TWM503630U - Daily physical activity surveillance system - Google Patents

Abstract

A daily physical activity surveillance system is provided. The system includes a wearable device worn by a tester, wherein the wearable device has a RFID tag capable of emitting a tag information; a plurality of RFID readers dispatched in rooms of a building and a zone near a door outside the building capable of receiving the tag information, wherein the tag information is received by one RFID readers closest to the RFID tag; a data transmitting system connected to the RFID reader capable of transmitting the tag information issued by every RFID readers; and a server connected to the data transmitting system capable of building a daily physical activity pattern of the tester in response to the tag information and corresponding RFID readers, thus detecting if an abnormal physical activity happens.

Description

Daily activity monitoring system

The present invention relates to a monitoring system, and more particularly to a daily activity monitoring system constructed using radio frequency identification (RFID) technology.

Due to factors such as the aging of the population and the work of children in the field, more and more silver-haired people are living alone. Although the atmosphere of living in nursing homes has gradually opened up, but in the traditional concept of society, most of the silver-haired people still choose to live in their own homes. Although there is special care for the nursing home, the quality of care has its limitations because of the limited staff and the need to care for the public. Living in your own home, due to lack of care, can be expected to cause more serious problems. Therefore, how to solve the extremely large care load and medical needs brought by it has become a social problem that is extremely headache but unavoidable in various countries.

In recent years, the progress of Information and Communication Technology (ICT) has become a hot research field. How to use ICT technology to mitigate the impact of population aging has become a hot research field. The sensor is used to develop a daily activity monitoring system that can observe the silver hair at any time, and automatically raise an alarm when an abnormal situation occurs. In this way, it will be able to achieve the purpose of caring for the solitary silver-haired people with the least amount of manpower.

For example, it is customary to use a camera as a sensor and to monitor the daily activities of the silver-haired family is widely adopted. It is the most direct and convenient way to perform monitoring with a camera. However, its shortcoming is that there is a violation of privacy, which will make the monitored person feel uncomfortable. That is, due to features such as the face and limbs It is not uncomfortable for the solitary people to feel repulsive, and there is a violation of privacy and ethical violations. In addition, it requires a lot of manpower surveillance, which is quite expensive.

In addition, in another monitoring system of the prior art, a Passive Infrared (PIR) Motion Sensor/Motion Detector can be utilized to monitor the daily activities of the silver-haired family. For example, a passive infrared human body motion sensor can be installed in the bedroom, living room, bathroom, study, kitchen, laundry room, front door, and back door to form a monitoring system that targets the daily life of the subject, such as sleep and Monitor using the toilet status, etc.

For example, using a passive infrared human body motion sensor installed indoors, the monitoring system can count the time of the elderly living alone indoors, such as the time of sleep, the number and time of using the toilet, the time of going out, and the like.

However, passive infrared human body motion sensors can only detect the movement of the human body, and cannot accurately estimate the amount of activity, so the information obtained is less valuable.

The present invention relates to a daily activity monitoring system for monitoring a subject, including: a wearable device, disposed on the subject, including a radio frequency identification tag, capable of actively transmitting a tag data; The radio frequency identification reader is disposed in a plurality of rooms in a building room and a door in the outdoor, wherein the radio frequency identification tag transmits the tag data to the door when the subject moves in the building The closest one of the plurality of RFID readers; a data transmission system coupled to the RFID readers for transmitting the tag data received by the RFID readers And a server connected to the data delivery system to receive the tag data corresponding to the RFID readers for establishing a daily activity data of the subject.

In order to better understand the above and other aspects of the present invention, the preferred embodiments are described below, and in conjunction with the drawings, the detailed description is as follows:

101~108‧‧‧Active RFID reader

110‧‧‧Subjects

120‧‧‧Regional Network

130‧‧‧ gateway

140‧‧‧Internet

150‧‧‧Server

Figure 1 is a diagram showing an embodiment of the present daily activity monitoring system.

Fig. 2 is a schematic diagram showing the definition of time intervals Ta, Tb, and Tc.

FIGS. 3A to 3C are diagrams showing the degree of membership of Ta, Tb, and Tc.

