System for determining a set of at least one cardiorespiratory descriptor of a person during sleep

Description

FIELD OF THE INVENTION

[0001] The present invention relates to the field of well-being, in particular cardio-respiratory dynamics during sleep.

[0002] It is applicable to the general public, independently of any medical structure, for example when a person needs to better understand the structure of their sleep and the events occurring during their sleep.

[0003] The present invention in fact allows an analysis of the physiology of sleep, and finds in particular an application in the collection and analysis of data which may subsequently be useful in establishing a diagnosis relating to the quality of sleep. Of course, the invention is not a diagnostic method insofar as no clinical picture is associated with the implementation thereof.

[0004] Traditional methods of estimating sleep quality are often a compromise between the design / manufacturing cost and the complexity of implementing a device implementing said method, and the robustness of the analysis and therefore the relevance of the evaluation.

[0005] Document EP0371424 is known. This document provides a set of sensors (cardiac electrodes, microphone) for equipping an individual and recording a plurality of signals such as snoring and breathing noises, as well as heartbeats during sleep in order to determine, using an index calculation, possible obstructive sleep apnea.

[0006] Document EP2621336 is known. In this document it is intended to record respiratory efforts from a pressure sensor of an awake individual and to determine a set of parameters in order to diagnose possible obstructive sleep apnea on the basis of an algorithm based on exhalation and comparison of the parameters to a set of frequency bands.

[0007] The document US9545227 is known. In this document, it is intended to determine the level of risk of sleep apnea in an individual. For this, the method is based on the detection of a physiological parameter converted into a pre-processed data stream from which characteristics are extracted. From these characteristics and information from the individual, a vector is constructed in order to determine the level of risk by comparison with a model constructed from machine learning.

[0008] Finally, document WO2008037820 is also known. This document provides a system for capturing and analyzing cardio-respiratory signals, characterized in that a single accelerometer is used to capture all of the physiological signals sought (cardiac, respiratory and snoring signals).

[0009] Thus, on the one hand, we find methods based, for example, on the analysis of signals captured by an inertial unit of a communicating object placed on the bed of a sleeping individual.

[0010] However, in addition to the fact that the components of a communicating object are not always of sufficient quality to obtain an optimal signal for analysis, the number of physiological parameters measured is also low. Only the absence of specific hardware for capturing physiological signals makes the solution inexpensive for the user, as well as the development not very complex for the designer.

[0011] On the other hand, there are medical or paramedical solutions, for example ventilatory polygraphy or polysomnography (PSG), based on a multimodal and multi-sensor approach. Often, these methods attempt to analyze, via a reading in full of a specialist, the cardio-respiratory dynamics of the individual (among others) to determine a state of sleep or identify typical events, representative of a sleep disorder. More specifically, the specialist seeks to determine respiratory events of the obstructive or central apnea and hypopnea type. These data are usually synthesized in the form of a single index called AHI for Apnea-Hypopnea Index and which corresponds to the sum of the events detected / unit of time, in this case the hour.

[0012] The sensors involved are typically of several types (ECG, respiratory plethysmography, naso-oral thermistor, etc.), mechanically increasing the manufacturing and design costs of the solutions. The signals obtained are also of variable quality, not always allowing for robust analysis because the individual may move during the night.

[0013] Furthermore, this multiplicity of sensors results in complexity of use for the user as well as reduced acceptability since the user must wear all of the sensors, on his torso, fingers, skull, etc. which in fine negatively affects the quality of sleep.

[0014] In this context, the applicant wished to resolve these problems of cost and complexity of implementation associated with existing solutions, while maintaining very good robustness of analysis.

[0015] In this sense, the solution proposed here consists of an approach based for example on a pair of accelerometers embedded within the same hardware device placed in a single location of the individual at the level of the individual's trunk, which makes the invention practical, compact and easy to implement. It is also possible to incorporate, in order to improve the robustness of the measurements explained below, a set of at least one additional sensor among a photoplethysmography (PPG) sensor, which allows indirect optical measurement of blood oxygenation and heartbeats; a microphone and a gyroscope, making it possible to track the angular position of said device.

[0016] Beyond the technological advances proposed by the present invention, it meets a dual need in public health and medical practice: certain populations are unaware of their sleep disorders and are therefore not guided in the health pathway. The present invention can make it possible to detect these populations and can therefore make it possible to meet a public health need, since it allows screening, possibly early, on these populations. In addition, standard techniques (PSG) require expertise that is complicated to obtain and practice and thus limit the number of experts available. The present invention makes it possible to promote the dissemination of a precise tool for the evaluation of cardio-respiratory sleep health among non-specialist practitioners.

