KR20130112158A - Apparatus and method for predicting or detecting a fall - Google Patents

Abstract

PURPOSE: A fall prediction or sensing device and a method therefor are provided to reinforce management for a user by providing the activity information of the user to a manager or a protector. CONSTITUTION: A fall sensing terminal (100) comprises an acceleration characteristic value extracting unit (121), an event classifying unit (122), and a control unit (123). The acceleration characteristic value extracting unit obtains acceleration data from a three-shaft acceleration sensor and extracts an acceleration characteristic value from the obtained acceleration data. The event classifying unit classifies the acceleration characteristic values into a plurality of events and generates data related to the event by accumulating the occurrence frequency of each event. The control unit determines fall occurrence rates by calculating fall danger rates based on the data related to the acceleration characteristic value or the event. [Reference numerals] (100) Fall sensing terminal; (110) Three axis acceleration sensor unit; (111) Wearing determination unit; (121) Acceleration characteristic value extracting unit; (122) Event classifying unit; (123) Control unit; (124,210) Communication unit; (125) Output unit; (126) User input button unit; (200) Fall sensing server; (220) Processor unit; (230) Display unit

Description

Fall prediction or detection device and method therefor {APPARATUS AND METHOD FOR PREDICTING OR DETECTING A FALL}

The present invention relates to a fall prediction or detection apparatus and a method therefor, and more particularly, to a device and a method for predicting a fall accident by not only detecting a fall accident but also calculating a risk of the fall accident.

In recent years, the aging has caused a lot of accidents caused by the fall of the elderly. About one in ten falls suffers from severe injury that requires hospitalization due to femoral fractures or head injuries, resulting in prolonged treatment and sometimes death. have. Therefore, prevention of fall of the elderly becomes important.

Falls, on the other hand, are likely to show signs or signs before they occur. In other words, the loss of balance over time and the fall of the balance will eventually cause falls, so if you can calculate the likelihood of falling or the risk of falling before falling, it will be very helpful in preventing falls.

Therefore, by monitoring the physical activity of the elderly to obtain physical characteristic values, and calculates the likelihood of falling or fall risk from them to prevent the fall accident in advance and also to detect the fall Development is needed.

The present invention relates to a device or method for preventing a fall accident in advance by calculating the likelihood of a fall or the likelihood of a fall, and also detecting whether a fall has occurred.

According to an embodiment of the present invention, there is disclosed a fall sensing terminal for predicting or detecting a fall using acceleration data measured by a 3-axis acceleration sensor, and the fall sensing terminal receives the acceleration data from the 3-axis acceleration sensor. An acceleration feature value extractor configured to sample and acquire at a predetermined time interval for a predetermined period and to extract an acceleration feature value from the obtained acceleration data; An event classification unit configured to classify the acceleration feature value into a plurality of events according to a size, and accumulate the occurrence frequency of each event to generate event related data; And a controller configured to calculate a fall risk or determine whether a fall occurs based on the acceleration feature value or the event-related data, wherein the event-related data includes a plurality of events classifying the acceleration feature value according to size; It may include information about the frequency of occurrence of each event.

The acceleration feature value may include at least one of an acceleration data magnitude, an acceleration data differential value, an acceleration data standard deviation, and a maximum acceleration data, and corresponding axis information of the maximum acceleration data.

The falling sensing terminal may further include a communication unit configured to periodically transmit the event related data to the falling sensing server in order to analyze the data for a predetermined period of time.

The controller determines that a fall has occurred when the acceleration characteristic value or the calculated fall risk rate exceeds a respective threshold value. When the fall occurrence is determined, if the user input is applied for a predetermined time, the fall decision is incorrect. And may increase each threshold by a certain amount.

According to an embodiment of the present invention, a fall sensing server for predicting or detecting a fall is disclosed, and the fall sensing server includes: a communication unit configured to periodically receive the event related data from the fall sensing terminal; And a processor configured to calculate a fall risk based on the received event-related data, wherein the event-related data includes a plurality of events classifying the acceleration feature values extracted from the sampled acceleration data according to size, and the respective events. It may include a frequency corresponding to the event.

The acceleration feature value may include at least one of an acceleration data magnitude, an acceleration data differential value, an acceleration data standard deviation, and a maximum acceleration data, and corresponding axis information of the maximum acceleration data.

The processor unit divides the events corresponding to the plurality of levels into the daily living area A and the impact area B so as to calculate the fall risk, so that the daily living area A and the impact area B It can be configured to calculate the rate of occurrence.

The processor unit may be configured to calculate a score for the entire event by assigning a differential score to each event to calculate the fall risk.

