EP1945098A1

EP1945098A1 - Method for analysis and processing of a parameter representative of the phases of sleep

Info

Publication number: EP1945098A1
Application number: EP06831006A
Authority: EP
Inventors: Loïc TRON
Original assignee: Labosserie
Current assignee: Labosserie
Priority date: 2005-11-09
Filing date: 2006-10-25
Publication date: 2008-07-23
Also published as: FR2892914A1; WO2007054629A1; FR2892914B1

Abstract

The present invention relates to a method for analysis and processing of a datum representative of the phases of sleep, characterized in that it comprises the steps of measuring a parameter representative of the phases of sleep at regular intervals, projecting all of the previously obtained data onto a set of basic periodic functions at each measurement, determining the most relevant basic periodic function, calculating the phase of the basic periodic function for a desired interval of time, and deducing from this the corresponding phase of sleep, and also to a corresponding waking method and an associated waking device.

Description

Method for analyzing and treating a parameter representative of the phases of sleep

The present invention relates to a method for analyzing and processing data representative of the sleep phases as well as to a method of waking up and waking up using said analysis method.

A night's sleep is broken down into several cycles, each lasting between 70 and 120 minutes and schematically comprising five distinct phases: two so-called slow-light sleep phases, two so-called deep slow-wave phases and a paradoxical sleep phase, each of these phases being manifesting by a characteristic motor and brain activity.

Slow sleep phases are characterized by low muscle tone and reduced brain activity. Body movements are rare and the electrical activity of the brain is slowed down. These phases represent almost half of total sleep.

Deep slow sleep phases are characterized by high muscle tone. The sleeper is usually still, fingers and fists clenched with a slow, steady pulse and breathing. These phases represent about a quarter of total sleep. The paradoxical sleep phase is characterized by intense electrical activity. The face is mobile and the eyes are animated by a rapid movement ("Rapid Eye Movement" or REM). The pulse and breathing are fast and irregular. The sleeper is in muscular hypotonia and is generally agitated with brief body movements. It is generally considered that an awakening in paradoxical phase or deep slow sleep phase is painful and causes a feeling of fatigue in the day. Thus, it is ideally desirable to wake up during the slow slow sleep phase. The phase in which the sleeper should preferably wake up may possibly vary depending on the crop. Although the duration of the cycles is in theory substantially constant during the night and more generally during the life of a human being, it is very difficult or impossible for a sleeper to determine the ideal time of his awakening even if the latter knows precisely the average duration of his sleep cycles. Differences in cycle time, night-to-night fatigue state, stress state and time falling asleep are the main reasons preventing such a determination.

However, there are various devices to analyze the sleep of a sleeper and can be combined with a wake up system to ensure awakening in good conditions.

First of all, one of the most well-known devices used in the medical field is sleep monitoring by electroencephalogram. This method makes it possible to follow the cerebral activity in a rather precise way and is used to diagnose possible pathologies of the sleep. However, this method requires a complex apparatus mainly used in a medical environment as well as the laying of electrodes on the sleeper, which disturbs the sleep and makes impossible a general public application.

Other less complex devices known as actigraphs have been developed for the purpose of studying circadian rhythms outside of a medical environment. An actigraphe measures the acceleration of the movements of a part of the body, usually the wrist using a kind of watch, with a band-pass filter. Actigraphy has a fairly good correlation with the electroencephalogram, especially for normal sleepers, but is mainly used to describe circadian rhythms and the quality of alertness of people with hazardous jobs. Although it can be applied to the general public outside of a medical environment, the presence of a watch that must firmly grip the wrist remains inconvenient to be used during the night.

Other devices still allow a less precise follow-up but less invasive to implement and are based on the muscular or body activity of the sleeper. It may be, for example, to measure, separately or in combination, cardiac activity, pulse, blood pressure or to measure body movements by sensors placed under the bed or at a distance. Such devices are described in EP 0 681 447 for blood pressure, EP 0 836 046 for muscle tone and FR 2 649 512 for pulse, among others.

Moreover, one of the main difficulties generally associated with these devices, lies in the analysis of sleep cycles and their modeling in order to anticipate the phases of sleep to calculate a satisfactory waking time. These methods of analysis are generally very briefly described. One of the main mathematical methods of cycle analysis that one might be tempted to use is to use a Fourier transform in order to find the central periodic signal. This method requires a large amount of calculation and is difficult to implement in a microcontroller. Moreover, it provides much more information than necessary because only the periodicity and its phase of the signal are really interesting.