[0017] FEKR ATENA ROSHAN ET AL: "Respiration Disorders Classification With Informative Features for m-Health Applications", IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, IEEE, PISCATAWAY, NJ, USA, vol. 20, no. 3, May 1, 2016 describe a system that uses motion sensors to detect changes in the anterior-posterior diameter of the chest wall during respiratory function as well as to extract informative respiratory features to be used for the classification of respiratory disorders. SUMMARY OF THE INVENTION

[0018] More specifically, the invention relates to a system for determining a set of at least one cardio-respiratory descriptor of an individual during their sleep according to claims 1 to 12.

[0019] It can be expected that the data filtering device is configured to: ∘ extracting a low frequency range, a medium frequency range and a high frequency range from the data from the first set of at least one accelerometer, and ∘ extracting at least one low frequency range from the data from the second set of at least one accelerometer.

[0020] In addition, we can provide: means for determining the position of said individual during his sleep, by data from at least one of said first set of at least one accelerometer and said second set of at least one accelerometer.

[0021] In addition, at least one of the following sensors may be provided: a photoplethysmography (PPG) sensor, a microphone and a set of at least one angular position measurement sensor, synchronized with the first set of at least one accelerometer and configured to be placed in the thoracic position of said individual; the data filtering device being further configured to: couple the data from a set of at least one angular position measurement sensor with the data from at least one of the first set of at least one accelerometer and the second set of at least one accelerometer, for the extraction of user position characteristics or atypical respiratory angular velocity variations; and / or average the data from the optical photoplethysmography sensor.

[0022] It can be provided that the calculator is configured to determine a set of at least one cardio-respiratory descriptor of the individual among: an apnea-hypopnea index (AHI), a first set of respiratory efforts in the data from the first set of at least one accelerometer, a second set of respiratory efforts in the data from the second set of at least one accelerometer, a thorax-abdomen desynchronization index, by synchronized comparison of the respiratory efforts of the first set of respiratory efforts and the respiratory efforts of the second set of respiratory efforts, heart rate variability (HRV), the oximetry profile or oxygen saturation, the respiratory entropy per unit of time, the percentage of snoring per unit of time; and a nighttime profile of oxygen saturation.

[0023] For example, the percentage of snoring per unit time is determined at least in part based on data from at least one of the first set of at least one accelerometer and the microphone.

[0024] For example, the nocturnal oxygen saturation profile is determined at least in part based on data from the thoracic PPG sensor.

[0025] It can be provided that the calculator is configured to select a set of characteristics from the set of characteristics determined, for example by an algorithm of the Sequential Feature Selection (SFS) or Fast Correlation-Based Filter (FCBF) type, prior to the comparison.

[0026] It can be provided that the calculator is configured to calculate at least one of: the temporal correlation between at least two events determined over the same time range or over several time ranges; a set of pairs (event / position), a set of pairs (event / characteristic), a set of triplets (event / characteristic / position), a set of quadruplets (event / event / characteristic / position).

[0027] It can be expected that the computer is configured to determine a chain of events which, during a predetermined time window, resulted in a cardio-respiratory descriptor of abnormal value.

[0028] Means for activating / deactivating optional accelerometers and sensors may also be provided.

[0029] It can be expected that the memory for recording the data from the accelerometers and optional sensors and the memory for recording the reference model are the same; said memory being: either remote or in wired computer connection with the first set of at least one accelerometer, the second set of at least one accelerometer as well as the optional sensors.

[0030] Other characteristics and advantages of the present invention will appear more clearly on reading the following description given by way of illustrative and non-limiting example and made with reference to the appended figures. DESCRIPTION OF THE DRAWINGS

[0031] there Figure 1 illustrates an embodiment of the system according to the invention, the Figure 2A illustrates a mode of positioning the accelerometers and optional sensors according to the invention, the Figure 2B illustrates another mode of positioning the accelerometers and optional sensors according to the invention, and the Figure 3 illustrates the frequency ranges within the meaning of the invention. DETAILED DESCRIPTION

[0032] By language convention, “individual”, “subject” or “user” means a natural person using a measuring device according to the invention.

