Data traffic volume analysis method and data traffic volume analysis system

Abstract

To make it possible to estimate a user's morbidity risk with a high degree of accuracy from information of data communication used by the user using a data communication volume analysis system.SOLUTION: A data communication volume analysis system includes: communication volume measuring means that measures the data communication volume of an information terminal used by a user; information processing means that processes the data communication volume of the information terminal measured by the communication volume measuring means by separating it into information of data communication volume received by the user and information of data communication volume transmitted by the user, and estimates the user's disease risk; and an output screen that displays information on the disease risk estimated by the information processing means.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a data traffic analysis method and a data traffic analysis system.

Background Art

[0002] The first stage of dementia diagnosis is carried out at a clinic outpatient department or the like. Naturally, the trigger for the visit depends on the awareness of the patient or family about the symptoms.

[0003] Initial symptoms are often non-specific and difficult to notice. Even if noticed, the timing of the first visit is generally delayed due to resistance to dementia tests. In fact, many patients visit a specialist after the symptoms have progressed.

[0004] If dementia can be detected early, appropriate intervention such as treatment drugs to suppress the progression of dementia and control of risk factors (diabetes, hypertension, lack of exercise, etc.) becomes possible, and prevention of progression can be achieved.

[0005] According to a report by the International Alzheimer's Disease Association, the proportion of dementia that is missed without being diagnosed reaches about 50% in developed countries and over 90% in developing countries.

[0006] The biggest reason for dementia to be missed without being diagnosed is that there is no simple test that can be used for screening. Screening at a stage even before the first stage (primary care physician examination) is important, and the development of a dementia test method with high objectivity in a short time is required.

[0007] As a technology related to screening, Patent Document 1 describes determining a person's behavior information by detection / measurement data such as traffic data and temperature data, identification information, and a learned behavior identification model constructed for each person.

Prior Art Documents

Patent Documents

[0008] [Patent Document 1] Japanese Patent Publication No. 2019-207604 [Overview of the project] [Problems that the invention aims to solve]

[0009] While various methods have been researched and commercialized as dementia screening tests, many require users to perform some kind of task before taking the test. However, the challenge lies in the fact that many users do not voluntarily take the test because it is time-consuming to perform the task.

[0010] To address the above challenges, research is being conducted on a new type of dementia screening test that utilizes data naturally collected in daily life.

[0011] However, many of these solutions were not suitable for a large number of users, as they were only applicable to a limited number of drivers or required the installation of sensors inside the home.

[0012] Patent Document 1 describes a method for determining a person's behavioral information using a pre-trained behavioral identification model, but it does not describe a specific means for determining behavioral information using only communication volume.

[0013] The present invention solves the problems of the prior art described above and provides a data traffic analysis method and a data traffic analysis system that enable the estimation of a user's disease risk with high accuracy from data communication information used by the user. [Means for solving the problem]

[0014] To solve the above-mentioned problems, the present invention provides a data traffic analysis system comprising: a traffic traffic measurement means for measuring the amount of data traffic from an information terminal used by a user; an information processing means for processing the data traffic from the information terminal measured by the traffic traffic measurement means, separating it into information on the amount of data traffic received by the user and information on the amount of data traffic transmitted by the user, and estimating the user's disease risk; and an output screen for displaying the disease risk information estimated by the information processing means. The information processing means includes an action estimation unit that identifies the type of action the user takes using the information terminal based on information about the amount of data communication received by the user and information about the amount of data communication sent by the user, and determines the activity level of the identified action; and a disease risk estimation unit that determines the user's disease probability based on data of the time-series change in the duration of the action performed by the action estimation unit and the user's disease probability based on data of the time-series change in the activity level of the action, and estimates the largest disease probability among the determined probabilities as the user's disease risk. It was constructed as follows.

[0015] Furthermore, in order to solve the above-mentioned problems, the present invention provides: The data traffic analysis system will be executed. In a data traffic analysis method, The means of measuring communication volume is Data usage of the information terminal used by the user Measure death, Information processing means, The data traffic volume of the information terminal measured by this communication volume measurement device is divided into information on the data traffic volume received by the user and information on the data traffic volume transmitted by the user. Process separately By doing so, the user's disease risk is estimated. The output screen is, Disease risk estimated by information processing means The system displays information, and the information processing unit's behavior estimation unit identifies the type of behavior the user engages in using the information terminal based on the data volume information received and the data volume information transmitted by the user. It then determines the activity level of the identified behavior. The information processing unit's disease risk estimation unit then determines the user's disease probability based on the time-series data of the duration of the behavior identified by the behavior estimation unit and the user's disease probability based on the time-series data of the activity level of the behavior. The highest of the determined disease probabilities is then estimated as the user's disease risk. I made it so. [Effects of the Invention]

[0016] According to the present invention, it has become possible to estimate a user's risk of contracting a disease with high accuracy by utilizing data communication volume information used by the user. [Brief explanation of the drawing]

