KR101565970B1 - Apparatus and method for determining stroke during the sleep - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a device and a method for determining a stroke during a sleep,
A stroke determination apparatus according to one aspect of the present invention includes a motion data acquisition unit arranged on a body part of a user and collecting motion data of a user, A feature extracting unit for extracting a user's motion feature data by comparing one motion data with the motion data and a motion limit point and monitoring the motion feature data and comparing the motion feature data with the stroke threshold, And determining a real-time sleep surface stroke determination unit.

Description

[0001] APPARATUS AND METHOD FOR DETERMINING STROKE DURING THE SLEEP [0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a device and a method for determining a stroke during a sleep,

Stroke is a condition in which the brain functions in a partial or complete manner, and this disorder lasts for a considerable period of time. In addition to the cerebrovascular disease, no other causes can be found. In addition to the aging of the population, Economic importance is growing.

According to statistics from the Health Insurance Review and Assessment Service, the number of stroke patients in 2011 is estimated at 520, 214 people, up 14% from 456,435 in 2007.

In addition, stroke is the second most common cause of cancer deaths in Korea, with 70.3 deaths per 100,000 population, accounting for 13.9% of all deaths.

Stroke is a term used to refer to a cerebral infarction (ischemic stroke) caused by a blockage of cerebral blood vessels and a cerebral hemorrhage (hemorrhagic stroke) caused by the blood leaking into the brain tissue due to rupture of cerebral blood vessels. The cerebral infarction is a transient ischemic (cerebral infarction in cardiogenic embolism), small vessel disease, or lacunar infarction (cerebral infarction in large vessel disease) do.

Stroke occurred in 3% to 5% of cases within 48 hours after transient ischemic attack (TIA), and 23% of patients with lacunar infarction were reported to have acute neurological symptoms within 48 hours.

Because thrombolysis using TPA (Tissue Plasminogen Activator) is the standard treatment for stroke, the early detection and early treatment of strokes will be better as it progresses early (preferably within 3 hours) after stroke. It is the most important principle in the treatment of stroke.

Stroke during sleep appears as a proportion of 6.4% to 20% of the total stroke. Such a stroke during sleep does not only cause the patient to feel symptoms, but it is also difficult for people around him to detect such a stroke, There is a difficulty in not being able to detect the occurrence of the disease.

Therefore, there is a problem that early thrombolysis can not be performed due to difficulty in early detection of stroke, and the prognosis of stroke symptoms is further deteriorated.

Therefore, there is a need to monitor the human body during sleep to enable early judgment of stroke occurring during sleep.

It is an object of the present invention to provide an apparatus and method for determining a stroke in a sleep, which determines whether or not a user is suffering from a stroke, by acquiring motion data from a user who is asleep and confirming whether or not a user has experienced hemiplegic symptoms .

Also, it is possible to analyze motion data obtained from a sleeping user using a sliding window technique, and adjust the sliding window time depending on whether the user is sleeping (REM), so that a sleeping user can be promptly judged whether or not a stroke occurs. The present invention relates to a stroke determination device and method.

The apparatus for determining a stroke in a sleep according to an embodiment of the present invention includes a limit threshold setting unit for setting a motion limit point and a stroke limit point, a motion data obtaining unit arranged on the body part of the user for collecting motion data of the user, A feature extraction unit for calculating motion data of at least one of a movement distance and a rotation distance per preset time and comparing the motion data with a motion limit point to extract motion feature data of the user, And a real-time underwater sleep stroke judgment unit for comparing the data with the stroke limit threshold to determine whether the user has a stroke.

According to another aspect of the present invention, there is provided a method for determining a stroke in a sleep, comprising: preparing a motion limit point and stroke limit point information; acquiring motion data from a predetermined body part; A second step of calculating motion data, comparing the motion data with a motion limit point to extract motion feature data, and a second step of monitoring the motion feature data and comparing the monitored motion feature data with a stroke limit point to determine whether stroke has occurred And three stages.

The apparatus and method for determining a stroke in the sleep according to the present invention determine whether a stroke occurs from a user's hemiplegic symptoms occurring during sleep using a left to right ratio obtained by analyzing motion data obtained from a body part that is bilaterally symmetrical, There is an effect that is possible.