A fall sensing system for predicting or detecting a fall is disclosed in accordance with an embodiment of the present invention, wherein the fall sensing system is worn by a user to detect acceleration data according to the user's movement or movement, and therefrom an acceleration characteristic value or event. A falling sensing terminal configured to generate related data; And a falling sensing server configured to periodically receive the event-related data from the falling sensing terminal and calculate a falling risk rate, wherein the event-related data includes a plurality of events classifying the acceleration feature value according to a magnitude; It may include information about the frequency of occurrence of the event.

The acceleration feature value may include at least one of an acceleration data magnitude, an acceleration data differential value, an acceleration data standard deviation, and a maximum acceleration data, and corresponding axis information of the maximum acceleration data.

The fall sensing terminal determines that a fall has occurred when the acceleration feature value exceeds a threshold value. When the fall occurrence is determined, the fall sensing terminal recognizes that the fall fall determination is wrong when the user input is applied for a predetermined time. Can be configured to increase by a certain amount.

According to an embodiment of the present invention, the fall risk of the user may be calculated and transmitted to the guardian or the administrator to provide the guardian or the administrator with the daily activity information of the user, and may be induced to strengthen the management protection for the user. In addition, the calculated fall risk enables more efficient and reliable detection of a fall.

In addition, by applying a learning algorithm to the application of the threshold for the fall determination algorithm can be adjusted to optimize the algorithm for each individual user.

1 is a block diagram of a falling sensing terminal and a falling sensing server according to an exemplary embodiment of the present invention.
2 illustrates a flowchart of a fall prediction or detection algorithm according to an embodiment of the present invention.
3 is a flowchart of a fall prediction or detection algorithm according to an embodiment of the present invention.
4 is a flowchart of a fall prediction or detection algorithm according to an embodiment of the present invention.
5 illustrates a fall sensor terminal and some functions thereof according to an embodiment of the present invention.
6 is a graph illustrating a fall risk calculation or a fall decision according to an embodiment of the present invention.
7 illustrates an operation related to falling prediction or detection of a falling sensing terminal according to an embodiment of the present invention.
8 illustrates an operation related to falling prediction or detection at the falling sensing server side according to an exemplary embodiment of the present invention.

Hereinafter, a mobile device related to the present invention will be described in detail with reference to the accompanying drawings. The suffix "module" and " part "for the components used in the following description are given or mixed in consideration of ease of specification, and do not have their own meaning or role.

In addition, terms or words used in the present specification and claims should not be construed as being limited to the common or dictionary meanings, but should be construed as meanings and concepts corresponding to the technical spirit of the present invention. Therefore, the embodiments described in the specification and the drawings shown in the drawings are only the most preferred embodiment of the present invention and do not cover all of the technical idea of the present invention, various modifications that can be replaced at the time of the present application It should be understood that there may be equivalents and variations.

1 is a block diagram of a falling sensing terminal 100 and a falling sensing server 200 according to an embodiment of the present invention.

The fall sensing terminal 100 is a device that is worn by a user who needs protection and management from falls, such as an elderly person or a patient, and is a device capable of detecting a user's movement or physical activity. In addition, the fall sensing terminal 100 may calculate a fall risk or determine whether a fall occurs from the detected movement or physical activity.

The fall sensing server 200 may receive, from the fall sensing terminal, event-related data classified into a plurality of events based on a certain criterion of the user's movement or physical activity, and calculate a fall risk or determine whether a fall has occurred. have. In addition, the fall risk or whether the fall occurs can be output through the built-in display or a speaker (not shown).

In addition, the falling sensing terminal 100 and the falling sensing server 200 may be configured as a single falling sensing system in pairs.

In addition, it will be described in the specification of the present invention that both the falling sensing terminal 100 and the falling sensing server 200 can determine whether a fall risk calculation or falling occurs. However, the distribution of functionality of each of these components may vary in specific embodiments and does not limit the scope of the present invention.

The falling sensing terminal 100 transmits the three-axis acceleration sensor unit 110 configured to detect the acceleration data, the data processing unit 120 for processing the detected acceleration data, and transmits the processed data to the falling sensing server 200. The configured communication unit 124, the output unit 125 for displaying or notifying the result related to data processing or whether the fall sensing terminal is properly worn by the user, and the user input button unit 126 for the user's control It may include.

The fall sensing terminal 100 is to be attached or worn on the user's body to be sensed, and is preferably light and small in size. The falling sensing terminal may be in the form of a wrist or an arm band.

Since the falling sensing terminal 100 measures or monitors the movement or physical activity of the user through a built-in 3-axis acceleration sensor, the falling sensing terminal should be properly worn by the user. Accordingly, the falling sensing terminal 100 may further include a wearing determination unit 111 for detecting whether the falling sensing terminal is properly worn by the user.