One way to partially remedy the amount of computation required could be to use fast Fourier transform. However, the sampling domain must then be power of two, which is difficult to adapt to nights of variable duration. In the same way, the discrete Fourier transform is accompanied by numerous constraints, in particular a constraint on the sampling interval.

It should also be noted that the calculations requested by these Fourier transforms are not recursive and must therefore be performed at each measurement over the entire sample.

Other methods have been described in many patent applications or patents attempting to simplify sleep analysis in order to wake up a person at the right time. These include statistical methods of autoregressive, autocorrelation (EP 0 597 032), spectral density, fuzzy logic or neural networks (DE 198 11 206). These methods all require a large number of calculations and are usually not recursive. Moreover, such methods, and more particularly the methods of spectral density and neural networks, require a large amount of data to give sufficiently precise results, and therefore generally present a problem of data adequacy to be exploited in the case of certain types of data. representative parameters of sleep phases such as movements for example. The object of the present invention is to propose a method that is both simple and precise for the analysis of sleep, allowing for easy implementation, and consists of doing so in a method for analyzing and processing data representative of the phases of sleep. , characterized in that it comprises the following recursive steps aimed at: - measuring at least one parameter representative of the sleep phases at regular intervals, to each measurement, project all the previously obtained parameter values onto a set of elementary periodic functions,

calculate the modulus of each projection obtained and retain the most relevant elementary periodic function having the highest modulus with respect to the number of parameter values measured,

- calculate the phase of the elementary periodic function selected for a desired time interval and deduce the corresponding sleep phase. By representative parameter of the phases of sleep, one understands any parameter varying according to the phases of sleep and whose variations are characteristic of a passage in paradoxical sleep. It may be, for example, pulse, blood pressure, movement of the eyes or limbs, separately or in combination. The representative parameter will be chosen so as to obtain a sufficiently discriminating measure of a paradoxical sleep phase.

Thus, by projecting the measured parameters on elementary periodic functions, it is easy to model the periodicity of the phases of sleep. Indeed, a projection calculation is recursive and the projection of an additional value of the parameter is done simply by adding the new projection. As a result, few calculations are necessary and the invention provides a rapid and reliable method for analyzing sleep cycles.

The method that is the subject of the invention therefore makes it possible to implement it in a simplified manner, requiring a minimum of computing and memory resources so that it can easily be implemented on a microcontroller, while maintaining a good accuracy without resorting to complex mathematical tools and very consumer of CPU resources.

Preferably, the periodic elementary functions are complex functions of the sinusoid type. The use of complex functions makes it possible to work simultaneously on a module representing the intensity of the movements and an argument representing the phase of the periodic functions.

Advantageously, each measurement of the representative parameter is a discrete value determined with respect to a threshold. Advantageously, the threshold is adjustable. The value of the threshold can be adjusted from one night to another automatically or by the user depending on the detection of the representative parameter. Advantageously, a measurement less than the threshold takes a value of 0 while a measurement greater than the threshold takes a value of 1. Although the use of a continuous value measurement is possible, the passage through a discrete measurement value makes it possible to adapt the statistical weight given to the characteristic measurement and improve the reliability of the process.

Preferably, an elementary function is defined for each period between 70 and 120 minutes and corresponding to a measurement interval.

Preferably, the time interval between each measurement of the representative parameter is less than or equal to 5 minutes.

Advantageously, the representative parameter measured is the movements of the sleeper. Indeed, such a parameter can be easily measured directly or indirectly, by contact or remotely, with sensors that do not significantly interfere with the sleeper. Advantageously, the method is a wake-up method and comprises the following additional steps aimed at

determining a waking time close to a set time and corresponding to an acceptable sleep phase,

- emit an audible and / or visual signal when the determined optimum alarm time is reached.

As explained above, a sleep phase is said to be acceptable when it corresponds to a light slow sleep phase. It is of course possible to choose to wake up during another phase of sleep. Advantageously, the step of determining an optimal waking time is performed only if the measurement time is greater than or equal to the set time minus the determined period.

Advantageously, the determined wakeup time is as close as possible to the set time. The present invention also relates to a wake-up device comprising at least one sensor of a parameter representative of the sleep phases connected to a microcontroller, characterized in that the microcontroller is able to execute the steps of the wake-up method according to the invention.