In analyzing the motion data resulting from processing the motion data, the motion data is classified into noise and motion feature data by using the motion limit data that is the classification reference of noise and motion feature data included in the motion data, The effect of the noise is excluded, and the motion feature data and the stroke threshold are compared with each other to improve the reliability of judging whether the hemiplegic symptom occurs or not.

By selecting stroke thresholds from normal (including users with no stroke symptoms) and stroke data from stroke patients with hemiplegic symptoms, it is possible to improve the reliability of stroke threshold selection for the occurrence of hemiplegic symptoms.

In extracting motion feature data, a sliding window technique is used. When the user is in the RAM sleep state, information on whether or not the user is in the RAM sleep state is used to exclude the RAM sleep interval in the sliding window section for extracting motion characteristic data Thereby making it possible to make a more accurate judgment as to whether the user is suffering from hemiplegic symptoms or stroke.

If the analysis of the user's motion data indicates that the sleeping period of the RAM is sustained for more than the predetermined time, there is an additional effect of enabling quick action in an emergency by providing an alarm notifying the dangerous situation.

The effects of the present invention are not limited to those mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the following description.

1 is a block diagram illustrating a device for determining a stroke during a sleep according to an embodiment of the present invention;
BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a device for determining a stroke in a sleep,
FIG. 3 is a block diagram showing a real-time sleep determining unit in a sleep determining apparatus according to an embodiment of the present invention. FIG.
4 is a block diagram illustrating a motion limit point and a stroke limit point selection process according to an embodiment of the present invention.
5 is a block diagram illustrating a device for determining a stroke during a sleep according to another embodiment of the present invention.
FIG. 6 is a block diagram showing a RAM sleep surface confirmation unit in the feature extraction unit of a stroke determination apparatus according to another embodiment of the present invention; FIG.
7 is a table showing motion feature data extracted by the feature extracting unit according to the embodiment of the present invention.
8 is a graph illustrating extraction of motion characteristic data using a sliding window technique according to an embodiment of the present invention.
FIG. 9 is a graph showing box-plotted left and right ratios using a Ulecridian distance value between a stroke patient and a normal person having hemiplegic symptoms according to an embodiment of the present invention.
FIG. 10 is a graph showing box-plotted left and right ratios based on rotation distance values of a stroke patient and a normal person having hemiplegic symptoms according to an embodiment of the present invention.
11 is a diagram illustrating stroke threshold selection according to an embodiment of the present invention.
12 is a graph showing an ROC curve for selecting a stroke threshold according to an embodiment of the present invention.
13 is a diagram illustrating TPR (True Positive Rate) and False Positive Rate (FPR) calculations for stroke threshold selection according to an embodiment of the present invention.
14 is a graph showing an ROC curve of an Euclidean distance value according to a sliding window parameter according to an embodiment of the present invention.
15 is a graph showing an ROC curve of a rotation distance value according to a sliding window parameter according to an embodiment of the present invention.
16 is a graph showing ROC curves of cumulative data (PIM, Proportional Integral Mode) of Euclidean distance values according to an embodiment of the present invention.
17 is a graph showing the ROC curve of the cumulative data (PIM, Proportional Integral Mode) of the rotation distance value according to the embodiment of the present invention.
18 is a graph showing TPR and FPR for cumulative data of Euclidean distance values and rotation distance values at an optimal point in accordance with an embodiment of the present invention.
FIG. 19 is a graph showing TPR for each stroke patient according to an embodiment of the present invention. FIG.
FIG. 20 is a table showing states of stroke patients of hemiplegia shown in FIG. 19; FIG.
FIG. 21 is a flowchart illustrating a method of determining a stroke during a sleep according to an embodiment of the present invention. FIG.
FIG. 22 and FIG. 23 illustrate an example of a device for determining a stroke in a sleeping state according to an embodiment of the present invention and a configuration of a device for determining a stroke during a sleep. FIG.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings.

In order to facilitate understanding of the present invention, in order to facilitate understanding of the present invention, the motion information of a user obtained by using a sensor disposed on a body part of a user is defined as motion data, the motion data is processed, Data.

In addition, a threshold, which is a reference for classifying noise and significant motion feature data, is defined as a motion limit point in the motion data, and a threshold value for determining whether the stroke has occurred using the motion feature data is defined as Stroke threshold.