[0033] By "sleep" is meant a period of time, possibly variable, during which the measuring device according to the invention is active; the individual being normally asleep, possibly discontinuously, for at least part of this period of time. Measuring device

[0034] A single, compact measuring device is planned, comprising at least two accelerometers.

[0035] In particular, a first set of at least one accelerometer is provided, intended to be placed in the thoracic position of the individual, preferably in the xiphoid position, which makes it possible to measure cardiac mechanical activity, thoracic ventilatory efforts and snoring activity.

[0036] A second set of at least one accelerometer is also provided, intended to be placed in the abdominal position of said individual, which makes it possible to measure the ventilatory efforts of the abdomen.

[0037] Each set of at least one accelerometer may comprise a plurality of accelerometers, for example for reasons of redundancy and robustness.

[0038] In this case, tests carried out by the applicant have proven that a single accelerometer per set may be sufficient.

[0039] It is also possible to provide for the addition of other optional sensors, for example at least one of a Photoplethysmography sensor, hereinafter PPG sensor, and a microphone, placed in this case in the thoracic position at the xiphoid level. It is also possible to provide a set of at least one angular position measurement sensor, in this case a gyroscope, in combination with a set of at least one accelerometer.

[0040] Therefore, the measuring device illustrated here notably comprises two accelerometers, and, for brevity, the first set of at least one accelerometer is called first accelerometer 110, and the second set of at least one accelerometer is called second accelerometer 120.

[0041] Each sensor (e.g. accelerometer) emits a respective signal comprising a set of data. Conciseness therefore means "signal" and "data" interchangeably.

[0042] Preferably, it is provided that the measuring device comprises means for activating / deactivating the sensors, for example in the form of an activation control switch, analog or digital.

[0043] In a first variant, illustrated Figure 2A , it is expected that the first accelerometer, the second accelerometer and the optional sensors are integral with a single support 100.

[0044] In a second variant, illustrated Figure 2B , it is provided that the first accelerometer and the second accelerometer are integral with a respective support 200, 210. The optional sensors may be integral with the support of the first set of at least one accelerometer.

[0045] Regardless of the variant, each support comprises a top face and a bottom face. The top face supports at least one accelerometer and the bottom face is typically coated with adhesive (glue), preferably repositionable. The advantage of the repositionable aspect is that the individual can remove the measuring device upon waking and reuse the same measuring device the following night, and this for several consecutive days. The device is powered by a battery, for example.

[0046] The first accelerometer and the second accelerometer may be identical. In this case, the first accelerometer and the second accelerometer are standard three-axis accelerometers, in this case with X the elongation axis of the individual and the Z axis which is orthogonal to the subject ( Figure 2A, Figure 2B ).

[0047] The data from each sensor is recorded in a computer memory.

[0048] In a first variant, the memory is in wired computer connection with the first and second accelerometers as well as the optional sensors. The memory can for example be arranged on the single support or any of the supports of the accelerometers.

[0049] Preferably, it is provided that the memory is in computer connection with an input / output (I / O) communication port, for example a USB port, which makes it possible to export the recorded data to a data processing device comprising data processing software, typically any communicating object, i.e. an electronic device comprising wired or wireless communication means, a calculator 300 (processor) and preferably a display screen, for example a personal computer, a smartphone, a touch tablet etc.

[0050] It is also possible to provide that the memory is removable, for example in the form of a data medium of the memory card type with USB connection, connectable to the first accelerometer and to the second accelerometer for recording the data from them, and then connectable to a computer equipped with filtering software making it possible to filter said data.

[0051] In a second variant, the memory is remote, i.e. without a wired connection between said memory and the sensors. For example, the memory is remote in the data processing device.

[0052] In this case, it is advantageous to provide that the measuring device comprises means of communication with said data processing device, whether wired or wireless communication.

[0053] The memory can therefore be embedded or remote, on a server, in particular a cloud computing system.

[0054] The accelerometers and optional sensors are advantageously mounted on an electronic card, comprising a battery, a computer, for example a processor (microcontroller) and a storage unit capable of storing the data for on-board processing. Alternatively, each accelerometer is mounted on an independent electronic card, each having a power supply battery, a computer (processor) and a storage unit. It is expected that the optional sensors are integral with the electronic card of the first set of at least one accelerometer. The synchronization of the signals is done by construction by the microcontroller.

[0055] Preferably, means for processing the signals from the sensors are also provided.

