Healthy management support system and healthy management support program

Abstract

To provide a health management support system and a health management support program that support the improvement of productivity of companies and organizations by clarifying the relationship between presenteeism and other health conditions that affect the presenteeism and linking it to the improvement of presenteeism.SOLUTION: In a health management support system in which a management server, a manager terminal, and user terminals owned by a plurality of users are communicably connected to each other via a communication network, a management server 1 includes: an index data calculation unit 320 that calculates each piece of index data associated with each index obtained by quantifying or discretizing each index of presenteeism, health awareness, stress, engagement, and happiness levels according to data acquired from the users; and a structural model generation unit 330 that outputs a structural model showing the relationship between each index of health awareness, stress, engagement and happiness levels, and the index of presenteeism, based on the index data.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to a health management support system and a health management support program. [Background technology]

[0002] In recent years, emphasis has begun to be placed on health management, which involves considering and strategically implementing employee health management from a management perspective. By practicing health management, companies and organizations can actively invest in the health of their employees based on their philosophy, which is expected to revitalize the organization by improving employee vitality and productivity, ultimately leading to improved business performance and stock prices. As management indicators for understanding the productivity of companies and organizations for health management, multiple indicators that grasp the physical and psychological health status of employees are sometimes used.

[0003] In particular, representative indicators of psychological health include health awareness, mental health (e.g., stress), engagement, and happiness. Furthermore, it is thought that a deterioration in psychological health, or a deterioration in both psychological and physical health, can lead to a decline in the productivity of companies and organizations. Absenteeism and presenteeism have been proposed as indicators for assessing the degree of impact on productivity. Absenteeism refers to the loss incurred by companies and other organizations due to employees' absences, lateness, early departures, etc., caused by health reasons, including mental health issues. Presenteeism refers to the loss incurred due to a decline in the work performance of employees, etc., caused by health reasons, including mental issues, even if these do not result in absence.

[0004] Patent Document 1 describes a system for evaluating employee health, which includes a loss information storage unit, an evaluation unit, an economic loss calculation unit, and an output unit. The loss information storage unit stores loss information associated with evaluation items for calculating an economic loss amount according to the value of the evaluation item, and the evaluation unit evaluates the employee's health condition based on the employee's responses to the evaluation items. The economic loss calculation unit calculates an economic loss amount based on the responses and the corresponding loss information, and the output unit outputs the economic loss amount.

[0005] The health assessment system in this document, for example, prompts the user to answer multiple questions (physical questions) based on their current physical condition and lifestyle habits, multiple questions (mental questions) based on their psychological state, and multiple questions (engagement questions) based on their motivation. A physical score, mental score, and engagement score are calculated based on the answers to each question (evaluation item), and a wellness score is further calculated based on these scores. The health assessment system also performs economic evaluations of the physical score, mental score, and engagement score. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Publication No. 2017-21660 Summary of the Invention [Problem to be solved by the invention]

[0007] In this way, it is possible to estimate a company's economic losses by scoring physical or psychological health conditions based on questionnaire responses, etc., and adding up the scores calculated for each category. However, in order to improve the productivity of a company or organization, it is important to correctly understand presenteeism, which is a subjective indicator of economic losses whose condition is difficult to detect and grasp, and to understand which factors can be effectively improved to alleviate presenteeism.

[0008] The present disclosure has been made in light of the above circumstances, and aims to provide a health management support system and a health management support program that can support improved productivity in companies and organizations by clarifying the relationship between presenteeism and other health conditions that may affect it and linking it to improvements in presenteeism. [Means for solving the problem]

[0009] The health management support system according to this embodiment includes an index data calculation unit that calculates index data associated with each of the indicators of presenteeism, health awareness, stress, engagement, and happiness by quantifying or discretizing the indicators based on data acquired from a user, and a structural model generation unit that outputs a structural model that shows the relationship between the indicator of presenteeism and each of the indicators of health awareness, stress, engagement, and happiness based on the index data.

[0010] In addition, a health management support system according to another embodiment of the present invention may further include an influence calculation unit that calculates the influence of each of the indicators of health awareness, stress, engagement, and happiness on the indicator of presenteeism.

[0011] In addition, in the health management support system according to another embodiment of the present invention, the influence calculation unit may further calculate the mutual influence of the indicators of health awareness, stress, engagement, and happiness.

[0012] In addition, in another form of the health management support system of this embodiment, when the data obtained from the user when a first measure and a second measure are implemented are data corresponding to the first measure and data corresponding to the second measure, respectively, the index data calculation unit calculates first index data and second index data, which are index data corresponding to each of the indexes of presenteeism, health awareness, stress, engagement, and happiness, corresponding to the data corresponding to the first measure and the data corresponding to the second measure, respectively, and the influence calculation unit may calculate the difference or the mutual change rate between the first index data and the second index data for each of the indexes.

[0013] In addition, in the health management support system according to another embodiment of the present invention, the first measure and the second measure may be implemented for different users.

[0014] In addition, in another form of the health management support system of this embodiment, when data obtained from the user at a first time is defined as data corresponding to the first time and data obtained from the user at a second time is defined as data corresponding to the second time, the index data calculation unit calculates first index data and second index data, which are each index data associated with each of the indexes of presenteeism, health awareness, stress, engagement and happiness, corresponding to the data corresponding to the first time and the data corresponding to the second time, respectively, and the influence calculation unit may calculate the difference or mutual change rate between the first index data and the second index data for each of the indexes.

[0015] In addition, in the health management support system according to another aspect of the present embodiment, the measure to be the target of effectiveness verification may be implemented between the first time and the second time.

[0016] In addition, in the health management support system according to another aspect of the present embodiment, the influence degree calculation unit may indicate the possibility of the existence of an interaction or a confounding factor between the indicators when calculating the influence degree.

[0017] In addition, in another form of the health management support system of this embodiment, the structural model generation unit outputs a structural model showing the relationship between each of the indicators of health awareness, stress, engagement, happiness, and physical health status and the indicator of presenteeism based on the indicators calculated by the indicator data calculation unit and indicator data in which the indicator of physical health status has been quantified or discretized, and the influence calculation unit may calculate the influence of each of the indicators of health awareness, stress, engagement, happiness, and physical health status on the indicator of presenteeism.

[0018] In addition, in the health management support system according to another embodiment of the present invention, the physical health condition index may be an index based on a medical checkup result, an index of lifestyle habits, or an index based on medical receipt data.

[0019] According to this embodiment, the computer program that causes a computer to execute processing to support health management calculates, based on data acquired from a user, index data associated with each of the indicators of presenteeism, health awareness, stress, engagement, and happiness, by quantifying or discretizing the indicators, and then executes processing to output, based on the index data, a structural model showing the relationship between the indicator of presenteeism and each of the indicators of health awareness, stress, engagement, and happiness.

[0020] In addition, according to another form of this embodiment, a computer program that causes a computer to execute processing to support health management may further cause the computer to execute processing to calculate the influence of each of the indicators of health awareness, stress, engagement, and happiness on the indicator of presenteeism. [Effects of the Invention]

[0021] According to this embodiment, by clarifying the relationship between presenteeism and other health conditions that may affect it and linking it to improvements in presenteeism, it is possible to provide a health management support system and health management support program that can help improve the productivity of companies and organizations. [Brief explanation of the drawings]

[0022] [Figure 1] 1 is a schematic diagram illustrating an example of the configuration of a health management support system according to an embodiment of the present invention. [Figure 2] FIG. 2 illustrates an example of a hardware configuration of a management server. [Figure 3] FIG. 2 illustrates an example of the software configuration of an auxiliary storage unit. [Figure 4] FIG. 2 illustrates an example of the software configuration of a management server. [Figure 5] FIG. 10 is an explanatory diagram showing an example of a questionnaire form. [Figure 6] FIG. 10 is an explanatory diagram showing an example of a questionnaire form. [Figure 7] FIG. 10 is a diagram illustrating an example of a structural model. [Figure 8] FIG. 10 is a diagram illustrating an example of a structural model. [Figure 9] FIG. 2 is a flowchart illustrating the processing content of the health management support system according to the present embodiment. [Figure 10] FIG. 10 is a diagram illustrating another example of the software configuration of the auxiliary storage unit. [Figure 11] FIG. 10 is a diagram illustrating an example of a structural model according to a modified example. [Figure 12] FIG. 10 is a diagram showing the degree of influence between each index before and after a measure. [Figure 13] 10 is an example of a table showing influence degrees. [Figure 14] FIG. 10 is a diagram illustrating the change in the conditional probability of psychological health status before and after the implementation of a measure. [Figure 15] 10 is another example of a table showing the degree of influence. DETAILED DESCRIPTION OF THE INVENTION

[0023] An example of the health management support system of the present disclosure will be described below with reference to the drawings, etc. However, the health management support system of the present disclosure is not limited to the embodiments and examples described below.