Figure 4 shows the activity density map (ADM) of the subject.

Figure 5 is a schematic diagram of the symbiosis matrix of the active density map.

Figures 6A to 6D are diagrams showing the activity density maps obtained by the subject at home from the first week to the fourth week.

Figure 7 is the result of the dissimilarity analysis of the normalized weighted Euclidean distance of the continuous four-week activity density map.

The present invention is a daily activity monitoring system constructed using radio frequency identification (RFID) technology. Using radio frequency identification (RFID) technology to further obtain the exact location of the silver-haired family, and because there may be silver-haired couples living together or having visitors, Radio Frequency Identification (RFID) can also help determine what the activity is, so it can be collected. More information on daily activity status is available for judgment.

In addition, the monitoring system records the activities of the daily activities of the silver-haired family (ADL) through a remote server installed in the monitoring center, including: Time Away From Home (TAFH), Activity Density Map (ADM) for analyzing the dissimilarity of daily, weekly, or monthly daily activity patterns, analyzing the dissimilarity of the activity density map, and verifying whether the silver-haired family goes out and calculates the time of travel. To detect if there is any abnormality in the body, authorized family members can log on to the monitoring center website to view the elderly status at any time. The detailed description of the present invention is as follows: Please refer to Figure 1, which is a new daily activity monitoring system. The embodiment of the system. First, install Active RFID Reader 101~108 and use the regional network in the front door and back door of the building's home interior (eg, bedroom, living room, bathroom, study, kitchen, laundry room). 120 transmits the data of the active RFID readers 101 to 108 to the gateway 130. The gateway 130 then transmits the data of the active RFID readers 101 to 108 to the server 150 of the remote monitoring center via the Internet 140 (also known as the Internet). In other words, the local area network 120, the gateway 130, and the Internet 140 can be regarded as a data transfer system for transmitting data of the active RFID readers 101 to 108 to the remote monitoring center. Server 150.

Furthermore, the wearable device on the subject 110 (eg, a silver hair) includes an active RFID tag 112. When the subject 110 walks to any room or door where the active RFID readers 101-108 are installed, the active RFID tag 112 will actively transmit the internal tag data to the nearest active RFID reader 101~ 108. The tag data inside the active RFID tag 112 includes the identity identifier (ID) of the subject 110. Therefore, even if there are other people or visitors in the 110 households, it will not affect the accuracy of the collected data.

In other words, a set of monitoring systems can be formed using active RFID readers 101-108 installed in the bedroom, living room, bathroom, study, kitchen, laundry, front door, and back door, etc., and for the subject 110 Daily life, such as sleep and use of toilets, etc., are monitored. Moreover, the remote server 150 can automatically issue an alarm message when an abnormal condition occurs. In this way, the purpose of caring for the person 110 (silver hair) who lives alone can be achieved with the least amount of manpower.

For example, using the active RFID readers 101-108 installed indoors, the monitoring system can count the activity time of the subject 110 every day, sleep time, the number and time of using the toilet, the time of going out, etc. Wait. And based on the establishment of daily activity data (ADL). The daily activity data (ADL) includes at least one time out (Time Away From Home, TAFH) and an activity density map. (Activity Density Map, ADM).

The following is an example of the daily activity monitoring system according to Fig. 1 as an example. And the remote 150 server is used to evaluate the collected daily activity data of the subject 110. The research methods used are as follows:

[Statistics of Outing Time (TAFH)]

First, the length of the subject's 110 travel time (TAFH) is recorded as an important characteristic parameter of the daily activity level based on the fuzzy logic. Outgoing time (TAFH) can use the indoor FRID reader 103 closest to the front door and the outdoor FRID reader 101 of the front door to complete the detection and time recording of exit and return, and as the time of going out (TAFH).