[0056] In this case, the signal processing means comprise filtering means, configured to filter the output signal of the first accelerometer, and configured to filter the output signal of the second accelerometer.

[0057] Similar to memory, the filtering means can be integrated into the measuring device, or transferred to the data processing device or even to a server in communication with the data processing device.

[0058] In this case, it is advantageous to provide that the measuring device comprises means of communication with said signal processing means, whether wired or wireless communication.

[0059] The Photoplethysmography sensor includes an optical transmitter and receiver to measure the oxygen saturation of the blood at the thoracic level by the difference in absorption between Hemoglobin and Oxyhemoglobin between red and infrared light.

[0060] Each signal generated by the accelerometers is processed to make it usable, then analyzed on several frequency bands to extract different values ​​of physiological parameters likely to indicate the occurrence of events typical of sleep-disordered breathing and therefore affecting the quality of sleep, as described below. Data processing

[0061] The data from the accelerometers are planned to be processed in the following manner.

[0062] The measuring device is installed on the individual, preferably by the individual, before bedtime, for continuous operation. Raw data

[0063] The data from the first accelerometer and the second accelerometer are called raw data.

[0064] The raw data is filtered as described later, either after being written to memory or on the fly.

[0065] As a non-limiting example, the raw data provided by the accelerometers are preferably sampled at a frequency of 10 KHz.

[0066] As a non-limiting example, the raw data provided by the PPG sensor are preferably sampled at a frequency of 25 Hz.

[0067] Advantageously, the acquisition is done continuously throughout the duration of sleep. Filtered data

[0068] Accelerometers have a frequency bandwidth preferably in the range of 0 to 500 Hz, illustrated Figure 3 .

[0069] Typically, the above frequency range can be subdivided into the following ranges and correspondence: the range from 0 to 1 Hz, hereinafter referred to as the “low frequency” range, corresponds to the individual’s respiratory activity, the range from 2 to 40 Hz, hereinafter referred to as the “medium frequency” range, corresponds to the individual’s cardiac activity, it can be subdivided into: ∘ a sub-range from 2 to 18 Hz which corresponds to cardiac mechanical activity linked to the contraction of the individual’s myocardium, and ∘ a sub-range from 18 to 40 Hz which corresponds to cardiac mechanical activity linked to the closing and opening of the individual’s heart valves, and the range from 40 to 500 Hz and above, hereinafter referred to as the “high frequency” range, corresponds to the individual’s snoring activity.

[0070] The low frequency range is the respiratory component of an accelerometer signal, it is induced by the movements of the individual's rib cage during breathing.

[0071] The mid-frequency range is the cardiac component of an accelerometer signal, it is induced by the individual's cardiac activity.

[0072] The high frequency range is the hum component of an accelerometer signal, it is induced by vibrations due to the snoring of said individual.

[0073] Each of the individual's activities (respiratory, cardiac and snoring) in fact makes the accelerometer(s) vibrate according to a specific and exclusive range of frequencies.

[0074] A filtering step is therefore advantageously provided consisting of filtering the signals from the accelerometers to extract from said signals at least one of the components, or indistinctly bands or ranges, among the “low frequencies”, the “medium frequencies” and the “high frequencies”.

[0075] In this case, it is planned to extract: the respiratory component by any known low-pass filter, and for example by principal component analysis (PCA) or by discrete Fourier transform; the cardiac component by any known band-pass filter, and for example by adaptive filtering, by envelope extraction, by learning, by cross-correlation, by k-means partitioning, or by principal component analysis (PCA), and the snoring component by any known high-pass filter, and for example by spectral analysis.

[0076] Preferably, it is planned to extract the respiratory component, the cardiac component and the snoring component from the raw data from the first accelerometer, placed in the individual's thoracic position.

[0077] Preferably, it is planned to extract at least the respiratory component from the raw data from the second accelerometer placed in the abdominal position of the individual.

[0078] Advantageously, with a three-axis X, Y and Z accelerometer arrangement as illustrated in Figure 2A and in Figure 2B , with XoZ the sagittal or median plane, XoY the coronal or frontal plane and YoZ the axial or transverse plane, we can extract: the respiratory component by a linear combination of the 3 axes, in this case in vector form; the cardiac component by using only the component along the Z axis of the signal from each accelerometer, the snoring component by a linear combination of the 3 axes, in this case in vector form.