[0024] The figures shown below are schematic illustrations. Therefore, the configuration, size, and shape of each part are appropriately exaggerated to facilitate understanding. Furthermore, hatching indicating the cross section of a member is omitted as appropriate in each figure. The numerical values such as dimensions of each member and the names of materials described in this specification are examples of embodiments and are not limited to these, and may be selected and used as appropriate. In this specification, terms specifying shapes and geometric conditions, such as parallel, orthogonal, and perpendicular, are intended to include not only their strict meanings but also substantially the same states.

[0025] 1. Embodiments of the present disclosure An embodiment of the health management support system of the present disclosure will be described. Health management support system 10 according to this embodiment is illustratively configured by a management server 1, an administrator terminal 2, and user terminals 3 owned by multiple users, each connected to each other so that they can communicate with each other via a communication network 20. However, health management support system 10 of the present disclosure may not have administrator terminal 2 and may be configured only by management server 1 and user terminals 3, or only by management server 1. Management server 1 is not limited to a server and may be any information processing device. The information processing device may be a simple computer device such as a personal computer. The functions of management server 1 may be realized as software functions using cloud computing or the like.

[0026] (a) Configuration of the health management support system

[0027] As shown in the hardware configuration example of Fig. 2, the management server 1 includes a control unit 11, a main memory unit 12, an auxiliary memory unit 13, an input unit 14, an output unit 15, and an interface unit 16, which are electrically connected to each other via a bus 17. The administrator terminal 2 and the user terminal 3 are simple computer devices such as personal computers, smartphones, and tablets that are capable of various operation inputs, result display, communication, etc.

[0028] The control unit 11 includes a CPU (Central Processing Unit), an MPU (Micro Processing Unit), etc., and receives input of various information, performs various calculations and decisions, and outputs various instructions, etc. The interface unit 16 includes, for example, a NIC (Network Interface Card), etc., and is a component that enables communication with external devices via the communication network 20. The interface unit 16 has the function of converting input signals received from various devices via the communication network 20 into data that can be recognized by the control unit 11, and also converting data output from the control unit 11 into an output signal that can be sent to the communication network 20.

[0029] The communication network 20 is configured as a wired or wireless network, or a combination of these. When the communication network 20 is wireless, the management server 1, the administrator terminal 2, and the user terminal 3 are connected to a wide area communication network such as a wireless LAN network, a mobile communication network, or the Internet via a Wi-Fi (registered trademark) access point or an LTE (Long Term Evolution) base station. As a result, it becomes possible to send and receive information to and from various external devices in nearby or remote locations connected to the communication network.

[0030] The main memory unit 12 includes a ROM that mainly stores a boot program for starting the control unit 11, an operating system, etc., and a RAM that is used for temporarily storing data during calculations performed by the CPU, etc., of the control unit 11. The main memory unit 12 also includes a non-volatile rewritable memory such as a flash memory that stores important data such as system parameters.

[0031] Meanwhile, auxiliary storage unit 13 mainly includes a nonvolatile rewritable memory such as a flash memory, and stores various application programs for the health management support system, parameters used therefor, etc. Furthermore, as shown in FIG. 3, it stores information on databases (DBs) of various questions that form the basis for calculating the scores of each index related to psychological health status used in the health management support system. Examples include a health awareness question information DB110 related to health awareness, a stress question information DB120 related to stress, an engagement question information DB130 related to engagement, and a happiness question information DB140 related to happiness. Furthermore, it stores a presenteeism question information DB150 related to presenteeism.

[0032] The auxiliary storage unit 13 also stores information in a questionnaire form DB210 that combines information from each of the above-mentioned databases to create a questionnaire for the user, and information in a response information DB220 that is the results of users' responses to the questionnaire form DB 210. Note that while the information from each of these databases is stored in the auxiliary storage unit 13 of the management server 1, it is also possible to provide an independent storage device separate from the management server 1 as a database server, and store the information in this.

[0033] The input unit 14 is a part that allows input operations to be performed on the management server 1, and includes input devices such as a mouse, keyboard, touch panel, and touch pen.

[0034] On the other hand, the output unit 15 is a part that outputs the results of judgments and calculations made by the management server 1 to the outside, for example by visualizing them, and includes output devices such as a display and a printer.

[0035] The bus 17 electrically connects the above-mentioned components and relays the transmission and reception of address signals, control signals, and data from one component to another.

[0036] Furthermore, the configuration of the management server 1 can be broken down into functional and software aspects. In this case, as shown in Fig. 4, the management server 1 includes a questionnaire creation unit 310, an index data calculation unit 320, a structural model generation unit 330, and an influence calculation unit 340. The questionnaire creation unit 310 has a function of extracting and assembling a set of questions to be actually asked of the user from each question information DB stored for each category related to psychological health state in the auxiliary storage unit 13 described above, and compiling the sets into a questionnaire.

[0037] For example, to ask questions about health awareness, engagement, and presenteeism as categories of psychological health status, the questionnaire creation unit 310 first accesses the health awareness question information DB 110, the engagement question information DB 130, and the presenteeism question information DB 150. Then, it reads out the necessary question information from these databases and assembles it into one or more questionnaires. The contents of the questionnaires are stored in the auxiliary storage unit 13 as a questionnaire form DB 210.

[0038] The questionnaire information created by the questionnaire creation unit 310 is transferred as a file from the management server 1 to the user terminal 3 via the communication network 20, and the user opens the downloaded file to display the questionnaire on the display screen of the user terminal 3. Alternatively, the user may launch an application program on their own user terminal 3 for viewing a web page provided by the management server 1, and view the questionnaire on the display screen of the user terminal 3 on the website of the management server 1. Alternatively, these questionnaires may be printed out on paper in advance and sent by mail or the like to the necessary users.

[0039] Examples of questionnaire forms are shown in FIGS. 5 and 6. For example, as a question regarding health awareness, the upper section of the questionnaire 211 in FIG. 5 lists a question item, "If you develop a lifestyle-related disease, how much do you know about the future diseases and risks you may face? (Your own assessment)" as question item "No. 3" under "(1) Questions Regarding Health Awareness." In response to this question, the user answers the question by selecting one of seven options: "I don't know at all," "I don't know," "I don't know very much," "I don't know either way," "I know somewhat," "I know," or "I know very well." In this embodiment, the user answers by checking the checkbox at the beginning of the corresponding answer section, but any selection method may be used.

[0040] Similarly, in the middle section of the questionnaire 211 in FIG. 5, the question "No. 12" under "(2) Questions about Stress" asks, "Have you ever felt depressed and like nothing could cheer you up?" The user answers this question by selecting one of five options. In addition, in the lower section of the questionnaire 211 in FIG. 5, the question "(3) Questions about Engagement" is listed, and in the upper section of the questionnaire 211 in FIG. 6, the question "(4) Questions about Happiness" is listed.

[0041] Furthermore, the lower section of the questionnaire 211 in Figure 6 lists a question item, "(5) Questions about presenteeism." As described in "(5) Questions about presenteeism," the answer in this case is to select one of multiple options, but the options are not limited to verbal expressions and may also be numerical selections or written numerical values or words. The answer results may be converted into scores as a quantitative variable, or may be divided into discrete groups as a qualitative variable.

[0042] Here, we will explain the indicators of psychological health status: health awareness, stress (mental health), engagement, happiness, and presenteeism, as well as the questionnaires related to each.

[0043] Health consciousness is an indicator of a user's subjective view of health and level of health awareness, i.e., their level of involvement in health, and may be related to presenteeism. Furthermore, when implementing measures for improvement, selecting target groups based on the level of health awareness makes it easier to evaluate the effectiveness of measures based on differences in health awareness levels. In addition to the questions shown in Figure 5, health consciousness questions include, for example, "Do you consider yourself healthy?", which can be answered with a four-point scale (choose one of "not good," "not very good," "somewhat good," or "good"). Other questions and answer options include the level of understanding health risks, the level of awareness of one's own health through diet and other means, the level of taking actions to improve one's health through exercise, and the level of continuing to take actions to improve one's health.