[0017] [Figure 1] This is a block diagram showing the configuration of the data traffic analysis system according to Example 1. [Figure 2] This is a block diagram showing the schematic configuration of the information processing device according to Example 1. [Figure 3] This is a block diagram showing the schematic configuration of the communication volume measurement device according to Example 1. [Figure 4]A diagram showing an example of communication information of a Wi-Fi (registered trademark) router measured by the traffic measurement unit of a traffic measurement device. (a) shows time-series data of the data traffic related to the upstream communication transmitted externally via the Wi-Fi router, and (b) shows an example of time-series data of the data traffic related to the downstream communication coming in from the outside via the Wi-Fi router. [Figure 5] A table showing a list of traffic feature amounts extracted by the traffic feature amount extraction unit of the information processing device according to Example 1. [Figure 6] A table showing a list of traffic feature amounts representing the action activity level obtained by the action estimation unit of the information processing device according to Example 1. [Figure 7] A table showing the types of actions estimated by the action estimation unit of the information processing device according to Example 1 and the activity level of those actions, associated with the time zone of the user's actions. [Figure 8] A graph showing the changes in the average daily execution time and action activity level of the user obtained by the change-over-time calculation unit of the information processing device according to Example 1 on a monthly basis. In (a), the change-over-time of the daily action execution time when the action type information is video viewing 810 is shown as a bar graph, and the change-over-time of the action activity level is shown as a line graph. In (b), the change-over-time of the daily action execution time when the action type information is an online call is shown as a bar graph, and the change-over-time of the action activity level is shown as a line graph. [Figure 9] A table summarizing the action execution time reduction rate and action activity level reduction rate per month as change-over-time feature amount data calculated by the change-over-time feature amount calculation unit of the information processing device according to Example 1 for each action type. [Figure 10] Showing the risk of disease probability for each action type estimated based on rules by the disease risk estimation unit of the information processing device according to Example 1. (a) is a table showing the relationship between the action execution time reduction rate and the risk of disease probability for each action type, and (b) is a table showing the relationship between the action activity level reduction rate and the risk of disease probability for each action type. [Figure 11] A flowchart showing the overall flow of the processing of the action estimation method using the data traffic according to Example 1. [Figure 12]This flowchart shows the detailed processing flow of the step for estimating user behavior in the behavior estimation method using data communication volume according to Example 1. [Figure 13] This flowchart shows the detailed processing flow of the step for determining the changes in user behavior over time in the behavior estimation method using data communication volume according to Example 1. [Figure 14] This flowchart shows the detailed processing flow of the step for estimating disease risk in the behavior estimation method using data communication volume according to Example 1. [Figure 15] In the behavior estimation method using data traffic volume according to Example 1, the initial setup step shows a screen for selecting the type of behavior to be estimated, which is displayed on the display screen of the display device of the traffic volume measuring device. [Figure 16] In the behavior estimation method using data traffic volume according to Example 1, the initial setup step shows a screen for selecting a method for calculating behavioral activity level, which is displayed on the display screen of the data traffic measurement device. [Figure 17] This screen shows the amount of data communication received by the information processing device according to Example 1, divided into uplink communication volume and downlink communication volume. [Figure 18] This screen shows the type of user behavior and the activity level of that behavior estimated by the behavior estimation unit of the information processing device according to Example 1, in relation to the time of day the user's behavior occurred. [Figure 19] The image shows a screen displaying the time-series feature data calculated by the time-series feature calculation unit of the information processing device according to Example 1, showing the monthly rate of decrease in activity duration and the rate of decrease in activity level for each type of activity. [Figure 20] This is a screen displaying the disease risk estimation results as a result of processing with the information processing device according to Example 1. [Figure 21] This is a block diagram showing the configuration of the data traffic analysis system according to Example 2. [Modes for carrying out the invention]

[0018] This invention relates to a data traffic analysis method and a data traffic analysis system that enable the estimation of a user's disease risk with high accuracy from data traffic information naturally obtained in daily life. The diseases to be estimated include dementia, mental illness, depression, and frailty.

[0019] Furthermore, the present invention relates to estimating disease risk from a user's behavioral history using a data communication volume analysis system comprising: an input device that receives the amount of data transmitted from a communication terminal used by a user as input; a computing device that predicts the type of user behavior using the terminal based on the correlation between the amount of data transmitted on the uplink and the amount of data transmitted on the downlink; and an output device that outputs the predicted type of behavior.

[0020] Embodiments of the present invention will be described in detail below with reference to the drawings. In all the drawings used to illustrate these embodiments, components having the same function will be denoted by the same reference numerals, and repeated explanations will be omitted in principle.

[0021] However, the present invention is not to be construed as being limited to the embodiments described below. It will be readily apparent to those skilled in the art that the specific configuration can be modified without departing from the spirit or purpose of the present invention. [Examples]

[0022] Figure 1 shows the configuration of the data traffic analysis system 10 according to Example 1.

[0023] This embodiment describes how to estimate the risk of developing dementia for users of the home-based system 100 shown in Figure 1.

[0024] The data communication volume analysis system 10 according to Example 1 has a configuration in which a home system 100 and an information processing device 200 are connected via a network 150.

[0025] The home system 100 consists of a personal computer 101, a television 102, a mobile terminal 103, IoT home appliances 104 used by user 1 within the home, a WiFi router 105 to which these are connected and which communicates with the outside, and a communication volume measurement device (terminal device) 300 that monitors the communication volume of the WiFi router 105.

[0026] The information processing device 200 receives and processes communication volume information via the network 150 regarding the uplink communication volume transmitted to the outside and the downlink communication volume received from the outside, which is monitored by the communication volume measuring device 300 via the WiFi router 105. The processing results are then output to the communication volume measuring device 300 via the network 150.

[0027] Figure 2 shows the configuration of the information processing device 200. The information processing device 200 includes a communication interface 201 for communicating with the communication volume measuring device 300 via the network 150, an output interface 202 for outputting to the outside, an input interface 203 for receiving input signals from the outside, a memory 204 for storing control programs and the like, a data processing unit 210 for processing signals sent from the communication volume measuring device 300 via the communication interface 201, and a storage unit 220 for storing signals sent from the communication volume measuring device 300 via the communication interface 201.

[0028] The data processing unit 210 includes a user information management unit 211, a communication volume feature extraction unit 212, an action estimation unit 213, a time-series change calculation unit 214, a time-series change feature calculation unit 215, a disease risk estimation unit 216, and a result processing unit 217.

[0029] The memory unit 220 includes a user information storage area 221 that stores information such as user 1's personal attribute values ​​2211, usage history information 2212, and user setting information 2213; a communication volume data storage area 222 that stores information such as uplink communication volume 2221 and downlink communication volume 2222; a communication volume feature data storage area 223 that stores information on communication volume feature 2231; an action estimation data storage area 224 that stores information such as action type 2241 and action activity level 2242; a time-series change data storage area 225 that stores information such as time-series changes in action type 2251 and time-series changes in action activity level 2252; a time-series change feature data storage area 226 that stores information such as time-series change feature 2261 and time-series change feature 2262 for action type; and a disease risk data storage area 227 that stores information such as disease risk (probability of occurrence or severity) 2271.

[0030] Communication information from the WiFi router 105, sent from the communication volume measuring device 300 via the network 150, is input to the information processing device 200 via the communication interface 201, and is classified by the user information management unit 211 into personal attribute values ​​2211 of user 1, usage history information 2212, user setting information 2213, etc., and stored in the user information storage area 221.

[0031] Furthermore, information about user 1 entered via the input interface 203, for example from a keyboard or input screen (not shown), is also input to the information processing device 200 and classified by the user information management unit 211 into personal attribute values ​​2211, usage history information 2212, user setting information 2213, etc., and stored in the user information storage area 221.