[0079] It is further advantageous to provide for weighting the extracted snoring component along at least one of the X, Y and Z axes in an adaptive manner, depending on the position of the individual, the determination of the position of the individual being described later.

[0080] A signal smoothing step can also be planned after extracting the components in each of the frequency bands, which allows artifacts to be removed.

[0081] At the PPG sensor, the difference in absorption between Hemoglobin and Oxyhemoglobin between red and infrared light is calculated. An average filter is then considered with an intermediate averaging time allowing a compromise between measurement accuracy and artifact rejection. This time is, as a non-limiting example, 1 second. Synchronization

[0082] The data (raw or filtered) from the second accelerometer are synchronized with the data (raw or filtered) from the first accelerometer and the data from the optional sensors, for example by an on-board computer (processor) (microcontroller) or an acquisition system (server).

[0083] For example, we can provide an acquisition system, in this case a CPU or a microcontroller, coupled with a clock and a sampler and an analog-digital converter.

[0084] As described later, the individual's position can be determined during sleep. This position can be synchronized with the characteristics (respiratory, cardiac, snoring) determined, as described below. Determination of characteristics

[0085] From the data from the accelerometers (raw and preferably filtered) and where appropriate also from at least one of a PPG sensor, a microphone, and a set of at least one gyroscope, it is possible to determine, per frequency band, a set of at least one characteristic representative of a sleep disorder which normally took place in the individual's sleep phase, as well as the moment, i.e. the instant or time range, at which said characteristic was extracted.

[0086] Indeed, sleep-disordered breathing is manifested by physical or physiological conditions of the individual: snoring, turning over, agitation, positioning of the individual, thoracic breathing (or absence of), abdominal breathing (or absence of), etc. The convergence of these events makes it possible to point towards a potentially pathological state.

[0087] These physical or physiological conditions in turn translate into measurable characteristics, which are extracted from the signals from the accelerometers and, where applicable, from the optional sensors.

[0088] The "low frequency" range of accelerometers is representative of respiratory efforts, which include, for example, the individual's inspiration and expiration.

[0089] We can extract characteristics such as: respiratory rate, i.e. both thoracic and abdominal respiratory rates, respiratory amplitude, respiratory entropy, etc.

[0090] The "mid-frequency" range is representative of the contractions of the individual's heart.

[0091] We can extract characteristics such as: the instantaneous heart rate, at least thanks to the first accelerometer, the time between two successive beats, etc.

[0092] The "high frequency" range is representative of the individual's snoring.

[0093] We can extract characteristics such as: the individual's snoring times or frequency, at least using the first accelerometer, spectral analysis of snoring (power, amplitude, etc.) etc.

[0094] For optional sensors, it is planned to extract features representing the individual's oxygen saturation at the thoracic level using the PPG sensor.

[0095] From at least one of the first accelerometer, the second accelerometer and at least one angular position sensor, typically a gyrometer (indistinctly gyroscope), it is possible to extract angular position characteristics preferentially allowing algorithmic optimization for: Calculation of the user's position, Detection, in association with at least one set of accelerometers, of any abnormal respiratory patterns (atypical angular variation during post-obstructive apnea respiratory recovery, for example).

[0096] The "low frequency", "mid frequency" and "high frequency" ranges can be processed in series or in parallel.

[0097] It is also possible to further plan to select a set of characteristics from the set of determined characteristics, in particular by an algorithm of the Sequential Feature Selection (SFS) or Fast Correlation-Based Filter (FCBF) type, which makes it possible to optimize the calculations. Sleep disorders - events

[0098] The individual's sleep disorders are characterized in particular by at least one of the following syndromes: obstructive sleep apnea syndrome (OSAS), and central sleep apnea syndrome (CSAS).

[0099] These syndromes typically correspond to the appearance of at least one of the following (physiological) events during the individual's sleep: obstructive apnea, central apnea, mixed apnea, hypopnea, hyperventilation and hypoventilation.

[0100] The above events can be characterized by a specific distribution over time of the determined characteristics.

[0101] From a physiological point of view, in most cases of OSA, air stops flowing to and from the lungs due to a blockage in the upper airway, in the nose or throat.

[0102] Unlike OSA, SASC can occur even when the airway is clear. SASC means that the individual stops breathing for a predetermined number of consecutive seconds during sleep, in this case at least 10 consecutive seconds.