[0044] Stress (mental health) is an indicator of the degree of mental health problems, including psychological stress, and is naturally related to presenteeism. A widely used method for assessing stress is the K6 method, which was developed by Kessler et al. in the United States for the purpose of screening for mental illnesses such as depression and anxiety disorders. Six questions, such as "Did you feel irritable?" and "Did you feel hopeless?" are answered with a five-point scale (choose one of "never," "a little," "sometimes," "most of the time," or "always"). Questions based on legally mandated surveys, such as "stress checks," may also be selected.

[0045] Furthermore, the stress index and its level may be determined based on the responses to such a questionnaire, as well as measurement data such as the user's heart rate, blood pressure, electrocardiogram, and lung capacity.

[0046] Engagement is an indicator of a user's positive and fulfilling psychological state related to their work. Highly engaged people feel pride and fulfillment in their work, work enthusiastically, derive energy from their work, and are in a lively state. This indicator is also thought to be related to presenteeism. A well-known questionnaire regarding engagement is the three-item version of the UWES (Urecht Work Engagement Scale).

[0047] Specifically, three questions are set: "When I work, I feel energized (energy)," "I am enthusiastic about my work (enthusiasm)," and "I am absorbed in my work (immersion)." In addition, seven answer options are provided (choose from "never," "hardly," "seldom," "sometimes," "often," "very often," or "always"). However, other questions and answer options may also be provided.

[0048] Happiness is also a measure that may be related to presenteeism. For example, as an indicator of subjective happiness, Diener's Satisfaction with Life Scale (SWLS) could be used, or perspectives such as life satisfaction and job satisfaction could be incorporated. When using the SWLS, in addition to the items listed in Figure 6, five questions are included: "In most ways, my life is close to my ideal," "My life is in a very good state," "I am satisfied with my life," and "If I could live my life over again, I would change almost nothing." Response options are provided on a seven-point scale (choose from "completely disagree," "mostly disagree," "somewhat disagree," "neither agree nor disagree," "somewhat agree," "quite agree," or "completely agree").

[0049] Furthermore, in terms of life satisfaction, for example, the question "Are you satisfied with your family life?" may be scored or grouped into four categories ("dissatisfied," "somewhat dissatisfied," "fairly satisfied," and "satisfied"). Furthermore, in terms of job satisfaction, for example, the question "Are you satisfied with your job?" may be scored or grouped into four categories ("dissatisfied," "somewhat dissatisfied," "fairly satisfied," and "satisfied"). However, question items and answer options other than those listed here may also be provided.

[0050] As mentioned above, presenteeism is an indicator of the subjective economic loss caused by a decline in employee performance due to health reasons, including mental health issues, that do not result in absenteeism. Proposed outcome assessment indicators for this include the WHO-HPQ (Health and Work Performance Questionnaire), the University of Tokyo single-item version, and the WLQ. In particular, the WHO-HPQ uses the WHO Health and Work Performance Questionnaire, a questionnaire used globally by the WHO, and is assessed with three questions, each of which is answered by selecting a score from 0 to 10. Scoring is displayed in two ways: absolute presenteeism and relative presenteeism. However, questions and answer options other than those listed here may also be used.

[0051] A questionnaire containing the necessary questions for each category related to psychological health status and answer options as described above is created by the questionnaire creation unit 310, and the contents are displayed on the user terminal 3. Alternatively, a questionnaire with the written items printed out on paper is sent to the user. The user inputs answers to the questionnaire by inputting them into the user terminal 3. Alternatively, the user writes answers by hand on the paper questionnaire. The answer information input from the user terminal 3 is sent to the management server 1 via the communication network 20 and stored in the answer information DB 220 of the auxiliary storage unit 13. Furthermore, if answers are written by hand on a paper questionnaire, the paper questionnaire is read using a scanner or the like, and the user's answer information is converted into digital data, which is then stored in the answer information DB 220 of the auxiliary storage unit 13.

[0052] Next, the index data calculation unit 320 in Fig. 4 will be described. As described above, the index data calculation unit 320 reads out the response information of each user from the response information DB 220, which stores the response information to the questionnaire regarding psychological health status created by the questionnaire creation unit 310. Then, the index data calculation unit 320 calculates index data by quantifying or discretizing each index regarding the psychological health status of each user. Taking the health consciousness index as an example, a predetermined score is assigned to each answer option for each question, and the scores of the answers to all questions are added up to calculate a numerical total score for the health consciousness index.

[0053] In other words, health awareness indicators can be quantified. Here, the higher the score, the higher the user's health awareness can be judged to be. However, there are some indicators, such as stress and presenteeism, where a lower score is generally considered to be better, but such scoring methods can be changed as desired.

[0054] Meanwhile, by assigning a predetermined ranking to each answer option for each question (for example, dividing it into four ranks: a, b, c, and d), and adding up the number of answers for each rank for all questions, the proportion of each rank for the health consciousness index can be calculated. In other words, the health consciousness index can be discretized by assigning it to mutually discrete qualitative ranks such as A, B, and C. Here, for example, the higher the proportion of the highest rank, a, or the lower the proportion of the lowest rank, d, the higher the user's relative health consciousness can be determined. This also applies to other indices such as stress, engagement, happiness, and presenteeism.

[0055] In this way, the index data calculation unit 320 has a function of calculating index data for each user by quantifying or discretizing necessary indexes from among the indexes of health awareness, stress, engagement, happiness, and presenteeism based on the user's response information to the questionnaire. However, the index data calculation unit 320 is not limited to collecting the user's response information to the questionnaire in a questionnaire-like format and quantifying each index based on the collected information. For example, the index data may be calculated based on collected, objective data such as physical measurement results and health checkup results of the user, such as stress and physical health indicators described below. Alternatively, information corresponding to the answers to the questions on the questionnaire may be extracted, processed, or estimated from the user's daily work report, attendance information, etc.

[0056] Next, the structural model generation unit 330 in Fig. 4 will be described. The structural model generation unit 330 has a function of generating and outputting a structural model that clarifies, for example, the relationship between each index, based on the index data for each user of health awareness, stress, engagement, happiness, and presenteeism calculated by the index data calculation unit 320. In particular, the structural model generation unit 330 generates and outputs a structural model that shows the relationship between each index of health awareness, stress, engagement, and happiness and the presenteeism index.

[0057] Here, the relationship necessarily refers to the presence or absence of a causal relationship between the indicators, and may also include the directionality of the influence between the indicators. That is, when there are indicators A and B, and a change in one of them causes a change in the other, it is considered that there is a causal relationship between the two indicators, and that indicators A and B have a relationship. Also, when a change in indicator A causes a change in indicator B, but a change in indicator A does not cause a change in indicator B, that is, when a change in indicator A is unrelated to a change in indicator B, it is considered that both indicators have a directionality from indicator A to indicator B.

[0058] Various known analytical methods can be used to generate a structural model that represents the relationships between the indicators based on the indicator data calculated from the response information of multiple users to questionnaires, etc., calculated by the indicator data calculation unit 320. Particularly preferred examples of structural modeling methods include Bayesian network analysis and structural analysis of covariance (SEM).

[0059] Bayesian network analysis is a type of probability model (graphical model) that can be used for predicting uncertain events, rational decision-making, and fault diagnosis that explores causes from observed results. In particular, when a causal relationship is recognized in which one indicator is the cause and the other is the result, a structural model showing the relationships between each indicator can be easily generated by representing each indicator as a random variable and creating a probability model that expresses the quantitative relationships between individual variables using conditional probability.

[0060] In a Bayesian network, each indicator having a random variable is called a node, and each directed link connecting the nodes and oriented along the direction of the causal relationship between the two nodes is called an edge. A Bayesian network has a structure in which each node having a random variable is linked by edges indicating the causal relationship, including the direction of the causal relationship, and there are no multiple paths from a first node to a second node along the edges. For example, Figure 7(a) shows an example of a graph representing a Bayesian network model. This graph has nodes N551, N552, and N553, which represent indicators that are random variables X1, X2, and X3, respectively, and edges E651 and E652, which are directed links from X1 and X2 to X3.