[0032] Furthermore, the user information management unit 211 classifies the communication information from the WiFi router 105 sent from the communication volume measuring device 300 into uplink communication volume 2221 and downlink communication volume 2222, and stores them in the communication volume data storage area 222.

[0033] The communication volume feature extraction unit 212 extracts communication volume features 2231 using the personal attribute values ​​2211, usage history information 2212, and user setting information 2213 of user 1 stored in the user information storage area 221 of the storage unit 220, as well as the uplink communication volume 2221 and downlink communication volume 2222 information stored in the communication volume data storage area 222, and stores them in the communication volume feature data storage area 223 of the storage unit 220.

[0034] The behavior estimation unit 213 uses the information on uplink traffic volume 2221 and downlink traffic volume 2222 stored in the communication volume feature data storage area 223 of the memory unit 220 to determine the behavior type 2241 and behavior activity level 2242, and stores this information in the behavior estimation data storage area 224.

[0035] The time-series change calculation unit 214 uses information such as the type of behavior 2241 and the activity level 2242 stored in the activity estimation data storage area 224 of the memory unit 220 to obtain the characteristic quantities 2261 of the time-series change in the type of behavior and the characteristic quantities 2262 of the time-series change in the activity level, and stores this information in the time-series change data storage area 225.

[0036] The time-series feature calculation unit 215 uses information such as the time-series feature quantities 2261 for behavior types and 2262 for behavior activity levels stored in the time-series data storage area 225 of the storage unit 220 to calculate the time-series feature quantities 2261 for behavior types and 2262 for behavior activity levels, and stores this information in the time-series feature quantity data storage area 226.

[0037] The disease risk estimation unit 216 uses information such as the time-dependent feature quantities 2261 for behavioral types and the time-dependent feature quantities 2262 for behavioral activity levels, which are stored in the time-dependent feature quantity data storage area 226 of the memory unit 220, to determine the disease risk 2271, and stores this information in the disease risk data storage area 227.

[0038] The results processing unit 217 transmits information such as behavior type 2241 and behavior activity level 2242 stored in the behavior estimation data storage area 224, information such as the feature quantities 2261 of the time-dependent change in behavior type and the feature quantities 2262 of the time-dependent change in behavior activity level stored in the time-dependent change data storage area 225, information such as the feature quantities 2261 of the time-dependent change in behavior type and the feature quantities 2262 of the time-dependent change in behavior activity level stored in the time-dependent change feature quantity data storage area 226, and information about disease risk 2271 stored in the disease risk data storage area 227 to the communication volume measuring device 300 via the network 150 from the communication interface 201.

[0039] Figure 3 shows the configuration of the communication volume measurement device 300. The communication volume measurement device 300 includes a communication interface 310 that connects to the network 150, a processor 320 that processes data, a data storage unit 330 that stores data, a memory 340 that stores the processing program of the communication volume measurement device 300, and a display device 350 that displays the processing results sent from the information processing device 200.

[0040] The data storage unit 330 stores data such as communication volume data 3301, behavior type data 3302, behavioral activity level data 3303, and disease risk data 3304 as processing results sent from the information processing device 200.

[0041] The processor 320 includes a result presentation unit 321 and a measurement unit 322 internally, and the display device 350 includes a display screen 351 and a touch sensor 352.

[0042] Based on the information specified by the touch sensor 352 of the display device 350, the processor 320 selects one of the data stored in the data storage unit 330, which is set in the result presentation unit 321, from the data containing communication volume data 3211, behavior type data 3212, activity level data 3213, or disease risk data 3214, and displays it on the display screen 351 to present to the user 1.

[0043] The measurement unit 322 includes a communication volume measurement unit 323, which measures communication information sent from the WiFi router 105 via the network 150, and sends this measured information from the communication interface 310 to the information processing device 200 via the network 150.

[0044] Next, a specific example of data processing in the data processing unit 210 of the information processing device 200 will be described.

[0045] Figure 4 shows an example of communication information of the WiFi router 105 measured by the communication volume measurement unit 323 of the communication volume measurement device 300. In Figure 4, (a) shows an example of time-series data of data communication volume (line communication volume) 411, 421, 431, and 441 related to uplink communication transmitted to the outside via the WiFi router 105, and (b) shows an example of time-series data of data communication volume (line communication volume) 412, 422, and 432 related to downlink communication coming in from the outside via the WiFi router 105.

[0046] The user information management unit 211 of the information processing device 200 receives communication information from the WiFi router 105 as shown in Figure 4, measured by the communication volume measurement unit 323 of the communication volume measurement device 300, measures time-series data of the communication volume for the uplink communication volume 2221 and downlink communication volume 2222 at unit time intervals (for example, every 0.1 seconds, every 1 second, every 1 minute, etc.), associates the timestamps of the uplink and downlink data, and stores them in the communication volume data storage area 222.

[0047] Next, the communication volume feature extraction unit 212 extracts communication volume features from the time-series data of uplink communication volume 2221 and downlink communication volume 2222 stored in the communication volume data storage area 222.

[0048] When the time-series data of uplink traffic 2221 and downlink traffic 2222 are combined, the pattern differs depending on the type of user 1's behavior. The combination of (a) uplink traffic 411 and (b) downlink traffic 412 shown in Figure 4 corresponds to the time-series data (video viewing time-series data 410) when user 1 is watching a video on the PC 101 or TV 102. It is characterized by a higher communication frequency and data volume for the downlink traffic 412 in (b) compared to the uplink traffic 411 in (a).

[0049] The combination of (a) uplink traffic volume 421 and (b) downlink traffic volume 422 corresponds to the time-series data (online call time-series data 420) when user 1 is making an online call on PC 101. The uplink traffic volume 421 and (b) downlink traffic volume 422 are characterized by having almost the same communication frequency and communication data volume.

[0050] The combination of (a) uplink traffic volume 431 and (b) downlink traffic volume 432 corresponds to time-series data (email usage time-series data 430) when user 1 is using email on a personal computer 101 or mobile terminal 103. The downlink traffic volume 412 in (b) is characterized by a large amount of communication data and a low communication frequency compared to the uplink traffic volume 411 in (a).