The data processor 120 is a component for processing the acceleration data measured by the three-axis acceleration sensor unit, and may include an acceleration feature value extractor 121, an event classifier 122, and a controller 123.

The acceleration feature value extractor 121 may use acceleration data obtained from the three-axis acceleration sensor unit to extract a feature such as a user's motion or physical activity. To this end, the acceleration feature value extractor may obtain and acquire the acceleration data from the three-axis acceleration sensor unit at predetermined time intervals during the first period. The acceleration feature value extractor may extract an acceleration feature value from the obtained acceleration data. Here, the acceleration feature value may include at least one of an acceleration data magnitude, an acceleration data differential value, an acceleration data standard deviation, minimum and maximum acceleration data, and corresponding axis information of the minimum and maximum acceleration data.

The event classifier 122 may classify the acceleration feature value obtained by the acceleration feature value extractor into a plurality of events. For example, the acceleration feature value may be data having a magnitude, and the acceleration feature value may be classified into a plurality of events according to the magnitude.

The fall may correspond to an event recognized as a sudden large movement or physical activity. Therefore, if the fall is detected through the sensing value of the 3-axis acceleration sensor, an event in which the sensing value is very large or the variation of the sensing value is very large is high. It would fall. In addition, a relatively large sensing value or a sensing value change may be judged as an indication or sign of a fall, and in such a case, it may be determined that the fall risk is high.

Therefore, if the acceleration characteristic value is classified into several sections (or events) according to the size and the frequency of occurrence of each section is accumulated, this information can be used as a basic data for calculating the fall occurrence rate or the fall risk. .

The controller 123 may calculate a fall risk or determine whether a fall occurs according to the accumulated frequency of each event.

In addition, the communication unit 124 is configured to periodically transmit to the fall sensing server 200 in order to analyze the accumulated occurrence frequency of the event by a predetermined period (for example, time, day, week, month, and year). ) May be included. The communication unit 124 may support wired or wireless communication, and may support a communication protocol for short-range or long-distance communication.

The output unit 125 may output a vibration or a warning sound according to a data processing result of the data processing unit 120 or whether the fall sensing terminal 100 is properly worn by a user. Therefore, the output unit 125 is not limited in kind as long as it can transmit information to any one of the five senses to the user, administrator, or guardian.

In addition, the fall sensing terminal may further include a user input button unit 126 for the interface with the user. For example, when the fall occurrence determination determined by the data processing unit 120 or the fall sensing server 200 is incorrect, the user input button unit 126 may press the button unit for a predetermined time by the user. The fall fall decision may be recognized as wrong.

In addition, the user input button unit 126 may be used for controlling the output unit 125.

If the output unit 125 is a display unit, when the user input button unit is pressed once, a function of switching the screen of the display unit may be performed. For example, if the user displays activity consumption, time, or steps of the user on the display unit, the displayed contents may be switched each time the user input button unit is pressed once.

The fall sensing server 200 may include a communication unit 210 configured to periodically receive event related data from the fall sensing terminal and a processor unit configured to calculate a fall risk or determine whether a fall occurs based on the received event related data ( 220).

Here, the event-related data may include a plurality of events classified according to magnitudes of acceleration feature values extracted from the acceleration data sampled by the falling sensing terminal, and information on frequency of occurrence of each event.

In addition, the fall sensing server may be mistaken as a complex and large-scale server system due to its name, but the fall sensing server may be a PC or a mobile terminal in a home.

In addition, the fall sensing server may include a display 230 for displaying the calculated fall risk. By displaying the fall risk calculated by the processor unit on the display unit, the guardian or administrator of the user can monitor the fall risk.

In addition, the fall sensing server may be provided with a speaker unit (not shown), and may output a warning sound when the fall risk rate exceeds a predetermined level and falls into a dangerous section.

The fall risk can be calculated using various mathematical methods. A plurality of events of the event related data and the frequency of occurrence of the events may be used to calculate the fall risk.

For example, assuming that the event is classified into six events according to the magnitude of the acceleration feature value, there may be a total of six event sections from E1 to E6 in order of smallest to largest according to the magnitude. Here, it can be seen that E5 and E6 have an acceleration characteristic value that can be seen as a sign or sign of a fall occurrence, and in this case, the ratio of the frequency of occurrence of E5 and E6 to the frequency of occurrence of all events can be regarded as the fall risk. have.

