a communication module, a chip, an antenna, etc. The communication device 13 may also be referred to as a "transmitter" or a "receiver," etc.

The input/output device 14 includes input devices such as a keyboard, a touch panel, a mouse, and/or a microphone, and output devices such as a display and/or a speaker. The input/output device may be called an "input unit" or an "output unit," for example.

The hardware configuration described above is merely an example.
Some of the hardware shown in Fig. 4 may be omitted, or hardware not shown in Fig. 4 may be included. Furthermore, the hardware shown in Fig. 4 may be configured with one or more chips.

Fig. 5 is a diagram showing the functional configuration of devices in the action support system 1. Note that Fig. 5 is merely an example, and it goes without saying that each device in the action support system 1 may have functions not shown.

5, the action support device 10 includes a setting unit 101, a message generation unit 102, a storage unit 103, a nudge cocktail generation unit 104, a transmission control unit 105, an acquisition unit 106, and an analysis unit 107. Note that at least a part of the functions realized by the acquisition unit 106 are implemented by the communication device 1.
3. The setting unit 101, the message generating unit 102, the nudge cocktail generating unit 104, the transmission control unit 105, and the analysis unit 107 can be realized by the processor 11:
This can be realized by executing a behavioral support program stored in the storage device 12. The behavioral support program can be stored in a storage medium. The storage medium storing the program is a computer-readable non-transitory storage medium (Non-transitory storage medium).
The non-transitory storage medium is not particularly limited, and may be, for example, a storage medium such as a USB memory or a CD-ROM. The storage unit 103 can be realized using the storage device 12.

In this embodiment, the action support device 10 is shown as an integrated device, but it may be divided into multiple devices. For example, the setting unit 101 and the message generation unit 102 may be configured as separate devices, and these separate devices may be connected to the action support device 10 via a wired and/or wireless network.

The setting unit 101 sets a frame in which a plurality of behavioral change factors and a plurality of behavioral change techniques that act on one or more of the plurality of behavioral change factors are associated, and generates a coordinate system whose axes are the plurality of behavioral change factors, and performs various settings such as setting a target area in the coordinate system and initial settings of the target person's coordinates.

The message generator 102 generates a plurality of message groups each associated with a plurality of BCTs, and stores the message groups in the storage unit 103. Each message group includes one or more messages that specifically express the associated BCT.

The message generator 102 may collect messages included in each message group via a wired or wireless communication network or a storage medium, or may automatically generate messages using machine learning. Alternatively, the message generator 102 may store messages that have been input as text via the input/output device 14, or may generate messages using templates prepared in advance.

The storage unit 103 stores a frame (see FIG. 1) that associates behavioral change factors with BCTs.
The coordinate system generated by 01 and various setting information in the coordinate system, as well as a group of nudge messages (see FIG. 2) linked to each BCT, etc. are stored.

The nudge cocktail generation unit 104 generates a nudge cocktail (MECE-type nudge cocktail, adaptive nudge cocktail) by extracting messages from the group of nudge messages stored in the memory unit 103 according to the stage of intervention.

The transmission control unit 105 controls the messages included in the nudge cocktail generated by the nudge cocktail generation unit 104 to be output to the subject once or multiple times. Under the control of the transmission control unit 105, the communication device 13 of the action support device 10 transmits the messages to the subject's terminal 20. The messages can be transmitted via email or short message (SM). Here, message transmission control may include, for example, control of the order in which multiple messages in the nudge cocktail are transmitted, the transmission date of each message, the transmission time period of each message, the time interval (transmission frequency) between transmissions of the multiple messages, whether to stop transmitting each message, the number of times each message is transmitted, etc.

The acquisition unit 106 acquires response information from the subject in response to the output message. The response information is information that indicates the reaction of the subject who received the message, and includes, for example, information such as whether or not a short message (SM) or email containing the message was viewed, the click-through rate (CTR) for links provided in the SM or email, and the length of time spent on the linked site. The acquisition unit 106 may acquire the response information from the subject's terminal 20 in real time. The response information may also include information (e.g., a message ID) that identifies one or more messages to which the subject responded,
This makes it possible to determine the message group to which the message to which the subject responded belongs.

The analysis unit 107 analyzes the response information from each subject and, depending on the analysis results, determines whether or not to continue the intervention, the type of nudge cocktail to be generated (MECE type or adaptive type), etc. Specifically, the analysis unit 107 extracts, from among multiple behavioral change factors, the behavioral change factor to which the subject, from whom response information was obtained, responded.