[0051] The combination of (a) uplink traffic of 441 and (b) downlink traffic of zero corresponds to the time-series data (IoT appliance usage time-series data 440) when user 1 is using the IoT appliance 104. It is characterized by having only uplink traffic, which is relatively small in amount of data, and zero downlink traffic.

[0052] On the other hand, if there is no data traffic for a long period (for example, 3 hours) or if the same pattern continues, the system will assume that User 1 is not using the service and will exclude that time period.

[0053] Since the patterns of uplink traffic 431 and downlink traffic 432, as well as the patterns of combinations thereof, have characteristics depending on the type of user 1's actions, the traffic traffic feature extraction unit 212 extracts traffic traffic features, as shown in Table 500 in Figure 5, from the time-series data of data traffic, as shown in Figure 4, in order to classify the type of user 1's actions.

[0054] Table 500 lists the features of the extracted data traffic in a table format for each feature. Feature No. 1 is the traffic volume per unit time for each uplink and downlink connection (501), Feature No. 2 is the duration of communication for each uplink and downlink connection (502), Feature No. 3 is the switching frequency between uplink and downlink communication (503), Feature No. 4 is the ratio of uplink to downlink traffic (504), and Feature No. 5 is the causal relationship between uplink and downlink communication (505). For Feature No. 5, the causal relationship between uplink and downlink communication (505), it is conceivable to use indicators that represent the correlation or causal relationship between two time series data, such as cross-correlation function, coherence function, moving entropy, and mutual information.

[0055] The communication volume feature extraction unit 212 extracts one or more of the following data communication volume features from the time-series data of data communication volume as shown in Figure 4: communication volume 501, communication duration 502, communication switching frequency 503, uplink to downlink ratio 504, and uplink to downlink causal relationship 505 (hereinafter referred to as data communication volume features 501 to 505), and stores them as communication volume feature 2231 in the communication volume feature data storage area 223 of the storage unit 220.

[0056] The data traffic volume features 501-505 shown in Table 500 of Figure 5 are just examples, and other features (for example, the rate of change of traffic volume per unit time during a continuous communication period, the difference between the maximum and minimum traffic volume during a continuous communication period, the variation in the time it takes to switch from the downlink to the uplink, etc.) may be added, or some of the data traffic volume features 501-505 may be deleted.

[0057] The behavior estimation unit 213 obtains information on the type of user 1's behavior 2241 and the activity level 2242 of that behavior based on the communication volume features 2231 extracted by the communication volume feature extraction unit 212 and stored in the communication volume feature data storage area 223, and stores this information in the behavior estimation data storage area 224.

[0058] Table 600 in Figure 6 shows a list of data communication volume features corresponding to each of User 1's actions, which are used by the action estimation unit 213 to calculate the activity level of each of User 1's actions. From among the multiple items shown in Table 600 in Figure 6, the features suitable for evaluating the activity level of User 1's actions are selected.

[0059] Table 600 shows several indicators representing the activity level of uplink communication, which signifies proactive communication from user 1. Feature No. 1 is the uplink communication volume per unit time (601), Feature No. 2 is the duration of uplink communication (602), Feature No. 3 is the switching frequency of uplink communication (603), and Feature No. 4 is the ratio of uplink to downlink communication volume (604). Additional features may be added from Feature No. 5 onwards.

[0060] Multiple traffic volume features are calculated for each predetermined time frame (e.g., 5 minutes, 10 minutes, 30 minutes), and the type of behavior is estimated from this data, as shown in Figure 7. There are two methods for estimating the type of behavior: a rule-based approach and an AI model.

[0061] In the rule-based method, for example, the following rules are predetermined for the extracted communication volume features shown in Table 500 of Figure 5. (1) If the amount of data transmitted per unit time (download) exceeds a predetermined value, it will be determined that video is being watched. (2) If the frequency of switching between communications (uplink / downlink) exceeds a predetermined number of times per unit of time, it will be determined to be an online call.

[0062] The method for building AI models involves performing discriminant analysis on a multi-class classification problem using a feature dataset and behavior type labels pre-assigned by humans as training data.

[0063] Multi-class classification frequently employs a method that combines two-class classifiers. For example, there is a method called Error Correcting Output Codes (ECOC). Existing methods such as Support Vector Machines (SVM) and linear discriminant analysis are used for the two-class classifiers.

[0064] There are several known methods for determining the two classes in a two-class classification. For example, if the number of classes is K, then in a one-to-one two-class classification method, K This would involve using C2 two-class discriminants. Alternatively, a method for one-to-other two-class discrimination would use K two-class discriminants.

[0065] In multi-class classification, the class with the smallest Hamming scale (the number of classifiers that produce different classification results for the class of interest) is selected as the multi-class classification result from the classification results of these two-class classifiers.

[0066] Next, for each estimated type of behavior, the activity level is calculated based on a method selected from Table 600 shown in Figure 6.

[0067] For the uplink traffic volume per unit time 601 of feature No. 1 in Table 600, an activity coefficient is set as an indicator of the activity level of user 1. For example, level 1 is set when the uplink traffic volume per hour is 0 to 2 Mbit, and the level is increased by 1 for every 1 Mbit increase thereafter, with level 10 being 10 Mbit or more.

[0068] Regarding feature No. 2, the duration of uplink communication (602), an indicator of the activity level of user 1's actions can be set as follows: for example, if the average duration of uplink communication is 3 seconds or less, it is set to level 1, and the level is increased by 1 for every 3 seconds thereafter, with level 10 being set for 27 seconds or more.

[0069] Regarding the uplink communication switching frequency 603 of feature No. 3, as an indicator of the activity level of user 1, for example, if the uplink communication switching frequency per hour is 2 times or less, it is set to level 1, and thereafter the level is increased by 1 for every 2 times, with level 10 being set for 19 times or more.

[0070] Regarding the uplink / downlink traffic ratio 604 for feature No. 4, as an indicator of user activity, for example, if the ratio of uplink traffic to the total of uplink and downlink traffic is between 0% and 10%, it is set as Level 1, and the level is increased by 1 in 10% increments thereafter, with Level 10 being set when it is 90% or more.

[0071] The definitions of the activity coefficients for each of the following features described above—uplink traffic volume per unit time (601), uplink communication duration (602), uplink communication switching frequency (603), and uplink-to-downlink traffic ratio (604)—are merely examples, and each activity coefficient may be defined using a different method.