[0061] Here, random variable X3 is influenced by both random variables X1 and X2, but random variables X1 and X2 are not influenced by random variable X3 and are independent of X3. In this case, X3 at the end of the directed link is also called the child node, and X1 and X2 at the end of the directed link are also called parent nodes. In this case, the joint probability distribution P(X1,X2,X3) of random variables X1, X2, and X3 is determined as follows using the prior probabilities P(X1) and P(X2) of X1 and X2, and the conditional probability P(X3 | X1,X2) of X3:

number

[0062] For example, when there are multiple parent nodes, the set of parent nodes of child node Xj is defined as Pa(Xj). In this case, the relationship between Xj and Pa(Xj), i.e., the dependency, can be expressed as P(Xj | Pa(Xj)). However, if Pa(Xj) is an empty set, it is treated as a prior probability distribution. Furthermore, if we consider each of the n random variables X1, ,Xn as a child node, the joint probability distribution P(X1, ,Xn) of all random variables can be expressed as follows:

number

[0063] If the observed value of any one of the random variables X1 through Xn is known, the posterior probability distribution of the other random variables can be calculated by substituting this observed value. In the example of Figure 7(a), the posterior probability distribution of the causes X1 and X2 can be calculated from the observed value of the result X3, and their expected values, maximum posterior probabilities, entropy, etc. can be obtained. Furthermore, quantitative information can be obtained to estimate the reasonable conditions for X1 and X2 to obtain the desired result of X3. Conversely, it is also possible to predict in advance how the probability distribution of the result X3 will change if the observed values of the causes X1 and X2 are changed, assuming specific measures.

[0064] 7(b) shows an example of a graph visualized as a structural model based on Bayesian network analysis, based on the index data for each user of the indexes of health consciousness, stress, engagement, happiness, and presenteeism calculated by the index data calculation unit 320. Here, it can be seen that the indexes that are the direct cause of the resulting index of presenteeism (P) are the three indexes of health consciousness (HS), happiness (W), and engagement (E), with their edge arrows pointing toward P. From nodes N502, N503, and N504 of HS, W, and E, there are edges E605, E608, and E607 that point toward node N505 of P.

[0065] Furthermore, the direct cause of the result, the health consciousness (HS) index is the stress (S) index, and the direct cause of the result, the engagement (E) index is also the stress (S) index. From the S node N501, there is an edge E601 pointing to the HS node N502, and an edge E603 pointing to the E node N504. Furthermore, the direct causes of the result, the happiness (W) index, are the three indexes, health consciousness (HS), stress (S), and engagement (E). From the HS, S, and E nodes N502, N501, and N504, there are edges E604, E602, and E606 pointing to the W node N503.

[0066] This shows that the result, presenteeism (P), has a direct or indirect causal relationship with all four other indicators. The strength of the relationship between each indicator can be determined by the magnitude of the conditional probability mentioned above.

[0067] In addition, when a first node indicating a first indicator of a structural model and a second node indicating a second indicator are linked by an edge from the first node to the second node, the first indicator may be referred to as an indicator that is a direct cause of the second indicator. Furthermore, when a first node indicating a first indicator of a structural model, a second node indicating a second indicator, and a third node indicating a third indicator are linked by an edge from the first node to the third node and an edge from the third node to the second node, the first indicator may be referred to as an indicator that is an indirect cause of the second indicator. In this case, multiple third indicators and third nodes may exist in series. However, each node must be linked by a unidirectional edge from the first node to the second node.

[0068] Here, in the graph of FIG. 7(b), in addition to the indicators HS, S, E, W, and P, a predetermined measure D is displayed as an additional indicator by a dashed line. D is an indicator that is thought to be a direct cause of S and E. From node N531 of D, there are edges E631 and E632 that point to node N501 of S and node N504 of E. Data such as responses to questionnaires regarding each indicator obtained from users before implementing measure D is defined as the first data, and data such as responses to questionnaires regarding each indicator obtained from users after implementing measure D is defined as the second data.

[0069] At this time, the index data calculation unit 320 calculates first index data and second index data, which are index data associated with each of the indexes P, HS, S, E, and W corresponding to the first data and second data, respectively. Furthermore, the influence calculation unit 340, which will be described later, can calculate the effect on each index of the implementation of measure D by calculating the difference between the first index data and the second index data of each index or the rate of change between them. This point will be described later.

[0070] FIG. 8 also shows a graph illustrating the relationship between the structural models when the same organization or user answers a first questionnaire and a second questionnaire at different times. Assume, for example, that the second questionnaire is answered m months after the first questionnaire. Based on the information in the answers to the first questionnaire, index data for the indices health awareness (HS1), stress (S1), engagement (E1), happiness (W1), and presenteeism (P1) are calculated. The symbols for the nodes HS1, S1, E1, W1, and P1 and the edges indicating the links between the nodes are the same as the symbols for the nodes HS, S, E, W, and P and the edges indicating the links between the nodes.

[0071] Similarly, index data for the indicators of health consciousness (HS2), stress (S2), engagement (E2), happiness (W2), and presenteeism (P2) are calculated based on the responses to the second questionnaire. The nodes for HS2, S2, E2, W2, and P2 are N522, N521, N524, N523, and N525, respectively. The edges of the structural model of the first responses corresponding to edges E601, E602, E603, E604, E605, E606, E607, and E608 are E621, E622, E623, E624, E625, E626, E627, and E628, respectively.

[0072] By examining the difference in the index data for each index, the relative degree of improvement or deterioration of each index over m months can be ascertained. For example, suppose that the index data for P2 has improved relative to P1. Furthermore, suppose that Bayesian network analysis reveals that the conditional probability value phs2 of node N522, which is HS2, is greater than the conditional probability value phs1 of node N502, which is HS1. Furthermore, suppose that the conditional probability value pp2 of node N525, which is P2, is greater than the conditional probability value pp1 of node N505, which is P1. In this case, it can be estimated that both the health awareness and presenteeism indexes have improved over the m months, and that the influence of the health awareness index on the presenteeism index has also increased.

[0073] Therefore, it is possible to further estimate which factors among the measures and environmental changes implemented over a period of m months are related to changes in the degree of impact. This means that even if changes occur in each indicator over a certain period of time, it is possible to assume that the effect is due to the implementation of measure D during that period, and calculate the degree of impact of each indicator by creating a structural model that adds measure D as an indicator, as shown in Figure 7(b). This is also true when using covariance structure analysis, which will be described later.

[0074] In Figure 8, the structural model corresponding to the response information of the first questionnaire and the structural model corresponding to the response information of the second questionnaire are shown as being the same structural model. However, when using Bayesian network analysis, the structural model itself, i.e., the link relationships between the edges of each node, may change. In this case, the above estimation is still possible.

[0075] Meanwhile, the structural model generation unit 330 may generate a desired structural model showing the relationships between each indicator using the Bayesian network analysis described above. Alternatively, a structural model may be generated using structural covariance analysis (SEM). SEM is a statistical approach that extends factor analysis and multiple regression analysis, introducing latent variables that are difficult to observe directly and identifying causal relationships between the latent variables and observed variables. This method assumes that a hypothesis of the causal relationships between each indicator is first formulated and modeled. Then, the validity of the hypothetical model, whether it can be modified, and the strength of the causal relationships can be estimated and tested.

[0076] Covariance structure analysis may be used instead of Bayesian network analysis to generate a structural model that shows the relationships between each indicator. However, as will be described later, the connections between each indicator (the relationships between each node and the edges connecting them) may be determined first using Bayesian network analysis, increasing the reliability of the structural model, and then covariance structure analysis may be used at the stage of quantifying the specific degree of influence between each indicator. This increases the reliability of the generation of structural model hypotheses compared to using covariance structure analysis alone, and allows for efficient quantification of the degree of influence between each indicator after determining a relatively accurate structural model.

[0077] Next, the influence calculation unit 340 of FIG. 4 will be described. The influence calculation unit 340 has a function of calculating the mutual influence of each index based on a structural model generated by the structural model generation unit 330, which visualizes or clarifies the relationships between the indexes of health awareness, stress, engagement, happiness, and presenteeism. The influence calculation unit 340 particularly calculates the influence of each index of health awareness, stress, engagement, and happiness on the presenteeism index. This is because quantitatively understanding the influence of each index that is a direct and indirect cause of the presenteeism index is important for improving the presenteeism index. Note that the influence is a quantitatively quantified or discretized measure of the degree to which the magnitude or fluctuation of index A affects the magnitude or fluctuation of index B, for example, when a relationship in which index A is the cause and index B is the result is considered in the structural model.