As shown in Fig. 2, it is a schematic diagram of the definition of time intervals Ta, Tb, and Tc. The time interval in which the subject 110 passes through the last indoor RFID reader 103 to the front door outdoor RFID reader 101 before leaving the home is defined as the pre-home preparation time Ta. The period of time during which all RFID readers 101-108 have not sensed an event after leaving home is defined as: home time Tb. And, the time interval after returning home through the front door outdoor RFID reader 101 to the indoor first indoor RFID reader 103 is defined as: sensing time Tc after returning home. Among them, the units of Ta, Tb, and Tc are counted in seconds. The present invention can further determine whether the subject 110 has an outing event and a calculated outing time (TAFH) by using the pre-home preparation time Ta, the home-time period Tb, and the home-back sensing time Tc.

For example, since the subject 110 is not necessarily leaving the home, the subject 110 simply takes a short time to open the door, take a newspaper, water the flowers, or just go out and say hello to the neighbor. The above actions are not incidents of going out. It is therefore necessary to use the time of travel (TAFH) to verify whether it is an out-of-office event.

The new model further utilizes the fuzzy logic to further test the time of the subject 110 (TAFH) and the outgoing event, first preparing the time before leaving home Ta, the time Tb after leaving home, and the time of sensing after returning home. Tc three TAFH characteristic parameters are input, use the trapezoidal function to define fuzzy membership, and judge Ta, Tb, Tc is It is a membership, short, long, and very long.

Please refer to FIGS. 3A to 3C , which are schematic diagrams for judging the membership degrees of Ta, Tb, and Tc. Basically, the membership is a value between 0 and 1. Taking the pre-home preparation time Ta as an example, when the Ta time is longer, the short membership degree will gradually decrease, and the long membership degree will gradually increase.

Furthermore, this new model uses the three membership values of Ta, Tb, and Tc as the input of the fuzzy rule, and then linearly combines the output trust values of multiple Takagi-Sugeno-Kang models. The output trust values of a Takagi-Konano-Kang model are defined between [0, 1].

Please refer to Table A, which is a schematic diagram of the TAFH confidence value. It can be used to characterize fuzzy rules for outbound events and their confidence value fuzzy rules. The TAFH trust value and membership can be obtained by using this set of fuzzy rules. Among them, the description of the TAFH trust value (Linguistic) and membership degree include: 1, which is excellent, 0, which represents very poor, 0.75, which represents good, and 0.25 which represents poor.

For example, the second fuzzy rule in Table A shows that when someone leaves the house or returns to the house, the time spent near the door should be much shorter than the duration of the apartment (Ta, Tc's membership) The degree is short and the membership of Tb is long). When the second fuzzy rule is met, the TAFH trust value is 1 representing excellent. Furthermore, the eighth fuzzy rule states that when someone leaves the apartment for a very long period of time (Tb's membership is very long), at this time, regardless of the length of time spent near the door, the TAFH trust value includes the representative. 1. According to the embodiment of the present invention, when the output trust value is higher than 0.75, it is regarded as a valid and true outgoing event, and the outgoing time (TAFH) is used as the reference data.

The out-of-office time (TAFH) and the out-of-office event are important data for the activity subject's 110 activity density map (ADM). Of course, the time of departure (TAFH) can also be expressed in hours. Furthermore, the average out-of-office time (TAFH) can also calculate the average out-of-day time (average TAFH).

[Active Density Map (ADM)]

In the activity density map (ADM), the number of steps taken by the subject in 110 unit time is used as an indicator to measure the amount of physical activity. Therefore, the wearable device on the subject 110 further includes a motion amount sensing component to obtain step count information. The information of the number of steps can be transmitted to the server 150 via the gateway 130 and the Internet 140 by using bluetooth communication technology.

First, as shown in Table B, healthy adults will not walk according to each day. The number of synchronizations is classified into groups of different activities: walking more than 12,500 steps per day, which is highly active; walking 10,000 to 12,499 steps per day, which is active (active); walking 7500 to 9999 steps per day, which is somewhat active (Somewhat Active); walking 5000~7499 steps per day, belonging to Low active; walking less than 5000 steps per day, belonging to Sedentary.