According to the embodiment of the present invention, it is possible to determine whether the stroke is caused through the confirmation of the hemiplegic symptom. Therefore, since the motion data is obtained from the left and right symmetrical parts of the user, the motion data processed and the motion data Motion characteristic data as meaningful data, rather than noise, is used. At this time, the ratio of the left motion characteristic data and the right motion characteristic data obtained from the left and right body parts (more precisely, a result value obtained by taking absolute values with respect to the log values of the left and right sliding window values, which will be described later) .

1 is a block diagram illustrating a device for determining a stroke during a sleep according to an embodiment of the present invention.

As shown in FIG. 1, the apparatus includes a motion data acquisition unit 100 that is disposed on a body part of a user and collects motion data of a user, A feature extraction unit 200 for extracting motion feature data of a user by comparing the motion data with a motion limit point, and a feature extraction unit 200 for monitoring motion feature data, comparing motion feature data with a stroke threshold, And a real-time underwater sleep determination unit 400 for determining whether or not the subject is a sleeping person.

The movement limit point and the stroke limit point are selected by the limit point selection unit 300. The selected movement limit point and stroke limit point are transmitted to the feature extraction unit 200 and the stroke determination unit 420 respectively, The stroke determination unit 420 extracts the motion data using the limit point, and the stroke determination unit 420 determines whether the stroke is caused by using the stroke limit point.

According to an embodiment of the present invention, it is determined whether or not the user has developed hemiplegia symptoms and the stroke is finally determined. Thus, the motion data obtaining unit 100 obtains the motion data from the left and right symmetrical body parts And an excelometer and a gyro sensor fixedly disposed on the body part of the user. The motion data acquisition unit 100 may include any one of an excel meter and a gyro sensor, and may include a gyro sensor, It is also possible to construct. According to an embodiment of the present invention, the sampling rate of the excel meter and the gyro sensor is preferably 100 Hz.

The feature extraction unit 200 according to an exemplary embodiment of the present invention classifies motion data less than the motion limit point into noise and extracts motion data that is equal to or larger than the motion limit point as motion feature data of the user, A feature extraction unit 210, a second feature extraction unit 220, and a third feature extraction unit 230.

The first feature extraction unit 210 processes the motion data acquired by the motion data acquisition unit 100 and calculates the movement distance of the data acquired by the three axis axis meter sensor per predetermined time (for example, 0.01 second) ED, and Euclidean Distance) (in this case, it is preferable to use the gravity component as the removed component), and the rotation distance (RD, Rotation The motion data corresponding to the motion vector is extracted.

The second feature extraction unit 220 calculates the movement distance motion data and the rotation distance motion data calculated by the first feature extraction unit 210 using cumulative data (PIM, Proportional Integral Mode).

The tertiary feature extraction unit 230 analyzes the ratio of the motion during sleep of the two arms in real time while moving the window of 30 minutes or 60 minutes for 5 minutes or 10 minutes in the sliding window period in the embodiment of the present invention 8 motion feature data as shown in the table shown in FIG. 7 by using the data interval in which the cumulative data per second calculated by the second feature extracting unit 220 is equal to or greater than the motion limit point. In this case, ED and RD are the results extracted by the first feature extracting unit 210, and ED_PIM and RD_PIM are the result values extracted by the second feature extracting unit 220.

8 is a graph showing motion feature data extraction using the sliding window technique according to an embodiment of the present invention. The RD_PIM (accumulated value per second of RD) of the tertiary feature extraction unit 230 according to an embodiment of the present invention (Using the sliding window technique based on the ED_PIM is applied in the same process). In this example, the sliding window technique is used to extract eight motion feature data. The upper graph of FIG. 8 shows accumulated values per second of RD obtained from the left body part during sleep, and the lower graph of FIG. 8 shows the accumulated values per second of RD obtained from the right body part. The box (window) shown in Fig. 8 extracts the left-to-right ratio for the eight motion feature data while moving.

The limit point selection unit 300 selects a motion limit point that is a classification criterion of noise and motion feature data and acquires a motion limit point that is obtained from a normal person (when a user selects a limit point using motion data acquired from a user, The motion characteristic data and the motion characteristic data obtained from stroke patients with hemiplegic symptoms are compared to determine whether the True Positive Rate indicated by the ROC (Receiver Operating Characteristic) curve calculation result for the motion characteristic data of the normal person and the hemiplegic stroke patient is Select a stroke threshold to be at or above the reference value (eg 90%).