In addition, in order to calculate the fall risk, a differential score may be assigned to the occurrence of each event E1 to E6, and a score for the entire event may be calculated. For example, 0.1, 0.3, 0.5, 0.7, 0.9, and 1.1 points may be assigned to E1, E2, E3, E4, E5, and E6, respectively, and the total score may be calculated by multiplying the occurrence frequency of each event. Assuming that the acceleration data or acceleration feature value corresponding to each event is collected 30 times during the first period, the maximum possible score is 30 * 1.1 = 33 points, and the fall risk is calculated as the ratio of the event occurrence score to the maximum possible score. Can be.

In addition, the plurality of events E1 to E5 may be divided into a daily living area A and an impact area B in order to calculate the fall risk. The division into the daily living area and the impact area may be divided according to the magnitude of the acceleration data or the acceleration feature value. For example, a relatively small amount of acceleration data or an acceleration feature may be considered a shock or acceleration change that may occur in a lifetime, and a relatively large amount of acceleration data or an acceleration feature may be dangerous to a user rather than a lifetime. This can be thought of as a possible shock or acceleration change. Accordingly, the events E1 to E4 may be classified into daily living areas, and the events E5 to E6 may be classified into shock areas. As such, the fall risk may be calculated by calculating the ratio of the frequency of occurrence of the daily living area A and the impact area B.

In addition, the aforementioned values for the fall risk calculation may be selectively combined to be used for the fall risk calculation. The description thereof will be described later with reference to FIG. 6.

On the other hand, the processor 220 may analyze the trend of the fall risk for a certain period of time. For example, the processor unit 220 may analyze the fall risk rate from the current trend to the future by collecting the fall risk rate by period, that is, time, daily, weekly, monthly, and yearly. More specifically, for example, if the fall risk of the last N (N = 1, 2, ..., n) weeks is gradually increasing, it is expected that the fall risk will increase in the future. May be displayed on the display unit 230 to request the attention of a guardian or an administrator.

Alternatively, the processor may analyze the trend by combining at least two of the fall risk trends for each period. For example, if both the daily fall risk and the weekly fall risk are collected, and both the trend of the daily fall risk and the change in the weekly fall risk are both falling, the fall risk of the user will increase. The prediction may be displayed on the display unit 230 to request the caregiver's or manager's attention.

In addition, if the daytime fall risk is decreasing but the daytime fall risk is increasing, the user is in need of protection or attention, so the daily fall risk is more important, so the fall risk of the user in the future may increase. In response to this, information on this may be displayed on the display unit 230 to request the attention of a guardian or an administrator.

2 illustrates a flowchart of a fall prediction or detection algorithm according to an embodiment of the present invention.

First, the controller 123 may set a timer for reading acceleration data from the 3-axis acceleration sensor unit 110 and start the timer (S210).

Then, the acceleration feature value extractor 121 may sample the acceleration data at a predetermined period or frequency (S220). Whether or not the acceleration data is sampled for a certain period may vary depending on the embodiment of the present invention. For example, the acceleration data may be read at 20 ms for 1.5 seconds.

Thereafter, the controller 123 may determine whether the set timer has expired (S230). In other words, it may be determined whether the acceleration data has been sampled for the period determined above.

If the timer has not expired, the algorithm must return to S220 to sample the acceleration data. If the timer has expired, the algorithm proceeds to S240 and is designed to perform a series of procedures for processing acceleration data to calculate the fall risk.

The acceleration feature value extractor 121 may extract the acceleration feature value from the acceleration data (S240). In addition, the event classification unit 122 classifies the acceleration feature value into a plurality of events according to the size (S250), and may calculate the frequency of occurrence of each event (S260).

The acceleration feature value may include at least one of an acceleration value obtained during an operation of the timer, an acceleration data differential value, an acceleration data standard deviation, and corresponding axis information of the maximum acceleration data and the maximum acceleration data.

Since most of the acceleration feature values have a magnitude, they can be classified into a plurality of events according to the magnitude. The acceleration characteristic value may use only one of the above-mentioned, and may use a plurality.

For example, if only the acceleration value among the acceleration feature values is used, 30 acceleration values would have been obtained since the acceleration data was sampled at 20 ms for 1.5 seconds according to an example. Accordingly, the 30 acceleration values may be divided into N groups according to their magnitudes, and each of them may be referred to as an event. For example, when divided into six group events, a total of six events of E1 to E6 may occur.

Accordingly, the frequency of occurrence of each event may be accumulated and recorded by classifying which of the 30 acceleration values corresponds to the events E1 to E6.

The controller 123 or the processor 220 may calculate a fall risk based on the classification of the event and the frequency of occurrence of each event (S270). There are several methods of calculating the fall risk using the acceleration value, and some methods are proposed here.

