Method for determining a person's sleeping phase which is favourable for waking up

Abstract

A pulse wave signal is registered and an occurrence of human limb movements detected during sleep using a pulse wave sensor and an accelerometer. The values of RR intervals and respiratory rate are measured at preset time intervals Δt.sub.i based on pulse wave signal. Mean P.sub.1, minimal P.sub.2, and maximal P.sub.3 values of RR intervals, the standard deviation of RR intervals P.sub.4, average respiratory rate P.sub.5 and average number of limb movements P.sub.6 are determined based on the above measured values. Function value F(Δt.sub.i) is determined thereafter as:
F(Δt.sub.i)=−K.sub.1P.sub.1−K.sub.2P.sub.2−K.sub.3P.sub.3+K.sub.4P.sub.4+K.sub.5P.sub.5+K.sub.6P.sub.6,
where K.sub.1-K.sub.6 are weight coefficients characterizing the contribution of the corresponding parameter to function value F(Δt.sub.i); whereat the onset and termination of sleep phase favorable to awakening is determined by increments of function F(Δt.sub.i).

Claims

1. A method for determining a sleep phase favorable for awakening a person from a sleep, the method comprising: attaching a motion sensor to an arm or leg of the person, the motion sensor being responsive to movements of the arm or leg; attaching a pulse wave sensor to a wrist or a forearm of the person, the pulse wave sensor being responsive to vascular blood filling of an area of the wrist or forearm; providing a measuring unit to which the motion sensor and the pulse wave sensor are connected; registering by the measuring unit signals from the pulse wave sensor and signals from the motion sensor; sending the signals from the pulse wave sensor and the signals from the motion sensor registered by the measuring unit to a CPU, operably associated with the measuring unit, for processing; calculating, by the CPU, values of RR intervals and a respiratory rate of the person, wherein RR intervals represent intervals between successive heartbeats, using the signals from the pulse wave sensor registered by the measuring unit; calculating, by the CPU, a number of movements of the arm or leg of the person using the signals from the motion sensor registered by the measuring unit; determining, by the CPU, a mean value P.sub.1 of the RR intervals over a time interval Δt.sub.i, wherein i is a serial number of the time interval Δt.sub.i; determining, by the CPU, a minimal value P.sub.2 of the RR intervals over the time interval Δt.sub.i; determining, by the CPU, a maximal value P.sub.3 of the RR intervals over the time interval Δt.sub.i; determining, by the CPU, a standard deviation value P.sub.4 of the RR intervals over a preceding time interval of 3 to 20 min; determining, by the CPU, a mean value P.sub.5 of the respiratory rate over the time interval Δt.sub.i; determining, by the CPU, an average number P.sub.6 of the arm or leg movements over a preceding time interval ranging from 0.5 to 10 minutes; evaluating, by the CPU, function values F(Δt.sub.i) over preset time intervals Δt.sub.i, wherein:
F(Δt.sub.i)=K.sub.1P.sub.1−K.sub.2P.sub.2−K.sub.3P.sub.3+K.sub.4P.sub.4+K.sub.5P.sub.5+K.sub.6P.sub.6, and wherein K.sub.1-K.sub.6 are weight coefficients characterizing contribution of parameters P.sub.1-P.sub.6 to the values of the function F(Δt.sub.i); evaluating, by the CPU, increment values of the function F(Δt.sub.i) over the time intervals Δt.sub.i; comparing, by the CPU, the increment values of the function F(Δt.sub.i) over the time intervals Δt.sub.i with a preset threshold value; determining, by the CPU, onset and/or termination of a sleep phase favorable for awakening a person from a sleep based on comparison between the increment values of the function F(Δt.sub.i) over the time intervals Δt.sub.i and a preset threshold value; outputting a signal by the CPU during the sleep phase favorable for awakening a person from a sleep to a vibrator based on comparison between the increment values of the function F(Δt.sub.i) over the time intervals Δt.sub.i and a preset threshold value, and generating by the vibrator a wake-up signal based on the signal outputted by the CPU during the sleep phase determined as favorable for awakening a person from a sleep based on comparison between the increment values of the function F(Δt.sub.i) over the time intervals Δt.sub.i and a preset threshold value.

2. The method of claim 1, further comprising selecting the time interval over which the value of P.sub.4 is calculated in a range from 4 to 6 minutes.

3. The method of claim 1, further comprising selecting the time interval over which the number P.sub.6 is calculated in a range from 4 to 6 minutes.

4. The method of claim 1, wherein: a value of K.sub.1 is selected in a range from 0.6 to 3 ms.sup.−1; a value of K.sub.2 is selected in a range of 0.1 to 0.7 ms.sup.−1; a value of K.sub.3 is selected in a range of from 0.01 to 0.3 ms.sup.−1; a value of K.sub.4 is selected in a range from 0.5 to 3 ms.sup.−1; a value of K.sub.5 is selected in a range from 1 to 10 min; and a value of K.sub.6 is selected in a range from 5 to 50.