[0072] The behavior estimation unit 213 uses the estimated user 1 behavior type 2241 and the behavior activity level 2242 information to create data that associates time zone 701 with behavior type 702 and behavior activity level 703, as shown in Table 700 in Figure 7. The behavior estimation unit stores the information of behavior type 2241 associated with time zone 701 and behavior type 702, and the information of behavior activity level 2242 associated with time zone 701 and behavior activity level 703, in the behavior estimation data storage area 224.

[0073] The time-series change calculation unit 214 uses the information on the type of behavior 2241 and the activity level 2242 stored in the activity estimation data storage area 224 to determine the average daily duration for each type of behavior. Since daily fluctuations are too large, it is preferable to look at time-series change data for behavior duration based on the average value over a certain period, such as weekly, monthly, quarterly, semi-annually, or yearly, as well as time-series change data for activity level corresponding to the average behavior duration.

[0074] Figure 8(a) shows the time-series changes in the monthly average daily activity duration 811 when activity type 2241 is video viewing 810, using bar graphs 812, and the time-series changes in activity level 813 corresponding to each bar graph 812 are shown using line graphs 814.

[0075] Furthermore, Figure 8(b) shows the time-series changes in the monthly average daily activity duration 821 when activity type 2241 is online phone calls 820, using bar graphs 822, and the time-series changes in activity level 823 corresponding to each bar graph 822 are shown using line graphs 824.

[0076] The data obtained in this way, such as the time-series change data of the monthly average value of the daily activity duration 811 in the case of video viewing 810 and the time-series change data of the monthly average value of the daily activity duration 821 in the case of online calls 820, are stored in the time-series change data storage area 225 as information on the time-series change of activity type 2251. The data on the time-series change of the monthly daily activity level 823 in the case of video viewing 810 and the time-series change data of the monthly average value of the daily activity level 823 in the case of online calls 820 are stored in the time-series change data storage area 225 as the time-series change of activity level 2252.

[0077] In addition to using the average daily activity duration over a certain period, as shown in Figure 8, as time-series data, information such as the number of times each activity type is performed in a day, or the number of different activity types performed in a day may also be used. Furthermore, an index that expresses the degree to which the above daily indicators vary over a certain period, such as a standard deviation, may also be used.

[0078] Similarly, for activity levels, in addition to the average value over a specified period, an indicator of variability over a certain period may also be used.

[0079] Next, the time-series feature calculation unit 215 calculates the monthly rate of decrease in activity time 902 and the rate of decrease in activity activity 903 for each activity type 901 from the time-series change 2251 information of activity type stored in the time-series change data storage area 225, as shown in Table 900 in Figure 9.

[0080] Specifically, the slope of the approximated line, obtained by linearly approximating the line connecting the peaks of the bar graph 812 representing the monthly average daily activity time 811 for video viewing 810 shown in Figure 8(a), is determined as the rate of decrease in activity time for video viewing 902. Similarly, for online calls 820 shown in Figure 8(b), the slope of the approximated line, obtained by linearly approximating the line connecting the peaks of the bar graph 822 representing the monthly average daily activity time 821, is determined as the rate of decrease in activity time for online calls 902.

[0081] Furthermore, from the time-series changes in activity level 2252 stored in the time-series data storage area 225, the rate of decline in activity level 903 for each type of activity 901 is determined, as shown in Table 900 in Figure 9. Specifically, the slope of the approximated line obtained by linearly approximating the graph of line 814 representing the monthly activity level 813 for video viewing 810 shown in Figure 8(a) is determined as the rate of decline in activity level 903 for video viewing. Similarly, for online calls 820 shown in Figure 8(b), the slope of the approximated line obtained by linearly approximating the graph of line 824 representing the monthly activity level 823 is determined as the rate of decline in activity level 903 for online calls.

[0082] Figure 9 shows an example of calculating the rate of decrease in activity duration 902 and the rate of decrease in activity level 903. However, if the number of activity sessions or the number of activity types are calculated in addition to the daily activity duration explained in Figure 8, the monthly rate of decrease and the rate of decrease in activity level may be calculated for these indicators.

[0083] The data obtained in this way, representing the rate of decrease in action duration 902 for each type of action 901, is stored in the time-series change feature data storage area 226 as the feature quantity 2261 for changes in the time-series change of the type of action, and the rate of decrease in action activity level 903 for each type of action 901 is stored in the time-series change feature data storage area 226 as the feature quantity 2262 for changes in action activity level.

[0084] The disease risk estimation unit 216 estimates the risk of illness for user 1 from the time-dependent feature quantities 2261 of behavioral type and 2262 of behavioral activity level, which are stored in the time-dependent feature quantity data storage area 226.

[0085] Disease risk assessment may involve estimating the probability of developing a disease and its severity. Methods for this estimation include rule-based approaches and AI-based modeling.

[0086] This method involves creating a correspondence table between the rate of decrease / decline and the probability of disease incidence and severity, and then determining the disease risk from that table. For example, the probability of disease incidence is calculated for each type of behavior from this correspondence table, and the result of determining the behavior with the highest probability of disease incidence is used.

[0087] A concrete example of determining disease risk using a rule-based approach is explained using Figure 10. In this example, the data on the rate of decrease in duration of each behavior type 901, shown in Table 900 of Figure 9, is used to determine the probability of disease from Table 1000, which is a pre-created rule-based table showing the rate of decrease in duration of each behavior type 1001 and the probability of disease 1002, as shown in Figure 10(a).

[0088] From this table, for example, if behavior type 1001 is video viewing, the probability of infection 1002 is determined to be 0-25% if the rate of decrease in behavior time per month is less than 3 minutes, and the probability of infection 1002 is determined to be 25-50% if the rate of decrease in behavior time per month is 3 minutes or more but less than 5 minutes. In this way, the probability of infection 1002 is calculated for each behavior type 1001 based on Table 1000 when using the data for the rate of decrease in behavior time 902.