1) the first draft

Events can be classified into E1 through E6 according to the magnitude of the acceleration feature value. Among these, since a relatively large acceleration feature value represents a relatively large physical activity or movement of the user, a rate of occurrence of an event corresponding to a relatively large acceleration feature value may be considered as a fall risk.

Referring to FIG. 6, the values are classified into L1 to L6 according to the magnitude of the acceleration characteristic value and the cumulative frequency of each occurrence is shown. L1 to L6 illustrated in FIG. 6 may correspond to events E1 to E6, respectively. Therefore, in FIG. 6A, E1 occurs seven times, E2 nine times, E3 four times, E4 five times, E5 three times, and E6 twice. In FIG. One time, three times E2, four times E3, five times E4, seven times E5 and ten times E6. These are shown in Tables 1 and 2 below, respectively.

E1 E2 E3 E4 E5 E6 Cumulative frequency 7 9 4 5 3 2

E1 E2 E3 E4 E5 E6 Cumulative frequency One 3 4 5 7 10

Since the users who are protected or managed through the fall sensing terminal or the server according to an embodiment of the present invention are generally elderly people with slow physical behavior and have a high degree of risk, E5 to E6 having high magnitudes of acceleration characteristic values are not easily monitored. It may be an event. Of course, the high risk event is only one example and other events such as E4 to E6 may be high risk events.

Thus, the occurrence or cumulative frequency of E5 and E6 assessed as relatively high risk events can be used to calculate the fall risk. As an example, the ratio of the occurrence frequency of the high risk event to the total frequency may correspond to the fall risk.

The fall risk according to the data in Table 1 corresponds to 5/30 * 100 (%), or about 16.7%. The fall risk according to the data in Table 2 corresponds to 17/30 * 100 (%), or about 56.7%.

In addition, since the events E1 to E4 correspond to events that may occur in everyday life, the events E1 to E4 are classified as events of daily living area A, and the events E5 to E6 are events that cannot occur in everyday life. The user can be classified as an impact area (B) event because it can be seen that the user corresponds to an event that is considered to have been shocked. Therefore, in order to calculate the fall risk, it is possible to calculate the occurrence rate of the daily living area A event and the impact area B event.

로 판단할 수도 있다. 여기서, α는 스케일링을 위한 계수에 해당한다.
In other words,

You can also judge. Where α corresponds to the coefficient for scaling.

2) second proposal

The content of Table 1 and Table 2 will be used as it is. Another way to calculate the fall risk is to assign differential scores to the frequency of each event and calculate their total score.

For example, E1 is 0.1 point, E2 is 0.3 point, E3 is 0.5 point, E4 is 0.7 point, E5 is 0.9 point, and E6 is 1.1 point. These are shown as a table | surface as follows.

E1 E2 E3 E4 E5 E6 system Cumulative frequency 7 9 4 5 3 2 30 Differential score 0.1 0.3 0.5 0.7 0.9 1.1 - score 0.7 2.7 2.0 3.5 2.7 2.2 13.8

E1 E2 E3 E4 E5 E6 system Cumulative frequency One 3 4 5 7 10 30 Differential score 0.1 0.3 0.5 0.7 0.9 1.1 - score 0.1 0.9 2.0 3.5 6.3 11.0 33.8

The respective total scores (13.8 and 33.8), shown in Tables 3 and 4, can also be used as fall risk. They can also be used as a fall risk as a percentage of the overall total score.

3) third proposal

In order to calculate the fall risk, we propose a new fall risk index using the mathematical connection of the acceleration characteristic values.

Vratio is a rate of occurrence of an event in which a user is dangerously shocked, and corresponds to, for example, a ratio of occurrence frequencies of E5 to E6 to the total frequency. A high Vratio value means that there are a lot of events that are big impacts, so the fall risk will rise (the fall risk is proportional to Vratio).

FPnum corresponds to the number of occurrences when the fall sensing terminal or the fall sensing server determines that a fall has occurred, but the user applies an input as a wrong decision. A high FPnum value means that the fall risk level is high, so the fall risk will rise (fall rate is proportional to FPnum).

L1STD corresponds to the standard deviation of the frequency of occurrence of L1 (ie, E1) at regular intervals during the duration of the timer. In other words, for example, if the duration T _s1 of the timer is 7 days (1 week), record the frequency of occurrence of L1 for a certain period (1 day, 1 day) to record their standard for the total duration (1 week). It means deviation. Higher L1STD values will increase the risk of falling because the frequency of occurrence of L1 is erratic (fall rate and L1STD are proportional to each other).

