An algorithm for estimating the incidence of chronic diseases using sleep snoring frequency

Abstract

The present invention relates to the technical field of home medical equipment, and discloses an algorithm for estimating the incidence of chronic diseases using the frequency of snoring during sleep, comprising the following steps: the tester wears an ADPD107 ring to fall asleep, and places a noise sensor at the head of the bed while sleeping; the sleep state of step one is repeated for one week, and fifty people are tested in the same way, and the data is fed back to a remote terminal. The algorithm for estimating the incidence of chronic diseases using the frequency of snoring during sleep collects data by using PPG signals, which are susceptible to poor blood perfusion and motion artifacts of peripheral tissues. In order to minimize the influence of these factors so as not to interfere with subsequent PPG analysis and heart rate estimation, a preprocessing stage is required. The on-demand algorithm will refuse to estimate based on such a data window. When there are enough data segments that pass the quality check continuously, the second stage of the algorithm will use a MUSIC-based algorithm to calculate the accurate heart rate.

Description

Technical Field

[0001] The present invention relates to the technical field of home medical equipment, and in particular to an algorithm for estimating the incidence of chronic diseases by using the frequency of sleep snoring. Background Art

[0002] Snoring refers to the snoring sound caused by the vibration of the uvula (uvula) due to the narrowing of the upper respiratory tract during sleep. Enlarged tonsils, an enlarged tongue and excessive drinking can also cause snoring. Snoring can lead to daytime sleepiness, fatigue, inattention, headaches, and decreased work ability. Snoring is an independent risk factor for hypertension. Severe snoring is often accompanied by sleep apnea syndrome. The cause of snoring needs to be identified and treated as soon as possible. Snoring is a common sleep phenomenon. Most people now think it is commonplace and do not take it seriously. Some people regard snoring as a sign of a good sleep. In fact, snoring is the enemy of health. Snoring causes repeated pauses in breathing during sleep, resulting in severe hypoxia of the brain and blood, forming hypoxemia, which can induce hypertension, cerebrocardia, cardiac arrhythmia, myocardial infarction, and angina pectoris. If the nighttime breathing pause lasts for more than 120 seconds, sudden death is likely to occur in the early morning.

[0003] Severe snoring, which we also call "malignant snoring", can cause repeated upper airway obstruction and interrupted respiratory airflow during sleep, affecting oxygen exchange and causing a decrease in blood oxygen saturation. It is clinically called obstructive sleep apnea syndrome. The disease can increase the incidence of chronic diseases such as hypertension, cardiovascular and cerebrovascular diseases, diabetes, Alzheimer's disease, respiratory diseases and children's growth and development disorders on the basis of causing metabolic disorders in the body. Therefore, for the calculation of snoring frequency during sleep, the possibility of chronic diseases is determined by the relationship between snoring frequency and blood oxygen content. Therefore, an algorithm is proposed to use the snoring frequency during sleep to infer the incidence of chronic diseases to achieve the above purpose. Summary of the invention

[0004] 1. Technical issues to be solved

[0005] In view of the shortcomings of the prior art, the present invention provides an algorithm for inferring the incidence of chronic diseases using the frequency of snoring during sleep. The algorithm has the advantages of intelligent calculation monitoring and detection and prevention of chronic diseases, and solves the problem of severe snoring, which we can also call "malignant snoring", which can cause repeated upper airway obstruction and interrupted respiratory airflow during sleep, affecting oxygen exchange and causing a decrease in blood oxygen saturation, which is clinically known as obstructive sleep apnea syndrome. This disease can increase the prevalence of chronic diseases such as hypertension, cardiovascular and cerebrovascular diseases, diabetes, Alzheimer's disease, respiratory diseases and children's growth and development disorders on the basis of causing metabolic disorders in the body. Therefore, for the calculation of snoring frequency during sleep, the possibility of chronic diseases is determined by the relationship between snoring frequency and blood oxygen content.

[0006] (II) Technical solution

[0007] In order to achieve the above-mentioned intelligent calculation monitoring and detection and prevention of chronic diseases, the present invention provides the following technical solution: an algorithm for estimating the incidence of chronic diseases using the frequency of sleep snoring, comprising the following steps:

[0008] 1) The tester wears the ADPD107 ring to sleep and places the noise sensor at the head of the bed;

[0009] 2) Repeat the sleep state of step 1 for one week, and test 50 people in the same way, and feed the data back to the remote terminal;

[0010] 3) The remote terminal records the tester's PPG signal through the ADPD107 ring and the tester's snoring frequency through the noise sensor;

[0011] 4) The relationship between the frequency of the tester's snoring and the heart rate harmonic signal model is calculated through the collected data, so as to determine whether the patient has a chronic disease through the heart rate harmonic signal. At the same time, the blood volume changes in the PPG signal can also be used to determine whether the tester has a chronic disease.