Advantageously, the alarm device comprises at least one remote motion sensor, in particular an infrared or ultrasonic sensor. The implementation of the invention will be better understood with the aid of the detailed description which is set out below with reference to the appended drawing in which:

Figure 1 is a representation of movement measurements over time.

Figure 2 is a representation of the relevance of elementary functions according to their periods.

FIG. 3 is a representation of the cycles determined by the method according to the invention. FIG. 4 is a curve of the phase of the periodic function selected at each time of measurement as a function of time.

An alarm device according to the invention comprises a remote motion sensor connected to an input of a microcontroller equipped with a clock component. The sensor is a sensor for measuring remote movements of the type infrared sensor or ultrasonic sensor. In the case of ultrasound, it will be ensured that the frequency of these ultrasounds is sufficiently high so that they are not disturbed by the environment.

The sensor is intended to measure the intensity of the movements of the sleeper at regular intervals of 5 minutes. A shorter time interval is of course possible. Below a certain threshold of measured movement intensity, the microcontroller considers that there is no movement and assigns the value 0 while the value 1 is assigned to the sample if the intensity of the detected motion exceeds the defined threshold. It is of course possible to work using continuous measurement values, the advantage of moving to discrete values residing mainly in the fact of assigning a greater statistical weight to the movements characteristic of a paradoxical sleep phase. This consideration can easily be adapted by those skilled in the art by means of tests, according to the representative parameter used.

An example of a set of measurements during one night is shown in Figure 1.

The intensity threshold of movement can be fixed or variable to take into account the distance to which the sensor is placed. In the case of a variable threshold, it can be set manually by the user or automatically by waking according to the number of movements detected during a night.

For each measurement, the cycle of the phases of sleep is modeled by projection of the movements on a set of elementary periodic functions.

The set of functions chosen is a set of sinusoidal elementary functions for the periods between 70 and 120 minutes with a period interval equal to the sampling interval, namely five minutes. The basic functions considered are complex functions and are written: lm_ _j 2x (.t '-' ^o) s _k {n) - e ^k = e _k ^θk with θ = kT, k integer, and t _n - t ₀ + nT where T is the sampling step, n is an integer representing the sample considered and to, t _n are respectively the analysis start time and the current sampling time.

The projection of the movements on the elementary function k is calculated according to the formula:

X _k (N) = Σx (t _n ) e ^{~ J ~} = X _k {N - \) + x (t _N ) e ^{~ J ~} n≈l where x (t _n ) is the measured motion value, 0 or 1, and N is the number of the sample at time t _N = t _o + NT.

It should be noted that this calculation is recursive and can therefore be easily performed with a minimum of computation and memory. Indeed, only the result of the projection is used and it is not necessary to keep the previous measurements of the representative parameter to be able to perform the next projection or calculation of relevance according to the method object of the invention.

Although the elementary functions are not unitary, they have the same module equal to the square root of the number of measurements made at a given moment. It is therefore possible to compare the module of each projection in order to deduce the most relevant elementary function and best modeling sleep cycles during the night. The projection modules are all smaller than the number of movements detected over the time considered, the most relevant elementary function having the highest module relative to the number of measured parameter values. The ratio between the module obtained squared and the number of movements measured makes it possible to obtain a probability. A projection is considered relevant if its probability is greater than 25%. Figure 2 represents the relevance of each elementary function as a function of its period as the abscissa.

Once the most relevant elementary function of period θ has thus been determined, the corresponding function is thus:

S _θ {t) = _θ X (N) .s _θ {t) II is the periodic complex function better then representing the cyclical nature of sleep movements. Its module is equal to the measure of relevance and its argument makes it possible to obtain the phase of the cycle of the movements at the moment considered. The sleep phase at the set time can therefore be easily calculated. This calculation can be done at the end of a set of measurements or at each measurement with the most relevant elementary function at the moment of measurement. It will be easy to display value curves for the function representing the most relevant periodicity calculated at the end of the night, in real time for the function most relevant to each measurement. It should be noted that the curve of values most relevant to each function joins the curve determined on the night at the end of the measurements. FIG. 3 represents each of these curves, curve 1 corresponding to the curve for the period determined at the end of the night, curve 2 corresponding to the curve for the period determined at each instant of measurement. After a certain number of measurements, typically when the measurement time corresponds to the set time minus the most relevant period chosen, it is then necessary to determine the appropriate alarm time. The appropriate wake-up time can be refined through subsequent calculations as the end of the night approaches and for which more measurements are available.