ADLratio is a rate of occurrence of an event to which the user is not dangerous, and corresponds to, for example, a ratio of occurrence frequencies of E1 to E4 to transpose frequencies. A high ADLratio value means that there are many events that correspond to small impacts, so the fall risk will decrease (the fall risk is inversely related to ADLratio).

Therefore, by using the proportional or inverse relationship of each factor, the fall risk can be calculated by the above equation. ω1 to ω4 are weight constants, and K corresponds to a unit conversion constant.

After calculating the fall risk, the controller 123 or the processor 220 may determine whether the calculated fall risk exceeds the first threshold value TH1 (S280). This corresponds to the step of determining whether the calculated fall risk is a value having a valid meaning. The calculated fall risk may have different meanings for each user. In other words, each user has a different physical or physical condition, so even if two different users have the same 60% risk of falling, it can be a sign or indication of a significant risk of falling for a first user and for a second user. It may be a sign or indication of a low level of falls. Therefore, by comparing the calculated fall risk with the first threshold, it is possible to determine whether the fall risk is a value that can support the user, that is, a value that can support the actual fall occurrence.

If the calculated fall risk in step S280 exceeds the first threshold, the algorithm proceeds to the next process, and if the calculated fall risk does not exceed the first threshold, the algorithm may return to step S210 and start again from the beginning.

On the other hand, since the user-specific meaning of the calculated fall risk is different, in the present invention can be corrected by adding a learning algorithm. This will be referred to in FIG. 4 to be described later.

In addition, although not shown, when the fall risk exceeds the first threshold value TH1, a vibration or warning alarm for outputting a high fall risk is output through the output unit 125 built in the fall sensing terminal 100. Can be. This has the advantage that if the fall sensing terminal 100 is used independently of the fall sensing server 200, a user or guardian (for example, a caregiver) who resides in a home can notify the user that a danger has occurred.

If the fall risk rate in step S280 exceeds the first threshold value TH1, the algorithm may proceed to step S310 of FIG. 3.

The controller 123 may set and start a second timer (S310). Again, setting a new timer and sampling the acceleration data newly may be to check whether a fall has occurred since it is determined in step S280 that the fall risk is relatively high. Or, since the fall risk was relatively high, the subsequent movement or physical activity of the user is monitored to prepare for a fall accident.

The controller 123 may sample the acceleration data at a predetermined interval for a predetermined period of time (S320). If the second timer expires (S330), the algorithm may proceed to step S340. The acceleration feature value extractor 121 may extract the acceleration feature value from the sampled acceleration data (S340). By sampling the acceleration data again, you are monitoring the movement or physical activity of the user with a high risk of falling.

The controller 123 may determine whether the acceleration feature value exceeds the second threshold value TH2 (S350). If the acceleration characteristic value does not exceed the second threshold, the algorithm is terminated, and although not shown in FIG. 3, it is preferable to return to step S210 again.

If the acceleration feature value exceeds the second threshold, the controller 123 may determine that a fall has occurred (S360).

An example of a method of performing steps S350 to S360 will be described in more detail. The acceleration feature value extractor 121 may calculate the maximum acceleration value C and the standard deviation D of the acceleration feature values among the acceleration feature values. The calculated maximum acceleration value C and the standard deviation D of the acceleration values may be transmitted to the falling sensing server 200. The controller 123 or the processor 220 may determine whether the calculated maximum acceleration value C and the standard deviation D exceed respective thresholds. As a result of the determination, when the maximum acceleration value C and the standard deviation D exceed respective thresholds, the controller 123 or the processor 220 may determine that a fall has occurred.

In addition, another example of a method of performing steps S350 to S360 will be further described. The acceleration feature value extractor 121 may extract acceleration data of each axis measured by the 3-axis acceleration sensor unit.

Since the three-axis acceleration sensor unit is affected by gravity, it is possible to determine which axis corresponds to the gravity axis by detecting the acceleration data of the three axes. Therefore, the controller 123 may repeat the sampling of the acceleration data and compare the recording of the acceleration data to recognize whether the gravity axis is changed to another axis. Since the fall of the user usually changes the posture of the user, it is possible to determine whether the fall has occurred through the variation of the gravity axis. If such a gravity axis variation is detected, it may be determined that a fall has occurred in step S360.

The algorithm may end with this, and may repeat the operations of steps S210 to S360 again. On the other hand, if the fall occurrence is determined (S360), the algorithm may proceed to step S410 of FIG.

Meanwhile, in step S350, the comparison relationship with the second threshold value TH2 may be different from that shown in FIG. 3. For example, it may be determined whether the acceleration feature value is smaller than the second threshold value TH2 in step S350. It is to be understood that such large and comparative relationships may be modified in accordance with specific embodiments of the present invention.