4, the threshold point selecting unit 300 according to an exemplary embodiment of the present invention includes a motion limit point checking unit 310 and a stroke threshold checking unit 320, The feature extraction unit 200a calculates motion feature data from the motion data, and the real-time monitoring unit 410a acquires the motion feature data of the normal person And transmits the result of the monitoring to the movement limit point confirmation unit 310. In addition, motion data is obtained from the excelometer and gyro sensors 130 and 140 disposed on the left and right body parts of a stroke patient of hemiplegic stroke, the feature extraction unit 200 calculates motion feature data from motion data, The monitoring unit 410 transmits the result of monitoring the motion characteristic data of the normal person to the movement limit point checking unit 310. [ In this case, the operation of the excelometer and gyro sensors 130 and 140 and the feature extraction unit 200 for obtaining motion data from a normal person and a stroke patient of hemiplegia are as described above.

The motion limit point checking unit 310 according to an embodiment of the present invention selects a motion limit point that is a classification criterion of noise and motion feature data included in the obtained motion data. In this case, the motion limit point confirmation unit 310 is a premise for accurately determining whether or not the person is experiencing hemiplegic symptoms from the motion feature data. The motion limit point confirmation unit 310 may be configured to exclude the influence of noise included in the obtained motion data And the data is defined as motion characteristic data is as described above).

FIG. 9 is a graph showing box-plotted left and right ratios using a Euclidean distance value (ED) between a stroke patient and a normal person having hemiplegic symptoms according to an embodiment of the present invention. (Left) and right (right) ratios based on the rotation distance values (RD) of a stroke patient and a normal person with symptoms.

Referring to FIGS. 9 and 10, it can be seen that the difference in the left-right ratio in the sleep of the stroke patients having hemiplegic symptoms is larger than that in the normal subjects. The movement limit point checking unit 310 according to an embodiment of the present invention selects a movement limit point as a threshold value at which the normal person and the stroke patient can be distinguished from each other.

11 to 19, the stroke limit point confirmation unit 320 according to the embodiment of the present invention determines whether or not the hit probability represented by the ROC (Receiver Operating Characteristic) curve calculation result for the motion characteristic data of the normal person and the stroke patient of the hemiplegia (True Positive Rate) is greater than or equal to the reference value (eg, 90%).

FIG. 11 is a diagram illustrating stroke threshold selection according to an embodiment of the present invention. The stroke threshold check unit 320 generates an ROC curve as the stroke threshold changes. FIG. The stroke threshold check unit 320 determines whether a stroke is caused by a false positive rate (FPR), which is a ratio of a normal person to a stroke, and a TPR True Positive Rate) from the minimum to the maximum of the box plot. 13, when a stroke patient is defined as S (stroke), a normal person as N (normal), a person expected to be a stroke as S '(Predict Stroke), and a person expected to be normal as N' TPR and FPR are calculated using the following [Equation 1] and [Equation 2].

[Equation 1]

[Equation 2]

FIG. 14 is a graph showing an ROC curve of an Euclidean distance value (ED) according to a sliding window parameter according to an embodiment of the present invention, and FIG. FIG. 15 is a graph showing an ROC curve of a rotation distance value RD according to a sliding window parameter according to an embodiment of the present invention.

FIG. 16 is a graph showing ROC curves of cumulative data (PIM, Proportional Integral Mode) of Euclidean distance values ED according to an embodiment of the present invention, and FIG. (RIM) curve of the cumulative data (PIM, Proportional Integral Mode) of the heartbeat RD. The stroke limit point confirmation unit 320 determines the optimal stroke end point among the data obtained from the excel meter and the gyro sensor Data is extracted. As the window time and the sliding window time become larger, it is found that the data is the optimal feature data drawn in the upper left corner. In this case, when the cut-off value is determined, it may vary depending on the applied system. For example, if the optimal characteristic data is selected when the false positive is 10%, the window time vs. sliding window time is 60/10 .