[0089] On the other hand, if we use the data on the rate of decline in activity level 903 for each type of activity 901 shown in Table 900 of Figure 9, we can determine the probability of illness from Table 1010, which shows the relationship between the rate of decline in activity level 1011 for each type of activity and the probability of illness 1012, as shown in Figure 10(b), which was created in advance based on a rule base. From this table, for example, if the type of activity 1011 is watching videos, we can determine that if the rate of decline in activity level is less than 0.4 per month, the probability of illness 1012 is 0-25%, and if the rate of decline in activity level is 0.4 or more but less than 0.6 per month, the probability of illness 1002 is 25-50%. In this way, when using the data on the rate of decline in activity level 903, we can determine the probability of illness 1012 for each type of activity 1011 based on Table 1010.

[0090] In the disease risk estimation unit 216, the largest of the disease probabilities 1002 or 1012 obtained for each type of behavior 1001 or 1011 is stored in the disease risk data storage area 227 as the disease risk 2271 for user 1.

[0091] On the other hand, when building an AI model to determine the risk of disease, disease risk is estimated using a time-series feature dataset as shown in Figures 10(a) and (b), and disease evaluation results pre-assigned by healthcare professionals as training data.

[0092] When estimating the probability of disease incidence as a disease risk, the results of the assessment by healthcare professionals regarding the presence or absence of the disease are used. In this case, by applying logistic regression analysis to the time-series feature data, a model can be obtained that outputs the probability of disease incidence in the range of 0% to 100%.

[0093] Furthermore, when estimating disease severity as a disease risk, a disease severity score is used as an assessment result by healthcare professionals. In this case, a model for estimating the severity score can be obtained by applying regression methods such as multiple regression analysis or support vector regression to time-series feature data.

[0094] Severity scores for dementia include the MMSE (Mini-Mental State Examination), the Hasegawa Dementia Scale, and FAST (Functional Assessment Staging). For depression, examples include the SQR-D (Self-Rating Questionnaire For Depression) and the Geriatric Depression Assessment Scale (GDS). For frailty, examples include the Japanese Frailty Criteria (J-CHS criteria).

[0095] In disease risk estimation, we explained a method using time-series features (decrease / decline rate) because we believe that relative changes among individual users are important. However, you can also use the values ​​of the time-series data at a specific point in time (such as the duration of the activity, the number of times the activity was performed, the number of types of activities, or the activity level).

[0096] The results processing unit 217 transmits information such as behavior type 2241 and behavior activity level 2242 stored in the behavior estimation data storage area 224, information such as the feature quantities 2261 of the time-dependent change in behavior type and the feature quantities 2262 of the time-dependent change in behavior activity level stored in the time-dependent change data storage area 225, information such as the feature quantities 2261 of the time-dependent change in behavior type and the feature quantities 2262 of the time-dependent change in behavior activity level stored in the time-dependent change feature quantity data storage area 226, and information about disease risk 2271 stored in the disease risk data storage area 227 to the communication volume measuring device 300 via the network 150 from the communication interface 201.

[0097] The process flow described above will be explained using Figures 11 to 14. Figure 11 shows the overall processing flow of the behavior estimation method using data traffic. The behavior estimation using data traffic volume according to this embodiment comprises the steps of: receiving information on the amount of WiFi 105 traffic used by user 1, input from the traffic volume measuring device 300 via the network 150, into the information processing device 200 to estimate user 1's behavior (S1110); determining the change in user 1's behavior over time based on the estimated user 1's behavior (S1120); and estimating disease risk from the change in user 1's behavior over time (S1130).

[0098] Figure 12 shows the details of the process (S1110) for estimating the actions of user 1. The process (S1110) for estimating the actions of user 1 includes an initial setup process (S1111), a communication volume feature calculation process (S1112), and an action type estimation process (S1113).

[0099] First, in the initial setup step (S1111), attribute information such as the age and gender of user 1, the type of behavior of user 1 to be estimated, and the method for calculating the activity level of behavior are input from an external source, such as a keyboard (not shown), via the touch sensor 352 of the display device 350 of the communication volume measuring device 300, or via the input interface 203 of the information processing device 200. This information is then stored in the user information storage area 221 of the storage unit 220 via the user information management unit 211.

[0100] Figure 15 shows an example in the initial setup process (S1111) where a screen 1500 for selecting the type of action to estimate is displayed on the display screen 351 of the display device 350 of the communication volume measurement device 300. User 1 performs the initial setup by touching the selection field 1502 corresponding to the action for which data communication volume is to be estimated from the action types 1501 displayed on this screen 1500, and then touching the OK button 1503. Multiple items may be selected.

[0101] Figure 16 also shows an example in the initial setup process (S1111) where a screen 1600 for selecting the method of calculating activity level is displayed on the display screen 351 of the display device 350 of the communication volume measuring device 300. Touching the OK button 1503 in Figure 15 switches to the screen in Figure 16. User 1 selects a method for calculating activity level from among several items related to the data communication volume feature quantity 1601 representing activity level displayed on this screen 1600, touches the selection button 1602, and then touches the OK button 1603 to complete the initial setup. Multiple items may be selected. Touching the OK button 1603 completes the initial setup process (S1111), and the communication volume feature quantity calculation process (S1112), in which the communication volume measuring device 300 measures the communication volume of user 1, begins.

[0102] In the communication volume feature calculation process (S1112), time-series data of communication information from the WiFi router 105, as shown in Figure 4, which is sent from the communication volume measuring device 300 via the network 150 and classified into uplink communication volume 2221 and downlink communication volume 2222 by the user information management unit 211 and stored in the communication volume data storage area 222, is processed by the communication volume feature extraction unit 212 using the information stored in the user information storage area 221. As a result, at least one of the communication volume features shown in Figure 5 is calculated as the communication volume feature 2231 for user 1 and stored in the communication volume feature data storage area 223.

[0103] Next, in the behavior type estimation step (S1113), the behavior estimation unit 213 estimates the behavior type 2241 of user 1 using the information of the communication volume feature 2231 of user 1 stored in the communication volume feature data storage area 223, and stores it in the behavior estimation data storage area 224.

[0104] Next, Figure 13 shows the process (S1120) for determining the changes in user 1's behavior over time. The process (S1120) for determining the changes in user 1's behavior over time consists of a process (S1121) for calculating the level of activity, a process (S1122) for calculating the changes in the type of activity over time, and a process (S1123) for calculating the changes in the level of activity over time.

[0105] First, in the step of calculating the activity level (S1121), the activity estimation unit 213 calculates the activity level 2242 for each type of activity 2241 of user 1 stored in the activity estimation data storage area 224 based on the communication volume feature quantity representing the activity level as shown in Figure 6, and stores it in the activity estimation data storage area 224.