[0078] Although a specific example will be described later, when the structural model generation unit 330 generates a structural model showing the relationship between each index by Bayesian network analysis, the influence calculation unit 340 may calculate the value of the conditional probability value between each index or the posterior probability value reflecting the observed value as the influence between each index. Furthermore, when the structural model generation unit 330 generates a structural model showing the relationship between each index by covariance structure analysis, the influence calculation unit 340 may calculate the value of the path coefficient between each index (a coefficient representing the strength of the causal relationship between both indexes) as the influence between each index.

[0079] (b) Processing contents of the health management support system Next, the processing content of health management support system 10 of this embodiment will be described, focusing mainly on the processing of management server 1.

[0080] First, the administrator uses the administrator terminal 2 to instruct the management server 1 to create a questionnaire, including the questions to be asked to the user (step S421 in FIG. 9). Specifically, the administrator selects the indicators to be used for calculating the final structural model and the impact. For example, only the indicators of health awareness, stress, engagement, and happiness, which indicate psychological health status, may be used as indicators that can affect the presenteeism indicator. In addition, indicators based on health checkup results, lifestyle habits, and prescription data may be added as indicators of physical health status. These indicators do not necessarily have to be all the questions asked in the questionnaire to the user, and information may be collected from performance data such as health checkups, physical fitness tests, and stress checks.

[0081] Next, the management server 1 creates a questionnaire to be sent to the user (step S401). Specifically, the questionnaire creation unit 310 of the management server 1 extracts and combines a set of necessary questions from a database of question information for each index, such as the health awareness question information DB 110, to form a specific questionnaire form. The completed questionnaire form is stored in the questionnaire form DB 210 of the auxiliary storage unit 13. The management server 1 then converts the questionnaire into a file and sends it to the user terminal 3, or uploads the questionnaire information to the management server 1's website so that it can be viewed from the user terminal 3 (step S402).

[0082] The user receives the questionnaire in file format from the management server 1 via the user terminal 3, or views the questionnaire on the website of the management server 1 (step S411). The user enters answers, such as selections, in the answer fields of the questionnaire form while viewing the questionnaire displayed on the screen of the display unit of the user terminal 3 (step S412). Answers may be entered by checking the desired answer from among the options, or by entering the number or wording indicated by the option. Alternatively, the questionnaire may not be displayed on the screen of the user terminal 3, but may be distributed in advance as a paper document by mail, and only the answer input may be performed on the user terminal 3.

[0083] Next, the user transmits the questionnaire information with the answers entered from the user terminal 3 to the management server 1 (step S413). When the user enters answers to the questionnaire on the website of the management server 1, the answer information is collected by the management server 1 at that time. The answer information is stored in the answer information DB 220 of the auxiliary storage unit 13 of the management server 1. Then, the index data calculation unit 320 of the management server 1 calculates each index data associated with each index by quantifying or discretizing each index based on the answer information from the user stored in the answer information DB 220 (step S403).

[0084] The user may handwrite the answers on the paper questionnaire that is provided to them. In this case, the information on the questionnaire may be read as image information using a scanner or the like, and the user's answer information may be identified and converted into digital data through image processing or the like, and this may be stored as answer information in the answer information DB 220 of the auxiliary storage unit 13 of the management server 1.

[0085] As described above, the index data is calculated by assigning a predetermined score to each answer option for each question and adding up the scores for all questions to calculate a total score for the target index. Alternatively, a predetermined ranking may be assigned to each answer option for each question, and the proportion of each rank for the target index may be calculated by adding up the numbers for each rank for all questions. In other words, the index data is calculated as quantified data or discretized data.

[0086] Next, based on the index data for each index, content analysis is performed to generate a structural model and calculate the degree of influence between each index (step S404). Specifically, the structural model generation unit 330 of the management server 1 generates and outputs a structural model showing the relationships between each index, particularly the relationships between the health awareness, stress, engagement, and happiness indexes and the presenteeism index, using Bayesian network analysis, covariance structure analysis, etc., from the index data for each index. Furthermore, the influence calculation unit 340 of the management server 1 calculates the degree of influence between each index based on the structural model, for example, in the case of Bayesian network analysis, based on the conditional probability or posterior probability between each index, or in the case of covariance structure analysis, based on the path coefficient between each index. In particular, the influence of the health awareness, stress, engagement, and happiness indexes on the presenteeism index is calculated.

[0087] Next, the analysis results, including the generated structural model and the calculated influence, are output as a display or printed document via output unit 15 of management server 1. These analysis results are also transmitted or uploaded to management server 1 so that they can be displayed or viewed on administrator terminal 2 or user terminal 3 (step S405). The administrator or user receives or makes the result information viewable and displayed via administrator terminal 2 or user terminal 3 (steps S414, S423). This completes a series of processes in health management support system 10.

[0088] (c) Health management support system according to an embodiment In summary, health management support system 10 of this embodiment includes index data calculation unit 320 that calculates index data associated with each index by quantifying or discretizing each of the indexes of presenteeism, health awareness, stress, engagement, and happiness according to data acquired from a user. Health management support system 10 also includes structural model generation unit 330 that outputs a structural model indicating the relationship between each of the indexes of health awareness, stress, engagement, and happiness and the index of presenteeism based on the index data.

[0089] When estimating a company's economic losses, it is important to understand the psychological health status of users, such as employees, which is difficult to grasp based solely on readily apparent information such as physical health status and attendance status. Among these, it is particularly important to correctly understand presenteeism, a subjective indicator of economic loss whose condition is difficult to identify and grasp, and to understand which other health factors should be improved in order to reduce presenteeism. In response to this, according to the present embodiment, the relationship between presenteeism and other health conditions that may affect it can be clarified, and by understanding the structural model, it can be clarified which specific health indicators require improvement to ultimately reduce presenteeism. As a result, a health management support system and a health management support program that can support the improvement of productivity in companies and organizations can be provided.

[0090] Furthermore, the health management support system 10 of this embodiment further includes an influence calculation unit 340 that calculates the influence of each of the indicators of health awareness, stress, engagement, and happiness on the presenteeism indicator. This not only clarifies the relationship between presenteeism and other health conditions that may affect it, but also allows the influence to be understood as the strength of the quantitative causal relationship between the indicators of health awareness, stress, engagement, and happiness on the presenteeism indicator. As a result, it is possible to quantitatively understand the extent to which presenteeism can be improved by improving specific health indicators to what extent. As a result, it is possible to provide a health management support system and a health management support program that can more effectively support improving the productivity of companies and organizations.

[0091] 2. Modifications of the embodiment Next, a modification of the health management support system of the present disclosure will be described.

[0092] (a) Configuration of the health management support system As shown in FIGS. 1, 2, and 4, a health management support system 10a and management server 1a according to the modified example have the same system configuration, hardware configuration, and software configuration as the health management support system 10 and management server 1 according to the previously described embodiment. However, management server 1a has a modified auxiliary storage unit 13a configuration so that physical health condition indicators can be added to each psychological health condition indicator to calculate the desired structural model and influence level. That is, the auxiliary storage unit 13a of management server 1a further includes a health checkup result DB 160 related to health checkup result indicators that serve as physical health condition indicators, and a lifestyle habit question information DB 170 related to lifestyle habit indicators that also serve as physical health condition indicators. This is different from the auxiliary storage unit 13 of management server 1.

[0093] The auxiliary storage unit 13a may include only one of the health check result DB 160 and the lifestyle question information DB 170, or may further include a database related to other indicators. For example, an indicator based on medical receipt data may be used as an indicator of physical health status. Medical receipt data may include, for example, data on tests, diagnoses, treatments, and prescriptions for illnesses, and may also include data on medical history from hospitals, sick leave histories from companies, etc.

[0094] The health checkup result DB 160 stores various data related to indicators of the user's health checkup data, such as the user's blood pressure, blood lipids, blood glucose level, obesity, and medical history, which are measured and recorded during regular health checkups and other events. However, this data may be stored in a server other than the management server 1a and acquired by accessing this information as needed. The indicator data calculation unit 320, structural model generation unit 330, and influence calculation unit 340 can handle indicators of objective physical health status in addition to subjective psychological health status using the information in the health checkup result DB 160. This enables the generation of rational structural models and calculation of influence levels with improved objectivity that do not rely solely on the user's subjectivity.

