A sleep data intelligent collection and processing method and system

Abstract

The present invention discloses a method and system for intelligently collecting and processing sleep data, and relates to the field of biomedical technology. The system realizes comprehensive acquisition and in-depth analysis of sleep data by combining a sleep signal detection module, a signal feature processing module, and a breathing and heart rate abnormality algorithm module. First, the sleep signal detection module uses relevant breathing signal detection equipment and heart rate signal detection equipment to capture the breathing signal and heart rate signal of the patient during sleep in real-time monitoring. These original signals are then transmitted to the signal feature processing module, where comprehensive preprocessing and feature extraction are performed. After preprocessing, the system can not only extract basic sleep features such as apnea, normal breathing and heart rate variability, but also analyze sleep data more deeply. This detailed feature extraction enables the system to more accurately understand the patient's sleep state and health status.

Description

Technical Field

[0001] The present invention relates to the field of biomedical technology, and in particular to a method and system for intelligently collecting and processing sleep data. Background Art

[0002] Sleep is an important part of human physiological activities and is essential for maintaining good health and improving the quality of life. Sleep apnea is a common sleep disorder characterized by repeated breathing pauses and recovery during sleep. By detecting the breathing status, the disease can be detected and diagnosed early, so that timely intervention can be carried out to prevent serious consequences such as high blood pressure, heart disease and stroke. At the same time, by monitoring the heart rate status, problems such as arrhythmia can be identified. These problems may not be obvious during the day, but they will appear during sleep. Early detection can help prevent more serious heart diseases.

[0003] In the Chinese invention application with application publication number CN107647846A, a new sleep apnea detection system is disclosed. In order to monitor sleep apnea hypopnea syndrome, a new sleep apnea detection system is invented based on a grating encoding sensor. The grating encoding sensor is embedded in a belly belt tied around the patient's waist. During the breathing process, the position of the grating in the belly belt changes due to the change in abdominal volume. The movement of the grating is detected by a photodiode sensor array, and the signal is analyzed and processed by a single-chip microcomputer to determine the process of exhalation and inhalation, and calculate the breathing cycle. If the cycle exceeds the time limit, it is apnea. After many experiments and comparison with the detection results of respiratory monitoring equipment, the average accuracy of this system is higher than 95%.

[0004] In combination with the prior art, the above application still has the following deficiencies:

[0005] Existing sleep monitoring technologies often have various deficiencies, which are mainly reflected in monitoring accuracy, data processing efficiency and comprehensive analysis capabilities. Most traditional methods can only monitor breathing or heart rate separately, and lack comprehensive analysis of the relationship between the two, resulting in an inability to comprehensively assess sleep health status. At the same time, many existing devices are inefficient in data processing, and are not easy to analyze and feedback sleep status in real time, delaying the diagnosis and treatment of potential sleep disorders. Especially for sleep apnea syndrome, traditional methods can often only judge by a single indicator, while ignoring the complex relationship between heart rate changes and apnea. Such limitations may lead to missed diagnosis or misdiagnosis. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a method and system for intelligently collecting and processing sleep data, which solves the problems mentioned in the background technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: including a sleep signal detection module, a signal feature processing module, a breathing abnormality algorithm module, a heart rate abnormality algorithm module and a comprehensive evaluation module;

[0008] The sleep signal detection module collects the breathing signal and heart rate signal during real-time sleep by using relevant breathing signal detection equipment and heart rate signal detection equipment;

[0009] The signal feature processing module is used to pre-process the collected breathing signal and heart rate signal, and then extract the features of the breathing signal and the heart rate signal to obtain the breathing feature signal and the heart rate feature signal;

[0010] The breathing abnormality algorithm module is used to perform data annotation based on the historical data of the extracted breathing characteristic signal, and use the extracted features and the annotated breathing signal data to train the SVM model. For each historical data of the breathing characteristic signal marked as Xi, the kernel function value of the historical data and the real-time collected breathing characteristic signal X is calculated. , and then calculate the weighted sum to obtain the respiratory abnormality coefficient Hxyc, and then use the symbol function Classify the respiratory abnormality coefficient Hxyc output;

[0011] The heart rate abnormality algorithm module triggers the heart rate abnormality algorithm when a sleep apnea event occurs, uses a signal processing algorithm to perform abnormality detection based on the extracted heart rate characteristic signal, calculates and obtains the heart rate variability HRV, and sets a heart rate abnormality detection threshold L to perform heart rate abnormality assessment;

[0012] The comprehensive evaluation module is used to perform comprehensive analysis and generate corresponding diagnosis recommendations based on the output results of the abnormal breathing coefficient Hxyc and the heart rate variability HRV.