In the apparatus for determining a stroke in a sleep according to an embodiment of the present invention, for example, an optimal TPR and an FPR are confirmed using experimental data. As a result, as shown in FIG. 18, Detection accuracy of 75.48% was confirmed from ED_PIM_sum motion feature data using Dian distance value and 97.12% from RD_PIM_sum motion feature data using rotation distance value obtained from gyro sensor. FIG. 19 is a graph showing TPR of four stroke patients according to an embodiment of the present invention, wherein 1 of the stroke detection rates shown in FIG. 19 means 100%. According to the condition of each patient, the detection probability is higher than 96% by using the data obtained from the excelometer and gyro sensor from the severe stroke patients. As a result, It is possible to secure a higher detection probability and improve the reliability in practice.

The real-time sleep determining unit 400 according to an embodiment of the present invention includes a real-time monitoring unit 410 for calculating left-to-right ratio, which is a ratio of motion characteristic data obtained from left and right symmetrical body parts, And a real-time stroke determination unit 420 for determining an occurrence of a stroke of the user and providing an alarm when the left-right non-operation result of the unit 410 is equal to or greater than the stroke threshold.

As shown in FIG. 3, the real-time monitoring unit 410 according to an embodiment of the present invention includes a left-right ratio comparing unit 403 that calculates left and right sliding window values by a symmetric comparison method and displays left and right ratios. In this case, the left-right ratio is calculated using the following equation (3).

[Equation 3]

| log (Left sliding window / Right sliding window) |

That is, the real-time monitoring unit 410 calculates the ratio of the sliding window value 401 of the motion characteristic data acquired from the left body part to the sliding window value 402 of the motion characteristic data acquired from the right body part, And calculates the absolute value of the absolute value with the left / right ratio.

The stroke determination unit 420 according to an embodiment of the present invention may calculate the left and right ratio as a hemiparesis symptom when the absolute value (left / right ratio) calculated by the left / right ratio comparison unit 403 of the real time monitoring unit 410 is greater than the stroke threshold And by providing an alarm by judging that the stroke is a stroke, it is possible to take prompt action to the user who has the stroke.

FIG. 22 is a diagram showing an example in which a device for determining a stroke during a sleep according to an embodiment of the present invention is fixedly disposed on a body (in the form of a band wound on a wrist), and FIG. 23 is a diagram Fig.

5, the apparatus for determining a stroke in the sleeping according to another embodiment of the present invention may be added to the apparatus for determining a stroke in the sleeping according to the embodiment of the present invention described above, And a sleep surface confirmation unit 50 for generating sleep surface information of the user whether or not the user is in the sleep state.

REM sleep is a period of rapid eye movement during several stages of sleep, suggesting that brain activity is occurring fairly, and when the voltage is low and rapid irregular waves appear, the positional muscles are more relaxed than other sleep phases .

The feature extraction unit 20 according to another embodiment of the present invention excludes the RAM sleep interval in the sliding window interval for extracting the motion characteristic data when the user has the REM level (interval in which there is no movement of both arms) from the sleeping information .

FIG. 6 is a block diagram showing a feature extraction unit 20 according to another embodiment of the present invention. The REM sleeping confirmation unit 50 corrects information about a RAM sleeping interval in which both arms of the user are not moved at all. And transmits it to the difference feature extraction unit 33. The modified tertiary feature extraction unit 33 uses the non-RAM sleep interval excluding the RAM sleep interval as a monitoring time interval and reduces the window slide time to 30 minutes or less, so that even if the error occurs in determining whether the stroke has occurred The data obtained from the RAM sleep interval is excluded, so that the data analysis and the determination of the onset of the stroke can be performed more quickly.

The RAM sleeping confirmation unit 50 according to another embodiment of the present invention provides an alarm to the stroke determination unit 42 when it is determined that there is no movement of the user for a preset time (e.g., one hour) And transmits a command signal.

FIG. 21 is a flowchart illustrating a stroke determination method according to another embodiment of the present invention. The stroke determination method includes a first step (S100) of acquiring motion data from a predetermined body part, a first step A second step S200 of selecting a motion limit point and a stroke limit point using the motion data acquired in step S100, and motion data obtained in the first step S100 are processed to calculate motion data, The motion feature data extracted in the third step S300 and the third step S300 of extracting the motion feature data are compared with the motion limit point selected in the second step S200, S200) to determine whether the stroke has occurred or not (S400).