[0095] Furthermore, the lifestyle question information DB 170 stores information on questions related to lifestyle indicators and their answer options. Lifestyle indicators are indicators related to a user's smoking habits, drinking habits, exercise habits, sleep / rest, breakfast, etc. Lifestyle indicators are not only closely related to stress, health awareness, and physical health status, but are also thought to be related to presenteeism.

[0096] Questions about lifestyle habits, for example, regarding smoking habits, may include a question asking "Do you currently smoke cigarettes habitually?" to which participants are asked to select either "yes" or "no." Regarding drinking habits, for example, a question asking "How much alcohol do you drink (sake, shochu, beer, western spirits, etc.)?" may include a question asking participants to select and enter the frequency of drinking and the "amount of alcohol consumed per day on drinking days." Regarding exercise habits, for example, questions asking "Do you exercise for at least 30 minutes at a time, breaking a light sweat, at least two days a week for at least one year?" or "Do you walk or engage in equivalent physical activity for at least one hour a day in your daily life?" may include a question asking participants to select either "yes" or "no." However, questions and answer options other than those listed here may also be provided.

[0097] The index data calculation unit 320, structural model generation unit 330, and influence calculation unit 340 can collect indices of physical health status as well as subjective psychological health status as a collective piece of information, that is, questionnaire response information from users, using information from the lifestyle habit question information DB 170. This makes it possible to easily generate accurate structural models and calculate influence levels from both psychological and physical health status indices.

[0098] (b) Processing contents of the health management support system The processing content of health management support system 10a is the same as the processing content of health management support system 10 described above, except for the addition of indicators, so detailed explanation will be omitted, but a brief explanation will be given including an actual analysis example.

[0099] First, the administrator uses the administrator terminal 2 to instruct the management server 1a to create a questionnaire (step S421 in FIG. 9). Here, the following indicators are used as indicators that may affect the presenteeism indicator: health awareness, stress, engagement, and happiness, which indicate psychological health status, and physical health status. An indicator of physical health status is, for example, lifestyle habits.

[0100] Next, the management server 1a creates a questionnaire (step S401), transmits the questionnaire file to the user terminal 3, etc. (step S402), and the user enters selected answers into the answer fields of the questionnaire form while viewing the questionnaire received, etc. via the user terminal 3 (steps S411 and S412). Next, the user transmits the questionnaire information with the entered answers from the user terminal 3 to the management server 1a, etc. (step S413). The answer information is stored in the answer information DB 220 of the auxiliary storage unit 13a of the management server 1. Then, the index data calculation unit 320 of the management server 1a calculates each index data associated with each index by quantifying or discretizing each index based on the answer information from the user stored in the answer information DB 220 (step S403). The index data also includes lifestyle habit indexes.

[0101] The index data is calculated by assigning a predetermined score to each answer option for each question and adding up the scores of the answers to all questions to calculate a total score for the target index. Alternatively, a predetermined ranking is assigned to each answer option for each question, and the proportion of each rank for the target index is calculated by adding up the numbers of each rank for the answers to all questions. Figure 13 shows an example of an influence table 341 that shows the influence between each index of the structural model output by the influence calculation unit 340.

[0102] Among these, the numerical values entered in the "index data" column are scores representing the respective index data before and after the implementation of each indicator. The index data may be represented as a numerical score like this, or may be discretized like a ranking. For example, when the "indicator" is "presenteeism," the "index data" before the implementation is "5235," and the "index data" after the implementation is "4092." The value of this "index data" is calculated by the index data calculation unit 320.

[0103] Next, based on the index data of each index, content analysis is performed to generate a structural model and calculate the influence between each index (step S404). First, the structural model generation unit 330 of the management server 1a generates and outputs a structural model that shows the relationships between each index, in particular the relationships between the indexes of health awareness, physical health status, stress, engagement, and happiness and the index of presenteeism, using a Bayesian network analysis or the like from the index data of each index.

[0104] The graph in Figure 11 illustrates a structural model generated by the structural model generation unit 330 using Bayesian network analysis or the like. The indicators of health consciousness (HS), physical health status (H), stress (S), engagement (E), happiness (W), and presenteeism (P) correspond to nodes N502, N506, N501, N504, N503, and N505, respectively. It can be seen that the indicators that directly cause the resulting presenteeism (P) indicator are the three indicators of physical health status (H), happiness (W), and engagement (E), whose edge arrows point toward P. Here, from nodes N506, N503, and N504 of H, W, and E, there are edges E611, E608, and E607 that point toward node N505 of P.

[0105] Furthermore, the direct cause of the result, health (H), is the indicator of health consciousness (HS). The direct cause of the result, health consciousness (HS), is the indicator of stress (S), and the direct cause of the result, engagement (E), is also the indicator of stress (S). From S node N501, there is edge E601 pointing to HS node N502, and edge E603 pointing to E node N504. Furthermore, the direct causes of the result, happiness (W), are the three indicators of physical health (H), stress (S), and engagement (E). From H, S, and E nodes N506, N501, and N504, there are edges E610, E602, and E606 pointing to W node N503.

[0106] This indicates that the resulting presenteeism (P) indicator has direct or indirect causal relationships with all five other indicators. Next, the influence calculation unit 340 of the management server 1a further performs covariance structure analysis on the structural model obtained here, calculating the path coefficients between each indicator as the influence. Figure 12(a) shows the path coefficients along the arrow direction between each node where an edge exists, displayed near each corresponding edge. Note that Figure 12(a) and other figures follow the method of displaying nodes and edges used in Bayesian network analysis, and do not follow the rules of covariance structure analysis. Note that the closer the absolute value of a path coefficient is to 1, the stronger the causal relationship between the causal indicator and the resulting indicator.

[0107] As shown in Figures 12(a) and 13, before the implementation of the measures, the indicators in the structural model that directly affect presenteeism (P) are physical health (H), engagement (E), and happiness (W). The path coefficients are -0.26, -0.21, and -0.46, respectively. Furthermore, the indicators that directly affect happiness (W) are physical health (H), stress (S), and engagement (E), with path coefficients of 0.31, -0.14, and 0.60, respectively. From these results, it can be estimated that, before the implementation of the measures, the indicator with the strongest direct effect on presenteeism (P) is happiness (W), and the indicator with the strongest direct effect on happiness (W) is engagement (E).

[0108] Such an evaluation of the degree of influence between indicators can be achieved through statistical methods such as generating a structural model once, Bayesian network analysis, or covariance structure analysis. However, it is also possible to predict the effectiveness of a measure in more detail by implementing a specific measure on a trial basis, or assuming that it has been implemented, and examining the extent to which the effect of that measure contributes to improving each indicator, particularly the presenteeism indicator.

[0109] For example, before a specific measure is implemented, the desired structural model shown in FIG. 11 is obtained through Bayesian network analysis based on the response information of the first questionnaire to each user in a specific organization. Furthermore, based on this, a covariance structure analysis is further performed to calculate path coefficients as the degree of influence between each indicator in the structural model, and the results are shown in the "Pre-Measure Impact" column in FIG. 12(a) and FIG. 13. After a specific measure is implemented, a second questionnaire is requested from each user in the specific organization, and Bayesian network analysis and covariance structure analysis are performed based on the response information. The structural model thus obtained and the path coefficients as the degree of influence are shown in the "Post-Measure Impact" column in FIG. 12(b) and FIG. 13.

[0110] In the influence table 341 of FIG. 13, in addition to displaying the values of the first index data before the implementation of the measure and the second index data after the implementation of the measure, calculated by the influence calculation unit 340, the difference between the values of the first index data and the second index data is calculated and displayed. Instead of the difference between the values of the first index data and the second index data, a mutual change rate may be calculated. Furthermore, the influence table 341 also calculates and displays the path coefficients and the differences in their absolute values calculated by covariance structure analysis for each index before and after the implementation of the measure. Again, instead of the difference, a mutual change rate may be calculated.