[0103] Hypopnea, or shallow breathing, is a decrease in respiratory flow, the limits of which are more or less well defined, for a predetermined number of consecutive seconds during sleep. In this case, a decrease of at least 50% in respiratory flow for 10 seconds is considered to define hypopnea. Hypopnea can also be defined as a decrease of 3% to 4% in saturation (oxygen in the blood).

[0104] In practice, apnea / hypopnea can therefore be characterized by repeated cessation of breathing or shallow breathing for short periods of time during sleep. In anatomical terms, there is an intermittent collapse of the upper airway which possibly results in a reduction in blood oxygen concentrations during sleep. Thus, a sleeping person becomes unable to breathe normally and most often wakes up with each collapse.

[0105] In reality, these definitions are relative and the boundaries are not clearly established due to different definitions between authors and the lack of harmonization. In other words, two different experts faced with the same results can make two different interpretations, including for example whether or not factors surrounding the individual (risk factors, medical history, etc.) are taken into account in their diagnosis.

[0106] The present invention makes it possible to escape from this constraint. Reference model

[0107] A reference model is provided. In this case, the reference model includes a first group of physiological data relating to individuals recognized as having sleep disorders, for example a set of polysomnographies.

[0108] The reference model also advantageously includes a second group of physiological data, relating to individuals recognized as not suffering from sleep disorders, for example a set of polysomnographies.

[0109] In particular, it is expected that at least the first group of physiological data comprises a correspondence between a set of time-distributed characteristics and a given event (obstructive apnea, central apnea, mixed apnea, hypopnea, hyperventilation or hypoventilation).

[0110] It is then possible to compare the characteristics (determined or selected) with the reference model, in this case at least with the data of the first group, to deduce the probable corresponding event(s) that the individual faced during the period of time considered, which makes it possible to detect a set of events potentially marking sleep disorders and to carry out a form of screening for sleep disorders instead of a diagnosis which aims to identify the nature and cause of the condition from which a patient is suffering.

[0111] Preferably, the comparison step is implemented by machine learning ( machine learningby Anglicism), which consists of constructing a decision-making model, for example by support vector machines (SVM by Anglicism), by Gaussian approach, by Bayesian approach, etc. in particular for the detection and discrimination of abnormal respiratory events.

[0112] The comparison is made for example on the following extracted characteristics: respiratory (rate, amplitude, thoraco-abdominal synchronization index), cardiac (rate, HRV), snoring (spectral composition). thoracic oxygen saturation (PPG sensor)

[0113] Learning at the processing level of the signals from the first and second accelerometers makes the extracted features more robust and therefore improves the performance of the comparison algorithm. For example, processing the signal from the "mid-frequency" range must result in a sufficiently accurate heart rate.

[0114] In this case, the comparison step is implemented by supervised learning using the reference model. Classification and determination of cardio-respiratory descriptors

[0115] From the determined characteristics, representative of a potential sleep disorder, a set of at least one cardio-respiratory descriptor of the individual can be determined, each descriptor being representative of the occurrence of event(s) representative of sleep disorder(s) during the individual's sleep; that is to say the type of disorder (apnea, hypopnea, hyper or hypoventilation), the time when the event occurs, and the number of times these events take place, as well as their frequency if they are regular.

[0116] A typical cardiorespiratory descriptor is, for example, the AHI, calculated from the number and type of respiratory events detected per hour of sleep (obstructive, central or mixed apneas / obstructive or central hypopneas).

[0117] Another cardiorespiratory descriptor is the thorax-abdomen desynchronization index per unit of time. This allows us to refine the central or obstructive nature of apneas.

[0118] In this case, we can compare the respiratory efforts from the filtered data of the second (abdominal) accelerometer and the respiratory efforts from the filtered data of the first (thoracic) accelerometer. The thorax-abdomen desynchronization index is calculated by estimating the phase shift between the signals from the first accelerometer and the signals from the second accelerometer. We can then estimate the delay of one relative to the other by comparing their respective passages through 0, then by calculating a wavelet transform for comparison of the spectral phase.

[0119] Another cardiorespiratory descriptor is heart rate variability (HRV), which is the fluctuation in the duration of the time intervals between two consecutive heartbeats.

[0120] It is essentially the result of extrinsic regulations and determines the heart rate. While the heart rate can be almost stable over a given time integration period, the time between two heartbeats can be very different and its informative value is greater.