At this time, the first step S100 acquires motion data using at least one of a normal stroke patient having hemiplegic symptoms and an excelometer and a gyro sensor disposed on a predetermined symmetrical body part of the user. In the first step (S100), it is preferable to acquire motion data for a population of stroke patients having a normal person, hemiplegic symptom, similar to the age, sex, age range, and possession disease of a user. In this case, So that a more reliable reference can be secured.

In a second step S200 according to another embodiment of the present invention, a motion limit point serving as a classification criterion of noise and motion characteristic data among the motion data is selected, and a motion limit data, which is a reference value for determining whether the motion characteristic data indicates the hemiplegic symptoms Stroke limit points are selected. In this case, the stroke limit point at which TP (True Positive) becomes equal to or greater than a predetermined value (for example, 90%) according to the calculation result of the ROC curve using the left-right ratio is as described above.

In a third step S300 according to another embodiment of the present invention, the motion data of the user is processed to calculate motion data of at least one of the movement distance ED and the rotation distance RD per preset time, In operation S400, the right-to-left ratio, which is a ratio of the motion feature data acquired from the left and right symmetrical body parts of the user, is calculated. If the right and left ratio is equal to or greater than the stroke threshold, Lt; / RTI > In this case, the left-right ratio calculated in the fourth step is calculated by using the above-mentioned equation (3).

The method of determining stroke during sleep according to another embodiment of the present invention further includes a RAM sleeping step of checking whether the user is in a sleep state using a motion data acquired from a user. By further including the RAM sleep confirmation step, it is possible to adjust the sliding window time, and more accurate stroke judgment can be performed by excluding the data in the RAM sleep interval from the data obtained for the confirmation of the hemiplegic symptom and the determination of the stroke occurrence . That is, in the third step, when the user is confirmed to be in the RAM sleep state in the RAM sleep state confirmation step, the sleep state is excluded from the sliding window section for extracting the motion characteristic data, and the RAM sleep confirmation step uses the motion data calculated in the third step And transmits a signal to provide a stroke onset alert when it is determined that the user does not move for a predetermined time (e.g., one hour).

The method of determining a stroke in a sleeping according to another embodiment of the present invention includes the steps of preparing a motion limit point and stroke limit point information, a first step of acquiring motion data from a predetermined body part, A second step of extracting motion feature data by comparing the motion data with a motion limit point, and a second step of monitoring motion feature data extracted in the second step and comparing the monitored motion feature data with the stroke limit point And a third step of determining whether or not a stroke has occurred.

The embodiments of the present invention have been described above. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. Therefore, the disclosed embodiments should be considered in an illustrative rather than a restrictive sense. The scope of the present invention is defined by the appended claims rather than by the foregoing description, and all differences within the scope of equivalents thereof should be construed as being included in the present invention.

10: motion data acquisition unit 20: feature extraction unit
30: Limit point selection unit 31: Primary feature extraction unit
32: Secondary feature extraction unit 33: Quadratic feature extraction unit
40: real-time sleep-stroke determining part 41: real-time monitoring part
42: stroke judgment unit 50: RAM sleep confirmation unit
100: motion data acquisition unit
110, 120, 130, 140: an excel meter and a gyro sensor
200: feature extraction unit 210: primary feature extraction unit
220: Secondary feature extraction unit 230: Third feature extraction unit
300: Limit point selection unit 310: Motion limit point confirmation unit
320: Stroke limit point confirmation unit
400: real-time sleep-stroke judgment unit
401, 402: Sliding window 403:
410: real-time monitoring unit 420: stroke determination unit

Claims (17)