5. The method of claim 4, further comprising selecting the value of K.sub.1 in a range from 0.9 to 1.05 ms.sup.−1.

6. The method of claim 4, further comprising selecting the value of K.sub.2 in a range from 0.1 to 0.2 ms.sup.−1.

7. The method of claim 4, further comprising selecting the value of K.sub.3 in a range from 0.02 to 0.05 ms.sup.−1.

8. The method of claim 4, further comprising selecting the value of K.sub.4 in a range from 1.3 to 1.5 ms.sup.−1.

9. The method of claim 4, further comprising selecting the value of K.sub.5 in a range from 1.5 to 2.3 min.

10. The method of claim 4, further comprising selecting the value of K.sub.6 in a range from 18 to 24.

11. The method of claim 1, wherein the pulse wave sensor-comprises a piezoelectric sensor, a strain gauge, or an optical sensor.

12. The method of claim 1, wherein the motion sensor comprises an accelerometer.

13. The method of claim 1, wherein the time intervals Δt.sub.i are selected in a range from 1 to 6 minutes.

14. The method of claim 1, further comprising identifying the onset of a sleep phase favorable for awakening a person from a sleep if the increment of the function F(Δt.sub.i) over the time interval Δt.sub.i exceeds the preset threshold value.

15. The method of claim 1, further comprising identifying the termination of a sleep phase favorable for awakening a person from a sleep if the increment of the function F(Δt.sub.i) over the time interval Δt.sub.i is smaller than the preset threshold value.

Description

BRIEF DESCRIPTION OF DRAWINGS

(1) The invention is illustrated by the following graphic materials:

(2) FIG. 1 shows an example of identifying REM sleep phase for one of the test subjects (8VAV), whereat FIG. 1a shows a graph of function F(Δt.sub.i) for one of the registered REM phases, while FIG. 1b shows a graph ΔF(Δt.sub.i) of function increment F(Δt.sub.i), shown in FIG. 1a;

(3) FIG. 2 shows a graph of function F(Δt.sub.i) over the entire sleep duration for the same test subject (8VAV) whose sleep is illustrated in FIG. 1, wherein the graph fragment shown in more detail in FIG. 1a is circled;

(4) FIG. 3 shows a graph of function F(Δt.sub.i) over the entire sleep duration for another test subject (7ESA);

(5) FIG. 4 shows a graph of function F(Δt.sub.i) over the entire sleep duration for yet another test subject (3SOR); and

(6) FIG. 5 and FIG. 6 schematically show the design of an exemplary portable device made in the form of a bracelet with sensors that implements the method in accordance with the present invention, whereat FIG. 5 gives the view of the device from its inner side contacting the wrist, and FIG. 6 shows the device from the outside, where the indicator is located.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

(7) A method for determining the sleep phase favorable to awakening can be implemented using two sensors: a pulse wave sensor and a sensor capable of responding to arm or leg movement, i.e., a motion sensor such as an accelerometer. The sensors can be mounted on a human body separately from each other. For example, the motion sensor can be attached to an arm or a leg, while the pulse wave sensor onto the wrist or forearm. Pulse wave sensors may be represented by piezoelectric sensors, strain gages, and optical sensors. The use of an optical sensor or photoplethysmographic sensor sensitive to vascular blood filling of bodily areas is preferable. It is more convenient for the user if both pulse wave sensor and motion sensor are mounted in a single device, such as shown in FIG. 5 and FIG. 6 and made in the form of bracelet 1 to be worn on the wrist.

(8) As shown in FIG. 5, the inner side of bracelet 1 carries pulse wave sensor 2 based, for example, on piezoelectric cell. Several pulse sensors may be used to ensure a reliable skin contact with the wrist area where pulse wave signal is detected. Bracelet 1 (see FIG. 6) may have indicator 3 which displays the initial settings and operation mode of the device. The device may also generate a wake-up signal during favorable sleep phase, for example, by means of a vibrator (not shown in the drawings) mounted in bracelet 1. An accelerometer (not shown in the drawings) may be mounted inside bracelet 1 for detecting arm movements of a sleeping person. Pulse wave sensor 2 and the accelerometer are connected to the measuring unit of bracelet 1, which registers pulse wave signals and accelerometer-generated signals. The registered signals are processed in a CPU which can be co-located with the measuring unit in bracelet 1 or made as a separate unit to be attached to human body or carried by person, whereat said CPU receives signals transmitted from the measuring unit by radio or some other means.