The present embodiment establishes an activity density map (ADM) of the subject 110 in units of time that are correlated with each other for one week. In general, the regularity of the subject 110 can be seen in units of one week. For example, the subject 110 goes to the hospital to see the clinic, and the outpatient doctor also schedules the work on a weekly basis. If the same time every week, such as Wednesday morning, go to the hospital to see the same clinic, it will show the week and week, Wednesday morning cycle. Sex and regularity.

Please refer to Figure 4, which is shown as the activity density map (ADM) of the subject. The activity density map (ADM) includes a date in a week and a statistical chart obtained by the number of walks of the subject 110 between 24 hours. The different color depths are assigned to indicate the number of steps taken by the subject 110 during the hour, and the calculated density value (steps/hour) is used to express the subject 110 at different times. Different activity densities. Of course, different color depths can also be changed to different gray levels, or different patterns to represent different density values (steps/hour). Of course, the average activity density (Average AD) can also be calculated based on the activity density map (ADM).

Use different color depths to quickly interpret the differences and adopt the theory of nonlinear quantification. The activity density map (ADM) and the time of departure (TAFH) in Fig. 4 can be used to estimate the activity of the subject 110 for a day. For example, the testee 110 gets up at about 6 o'clock every day. From 7 o'clock, he can often infer that he should go out for sports or shopping, exercise around 4 pm, and go to bed after 8 pm.

That is, the high density value (steps/hour) pattern is matched with the outgoing event, which means that the subject 110 is more frequent from home and tends to exhibit a higher degree of vitality. We can use this to observe or confirm whether the subject 110 continues to maintain a healthy trend. In other words, the activity density map (ADM) can be used to interpret the regularity of daily life of the subject 110, and to continuously observe changes in his health.

[activity density map co-occurrence matrix (ADMCM)]

Although the average daily activity density (average AD) and the average out-of-day time (average TAFH) of the subject 110 can be used to provide an average degree of living of the subject 110, it cannot be used as a regularity of the daily life of the subject 110. index.

The present invention further directly generates an activity density map co-occurrence matrix (ADMCM) based on the data of the activity density map (ADM) and the out-of-time time (TAFH) of the subject 110. In other words, the texture information of the co-occurrence matrix (or co-occurrence matrix) of the activity density map (ADM) is used to capture the regular life pattern of the daily activity of the subject 110. The dissimilarity measure based on the symbiotic matrix can be used to calculate the change of the daily life pattern in the activity density map.

In the (7 × 24) matrix of the activity density map (ADM), the horizontal direction points to the number of hours of the day (24 hr/day), and the vertical direction points to the number of days of the week (ie, the week) 1. Tuesday, Wednesday, Thursday, Friday, Saturday, and Sunday). And each matrix element represents the density of the home activity of the subject 110 during the hour of the day. Next, the matrix elements are quantized with M levels, ie, an active density map co-occurrence matrix is generated. When M=8 is the activity density map (ADM) as shown in Fig. 6. That is to say, the matrix element of the activity density map (ADM) is a visualized activity density map if eight patterns are used to represent the eight quantized frequencies from small to large.

[Identity analysis of activity density map]

Please refer to FIG. 5, which is a schematic diagram of the active density map co-occurrence matrix. The co-occurrence matrix includes: a matrix element A(k, l) and its eight adjacent matrix elements. Furthermore, the process of extracting texture information from the active density map co-occurrence matrix is as follows. As shown in Fig. 7, the symmetry matrix marks the directional axes of four different angles, including 0 degrees, 45 degrees, 90 degrees, and 135 degrees.

Furthermore, let P ₀ (i, j) be the adjacent two elements A(k, l) = i (steps / hour) and A (k, l + 1) = j (number of steps / hour).

Similarly, let P ₄₅ (i,j) be two adjacent elements A(k,l)=i (steps/hour) and A(k-1,l+1)=j (steps) in the 45-degree axis direction. Number / hour).

P ₉₀ (i, j) is the adjacent two elements A(k, l) = i (steps / hour) and A (k-1, l) = j (number of steps / hour) in the direction of the 90 degree axis.

P ₁₃₅ (i,j) is the adjacent two elements A(k,l)=i (steps/hour) and A(k-1,l-1)=j (steps/hour) in the direction of the 135 degree axis. ).