(Operation of the behavior support system)
Fig. 6 is a flowchart showing the operation of the action support system according to this embodiment. Note that the operation of the action support system shown in Fig. 6 is merely an example and is not limited to that shown. Figs. 7 and 8 are schematic diagrams showing specific examples of MECE-type nudge cocktails. Fig. 9 is a schematic diagram showing a specific example of an adaptive nudge cocktail.

First, by using various surveys and papers, multiple factors for behavior change in the area of interest for intervention are identified in advance through the persona of the subject and a behavioral process map. However, the identification of one or more factors for behavior change is not limited to the above. For example, one or more factors for behavior change may be identified by using a MECE-type nudge cocktail with a high valence. In other words, MECE
The first intervention using the MECE-type nudge cocktail may narrow down the factors for behavior change (determine the number of coordinate axes), and the second intervention using the MECE-type nudge cocktail may set the initial coordinates. Furthermore, the subject's persona and/or behavioral process map may be set based on data collected through a payment app or a social networking service (SNS).

6, the behavior support device 10 generates a coordinate system and performs various settings based on the identified behavior change factors (step S101). That is, the device 10 defines a BCT vector based on a frame (see FIG. 1) that associates the behavior change factors with the BCT, generates a coordinate system with multiple behavior change factors as its coordinate axes, and sets the initial coordinates of the target area and the subject in the coordinate system.

Next, the behavior support device 10 generates a plurality of messages corresponding to each BCT (behavior change technique) (step S102). The generated messages are stored in the storage unit 103 in association with the corresponding BCT.

Next, the action support device 10 creates an MECE-type nudge cocktail for each subject (step S103). As described above, the MECE-type nudge cocktail includes multiple BCs.
Messages are randomly drawn from each of the message groups associated with T, resulting in a different message set for each target person.

Specifically, as shown in Figure 7, for subject X, an MECE-type nudge cocktail is generated, which consists of text B extracted from the message group linked to "action instructions" among the nudge message group shown in Figure 2, text D extracted from the message group linked to "social support," and text J extracted from the message group linked to "social norms." Similarly, for subject Y, an MECE-type nudge cocktail is generated, which consists of text B extracted from the message group linked to "action instructions," text E extracted from the message group linked to "social support," and text H extracted from the message group linked to "social norms." The same is true for subject Z.

Next, the action support device 10 transmits the generated MECE-type nudge cocktail to each subject (step S104). At this time, all messages included in one nudge cocktail may be transmitted at once, or one or more messages may be transmitted sequentially with a certain interval between them.

Note that there may be cases where the number of BCTs constituting a nudge cocktail is small relative to the number of message transmissions set for a single intervention (opportunity for generating a nudge cocktail). In such cases, the number of messages extracted from the message group associated with one type of BCT may be increased. For example, if the set number of transmissions is three and there are two BCTs, two messages may be randomly extracted from the message group associated with the BCT that is considered to have a relatively large impact. Alternatively, one message may be randomly extracted from the message group associated with each BCT, and then one message may be randomly extracted from the entire two message groups (excluding messages that have already been extracted).

Furthermore, instead of sending the message contained in the nudge cocktail via email or SM, the nudge cocktail may be sent in the form of a video or website that allows for heat map analysis.

If there is no further opportunity to send a message thereafter (step S105: No), the action support apparatus 10 ends the intervention action.
Step S105: Yes), the action support apparatus 10 converts the transmitted MECE
The response to the behavioral change factors of the type nudge cocktail is confirmed (step S106).

For a subject who does not respond to the MECE-type nudge cocktail (step S107: No), the action support device 10 creates a MECE-type nudge cocktail again (step S103). However, in the MECE-type nudge cocktail created the second time or later,
From each message group to be extracted, one message is randomly extracted, excluding messages that have already been sent to the target person. For example, if no response information is obtained from target person Y, an MECE-type nudge cocktail is generated for target person Y, as shown in Figure 8, consisting of one message (e.g., text C) randomly extracted from messages linked to "action instructions" (see Figure 2) other than sent text B, one message (e.g., text F) randomly extracted from messages linked to "social support" other than sent text E, and one message (e.g., text J) randomly extracted from messages linked to "social norms" other than sent text H.

On the other hand, the action support device 10 performs the following for the subject from whom some kind of reaction was obtained (step S10
7: Yes), and update the coordinates in the coordinate system (step S108).