(9) The values of RR intervals and respiratory rate are determined in human sleep based on registered pulse wave signal. Since a pulse wave signal is a periodic signal that varies in synchronism with heartbeat, the time intervals between any characteristic points on pulsogram (e.g., peak value of the signal or its derivative) correspond exactly to RR intervals. Instrumental methods for determining heart rate or RR intervals from a pulse wave signal are well known to those skilled in the art. It is also known that, alongside with the above-mentioned periodic variations corresponding to blood filling dynamics at each cardiac cycle, pulse wave signal includes a low frequency component corresponding to respiratory cycle. Instrumental methods of determining the respiratory rate based on low-pass filtering of respiratory component out of pulse wave signal are well known to those skilled in the art.

(10) Thereafter, using the obtained data, i.e., values of RR intervals and respiration rate, the following parameters are periodically measured at in preset time intervals Δt.sub.i:

(11) P.sub.1—the mean value of RR intervals;

(12) P.sub.2—the minimum value of RR intervals;

(13) P.sub.3—the maximum value of RR intervals;

(14) P.sub.5—the mean respiratory rate.

(15) The time interval Δt.sub.i over which said parameters are measured is selected in the range from 1 minute to 6 minutes. Here, i is the serial number of i-th time interval.

(16) Furthermore, parameter P.sub.4 is determined as the standard deviation of RR intervals over the preceding time interval of 3 minutes to 20 minutes, preferably from 4 minutes to 6 minutes.

(17) The mean number of limb movements P.sub.6 over the preceding time interval from 0.5 minutes to 10 minutes, preferably from 4 minutes to 6 minutes, is another parameter needed for final identification of REM sleep phase. Since the occurrence of motor activity is informative by itself for identification of REM sleep, all limb movements detected by accelerometer over a 10 seconds period are taken for one movement.

(18) Thereafter, function value F(Δt.sub.i) is determined by formula:
F(Δt.sub.i)=−K.sub.1P.sub.1−K.sub.2P.sub.2−K.sub.3P.sub.3+K.sub.4P.sub.4+K.sub.5P.sub.5K.sub.6P.sub.6,

(19) where: K.sub.1-K.sub.6 are weight coefficients characterizing the contribution of corresponding parameter P.sub.1-P.sub.6 to the value of F(Δt.sub.i).

(20) Table 1 below shows the value ranges of weight coefficients K.sub.1-K.sub.6, as well optimal value thereof.

(21) TABLE-US-00001 TABLE 1 Weight Coefficient Values Parameters, Weight coefficients units of Weight coefficient values measurement Designation min max optimal P.sub.1, ms K.sub.1 0.6 ms.sup.1 3 ms.sup.1 1 ms.sup.1 P.sub.2, ms K.sub.2 0.1 ms.sup.1 0.7 ms.sup.1 0.14 ms.sup.1 P.sub.3, ms K.sub.3 0.01 ms.sup.1 0.3 ms.sup.1 0.03 ms.sup.1 P.sub.4, ms K.sub.4 0.5 ms.sup.1 3 ms.sup.1 1.4 ms.sup.1 P.sub.5, min.sup.−1 K.sub.5 1 min 10 min 2 min P.sub.6 K.sub.6 5 50 22

(22) Informative parameters P.sub.1-P.sub.6 were established, and their weight coefficients K.sub.1-K.sub.6 for healthy people were obtained experimentally based on polysomnographic clinical studies. Statistically valid methods accepted in medical practice and described, for example, in the article “Polysonmography” (http://www.zonasna.ru/serv002.html) were used for checking the accuracy of REM sleep identification. Weight coefficients K.sub.1-K.sub.6 were selected so that the function values F(Δt.sub.i) in REM and non-REM phases display a maximum difference from each other.

(23) The increment ΔF(Δt.sub.i) of function F(Δt.sub.i)) over time Δt.sub.i is used to identify the onset and termination of REM sleep. If the difference between the current function value F(Δt.sub.i) and its previous value F(Δt.sub.i-1) exceeds the first preset threshold value, the onset of REM sleep is identified. If said difference is less than the second preset threshold value, the termination of REM sleep is identified.

(24) FIG. 1-FIG. 4 show examples of function F(Δt.sub.i) obtained for different test subjects during their sleep. Optimal weight coefficients K.sub.1-K.sub.6 given in Table 1 were selected in the course of studies to calculate function values F(Δt.sub.i). FIG. 1 FIG. 4 demonstrate a smoothed form of function F(Δt.sub.i)).

(25) The measuring resolution of accelerometer and pulse wave sensor signals amounted 0.1 in the testing process. All limb movements detected over 10-second time interval were considered to be a single movement and were averaged over the period of 5 min Function values F(Δt.sub.i) were calculated every minute, in other words, value Δt.sub.i was taken to be 1 minute for each i-th time interval. The first threshold value L.sub.1 was selected in the range from 20 to 30, while the second threshold value L.sub.2 was selected in the range from −30 to −20.