A motion data acquisition unit arranged on the body part of the user and collecting motion data of the user;
A feature extraction unit for calculating motion data of at least one of a movement distance and a rotation distance per predetermined time by processing the motion data and extracting the motion feature data of the user by comparing the motion data with a motion limit point; And
Wherein the motion feature data is monitored and the motion feature data is compared with a stroke threshold to determine whether the user has stroke,
Wherein the stroke determination device comprises:
The method according to claim 1,
Wherein the motion limit point is selected as a reference point serving as a classification criterion for excluding noise from the motion data, and the stroke threshold is defined as a hit probability represented by a ROC (Receiver Operating Characteristic) curve calculation result on motion characteristic data of a normal person and a stroke victim of a hemiplegic stroke (True Positive Rate) is selected to be equal to or greater than the reference value
A device for determining stroke during sleep.
The apparatus of claim 1, wherein the motion data obtaining unit
And an excelrometer and a gyro sensor fixedly arranged on a body part of the user which is a predetermined left and right symmetrical body part
A device for determining stroke during sleep.
The apparatus of claim 1, wherein the feature extraction unit
Classifying the motion data less than the motion limit point into noise and extracting motion data not less than the motion limit point as the motion feature data of the user
A device for determining stroke during sleep.
2. The apparatus according to claim 1, wherein the real-
A real time monitoring unit for calculating a left to right ratio, which is a ratio of motion feature data obtained from the left and right symmetrical body parts, and transmitting the result; And
And a real-time stroke determination unit for determining an occurrence of a stroke of the user and providing an alarm when the left-right non-operation result of the real-time monitoring unit is greater than or equal to the stroke threshold
A device for determining stroke during sleep.
6. The apparatus of claim 5, wherein the real-
Calculating the absolute value of the logarithmic value of the ratio of the sliding window value of the motion characteristic data obtained from the left body part to the sliding window value of the motion characteristic data obtained from the right body part with the left to right ratio
A device for determining stroke during sleep.
6. The method of claim 5,
A RAM sleep confirmation unit for receiving the motion data calculated by the feature extraction unit and generating sleep information of the user as to whether the user is in a RAM sleep state,
Further comprising: means for determining a stroke of the subject.
8. The apparatus of claim 7, wherein the feature extraction unit
If the user is determined to be in the RAM sleep state from the sleep information, exclude the section in the RAM sleep state from the sliding window section for extracting the motion feature data
A device for determining stroke during sleep.
8. The apparatus according to claim 7, wherein the RAM sleep confirmation unit
And transmitting an alarm providing command signal to the stroke determining unit when it is determined that the user does not move for a predetermined time or longer from the inputted motion data
A device for determining stroke during sleep.
Preparing a movement limit point and stroke limit point information;
A first step of obtaining motion data from a predetermined human body part;
A second step of processing the motion data acquired in the first step to calculate motion data, and comparing the motion data with the motion limit point to extract motion feature data; And
A third step of monitoring the motion feature data extracted in the second step and comparing the monitored motion feature data with the stroke threshold to determine whether or not stroke has occurred;
Wherein the at least one stroke is selected from the group consisting of:
11. The method of claim 10, wherein the first step
Obtaining the motion data using at least one of a normal person, a stroke patient having hemiplegic symptoms, and an excelometer and a gyro sensor disposed on a predetermined symmetrical body part of a user
How to Determine Stroke During Sleep.
11. The method of claim 10, wherein preparing the motion limit point and stroke limit point information comprises:
Preparing a motion limit point that is selected to be a classification criterion of noise and motion feature data among the motion data and preparing a stroke limit point that is a reference value for determining whether or not the motion feature data indicates a hemiplegic symptom
How to Determine Stroke During Sleep.
12. The method of claim 11, wherein the second step comprises:
Processing the motion data of the user to calculate motion data of at least one of a moving distance and a rotation distance per preset time and extracting motion data of the motion limit point or more with the motion feature data
How to Determine Stroke During Sleep.
14. The method of claim 13, wherein the third step comprises:
Calculating a left to right ratio, which is a ratio of motion feature data acquired from the left and right symmetrical body parts of the user, and providing an alarm for the stroke occurrence of the user when the right and left ratio is equal to or greater than the stroke threshold
How to Determine Stroke During Sleep.
15. The method of claim 14,
A RAM sleep confirmation step of checking whether the user is in the RAM sleep state using the motion data acquired from the user
Further comprising the step of determining a stroke in the sleep.
16. The method of claim 15, wherein the second step comprises:
If the user is confirmed to be in the RAM sleep state in the RAM sleep confirmation step, exclude the section that is in the RAM sleep state in the sliding window section for extracting the motion feature data
How to Determine Stroke During Sleep.
16. The method according to claim 15, wherein the RAM sleep confirmation step
A step of confirming whether or not the motion of the user is generated using the motion data calculated in the third step and transmitting a signal of providing a stroke onset alert when it is confirmed that the user does not move for a predetermined time or more
How to Determine Stroke During Sleep.

Images