[0111] According to this, for example, the path from stress (S) to happiness (W) showed the largest difference in absolute value of the path coefficient before and after the measure, at 0.09. The second largest difference was for the path from happiness (W) to presenteeism (P), at 0.08. Conversely, the path from engagement (E) to happiness (W) showed the smallest difference in absolute value of the path coefficient, at -0.34. From this, it can be inferred that happiness (W), which has the strongest direct relationship to presenteeism (P), should be improved, and that to do so, it would be effective to first improve stress (S), which has the strongest direct relationship to happiness (W).

[0112] The second questionnaire request for each user in a specific organization after the implementation of the policy may be actually sent, or alternatively, expected questionnaire responses may be predicted and modified based on certain rules to the questionnaire responses of each user in the first questionnaire. In this way, by generating the questionnaire responses of each user in the second questionnaire after the implementation of the policy while anticipating the effects of the policy, it is possible to easily simulate the effectiveness of the policy without actually requesting each user to respond to a questionnaire. For example, in the case of a direct policy related to health awareness, the second questionnaire responses for each specific user may be generated by assuming that the answers to the questions related to health awareness in the first questionnaire from each user are answered in a way that improves the scores or ranks of all or some of the options by one or two levels.

[0113] Furthermore, the second request for questionnaire responses from each user in the specific organization after such a measure or the generation of hypothetical responses is not limited to being made to all target users in the specific organization in the first measure. For example, the structural model may be generated or the influence calculated by limiting the responses to the questionnaire from each user in the specific organization to a certain group of users in the specific organization, based on the responses to the questionnaire from each user in the specific organization before the measure. Furthermore, the second request for questionnaire responses from each user in the specific organization or the generation of hypothetical responses may be made to this certain group of users. The accuracy with which the obtained structural model and influence can be applied to the original population may then be statistically estimated or tested.

[0114] As mentioned above, the method of assessing the effectiveness of a measure by comparing the response information of each user to the first questionnaire before the measure with the response information of each user to the second questionnaire after the measure, or the response information assumed to be such, can also be applied to the mere passage of time. In other words, the above comparison may be performed assuming that the implementation of the measure signifies the passage of a predetermined period of time without any particular measure being taken. This makes it possible to estimate whether or not any measure was taken during the period and its effects, as well as the effects of other factors such as environmental changes.

[0115] Next, we will explain how to determine the effectiveness of implementing measures using only Bayesian network analysis for a structural model like the one shown in Figure 11. The indicator data for each indicator - health consciousness (HS), physical health status (H), stress (S), engagement (E), happiness (W), and presenteeism (P) - is expressed as a five-level or four-level discretized rank of a, b, c, d, and e, or a, b, c, and d. A indicates the state in which the target indicator has improved the most, and the closer it is to e or d, the more the target indicator has deteriorated or worsened.

[0116] Here, for each indicator, the structural model generation unit 330 obtains the desired structural model shown in Fig. 11 through Bayesian network analysis using response information, etc., from the first questionnaire given to each user in the specific organization before the implementation of the specific measure. After the specific measure is implemented, responses are again requested from each user in the specific organization for the second time, or responses are generated based on hypotheses, and Bayesian network analysis is performed using this response information, etc.

[0117] The impact calculation unit 340 calculates the impact of each index obtained in this way before and after the measure as the conditional probability value of a through e or a through d. Figures 14(a), 14(b), and 14(c) show the conditional probability values before and after the measure for each index of presenteeism (P), engagement (E), and health consciousness (HS) as horizontal stacked bar graphs with a 100% value. As shown in these figures, the implementation of the measure has increased the conditional probability of rank a for presenteeism (P) from 7.5% to 11.8%, while the conditional probabilities of ranks c, d, and e have decreased.

[0118] Therefore, it can be seen that the measures are effective in reducing presenteeism. Furthermore, as a result of implementing the measures, the conditional probability of achieving an A rank in engagement (E) rose dramatically from 4.0% to 22.5%, while the conditional probability of achieving a C rank fell. Therefore, it can be seen that the measures are also extremely effective in improving engagement.

[0119] On the other hand, the implementation of the measures slightly increased the conditional probability of achieving an A rank in health consciousness (HS) from 21.6% to 21.9%, and the conditional probability of achieving a B rank from 29.6% to 29.9%, but these were barely noticeable differences, and the same was true for C and D ranks. Therefore, it can be seen that the measures are effective in improving presenteeism and engagement, but have little impact on improving health consciousness.

[0120] In this way, by viewing the change in the impact on the structural model before and after the implementation of a measure as a change in the conditional probability value using Bayesian network analysis, it is possible to clearly visualize which indicators, including presenteeism, the measure will have an effect on and which will not.By changing the measure in various ways and observing the change in conditional probability, it is possible to estimate the effectiveness of the measure and the impact on effective indicators, and ultimately to objectively determine the direction of effective measures to improve presenteeism by comparing quantitative impacts.

[0121] In summary, data acquired from a user at a first time is defined as data corresponding to the first time, and data acquired from a user at a second time is defined as data corresponding to the second time. Between the first time and the second time, a measure to be verified for effectiveness may be implemented, or no particular measure may be implemented, and only a predetermined amount of time may have passed between the first time and the second time. Here, the index data calculation unit 320 calculates first index data and second index data, which are index data associated with each of the indices of presenteeism, health awareness, stress, engagement, and happiness, based on the data corresponding to the first time and the data corresponding to the second time, respectively.

[0122] The impact calculation unit 340 calculates the difference between the first index data and the second index data for each index or the rate of change between them. This makes it possible to quantitatively grasp the degree of improvement in presenteeism and related indexes before and after the implementation of a measure, and objectively verify the effectiveness of the measure.

[0123] On the other hand, data acquired from a user when a first measure and a second measure are implemented as mutually different measures is defined as data corresponding to the first measure and data corresponding to the second measure, respectively. In this case, the timing of the implementation of the first measure and the second measure is not taken into consideration, and both may be implemented simultaneously or one may be implemented first. Here, the index data calculation unit 320 may calculate first index data and second index data, which are index data associated with each of the indices of presenteeism, health awareness, stress, engagement, and happiness, respectively, according to the data corresponding to the first measure and the data corresponding to the second measure.

[0124] In this case, the influence calculation unit 340 also calculates the difference or the mutual change rate between the first index data and the second index data of each index. Furthermore, in this case, the target users for implementing the first and second measures may be different from each other. For example, a specific organization is divided into Group A and Group B, and measures 1 and 2 are implemented for users in each group. Then, based on the index data calculated from the data of users in Group A corresponding to measure 1 and the index data calculated from the data of users in Group B corresponding to measure 2, the difference or the mutual change rate between the two can be calculated. As a result, the effects of measures 1 and 2 can be estimated or tested for the entire specific organization, which is the population.

[0125] In the above, it is also possible to assume that no specific measures are implemented for either or both of Measure 1 and Measure 2. That is, in the above example, it is also possible to calculate the difference in the index data or the mutual change rate between Measure 1 and Measure 2 without implementing any measures. In this way, if only Measure 1 or Measure 2 is implemented, the effectiveness of the implemented measure can be verified, and if neither is implemented, it is possible to analyze potential factors that cause differences in index data between the two groups. Furthermore, Measure 1 and Measure 2 may be the same measure. In this way, it is possible to analyze potential factors that cause differences in index data between the two groups as a result of implementing the same measure.

[0126] When the influence of each indicator on a structural model is evaluated using conditional probability values obtained by Bayesian network analysis, the influence calculation unit 340 can indicate the possibility of an interaction or confounding factor between the indicators, a feature of Bayesian network analysis. An interaction refers to a synergistic effect that only appears when two factors are combined. For example, when only one of two causal indicators acts, the resulting increase in the indicator is defined as r, and when only the other indicator acts, the resulting increase in the indicator is defined as s. In this case, an interaction means that when both indicators act, the resulting increase in the indicator is greater or smaller than r + s.

[0127] On the other hand, let t be the causal indicator, u be the resulting indicator, and let v be another factor. In this case, if indicator v is related to indicator t, also affects indicator u, and is not an intermediate factor between indicators t and u, then indicator v is said to be a confounding factor. Even if such interactions or confounding factors are included in the relationships between indicators, Bayesian network analysis can indicate the possibility of their existence under certain conditions, thereby reducing the risk of overlooking the existence of synergistic effects between indicators and reducing the risk of erroneously estimating causal relationships between indicators.