[0012] Furthermore, the ADPD107 ring is an ADI multi-sensing ring, which aims to measure multiple vital signs on the human wrist. The ADI ring has PPG, electrocardiogram (ECG), electrodermal activity (EDA), acceleration (ACC) and temperature sensors.

[0013] Furthermore, the noise sensor collects the noise emitted by the tester's snoring and provides noise subspace data for the harmonic signal model. The key factor of the harmonic signal model is the orthogonality of the noise subspace and the signal subspace. The zero point of the noise subspace indicates the signal frequency.

[0014] Furthermore, in step 2), fifty people for one week are used as reference data, so that the sample of the experiment is increased. The influence of external factors on the experimental results can be reduced through a large sample size and a certain period of time. More data can make the test more accurate.

[0015] Furthermore, in the step 4, the calculation of the heart rate harmonic signal model requires obtaining the left eigenvector of the covariance matrix, the diagonal matrix of the eigenvalues, the right eigenvector, the signal and the noise subspace.

[0016] Furthermore, in step 4, the frequency points within the target frequency range of the heart rate can be obtained through the formula, thereby obtaining a waveform diagram of the harmonic signal model, and the change in the heart rate of the tester can be determined through the change in the waveform diagram.

[0017] Furthermore, the blood oxygen data of the PPG signal test axis will be compared with the data in the medical database. After the comparison, the big data will be used to determine whether the test subject has a chronic disease. If the type of chronic disease exists, the detection point will return data; if not, it will be displayed as healthy.

[0018] Furthermore, in step four, ADI provides software and algorithm functions for sensors (embedded) and edge nodes, enabling them to obtain valuable information from the data and send only the most important data to the cloud. The results are very important for the results of the test. This test is based on the on-demand algorithm evolution based on MUSIC.

[0019] (III) Beneficial effects

[0020] Compared with the prior art, the present invention provides an algorithm for estimating the incidence of chronic diseases using the frequency of snoring during sleep, which has the following beneficial effects:

[0021] 1. The algorithm that uses the frequency of sleep snoring to infer the incidence of chronic diseases uses PPG signals for data collection. PPG signals are easily affected by poor blood perfusion of peripheral tissues and motion artifacts. In order to minimize the impact of these factors and avoid interfering with subsequent PPG analysis and heart rate estimation, a preprocessing stage is required. A bandpass filter is required to eliminate the high-frequency components (such as power) and low-frequency components (such as changes in capillary density and venous blood volume, temperature changes, etc.) of the PPG signal. After filtering, a set of signal quality indicators are used to find the first window of the PPG signal suitable for the on-demand algorithm. The first check involves ACC data and PPG signals to determine whether a section of motion-free data can be detected, and then other signal quality indicators are measured. If there is motion in three directions that is higher than a specific threshold of the absolute value of the ACC data, the on-demand algorithm will refuse to estimate based on such a data window. When there are enough data segments that pass the quality check continuously, the second stage of the algorithm will use a MUSIC-based algorithm to calculate the accurate heart rate.

[0022] 2. The algorithm for inferring the incidence of chronic diseases by using the frequency of sleep snoring is calculated by using MUSIC. MUSIC is a subspace-based method that uses a harmonic signal model to estimate the frequency with high accuracy. For PPG signals corrupted by noise, a high-resolution heart rate estimation algorithm is used. In addition, FT evenly distributes time domain noise to the entire frequency domain, which limits the certainty of the estimation. It is difficult to observe a small peak near a larger peak using FT. In this study, a MUSIC-based algorithm is used to estimate the frequency of heart rate. The key idea behind MUSIC is that the noise subspace is orthogonal to the signal subspace, so the noise subspace is orthogonal to the signal subspace. The zero point of space will indicate the signal frequency. When a light beam of a certain wavelength is irradiated onto the surface of the skin at the fingertips, the contraction and expansion of blood vessels will affect the transmission or reflection of light with each heartbeat. When light passes through the skin tissue and then reflects to the photosensor, the light will be attenuated to a certain extent. However, the arteries will be different. Due to the pulsation of blood in the arteries, the absorption of light will naturally change. When we convert light into electrical signals, it is precisely because the absorption of light by the arteries changes while the absorption of light by other tissues remains basically unchanged that the obtained signals can be divided into direct current (DC) signals and alternating current (AC) signals. Extracting the AC signal can reflect the characteristics of blood flow. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a schematic diagram of ECG signal waveform points of an algorithm for estimating the incidence of chronic diseases using sleep snoring frequency according to the present invention. DETAILED DESCRIPTION