Let Q(i,j) be equal to the sum of P ₀ (i,j), P ₄₅ (i,j), P ₉₀ (i,j), and P ₁₃₅ (i,j). That is, Q(i,j)=P ₀ (i,j)+P ₄₅ (i,j)+P ₉₀ (i,j)+P ₁₃₅ (i,j)

In the dissimilarity analysis, the novel uses five characteristic parameters to measure the regular behavior in the activity and capture its changes. The five characteristic parameters include: average activity density (average AD) x1, average out-of-time time (average TAFH) x2, angular second moment feature x3, contrast feature parameter (The contrast feature) x4, and entropy (entropy) x5. Wherein, the second-order momentum characteristic parameter x3 can be used to measure the homogeneity of the activity density, defined as:

The contrast characteristic parameter x4 can be used to measure the amount of local variation in activity density, defined as:

Entropy x5 is used to measure the degree of disorder of activity density, defined as:

Finally, the weighted normalized Euclidean distance is used to calculate the dissimilarity between the two active density maps. The value falls between 0 and 1. The more similar the active density map is mapped to the feature. The distance in space (d _Euclidean ) is also smaller. among them,

ω _i is a weighting factor, which can be set to 1/5 first, and the optimal value is estimated by experimental data. x _i,c is the current characteristic map of the five activity parameters used to calculate the dissimilarity. x _i,r is the characteristic parameter of the reference activity map. Furthermore, the normalized characteristic parameters and The definition is as follows. Among them, the above i = 1, 2, 3, 4, 5.

Please refer to FIG. 6A to FIG. 6D, which are diagrams showing the activity density map obtained by the subject in the home from the first week to the fourth week. Figure 7 is the result of the dissimilarity analysis of the normalized weighted Euclidean distance of the above-mentioned continuous four-week activity density map. Among them, the activity density map of the first week is the baseline.

Obviously, the second and third weeks are less different from the first week that is referred to as the baseline, which has significant regularity and periodic behavior. Therefore, as shown in Fig. 7, the degree of dissimilarity of the normalized weighted Euclidean distance is small. However, the activity density map between the fourth week and the first week is very different, and the fourth week is live. The dynamic density is much smaller than the activity density of the first week. Therefore, the degree of dissimilarity of the normalized weighted Euclidean distance in the fourth week becomes large, that is, the degree of dissimilarity between the first week and the fourth week suddenly becomes large. Therefore, the remote server 150 can be used as a warning to automatically alert the subject, the physician, and his/her family.

[Daily Dissimilarity Analysis Based on Gaussian Mixture Model]

In order to enable the remote server 150 to detect the abnormality of the subject's activity density earlier, we will propose a Gaussian mixture model (GMM) dissimilarity analysis in days. The description is as follows: The novel will use the observation sequence of the 24-hour activity density (steps/hour) of the subject as a feature vector benchmark of the training database for a period of time, such as one month or longer, and use 24 The probability function of a continuous random variable is usually expressed as a set of linear combinations containing K Gaussian distributions, namely Gaussian mixture density.

Where o is the observed activity density (steps/hour) of the subject at 24 hours a day; μ _k is the average vector of the kth Gaussian distribution; Σ _k is the covariation matrix of the kth Gaussian distribution; N( o, μ _k , Σ _k ) is the probability density function value of the subject at the kth Gaussian distribution on the day; c _k is the weighted value of the Gaussian distribution; b(o) is the subject's day at this Gaussian The probability of occurrence under the hybrid model; and must be consistent with:

For a more accurate assessment, we also added the length of time the subject went out (TAFH) as another important random variable, also expressed as a Gaussian distribution probability density function. The probability b(o) that will occur is corrected to:

Where ω and (1-ω) are weighted values, μ _TAFH is the average of the daily outing time, and Σ _TAFH is the variation of the daily outing time. The smaller the probability b(o) will occur, the greater the dissimilarity. If the activity is smaller, the remote server 15 will send a warning to remind the relevant person.