For example, when a fall occurs, there is a possibility that there is little user movement. Therefore, it is necessary to determine whether the maximum acceleration data among the acceleration characteristic values is smaller than the second threshold or the acceleration data standard deviation is smaller than the second threshold. The occurrence can be determined.

In step S410 of FIG. 4, the controller 123 may control to generate a vibration or an alarm through the output unit 125 for a predetermined time. Then, the controller 123 may determine whether the user input button 126 is pushed for a predetermined time (S420). The user input button is a button for interfacing with a user and may have various functions as described above. That is, when the fall occurs, the vibration or alarm is generated through the output unit 125 of the fall sensing terminal 100. If this is a wrong decision, the user determines a predetermined time for the user input button 126. Should be guided to push.

Therefore, if the fall occurrence is determined and the user's input is applied for a predetermined time, since the decision regarding the fall occurrence is wrong, the controller 123 or the processor 220 reports the algorithm according to FIG. 2 or 3. You may be able to recognize that this is working incorrectly.

In addition, the cause of such a malfunction may be that the setting of the second threshold value TH2 of S350 of FIG. 3 is incorrect. Therefore, when the user input button is pushed for a predetermined time, the second threshold value TH2 is increased by a predetermined size. Or reduce.

For example, when it is determined whether the acceleration feature value exceeds the second threshold value as shown in S350 of FIG. 3, it is preferable to increase the second threshold value by a predetermined magnitude. On the contrary, when it is determined whether the acceleration feature value is smaller than the second threshold, it is preferable to reduce the second threshold by a predetermined size.

Meanwhile, the algorithms in the flowchart shown in FIG. 2 or FIG. 3 may operate individually. In other words, although FIG. 2 and FIG. 3 have been described as sequentially connected to form one algorithm, FIG. 2 and FIG. 3 may function as one independent algorithm. For example, FIG. 2 may be used independently as a fall risk calculation algorithm and FIG. 3 may be used independently as a fall algorithm.

5 illustrates a fall sensor terminal and some functions thereof according to an embodiment of the present invention. The fall sensing terminal 100 is largely composed of a clip 510 and a main body 520, and the clip may be fixed and worn on clothing such as a user.

Referring to FIGS. 5B and 5C, the falling sensing terminal 100 includes a button 531 and a button pressing sensing unit 532 to determine whether the falling sensing terminal 100 is properly worn. Can be determined. The button 531 and the button pressing sensing unit 532 correspond to the wearing determining unit 111 of FIG. 1, and the example illustrated in FIG. 5 is only an example, and the present disclosure is not limited thereto.

As illustrated in FIG. 5C, when the user's clothing C is caught between the clip 510 and the main body 520, the button press sensing unit 532 is pressed by pressing the button 531. You can detect this.

6 is a graph illustrating a fall risk calculation or a fall decision according to an embodiment of the present invention. It is data that accumulates and records the frequency of occurrence of six events classified according to the magnitude of the acceleration feature value in order to calculate the fall risk or determine the fall.

L1 to L6 of the x-axis (horizontal axis) of FIG. 6 (a) or (b) are divided into sections according to the magnitude of the acceleration characteristic value, and are displayed as sections corresponding to small values and sections corresponding to large values.

Therefore, as the cumulative bar graph from L1 to L6 increases, the fall risk will increase, and it can be determined based on the fall.

7 illustrates an operation related to falling prediction or detection of a falling sensing terminal according to an embodiment of the present invention.

The acceleration feature value extractor 121 of the falling sensing terminal 100 may sample the acceleration data of each axis of the 3-axis acceleration sensor 110 (S710). The sampling frequency may vary depending on the usage environment, and the sampling period may also vary depending on the usage environment.

The acceleration feature value extractor 121 of the falling sensing terminal 100 may extract an acceleration feature value from the sampled acceleration data (S720). The acceleration feature value may include at least one of an acceleration value, an acceleration data differential value, an acceleration data standard deviation, minimum and maximum acceleration data, and corresponding axis information of the minimum and maximum acceleration data.

The event classifier 122 may classify the acceleration feature value into a plurality of sections, that is, a plurality of events, according to the size (S730). In addition, the event classification unit 122 may cumulatively calculate a frequency of occurrence of each event (S740). That is, the event classification unit 122 may generate event related data, and the event related data is necessary for calculating a fall risk or determining whether a fall occurs.

The controller 123 may control the communicator 124 to transmit the event related data to the fall sensing server 200. In addition, the controller 123 may calculate a fall risk or determine whether a fall occurs through the event-related data (S750). For a detailed method of calculating a fall risk or determining whether a fall has occurred, refer to the aforementioned contents of FIGS. 2 to 4.