[0106] Next, in the step of calculating the change in behavior type over time (S1122), the change over time calculation unit 214 uses the information on the behavior type 2241 of user 1 stored in the behavior estimation data storage area 224 to determine the change in behavior type over time 2251 and stores it in the change over time data storage area 225.

[0107] Next, in the step of calculating the change in activity level over time (S1123), the time-series change calculation unit 214 uses the information on the activity level 2242 for each type of user 1 activity 2241 stored in the activity estimation data storage area 224 to determine the change in activity level over time 2252 and stores it in the time-series change data storage area 225.

[0108] The process for estimating disease risk (S1130) is shown in Figure 14. The process for estimating disease risk (S1130) consists of a process for calculating time-dependent features (S1131), a process for estimating disease risk (disease probability or severity) (S1132), and a process for outputting the estimated results (S1133).

[0109] First, in the step of calculating time-dependent feature quantities (S1131), the time-dependent feature quantity calculation unit 215 obtains the time-dependent feature quantity 2261 of the time-dependent change in behavior type from the time-dependent change in behavior type 2251 stored in the time-dependent change data storage area 225, and obtains the time-dependent feature quantity 2262 of the time-dependent change in behavior activity level from the time-dependent change in behavior activity level 2252, and stores them in the time-dependent feature quantity data storage area 226.

[0110] Next, in the step of estimating disease risk (disease probability or severity) (S1132), the disease risk estimation unit 216 estimates the highest disease risk (disease probability or severity) 2271 from the disease risk (disease probability or severity) 2271, based on the information of the time-dependent feature quantities 2261 of the behavior type and the time-dependent feature quantities 2262 of the behavior activity level stored in the time-dependent feature quantity data storage area 226, and the disease risk (disease probability) 2271 from the relationship between the rate of decrease in the duration of behavioral performance for each behavior type of time-dependent feature quantity data as shown in Figure 10(a), and the relationship between the rate of decrease in the behavior activity level for each behavior type of time-dependent feature quantity data as shown in Figure 10(b), and stores it in the disease risk data storage area 227.

[0111] Finally, in the step of outputting the estimated results (S1133), the result processing unit 217 transmits information such as the behavior type 2241 and behavior activity level 2242 stored in the behavior estimation data storage area 224, information such as the feature quantities 2261 of the behavior type's time-dependent change and the feature quantities 2262 of the behavior activity level's time-dependent change stored in the time-dependent change data storage area 225, information such as the feature quantities 2261 of the behavior type's time-dependent change and the feature quantities 2262 of the behavior activity level's time-dependent change stored in the time-dependent change feature quantity data storage area 226, and information on disease risk (probability of occurrence or severity) 2271 stored in the disease risk data storage area 227 to the communication volume measuring device 300 via the network 150 and outputs it to the display screen 351.

[0112] The display screen 351 shows screen 1700, which displays the communication volume as shown in Figure 17. User 1 can confirm that the information displayed on screen 1700 is about them by using the user ID: 1701 displayed on screen 1700. Screen 1700 displays information on the type of action 1704 corresponding to the characteristics of the uplink communication volume 1702 and the downlink communication volume 1703.

[0113] When User 1 touches the OK button 1705 displayed on screen 1700, the display screen 351 switches to screen 1800 which displays the behavior estimation results as shown in Figure 18. When User 1 enters a date in the date input field 1801, a list of information regarding the types of home activities 1803 and cognitive activity levels 1804 for each time period 1802 on the entered date is displayed.

[0114] By touching the OK button 1805 on screen 1800, the display screen 351 will show screen 1900, which displays the behavior estimation results and changes over time, as shown in Figure 19. If you set the period to be displayed in the period input field 1901 on screen 1900 (Figure 19 shows the case where May 2022 to September 2022 is set), the display area 1902 will display data on the daily duration of the behavior and data on the level of activity of the behavior, for each type of behavior.

[0115] Figure 19 shows an example where, for video viewing 1903, the monthly average daily activity time is displayed as a bar graph 1904, and the activity level is displayed as a line graph 1905; and for online calls 1906, the monthly average daily activity time is displayed as a bar graph 1907, and the activity level is displayed as a line graph 1908. The types of activities displayed in the display area 1902 can be changed by moving the button 1909 next to the display area 1902 up or down.

[0116] In the example shown in Figure 19, the period input field 1901 is set to May 2022 to September 2022, and the display area 1902 shows a bar graph of the average daily activity time for each month. However, the period set in the period input field 1901 may span multiple years. In this case, the display area 1902 may be set to display a bar graph of the average daily activity time in one-year, six-month, or three-month increments.

[0117] By touching the OK button 1910 on screen 1900, the display screen 351 switches to screen 2000, which displays the disease risk estimation results as shown in Figure 20. The display area 2001 of screen 2000 displays information regarding the probability of developing dementia as a result of the disease risk estimation.

[0118] Figure 20 shows an example where information regarding the probability of developing dementia is displayed in display area 2001. Alternatively, or together with this, a dementia severity score may be displayed. In this case, the display area 2001 would display it in the form of [Dementia Severity Score 28 / 30].

[0119] By touching the OK button 2002 on screen 2000, a signal is sent to the information processing device 200 indicating that the information displayed on screen 2000 has been confirmed, and this confirmation information is recorded in memory 204.

[0120] The embodiments described above describe a method for estimating the risk of developing dementia, but the present invention is not limited thereto and can also be applied to determining the risk of other symptoms.

[0121] In the embodiment described above, the case where user 1 occupies the personal computer 101, television 102, mobile terminal 103, IoT home appliance 104, and the WiFi router 105 to which they are connected and communicate with the outside world was explained. However, this embodiment can also be applied when multiple users share the WiFi router 105. In that case, by identifying user 1 from the characteristics of the communication pattern of the WiFi router 105 corresponding to the mobile terminal 103 occupied by user 1 and the IP address information of the mobile terminal 103 occupied by user 1, the communication volume of the WiFi router 105 corresponding to the mobile terminal 103 occupied by user 1 can be monitored and processed in the manner described above, thereby inspecting or monitoring the status of user 1.

[0122] As described above, according to this embodiment, by using the amount of data communication on the communication terminal used by the user as input, it has become possible to estimate the user's behavior using the amount of data communication that can be naturally obtained in daily life and to inspect or monitor the user's state.