[0024] The following will be combined with the embodiments of the present invention and the accompanying drawings to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] An algorithm for estimating the incidence of chronic diseases using the frequency of snoring during sleep includes the following steps:

[0026] 1) The tester wears the ADPD107 ring to sleep and places the noise sensor at the head of the bed;

[0027] 2) Repeat the sleep state of step 1 for one week, and test 50 people in the same way, and feed the data back to the remote terminal;

[0028] 3) The remote terminal records the tester's PPG signal through the ADPD107 ring and the tester's snoring frequency through the noise sensor;

[0029] 4) The relationship between the frequency of the tester's snoring and the heart rate harmonic signal model is calculated through the collected data, so as to determine whether the patient has a chronic disease through the heart rate harmonic signal. At the same time, the blood volume changes in the PPG signal can also be used to determine whether the tester has a chronic disease.

[0030] In this embodiment, the ADPD107 ring is an ADI multi-sensing ring, which aims to measure multiple vital signs on the human wrist. The ADI ring has PPG, electrocardiogram (ECG), electrodermal activity (EDA), acceleration (ACC) and temperature sensors.

[0031] In this embodiment, the noise sensor collects the noise emitted by the tester's snoring and provides noise subspace data for the harmonic signal model. The key factor of the harmonic signal model is the orthogonality of the noise subspace and the signal subspace. The zero point of the noise subspace indicates the signal frequency.

[0032] In this embodiment, fifty people for one week are used as reference data in step 2), so that the sample of the experiment is increased. The influence of external factors on the experimental results is reduced by a large sample size and a certain period of time. More data can make the test more accurate.

[0033] In this embodiment, the calculation of the heart rate harmonic signal model in step 4 requires obtaining the left eigenvector of the covariance matrix, the diagonal matrix of the eigenvalues, the right eigenvector, the signal and noise subspaces;

[0034] The sample covariance matrix format is as follows:

[0035]

[0036] Among them, U is the left eigenvector of the covariance matrix, Λ is the diagonal matrix of eigenvalues, V is the right eigenvector, and the subscripts s and n represent the signal and noise subspaces respectively.

[0037] In this embodiment, in step 4, the frequency points within the target frequency range of the heart rate can be obtained through the formula, thereby obtaining a waveform diagram of the harmonic signal model, and the change of the heart rate of the tester can be determined by the change of the waveform diagram;

[0038] When P is 2 model number, the search frequency vector is defined as:

[0039]

[0040] Wherein, k is the frequency point within the target frequency range of the heart rate, and L is the window length of the data in xm(t).

[0041] In this embodiment, the blood oxygen data of the PPG signal test axis will be compared with the data in the medical database. After the comparison, the big data will be used to determine whether the test subject has a chronic disease. If the type of chronic disease exists, the detection point will return data; if not, it will be displayed as healthy.

[0042] In this embodiment, step four ADI provides software and algorithm functions for sensors (embedded) and edge nodes, enabling them to obtain valuable information from data and send only the most important data to the cloud. Its achievements are very important for the results of the test. This test is based on the evolution of on-demand algorithms based on MUSIC;

[0043] The peak value of MUSIC is:

[0044]

[0045] refer to Figure 1 middle:

[0046] (1) P wave: The first deviated wave is named P wave, which reflects the potential change of atrial depolarization and represents the depolarization of the two atria.

[0047] (2) QRS complex: the largest wave produced by ventricular activation, which reflects the potential changes during ventricular muscle depolarization. The normal interval is 0.08-0.12 seconds. A typical QRS complex refers to three closely connected waves.

[0048] (3) PR interval: from the beginning of the P wave to the beginning of the QRS complex wave, it represents the time from the beginning of atrial muscle depolarization to the beginning of ventricular muscle depolarization. The normal interval is 0.12-2.0 seconds;

[0049] (4) ST segment: It is a flat line after the QRS complex and before the T wave, representing the period from the completion of depolarization of the left and right ventricles to the beginning of repolarization.

[0050] (5) T wave: represents the potential change caused by the repolarization process of ventricular muscle;

[0051] (6) QT interval: represents the total time from the beginning of depolarization to the completion of repolarization of the entire ventricular muscle.

[0052] The beneficial effects of the present invention are as follows: the algorithm for inferring the incidence of chronic diseases using the frequency of sleep snoring uses PPG signals for data collection. The PPG signals are susceptible to poor blood perfusion and motion artifacts of peripheral tissues. In order to minimize the impact of these factors so as not to interfere with subsequent PPG analysis and heart rate estimation, a preprocessing stage is required. A bandpass filter is required to eliminate high-frequency components (such as power supply) and low-frequency components (such as changes in capillary density and venous blood volume, temperature changes, etc.) of the PPG signal. The filtered PPG signal uses a set of signal quality indicators to find the first window of the PPG signal suitable for the on-demand algorithm. The first check involves ACC data and PPG signals to determine whether a section of data without motion can be detected, and then other signal quality indicators are measured. If there is motion in three directions that is higher than a specific threshold of the absolute value of the ACC data, the on-demand algorithm will refuse to make estimates based on such a data window. When there are enough data segments that pass the quality check continuously, the second stage of the algorithm will use a MUSIC-based algorithm to calculate the accurate heart rate.