It can be seen from the above description that the present invention is a daily activity monitoring system constructed by using radio frequency identification (RFID) technology. Radio frequency identification technology (RFID) is used to further obtain the exact location of the silver-haired family, and an activity density map (ADM) is established as a daily activity status information for judgment.

Furthermore, the monitoring system records the daily activity data (ADL) of the silver hair for a long time through a remote server installed in the monitoring center, including: Time Away From Home (TAFH), Activity Density Map (Activity Density Map, ADM), for analyzing the dissimilarity of the day, week, or month of the daily activity pattern, and analyzing the dissimilarity of the activity density map. When a situation of great dissimilarity is detected, a warning is issued to inform the family or doctor.

In summary, although the present invention has been disclosed above in the preferred embodiments, it is not intended to limit the present invention. Those skilled in the art can make various changes and refinements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of this new type is subject to the definition of the scope of the patent application.

101~108‧‧‧Active RFID reader

110‧‧‧Subjects

112‧‧‧Active RFID tag 112

120‧‧‧Regional Network

130‧‧‧ gateway

140‧‧‧Internet

150‧‧‧Server

Claims (10)

A daily activity monitoring system for monitoring a subject, including: a wearable device, disposed on the subject, and the wearable device includes a radio frequency identification tag, and can actively transmit a tag data; a plurality of wireless devices The RFID reader is disposed in a plurality of rooms in a building room and a door outside the building, wherein the RFID tag is the tag data when the subject moves within the building Passing to one of the plurality of plurality of radio frequency identification readers; a data transmission system coupled to the radio frequency identification readers for transmitting the plurality of radio frequency identification readers And a server connected to the data transmission system to receive the label data corresponding to the RFID readers for establishing a daily activity data of the subject. The daily activity monitoring system of claim 1, wherein the data transmission system comprises a regional network connected to the wireless RFID readers; a gateway connected to the regional network; and an internet The network is connected to the server; wherein the tag data corresponding to the RFID readers is transmitted to the server via the gateway. The daily activity monitoring system of claim 1, wherein the radio frequency identification reader comprises a first radio frequency identification reader located at the door of the building outdoor and located in the building interior a second RFID reader close to the doorway, and the daily activity data includes an outgoing time, wherein the outgoing time includes a pre-home preparation time, a departure time, and a home return sensing time. The time interval defined by the second radio frequency identification reader and the first radio frequency identification reader is defined as the pre-home preparation time, and the radio frequency identification readers are not The time interval in which the subject is sensed is defined as the departure time, and the subject is sequentially sensed by the first RFID reader and sensed by the second RFID reader. The time interval is defined as the time after the return home. The daily activity monitoring system according to claim 3, wherein the server uses a fuzzy rule to determine a relationship between the pre-home preparation time, the departure time, and the sensing time after returning home. A trust value and a membership degree that determine the time of the outing are used to determine whether an outing event occurs. The daily activity monitoring system of claim 1, wherein the wearable device further comprises a motion sensing component to transmit the step information of the subject to the server. For example, the daily activity monitoring system described in claim 5, wherein the daily activity data further includes an activity density map. For example, the daily activity monitoring system described in claim 6 wherein the activity density map is in units of one week, and counts the number of steps of the subject per hour as an activity density within 24 hours a day, and the sister is The activity density can correspond to one of a plurality of patterns. The daily activity monitoring system of claim 7, wherein an active density map symbiosis matrix is generated according to the daily activity data of the subject, and a first week activity density map and a second are performed accordingly. A dissimilarity analysis between weekly activity density maps. The daily activity monitoring system of claim 8, wherein the dissimilarity analysis is performed by using a plurality of characteristic parameters, and the characteristic parameters include: an average activity density, an average elapsed time, and a second order The momentum characteristic parameter, a contrast characteristic parameter, and an entropy; and using a normalized weighted Euclidean distance to calculate a dissimilarity between the first week activity density map and the second activity density map. The daily activity monitoring system according to claim 7, wherein the server uses the daily activity data and the activity density of 24 hours a day as a feature vector reference of the training database, and is a 24-dimensional continuous random variable. The probability function is expressed and written as a Gaussian mixture density that includes a linear combination of a plurality of Gaussian distributions.