[0128] Health management support system 10a may further clarify changes in the presenteeism index and changes in the impact of each index when each of multiple possible measures is implemented. The system may also have a function for comparing these to indicate which of the multiple possible measures is estimated to be the most effective. For example, impact table 342 shown in FIG. 15 ranks the multiple presented measures in descending order of the degree of improvement in the presenteeism index and displays the degree of change (improvement or decrease) in the impact of each measure.

[0129] The impact calculation unit 340 calculates the difference in the presenteeism (P) index data before and after the implementation of each measure, and extracts the measure with the largest difference, i.e., the measure with the highest degree of improvement in presenteeism after the measure, as the highest priority. For example, "Measure 5" listed in the "Measures" column is extracted because the difference in the index data displayed in the "(P) Difference in Improved Index Data" column is "284." In addition, the indicators that are direct causes of presenteeism are displayed in order of the largest and smallest change before and after each measure. Note that a similar display may be made not only for indicators that are direct causes of presenteeism, but also for indicators that are indirect causes.

[0130] For example, when Measure 5 was implemented, Engagement (E), ranked first, improved by a difference of 0.34 in its impact indicator value, and Stress (S), ranked second, improved by a difference of 0.25 in its impact indicator value. The impact indicator value can be expressed as, for example, a conditional probability value or the absolute value of a path coefficient. Furthermore, Health Awareness (HS), ranked first in the worst ranking, decreased by a difference of -0.14 in its impact indicator value. The same is true for Measure 2 and Measure 7, which were extracted as the second and third priorities.

[0131] In this way, when multiple candidate measures are considered, the impact calculation unit 340 can rank the candidate measures in order of their effectiveness on presenteeism indicators by implementing or simulating each measure. In addition, at the same time, it can also grasp changes in the impact (improvement or decline) of other indicators related to presenteeism. This makes it easier to determine which measures are useful for improving presenteeism and which indicators that affect presenteeism improvement should be prioritized for improvement.

[0132] The indicators of presenteeism and psychological or physical health conditions that may affect presenteeism exemplified in the embodiments and modifications of the present disclosure are merely examples, and include all other names and categories that may serve as indicators with a similar purpose, regardless of the exemplified names and categories. Furthermore, while Bayesian network analysis and covariance structure analysis are exemplified for calculating the structural model and the influence, these are merely preferred examples, and other known analytical methods such as Markov network analysis, decision tree analysis, and Kalman filter analysis may also be used.

[0133] In addition, the control processing and other processing by an information processing device such as a management server used in a health management support system relating to an embodiment of the present disclosure or its modified example, as described above, can also be achieved by the information processing device reading and executing the code of a computer program, such as a processing program, that constitutes software for realizing each function of the embodiment, etc.

[0134] Therefore, the computer program code itself and a storage medium on which the computer program code is recorded also constitute the present disclosure. Examples of storage media on which the computer program code is recorded include, but are not limited to, hard disks, USB memory, SD (registered trademark) cards, CD-ROMs, CD-Rs, IC cards, DVD-ROMs, and DVD-Rs. [Explanation of symbols]

[0135] 1, 1a Management Server 2. Administrator terminal 3. User terminal 4. Store terminals 10, 10a Health Management Support System 11 Control section 12 Main memory 13, 13a Auxiliary storage section 14 Input section 15 Output section 16 Interface section 17 Bus 20. Communication Networks 110 Health awareness question information DB 120 Stress Question Information DB 130 Engagement Question Information DB 140 Happiness Question Information DB 150 Presenteeism Question Information DB 160 Health checkup results DB 170 Lifestyle question information DB 210 Questionnaire Form DB 211 Questionnaire 220 Answer information DB 310 Questionnaire Creation Department 320 Index data calculation unit 330 Structural Model Generation Unit 340 Impact calculation part 341 Impact Table 342 Impact Table E601, E602, E603, E604, E605, E606, E607, E608, E609, E610, E611, E621, E622, E623, E624, E625, E626, E627, E628, E631, E632, E641, E642, E651, E652 Edge N501, N502, N503, N504, N505, N506, N521, N522, N523, N524, N525, N531, N541, N551, N552, N553 nodes

Claims

1. an index data calculation unit that calculates index data associated with each of the indices, which are presenteeism, health awareness, stress, engagement, and happiness, by quantifying or discretizing each of the indices according to data acquired from the user; a structural model generation unit that outputs a structural model indicating a relationship between each of the indexes of health awareness, stress, engagement, and happiness and the index of presenteeism based on the index data; The data includes information about the user's responses to a questionnaire that includes questions about psychological health and answer options for the questions; A health management support system in which the index data is calculated by assigning a predetermined score to each answer option for each question and adding up the scores of the answers to all of the questions to calculate a total score for the target index, or by assigning a predetermined ranking to each answer option for each question and adding up the numbers of each rank for the answers to all of the questions to calculate the proportion of each rank for the target index.

2. an influence calculation unit that calculates an influence of each of the indices of the health awareness, the stress, the engagement, and the happiness level on the index of presenteeism, 2. The health management support system of claim 1, wherein when the structural model generation unit generates a structural model showing the relationship between each of the indicators by Bayesian network analysis, the influence calculation unit calculates the value of a posterior probability value reflecting the conditional probability value or observed value between each of the indicators as the influence of each of the indicators, and when the structural model generation unit generates a structural model showing the relationship between each of the indicators by covariance structure analysis, the influence calculation unit calculates the value of a path coefficient between each of the indicators as the influence of each of the indicators.

3. The health management support system according to claim 2 , wherein the influence calculation unit further calculates a degree of mutual influence between the indices of health awareness, stress, engagement, and happiness.

4. When the data acquired from the user when a first measure and a second measure, which are mutually different measures, are implemented, are defined as data corresponding to the first measure and data corresponding to the second measure, respectively, the index data calculation unit calculates first index data and second index data, which are index data associated with each of the indexes of presenteeism, health awareness, stress, engagement, and happiness, in accordance with the data corresponding to the first measure and the data corresponding to the second measure, respectively; The health management support system according to claim 2 or 3, wherein the influence calculation unit calculates a difference or a mutual change rate between the first index data and the second index data for each of the indices.

5. The health management support system according to claim 4 , wherein the first measure and the second measure are implemented for different users.

6. When data acquired from the user at a first time is defined as data corresponding to a first time, and data acquired from the user at a second time is defined as data corresponding to a second time, the index data calculation unit calculates first index data and second index data, which are index data associated with each of the indexes of presenteeism, health awareness, stress, engagement, and happiness, according to the data corresponding to the first time period and the data corresponding to the second time period, respectively; The health management support system according to claim 2 or 3, wherein the influence calculation unit calculates a difference or a mutual change rate between the first index data and the second index data for each of the indices.

7. The health management support system according to claim 6 , wherein a measure to be the subject of effectiveness verification is implemented between the first time and the second time.

8. The data further includes information on the user's responses to a questionnaire containing questions about lifestyle habits obtained from the user and answer options for the questions, the structural model generation unit outputs a structural model indicating a relationship between each of the indices of the health awareness, the stress, the engagement, the happiness level, and the physical health condition and the index of presenteeism, based on index data obtained by quantifying or discretizing the index calculated by the index data calculation unit and the index of data based on the user's response information regarding lifestyle habits obtained from the user; 8. The health management support system according to claim 2, wherein the influence calculation unit calculates the influence of each of the indicators of health awareness, stress, engagement, happiness, and physical health status on the indicator of presenteeism.

9. A computer program that causes a computer to execute processes that support health management hand, Calculating index data associated with each of the indicators of presenteeism, health awareness, stress, engagement, and happiness by quantifying or discretizing the indicators according to the data acquired from the user; executes a process of outputting a structural model showing a relationship between each of the indexes of health awareness, stress, engagement, and happiness and the index of presenteeism based on the index data; The data includes information about the user's responses to a questionnaire that includes questions about psychological health and answer options for the questions; The index data is a computer program in which a predetermined score is assigned to each of the answer options for each of the questions, and the scores of the answers to all of the questions are added together to calculate a total score for the target index, or a computer program in which a predetermined ranking is assigned to each of the answer options for each of the questions, and the proportion of each rank for the target index is calculated by adding together the numbers of each rank for the answers to all of the questions.

10. The computer program according to claim 9 , further comprising the step of calculating an influence of each of the indices of health awareness, stress, engagement, and happiness on the index of presenteeism.

Images