[0013] Preferably, the sleep signal detection module includes a breathing signal detection unit and a heart rate signal detection unit;

[0014] The respiratory signal detection unit collects the respiratory signal during sleep in real time by using a polysomnography (PSG) system, wherein the respiratory signal includes nasal airflow, oral airflow and blood oxygen saturation;

[0015] The heart rate signal detection unit collects the heart rate signal during sleep by using a heart rate sensor, and the heart rate signal includes the number of heartbeat intervals N and the number of heartbeats per minute R.

[0016] Preferably, the signal feature processing module includes a preprocessing unit and a feature extraction unit;

[0017] The preprocessing unit is used to perform preprocessing based on the collected respiratory signal and heart rate signal. The specific preprocessing scheme is as follows: use low-pass filtering and high-pass filtering to eliminate high-frequency noise from the respiratory signal and heart rate signal, and remove interference caused by body movement factors, then divide the continuous respiratory signal data into windows of fixed length, and finally perform normalization processing on the collected respiratory signal and heart rate signal;

[0018] The feature extraction unit includes a breathing feature extraction unit and a heart rate feature extraction unit;

[0019] The respiratory feature extraction unit is used to extract time domain features from the pre-processed respiratory signal by analyzing and counting the time series data, and the time domain features include the average respiratory frequency Pl, the respiratory time Hj, the number of blood oxygen saturation decreases Xj, and the blood oxygen saturation decrease amplitude Xf;

[0020] The calculation method is as follows:

[0021] The average respiratory rate Pl is calculated by recording the nasal and oral airflows of the respiratory signal, identifying the respiratory cycle, one inhalation and one exhalation constitute one cycle, calculating the number of breaths per unit time, and then taking the average value;

[0022] The breathing time Hj records the breathing signal, identifies the start and end of each breathing cycle, and calculates the time difference between the start time points of two adjacent breathing cycles;

[0023] The number of blood oxygen saturation decreases Xj and the amplitude of blood oxygen saturation decrease Xf are determined by recording the blood oxygen saturation signal, determining the baseline blood oxygen saturation value and the decrease threshold, identifying each blood oxygen saturation decrease event, and recording the start time and end time of the decrease to calculate the amplitude of each decrease;

[0024] The heart rate feature extraction unit is used to extract features from the preprocessed heart rate signal, and obtain the intervals between two adjacent heart beats by performing time domain analysis based on the heart beat interval number N and the heart beats per minute number R. , and then by averaging The interval uses the average value algorithm to obtain the average heartbeat interval .

[0025] Preferably, the abnormal breathing algorithm module includes a kernel function algorithm unit and a SVM model algorithm unit;

[0026] The kernel function algorithm unit is used to use the sleep apnea events manually annotated by doctors as the annotations of training data, which are annotated as "apnea" and "normal breathing" category labels, and the data after the historical respiratory signal features are extracted are marked as historical respiratory feature vectors Xi, wherein the historical respiratory feature vector Xi={Pl i , Hj i, Xj i , Xf i}, the respiratory feature vector collected in real time is marked as the respiratory feature vector X, and the radial basis function kernel is constructed based on the historical respiratory feature vector Xi and the respiratory feature vector X. ;

[0027] The radial basis function kernel Obtained by the following formula;

[0028] ;

[0029] Where exp represents the exponential function, Represents the parameters of the radial basis function kernel, which controls the width of the kernel function.

[0030] Preferably, the SVM model algorithm unit includes a classification prediction unit and a classification decision unit;

[0031] The classification prediction unit is used to train the SVM model using the extracted respiratory signal features and the annotated labels, evaluate the model performance using the cross-validation technique, and adjust the SVM model parameters, including the penalty parameter C and the kernel function parameter γ to optimize the SVM model, and then extract the same features from the respiratory feature vector X, use the trained SVM model to construct the respiratory abnormality algorithm formula, output the respiratory abnormality coefficient Hxyc, and analyze the "apnea" or "normal breathing" category;

[0032] The abnormal breathing coefficient Hxyc is obtained by the following algorithm formula;

[0033] ;

[0034] In the formula, Represents the Lagrange multiplier, which is a non-negative number obtained through the training process, indicating the importance of each training sample in the final model. corresponds to the support vectors, which are points on or between the boundaries of the model, represents the category label of the i-th sample in the training data set. For binary classification problems, the label is usually -1 or 1; b represents the bias term, a constant obtained through the training process, which is used to adjust the position of the decision boundary.

[0035] Preferably, the classification decision unit is used to use a symbolic function The formula classifies the output results of the output breathing abnormality coefficient Hxyc;

[0036] The symbol function Obtained by the following formula;

[0037] ;

[0038] In the formula, represents the category label predicted by the SVM model, and sign represents the sign function used to determine the predicted category.