8 illustrates an operation related to falling prediction or detection at the falling sensing server side according to an exemplary embodiment of the present invention.

The falling sensing server 200 may periodically receive the event related data from the falling sensing terminal (S810). The event related data may include information about the categorized event and its occurrence frequency.

The processor unit 220 of the fall sensing server 200 may calculate a fall risk or determine whether to fall based on the received event-related data (S820).

For a detailed method of calculating a fall risk or determining whether a fall has occurred, refer to the aforementioned contents of FIGS. 2 to 4.

Having described the embodiments of the present invention above, those of ordinary skill in the art will recognize that these embodiments are illustrative rather than limiting, and that various changes and modifications may be made without departing from the scope or spirit of the invention Variations, and modifications may be made without departing from the scope of the present invention.

100: fall sensing terminal 110: 3-axis acceleration sensor unit
121: acceleration feature value extractor 122: event classifier
123: control unit 124: communication unit
125: output unit 126: user input button unit
200: fall sensing server 210: communication unit
220: processor unit 230: display unit

Claims (11)

A fall sensing terminal for predicting or detecting a fall using acceleration data measured by a 3-axis acceleration sensor,
An acceleration feature value extraction unit configured to sample and acquire the acceleration data from the three-axis acceleration sensor at a predetermined time interval for a predetermined period of time, and extract an acceleration feature value from the acquired acceleration data;
An event classification unit configured to classify the acceleration feature value into a plurality of events according to a size, and accumulate the occurrence frequency of each event to generate event related data; And
And a controller configured to calculate a fall risk or determine whether a fall occurs based on the acceleration feature value or the event related data.
The event-related data includes a plurality of events in which the acceleration feature values are classified according to size, and information on a frequency of occurrence of each event.
The falling sensing terminal of claim 1, wherein the acceleration feature value comprises at least one of an acceleration data magnitude, an acceleration data differential value, an acceleration data standard deviation, and a maximum acceleration data, and corresponding axis information of the maximum acceleration data.
The falling sensing terminal of claim 1, further comprising a communication unit configured to periodically transmit the event related data to the falling sensing server to analyze the event-related data for a predetermined period of time.
The method of claim 1, wherein the control unit:
If the acceleration characteristic value or the calculated fall risk exceeds each threshold, it is determined that a fall has occurred.
And when the fall occurrence is determined, if the user input is applied for a predetermined time, recognizing that the fall occurrence determination is wrong, and configured to increase the respective threshold by a predetermined amount.
A fall sensing server for predicting or detecting a fall,
A communication unit configured to periodically receive event related data from a fall sensing terminal; And
A processor unit configured to calculate a fall risk or determine whether a fall based on the received event related data,
And the event-related data includes a plurality of events classified according to magnitudes of acceleration feature values extracted from sampled acceleration data, and information on a frequency of occurrence of each event.
6. The falling sensing server of claim 5, wherein the acceleration feature value comprises at least one of an acceleration data magnitude, an acceleration data differential value, an acceleration data standard deviation, and a maximum acceleration data and corresponding axis information of the maximum acceleration data.
The method of claim 5, wherein the processor unit is configured to calculate the fall risk:
Falling sensing server, configured to calculate the rate of occurrence of the daily living area (A) and the impact area (B) by dividing the plurality of events into a daily living area (A) and the impact area (B).
The method of claim 5, wherein the processor unit is configured to calculate the fall risk:
A fall sensing server, configured to calculate a score for the entire event by assigning a differential score to each occurrence of the event.
A fall sensing system for predicting or detecting falls,
A fall sensing terminal, worn by a user, configured to detect acceleration data according to the movement or movement of the user and to generate acceleration characteristic values or event related data therefrom; And
A fall sensing server configured to periodically receive the event related data from the fall sensing terminal to calculate a fall risk;
And the event-related data includes a plurality of events in which the acceleration feature values are classified according to size, and information on a frequency of occurrence of each of the events.
10. The fall sensing system of claim 9, wherein the acceleration feature value comprises at least one of acceleration data magnitude, acceleration data derivative, acceleration data standard deviation, and maximum acceleration data and corresponding axis information of the maximum acceleration data.
The apparatus of claim 9, wherein the falling sensing terminal is:
If the acceleration characteristic value exceeds a threshold, it is determined that a fall has occurred.
And when the fall occurrence is determined, recognize that the fall occurrence determination is wrong when an input of a user is applied for a predetermined time, and increase the threshold by a certain amount.

Images