[0121] In this case, it is expected that the extraction of the cardiac component further comprises a step of determining the variability of the heart rate, in this case by temporal and spectral analysis of the signal from the first accelerometer in the “medium frequency” range.

[0122] Another cardiorespiratory descriptor is the oxygen desaturation index for the measurement night. It reports the number of desaturations greater than 3% during the night, the greatest oxygen desaturation during the night, and the average profile during the night.

[0123] Another cardiorespiratory descriptor is respiratory entropy per unit time.

[0124] Another cardiorespiratory descriptor is the percentage of snoring per unit of time.

[0125] Advantageously, at least one of the determined cardio-respiratory descriptors is coupled to the position of the individual. It is indeed possible to associate the occurrence of sleep disorder marker events, determined by the comparison step, with the position of the individual at that time or in a predetermined preceding time window, which makes it possible to determine cardio-respiratory descriptors for which the physiological data of the individual are coupled to his posture in a synchronized manner.

[0126] We can thus determine for example: a set of pairs snoring / position; HRV / position; snoring / heart rate; HRV / heart rate; etc. a set of triplets snoring / heart rate / position; etc. a set of quadruplets snoring / heart rate / HRV / position; etc.

[0127] This makes it easier to determine the chain of events which, during a predetermined time window, resulted in an abnormal cardio-respiratory characteristic, for example an increase in heart rate, an increase in snoring amplitude, etc. To determine whether a cardio-respiratory characteristic is "abnormal", it is typically planned to compare the value of said characteristic with a reference value, preferably included in the reference model.

[0128] In addition to the determined or selected events, the decision-making model can use as input other information relating to the individual, including but not limited to: their age, weight, height, neck circumference, medical history, eating habits, consumption of products known to be likely to disrupt sleep (alcohol and cigarettes in particular).

[0129] It is also advantageous to plan to calculate the temporal correlation between at least two physiological events of the individual over the same time range (simultaneity of events) or over several time ranges (diachronism of events, for example turning on one side which induces snoring). Determining the position

[0130] The position of the individual during sleep is then determined by quantifying the variations in acceleration on the different axes. Typically, the signals from at least one of the accelerometers and optionally from a set of at least one gyroscope make it possible to determine whether the individual is in one of the following positions: lying on the back, lying on the stomach, lying on the right side, lying on the left side, sitting or semi-sitting, and standing.

[0131] Thanks to the invention, it then becomes possible to carry out a statistical analysis by cross-correlation, which makes it possible to better define the conditions in which the individual presents sleep disorders, for example an accelerated heart rate when snoring on the left side.

[0132] The data from the first and second accelerometers and, where appropriate, the optional additional sensors are analyzed (in real time or a posteriori) by a succession of sliding windows whose duration is predetermined (for example, a few seconds) during a predetermined total period (for example, a few minutes) and the characteristics determined during this predetermined total period are compared with the data from the reference model.

[0133] Data from the first and second accelerometers, the PPG sensor, the microphone, or the individual's position can also be used as a discriminator for the determined or selected events. For example, a cardiac acceleration associated with a vertical acceleration indicates that the individual is certainly getting up, therefore awake, and the corresponding events can then be discriminated from the comparison stage to the reference model.

[0134] Thanks to the present invention, it is possible to obtain dynamic measurements over several nights whereas the prior art only allows a single night.

[0135] The analysis time according to the prior art solutions is generally of the order of one to two hours of work to analyze one night of an individual. On the contrary, according to the invention the calculation time is that of a calculator (processor), i.e. a few seconds. In addition, the calculation is automatic, that is to say independent of any human intervention, in particular from a medical specialist.

[0136] The present invention allows the determination of correlated and abnormal events, such as monitoring treatment of sleep disorders. It is non-intrusive, does not require any bulky traditional equipment (headset, EEG, EMG, oxygenation measurement device, air flow measurement, etc.) and may require only two accelerometers, for a precision of measurement and analysis adapted to the early screening approach in the general or pathological population.