(26) FIG. 1a shows a fragment of function F(Δt.sub.i)) which includes one of REM phases registered during the sleep of one of the test subjects (8VAV). As is seen, function value F(Δt.sub.i)) rises sharply at 202-th minute of sleep, which indicates the onset of REM sleep, whereas at 210-th minute said function value F(Δt.sub.i) falls abruptly, which indicates the termination of REM sleep.

(27) FIG. 1b shows a graph of increment ΔF(Δt.sub.i) of function F(Δt.sub.i) from FIG. 1a. As is seen, the increment value ΔF(Δt.sub.i) considerably exceeds the first threshold value L.sub.1 with the onset of REM sleep, and becomes noticeably lower than the second threshold value L.sub.2 with REM sleep termination.

(28) The example illustrated in FIG. 1 is presented in Table 2 in the form of parameter values P.sub.1-P.sub.6, function values F(Δt.sub.i) and function increment ΔF(Δt.sub.i). The lines with parameter values presented in bold type in Table 2 correspond to REM sleep onset and termination in test subject.

(29) TABLE-US-00002 TABLE 2 Sleep Number Duration, P.sub.1, P.sub.2, P.sub.3, P.sub.4, P.sub.5, of P.sub.6 over in min. ms ms ms ms min Movements 5 min. F (Δt.sub.i) ΔF (Δt.sub.i) 185 92 1201 1422 — 14 0 — — 186 1273 1200 1421 — 15 1 — — 187 1272 1199 1420 — 15 0 — — 188 1272 1198 1418 — 15 0 — — 189 1272 1199 1418 92 15 0 0.2 −1319 190 1274 1198 1419 92 15 0 0.2 −1321 −1.9 191 1273 1201 1419 94 14 0 0 −1324 −3.0 192 1272 1202 1421 92 14 0 0 −1326 −2.0 193 1272 1200 1422 92 14 0 0 −1326 0.3 194 1271 1202 1421 92 14 0 0 −1325 0.8 195 1272 1201 1421 92 14 0 0 −1326 −0.9 196 1272 1202 1422 92 15 0 0 −1324 1.8 197 1272 1202 1420 93 15 0 0 −1323 1.5 198 1271 1198 1422 92 15 0 0 −1323 0.1 199 1272 1199 1421 92 15 0 0 −1324 −1.1 200 1272 1200 1418 92 15 0 0 −1324 0.0 201 1273 1197 1418 92 16 0 0 −1322 1.4 202 1206 1015 1290 89 18 0 0 −1226 96.1 203 1207 1011 1290 88 19 0 0 −1226 0.2 204 1207 1012 1290 89 19 0 0 −1225 1.3 205 1208 1012 1290 89 19 0 0 −1226 −1.0 206 1207 1010 1290 90 18 0 0 −1225 0.7 207 1207 1012 1290 89 19 0 0 −1225 0.3 208 1206 1013 1290 88 19 0 0 −1225 −0.5 209 1207 1012 1290 89 19 0 0 −1225 0.5 210 1367 1290 1500 97 14 1 0.2 −1424 −199.6 211 1369 1300 1505 98 16 0 0.2 −1422 1.9 212 1369 1290 1501 99 15 0 0.2 −1421 0.9 213 1367 1290 1498 100 14 0 0.2 −1420 1.5 214 1367 1285 1498 99 13 0 0.2 −1422 −2.7 215 1367 1290 1500 100 15 0 0 −1422 0.2

(30) FIG. 2 is a graph of function F(Δt.sub.i) over the entire sleep duration for the same test subject (8VAV). As follows from function values F(Δt.sub.i), there occurred four REM phases during the sleep of the test subject.

(31) FIG. 3 shows a graph of function F(Δt.sub.i) for another test subject (7ESA). As follows from the graph, four REM phases favorable to awakening were similarly registered during subject's sleep. The subject woke up by himself during the last REM phase.

(32) The number of REM phases may vary during sleep. For example, FIG. 4 shows that three REM phases occurred during the sleep of another test subject (3SOR).

(33) The graph also shows that different REM phases feature different absolute values of function F(Δt.sub.i) throughout sleep duration and that REM sleep onset and termination can be reliably identified only by the increment of said function.

(34) A series of tests showed that the method according to the present invention enabled the identification of 73 out of 76 REM sleep phases in 20 test subjects, which testifies to its high reliability of identification of human sleep phase favorable to awakening. The parameters of function F(Δt.sub.i) selected therein were also defined by the necessity to use a minimum number of sensors fixed on the wrist to provide comfortable sleeping conditions.