Next, the action support device 10 generates an adaptive nudge cocktail for the subject (
Step S109). For example, based on the updated coordinates of the subject X, the intervention vector <+2, +2, +
2> is calculated, and the set of BCT vectors equivalent to this intervention vector [<+2,+1,0>, <0,+1,0>
, <0,0,+1>×2] is selected for subject X, as shown in Figure 9, one message (e.g., message E) randomly extracted from the messages associated with "social support" (see Figure 2) other than the sent message D, and two messages (e.g., message J) randomly extracted from the messages associated with "social norms" other than the sent message J.
An adaptive nudge cocktail is generated, consisting of two messages (e.g., texts H and I) and one message (e.g., text K) randomly extracted from the messages linked to the "stepwise task setting." The valence n of the intervention vector may be equal to the number of types of BCT vectors (i.e., the behavioral change factors indicated by the BCT vectors) included in the collection. In this way, when the collection contains multiple BCT vectors of the same type (e.g., the above <0,0,+1>), from the perspective of the diminishing returns effect and avoiding spam detection, the number of messages extracted may be equal to or less than the number of BCT vectors.

Next, the action support device 10 transmits the generated adaptive nudge cocktail to each subject (
Step S110) The number of times and method of transmission are the same as in step S104.

If there is no further opportunity to send a message thereafter (step S111: No), the action support apparatus 10 ends the intervention action.
Step S111: Yes), the action support device 10 checks the response of each subject to the behavior change factors of the transmitted adaptive nudge cocktail (Step S112).

For a subject who does not respond to the adaptive nudge cocktail (step S113: No), the behavioral support device 10 checks the subject's response to the behavioral change factors of the MECE nudge cocktail sent to that subject (step S106).

On the other hand, for the subject who has responded to the adaptive nudge cocktail (step S113: Yes), the action support device 10 updates the coordinates in the coordinate system (step S114). If the updated coordinates reach the target area (step S115:
On the other hand, if the updated coordinates have not reached the target area (step S115: No), the action support device 10 creates an adaptive nudge cocktail for the subject again (step S109).

As described above, according to this embodiment, it is possible to effectively encourage a subject to change their behavior without using personal information of the subject.

Furthermore, according to this embodiment, the MECE-type nudge cocktail allows for comprehensive intervention without overlap from the start, making it possible to bring about behavioral change in subjects in a shorter time than conventional methods.

Furthermore, according to this embodiment, even if the passage of time or changes in the environment affect factors that change behavior, such influences can be flexibly incorporated using an adaptive nudge cocktail.

Furthermore, according to this embodiment, by incorporating randomness into the process of generating nudge cocktails, a more stable intervention effect can be obtained compared to conventional intervention designs in which the intervention designer selects nudges based on intuition and experience.

Furthermore, according to this embodiment, the factors for behavioral change are identified using the COM-B model and factors defined by IBM, thereby ensuring quality from the perspective of behavioral science.
Furthermore, by randomly extracting messages from each message group to generate a nudge cocktail, it is possible to diversify risks caused by uncertainty that cannot be fully controlled by behavioral science.

The above-described embodiments are intended to facilitate understanding of the present invention and are not intended to limit the present invention. The flowcharts, sequences, elements included in the embodiments, as well as their arrangements, materials, conditions, shapes, sizes, etc., are not limited to those exemplified and may be modified as appropriate.

1... Behavior support system, 10... Behavior support device, 11... Processor, 12... Storage device, 1
3...communication device, 14...input/output device, 20...terminal, 101...setting unit, 102...message generation unit, 103...storage unit, 104...nudge cocktail generation unit, 105...transmission control unit, 106...
Acquisition section, 107... Analysis section

Claims (4)

a coordinate system generation unit that generates a coordinate system based on one or more behavior change factors;
a target area setting unit that sets target coordinates or a target area in the coordinate system;
a subject coordinate update unit that updates where the subject is located in the coordinate system;
an intervention action determination unit that determines an intervention action for the subject,
the intervention action determination unit determines an intervention action for the subject based on a positional relationship between the target area and the position of the subject in the coordinate system;
Behavioral support system. The target area setting unit appropriately changes the set target area based on the passage of time or changes in the environment.
The behavior support system according to claim 1 . the intervention action for the subject is outputting a message to the subject,
calculating an intervention vector representing the shortest route from the subject's current coordinates to the nearest target area;
generating a collection of BCT vectors whose sum equals said intervening vector;
Selecting a message to be output to the target person from a group of messages linked to the BCT vectors that make up the collection;
The behavior support system according to claim 1 . Calculating an intervention vector for each of the plurality of subjects, and selecting subjects for whom the intervention action will be continued from the plurality of subjects according to each intervention vector;
The behavior support system according to claim 2 .