Claims

1. System for determining a set of at least one cardio-respiratory descriptor of an individual during sleep, comprising: - a measuring device comprising: o a first set of at least one accelerometer, configured so as to be placed in a thoracic position on the individual; and o a second set of at least one accelerometer, synchronized with the first set of at least one accelerometer and configured so as to be placed in an abdominal position on said individual; and o optionally, at least one from amongst a photoplethysmographic sensor (PPG), a microphone and a set of at least one gyroscope synchronized with the first set of at least one accelerometer; - a memory for recording the data coming from the accelerometers and from the optional sensors; characterized in that it further comprises: - a reference model recorded in a memory, the model comprising a correspondence between a set of characteristics distributed over time and a set of given physiological events, each event preferably being representative of a potential sleep disorder; - a device for filtering the data coming from the accelerometers configured for extracting low-frequency, medium-frequency and high-frequency ranges, and - a computer configured for: o determining, per extracted frequency band, a set of at least one characteristic representative of a cardio-respiratory and physiological state, together with the time at which said characteristic was extracted; o comparing the temporal distribution of said set of at least one determined characteristic with the temporal distribution of similar characteristics coming from said reference model; o deducing therefrom a set of at least one probable corresponding event which said individual experienced during a predetermined period of time; o determining, based on at least one from amongst said events, a set of at least one cardio-respiratory descriptor of the individual.

2. System according to Claim 1, in which the device for filtering the data is configured for: o extracting a low-frequency range, a medium-frequency range and a high-frequency range from the data coming from the first set of at least one accelerometer, and o extracting at least one low-frequency range from the data coming from the second set of at least one accelerometer.

3. System according to either one of the preceding claims, furthermore comprising - means for determining the position of said individual during sleep, from the data coming from at least one from amongst said first set of at least one accelerometer and said second set of at least one accelerometer.

4. System according to any one of the preceding claims, furthermore comprising at least one of the sensors from amongst - a photoplethysmographic sensor (PPG), - a microphone and - a set of at least one sensor for measuring angular position, synchronized with the first set of at least one accelerometer and configured so as to be placed in a thoracic position on said individual; - the device for filtering the data being furthermore configured for: - coupling the data from a set of at least one sensor for measuring angular position with the data of at least one from amongst the first set of at least one accelerometer and the second set of at least one accelerometer, for the extraction of characteristics on position of the user or on atypical variations in respiratory angular speed; and / or - averaging the data coming from the optical photoplethysmographic sensor.

5. System according to any one of the preceding claims, in which the computer is configured for determining a set of at least one cardio-respiratory descriptor of the individual from amongst: • an apnea-hypopnea index (AHI), • a first set of respiratory forces in the data coming from the first set of at least one accelerometer, • a second set of respiratory forces in the data coming from the second set of at least one accelerometer, • a thorax-abdomen desynchronization index, by synchronized comparison of the first set of respiratory forces and of the second set of respiratory forces, • the heart rate variability (HRV), • the oxymetric profile or oxygen saturation, • the respiratory entropy per unit time, • the percentage of snoring per unit time; and • a night-time profile of the oxygen saturation.

6. System according to any one of the preceding claims, in which the computer is configured for selecting a set of characteristics from amongst the set of the determined characteristics, for example by an algorithm of the Sequential Feature Selection (SFS) or Fast Correlation-Based Filter (FCBF) type, prior to the comparison.

7. System according to any one of the preceding claims, in which the computer is configured for calculating at least one from amongst: - the temporal correlation between at least two events identified over the same range of time or over several ranges of time; - a set of (event / position) pairs, - a set of (event / characteristic) pairs, - a set of (event / characteristic / position) triplets, - a set of (event / / event / characteristic / position) quadruplets.

8. System according to any one of the preceding claims, in which the computer is configured for identifying a chain of events which, during a predetermined time window, resulted in a cardio-respiratory descriptor of abnormal value.

9. System according to any one of the preceding claims, furthermore comprising means of enabling / disabling the accelerometers and optional sensors.

10. System according to any one of the preceding claims, in which the memory for recording the data coming from the accelerometers and from the optional sensors and the memory for recording the reference model are the same; said memory being: - either remote, - or connected via a wired data link with the first set of at least one accelerometer, the second set of at least one accelerometer and also with the optional sensors.

11. System according to any one of the preceding claims, comprising: - a low-pass filter, configured to extract a respiratory component; - a bandpass filter, configured to extract a cardiac component; and - a high-pass filter, configured to extract a snoring component.

12. System according to Claim 11, in which: - the low-pass filter is configured to implement a principal component analysis (PCA) or an analysis based on a discrete Fourier transform; - the bandpass filter is configured to implement adaptive filtering, via envelope extraction, learning, cross-correlation, k-mean clustering, or principal component analysis (PCA), and - the high-pass filter is configured to implement spectral analysis.