The movements of the sleeper occur mainly at the beginning and the end of paradoxical sleep. By comparing various electroencephalograms on which the method of the invention has been applied, it has been found that paradoxical sleep is situated for arguments of the elementary complex function retained between 270 and 90 ° before the light slow sleep and sleep phases. slow deep. The argument of the most relevant complex periodic function is therefore an indicator of the corresponding sleep phase.

The range of values representative of the phases of sleep can be refined in clinical studies. In addition, it is possible for the sleeper to choose his waking phase and to determine its range of acceptable values. It is also possible to add a function for learning this range of values from information on the quality of the alarm entered by the user.

To determine an appropriate wake-up time, it is sufficient first to calculate the argument of the complex function held at the set time. If this argument belongs to the range of acceptable values outside REM sleep, that is to say between 90 and 270 °, the set time is not changed.

If this is not the case, then look for the time closest to the set time for which the function argument belongs to the range of acceptable values. Typically, the closest time will correspond to an argument of the complex function held by 270 °. A second solution consists in calculating at each instant of measurement if the argument of the selected function is passed higher than the upper limit of the range of acceptable values, namely 270 °.

FIG. 4 is a curve representing the phase of the most relevant period elementary function at each time of measurement as a function of time. Such a curve may possibly be displayed at any time on a screen and may allow, for example, to monitor the sleep of a child.

The new set time thus determined, it will be sufficient to trigger the alarm clock at the said time.

It should be noted that during sleep, fatigue is recovered early in the night priority over paradoxical sleep. As a result, the phases of slow light sleep and slow deep sleep are more important at the beginning of the night than at the end of the night. In the case where the number of cycles before waking is less than three generally, then it will be necessary to adjust the range of acceptable waking values according to the cycle in which the sleeper must wake up. To do this we can define a range of acceptable values for the first cycle, a range of acceptable values for the second cycle, etc. The number of elapsed cycles can be determined very simply by dividing the sleep time elapsed by the calculated relevant period.

An alternative embodiment may consist in using a mean period of the sleep cycles determined from the relevant periods obtained for each night. In the case where a night does not allow to calculate a sufficiently relevant period, it will be possible to use the average period determined from the history of the values.

An advantageous improvement of the method will consist in affecting the measurements made of a damping coefficient α as a function of the instant of measurement.

Although the invention has been described in connection with particular embodiments, it is obvious that it is not limited thereto and that it comprises all the technical equivalents of the means described and their combinations if they are within the scope of the invention.

Claims

1. A method for analyzing and processing data representative of the phases of sleep, characterized in that it comprises the following recursive steps aimed at:

to measure at least one parameter representative of the phases of sleep at regular intervals,

to each measurement, project all the previously obtained parameter values onto a set of elementary periodic functions,

calculating the modulus of each projection obtained and retaining the most relevant elementary periodic function having the highest modulus with respect to the number of measured parameter values, calculating the phase of the elementary periodic function selected for a desired time interval and deduce the corresponding sleep phase.

2. Method according to claim 1, characterized in that the periodic elementary functions are complex functions of the sinusoid type.

3. Method according to any one of claims 1 or 2, characterized in that each measurement of the representative parameter is a discrete value determined with respect to a threshold.

4. Method according to claim 3, characterized in that the threshold is adjustable.

5. Method according to any one of claims 1 to 4, characterized in that an elementary function is defined for each period between 70 and 120 minutes and corresponding to a measurement interval.

6. Method according to any one of claims 1 to 5, characterized in that the time interval between each measurement of the representative parameter is less than or equal to 5 minutes.

7. Method according to any one of claims 1 to 6, characterized in that the representative parameter measured are the movements of the sleeper.

8. Awakening method according to any one of claims 1 to 7, characterized in that it comprises the following additional steps to

9. Awakening method according to claim 8, characterized in that the step of determining an optimal waking time is performed only if the measurement time is greater than or equal to the set time minus the period determined.

10. Wake-up device comprising at least one sensor of a parameter representative of the phases of sleep connected to a microcontroller, characterized in that the microcontroller is able to perform the steps of a method according to any one of claims 8 or 9.

11. Awakening device according to claim 10, characterized in that it comprises at least one remote motion sensor including an infrared sensor or ultrasound.