[0053] Moreover, by adopting MUSIC to calculate the incidence of chronic diseases, MUSIC is a subspace-based method that uses a harmonic signal model to estimate the frequency with high precision. For PPG signals corrupted by noise, a high-resolution heart rate estimation algorithm is used. In addition, FT evenly distributes the time domain noise to the entire frequency domain, which limits the certainty of the estimation. It is difficult to observe a small peak near a larger peak using FT. In this study, a MUSIC-based algorithm is used to estimate the frequency of the heart rate. The key idea behind MUSIC is that the noise subspace is orthogonal to the signal subspace, so the zero point of the noise subspace indicates the signal frequency. When a light beam of a certain wavelength is irradiated onto the surface of the fingertip skin, the contraction and expansion of the blood vessels will affect the transmission of light with each heartbeat. The radiation or reflection of light, when the light passes through the skin tissue and then reflects to the photosensitive sensor, the light will be attenuated to a certain extent, but the artery will be different. Because there is blood pulsation in the artery, the absorption of light will naturally change. When we convert light into electrical signals, it is precisely because the absorption of light by the artery changes while the absorption of light by other tissues remains basically unchanged. The obtained signal can be divided into a direct current (DC) signal and an alternating current (AC) signal. Extracting the AC signal can reflect the characteristics of blood flow and solve the problem of severe snoring, which we can also call "malignant snoring". It can cause repeated upper airway obstruction during sleep, interrupted respiratory airflow, affect oxygen exchange, and cause a decrease in blood oxygen saturation. It is clinically called obstructive sleep apnea syndrome. This disease can increase the prevalence of chronic diseases such as hypertension, cardiovascular and cerebrovascular diseases, diabetes, Alzheimer's disease, respiratory diseases, and children's growth and development disorders on the basis of causing metabolic disorders in the body. Therefore, for the calculation of snoring frequency during sleep, the relationship between snoring frequency and blood oxygen content is used to determine the possibility of chronic diseases.

[0054] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An algorithm for estimating the incidence of chronic diseases using the frequency of snoring during sleep, characterized in that: The following steps are involved: The PPG signal of the tester is recorded by the ADPD107 ring. The ADPD107 ring is an ADI multi-sensing ring, which aims to measure multiple vital signs on the human wrist. The ADI ring has PPG, electrocardiogram (ECG), electrodermal activity (EDA), acceleration (ACC) and temperature sensors; The frequency of heart rate is estimated using a MUSIC-based algorithm. The key idea of ​​MUSIC is that the noise subspace is orthogonal to the signal subspace, and the zero point of the noise subspace indicates the signal frequency. MUSIC is a subspace-based method that uses a harmonic signal model to estimate the frequency with high accuracy. It is a high-resolution heart rate estimation algorithm for PPG signals corrupted by noise. The snoring frequency of the tester is recorded by a noise sensor; the noise sensor collects the noise emitted by the tester's snoring and provides noise subspace data for the harmonic signal model. The key to the harmonic signal model is the orthogonal value of the noise subspace and the signal subspace, and the zero point of the noise subspace indicates the signal frequency; The relationship between the tester's snoring frequency and the heart rate harmonic signal model is calculated through the collected data, so as to determine whether the patient has a chronic disease through the heart rate harmonic signal.

2. The algorithm for estimating the incidence of chronic diseases using the frequency of snoring during sleep according to claim 1, characterized in that: The calculation of the heart rate harmonic signal model requires the left eigenvector of the covariance matrix, the diagonal matrix of the eigenvalues, the right eigenvector of the covariance matrix, the signal subspace and the noise subspace.

3. The algorithm for estimating the incidence of chronic diseases using the frequency of snoring during sleep according to claim 1, characterized in that: By obtaining the frequency points within the target frequency range of the heart rate, a waveform diagram of the harmonic signal model is obtained, and changes in the heart rate of the tester are determined by changes in the waveform diagram.

4. The algorithm for estimating the incidence of chronic diseases using the frequency of snoring during sleep according to claim 1, characterized in that: Also includes: The blood oxygen data of the PPG signal test axis is compared with the data in the medical database. After the comparison, the big data is used to determine whether the test subject has a chronic disease. If the type of chronic disease exists, the data will be returned. If not, it will be displayed as healthy.

Images