[0123] Furthermore, it is possible to perform dementia screening tests based on the estimated user status derived from data traffic.

[0124] Furthermore, according to the present invention, data usage information for data communications within the home used by the user in daily life can be obtained naturally without burdening the user, making it possible to perform tests to estimate the risk of illness at an even earlier stage than the first stage (consultation with a family doctor). [Examples]

[0125] Figure 21 shows the configuration of a data traffic analysis system 20 according to a second embodiment of the present invention.

[0126] In Example 1, the communication volume of a WiFi router 105 connected to a personal computer 101, television 102, mobile terminal 103, and IoT appliance 104 used by user 1 in their home was monitored by a communication volume measuring device (terminal device) 300. In Example 2, however, a portable or wearable terminal device such as a smartwatch or smartphone was attached to user 1 as a communication volume measuring device 2100, and the communication volume measuring device 2100 was linked to user 1.

[0127] The communication volume measuring device 2100 has the configuration described with reference to Figure 3 in Example 1. The information processing device 200 has the same configuration and functions as described with reference to Figures 2 and 4 to 20 in Example 1.

[0128] According to this embodiment, in addition to the effects described in Embodiment 1, by using the amount of data transmitted on the communication terminal used by the user as input to monitor the amount of data transmitted, it is possible to estimate the user's behavior using the amount of data transmitted naturally in daily life and to inspect or monitor the user's state.

[0129] Furthermore, it is possible to perform dementia screening tests based on the estimated user status derived from data traffic.

[0130] According to this embodiment, by using the communication volume of a terminal device linked to the user as input, it is possible to estimate the user's behavior using the amount of data communication that can be naturally obtained in daily life, and to inspect or monitor the user's state.

[0131] Furthermore, it is possible to perform dementia screening tests based on the estimated user status derived from data traffic.

[0132] The present invention has been described in detail above based on examples, but it goes without saying that the present invention is not limited to the above examples and can be modified in various ways without departing from its essence. For example, the above examples are described in detail in order to explain the present invention in an easy-to-understand manner and are not necessarily limited to those having all the described configurations. Furthermore, it is possible to add, delete, or replace some of the configurations in each example with other configurations. [Explanation of Symbols]

[0133] 10, 20 Data traffic analysis system 100 Home-based systems 105 WiFi Router 150 Networks 200 Information Processing Devices 210 Data Processing Unit 211 User Information Management Department 212 Communication volume feature extraction unit 213 Behavior Estimation Department 214 Time-dependent change calculation unit 215 Time-dependent feature calculation unit 216 Disease Risk Estimation Department 217 Result Processing Unit 220 Storage section 221 User information storage area 222 Data storage area for data usage 223 Communication volume, feature data, storage area 224 Action Estimation Data Storage Area 225 Time-series data storage area 226 Time-dependent feature data storage area 227 Disease risk data storage area 300 Communication volume measuring device 310 Communication Interface 320 processors 321 Results presentation section 322 Measurement Unit 330 Data Storage Unit 350 Display equipment

Claims

1. A means for measuring the amount of data traffic used by an information terminal used by a user, Information processing means for estimating the user's disease risk by processing the data communication volume of the information terminal measured by the communication volume measurement means, by separating it into information on the data communication volume received by the user and information on the data communication volume transmitted by the user. The system includes an output screen that displays information on the disease risk estimated by the information processing means, The information processing means includes an action estimation unit that identifies the type of action taken by the user using the information terminal based on the data communication volume information received by the user and the data communication volume information transmitted by the user, and determines the activity level of the identified action; A data communication volume analysis system characterized by comprising: a disease risk estimation unit that determines the user's disease probability based on data of the time-series change in the duration of the behavior identified by the behavior estimation unit and the user's disease probability based on data of the time-series change in the activity level of the behavior, and estimates the largest disease probability among the determined disease probabilities as the user's disease risk.

2. A data traffic analysis system according to claim 1, The data traffic analysis system is characterized in that the disease risk estimation unit estimates the user's disease risk from the data on the rate of decrease over time regarding the duration of the user's actions using the information terminal, and the data on the rate of decrease over time regarding the activity level of the actions.

3. A data traffic analysis system according to claim 1, A data traffic analysis system further comprising an input screen, characterized in that the user selects and inputs from a plurality of items displayed on the input screen the type of action performed by the user using the information terminal and the method for calculating the activity level of the action.

4. A data traffic volume analysis method performed by a data traffic volume analysis system, The data usage measurement device measures the amount of data transmitted by the information terminal used by the user. The information processing means estimates the user's disease risk by processing the data communication volume of the information terminal measured by the communication volume measurement means, by separating it into information on the data communication volume received by the user and information on the data communication volume transmitted by the user. The output screen displays the disease risk information estimated by the information processing means. The behavior estimation unit of the information processing means identifies the type of behavior the user engages in using the information terminal based on the data communication volume information received by the user and the data communication volume information transmitted by the user, and determines the activity level of the identified behavior. A data communication volume analysis method characterized in that the disease risk estimation unit of the information processing means determines the user's disease probability based on data of the time-series change in the duration of the behavior identified by the behavior estimation unit and the user's disease probability based on data of the time-series change in the activity level of the behavior, and estimates the largest disease probability among the determined disease probabilities as the user's disease risk.

5. A data traffic analysis method according to claim 4, A data traffic analysis method characterized in that the disease risk estimation unit estimates the user's disease risk from the data relating to the rate of decrease over time regarding the duration of the user's actions using the information terminal, and the data relating to the rate of decrease over time regarding the activity level of the actions.

6. A data traffic analysis method according to claim 4, A data traffic analysis method characterized in that the information processing means selects a type of action performed by the user using the information terminal from among a plurality of types of actions performed using the information terminal displayed on the input screen, selects a method for calculating the activity level of the action performed by the user using the information terminal from among a plurality of methods for calculating the activity level of the action performed using the information terminal displayed on the input screen, and estimates the user's disease risk by processing the data traffic volume of the information terminal measured by the traffic traffic measurement means into information on the data traffic volume received by the user and information on the data traffic volume transmitted by the user, based on the selected type of action performed by the user using the information terminal and the selected method for calculating the activity level of the action performed by the user using the information terminal.

Images