[0039] Preferably, the abnormal breathing coefficient Hxyc is calculated by the SVM model output, and the symbol function is applied Determine the prediction category. The specific prediction category method is as follows;

[0040] When the respiratory abnormality coefficient Hxyc>0, it means =1, indicating that a sleep apnea event occurred in the respiratory signal during the predicted time period, which is considered as abnormal breathing. In this case, further abnormal heart rate detection is initiated;

[0041] When the respiratory abnormality coefficient Hxyc < 0, it means =-1, indicating that there is no sleep apnea event in the respiratory signal during the predicted time period, and the breathing is judged to be normal, that is, the respiratory characteristics during this time period are normal and do not meet the apnea standard.

[0042] Preferably, the abnormal heart rate algorithm module includes an abnormal heart rate algorithm unit and an abnormal heart rate evaluation unit;

[0043] The heart rate abnormality algorithm unit is triggered by monitoring the apnea event, and performs heart rate abnormality detection by using a signal processing algorithm based on the extracted real-time heart rate features, and calculates and obtains the heart rate variability HRV;

[0044] The heart rate variability HRV is obtained by the following algorithm formula;

[0045]

[0046] The heart rate abnormality evaluation unit is used to compare and evaluate the preset heart rate abnormality detection threshold L with the result of the heart rate variability HRV output, and analyze the heart rate situation during apnea. The specific evaluation results are as follows;

[0047] When the heart rate variability HRV>heart rate abnormality detection threshold L, it is judged as heart rate abnormality;

[0048] When heart rate variability HRV = heart rate abnormality detection threshold L, the heart rate is judged to be normal;

[0049] When the heart rate variability HRV is less than the heart rate abnormality detection threshold L, it is determined that the heart rate is abnormal.

[0050] Preferably, the comprehensive evaluation module is used to comprehensively analyze the results of the abnormal breathing coefficient Hxyc and the heart rate variability HRV output, and generate relevant diagnostic suggestions. The specific comprehensive analysis content is as follows;

[0051] When the heart rate is normal but the breathing is abnormal, the individual is assessed to have respiratory problems. Although the heart rate is normal, the abnormal breathing requires further formal examination of the respiratory system to rule out potential respiratory problems.

[0052] When the heart rate is normal but breathing is abnormal, the patient is at potential health risk. A high heart rate may indicate emotional stress and insufficient physical activity, while abnormal breathing may indicate poor sleep quality or potential respiratory problems. Further medical examination is required at this time.

[0053] A method for intelligently collecting and processing sleep data comprises the following steps:

[0054] S1. First, according to the signal type to be collected, select a sleep monitoring device and sensor with corresponding functions to record the changes of breathing signals and heart rate signals in real time during sleep;

[0055] S2. After the signal is collected, the raw data is preprocessed to eliminate possible noise and interference, and then the characteristic signal and the heart rate characteristic signal are extracted from the preprocessed breathing signal and heart rate signal for subsequent abnormality detection and analysis;

[0056] S3, extract features from the preprocessed respiratory signal, manually label the sleep apnea event category labels through doctors, and construct the radial basis function kernel , and then further calculate and analyze the respiratory abnormality coefficient Hxyc, and use the symbolic function The formula classifies the output results of the output breathing abnormality coefficient Hxyc, and predicts and classifies the abnormalities produced by breathing during sleep;

[0057] S4. When a breathing pause event is detected, a heart rate abnormality detection is further performed. Based on the features extracted from the pre-processed heart rate signal, a signal processing algorithm is used to perform heart rate abnormality detection calculation to obtain the heart rate variability HRV, and then the degree of the heart rate variability HRV is evaluated to evaluate the abnormal and normal conditions of the heart rate;

[0058] S5. Finally, a comprehensive analysis is performed based on the results of abnormal detection of respiratory signals and heart rate signals, and relevant diagnostic suggestions are provided.

[0059] Beneficial Effects

[0060] The present invention provides a method and system for intelligently collecting and processing sleep data, which has the following beneficial effects:

[0061] (1) The system realizes comprehensive acquisition and in-depth analysis of sleep data by combining the sleep signal detection module, the signal feature processing module, and the breathing and heart rate abnormality algorithm module. First, the sleep signal detection module uses relevant breathing signal detection equipment and heart rate signal detection equipment to capture the patient's breathing signal and heart rate signal during sleep in real-time monitoring. These raw signals are then transmitted to the signal feature processing module, where comprehensive preprocessing and feature extraction are performed. This process includes low-pass filtering and high-pass filtering to eliminate noise, and normalization processing to ensure data consistency and comparability. After preprocessing, the system can not only extract basic sleep features such as breathing frequency and heart rate variability, but also analyze sleep data more deeply. This detailed feature extraction enables the system to more accurately understand the patient's sleep state and health status. Through the breathing and heart rate abnormality algorithm module, the system further uses the extracted features for anomaly detection and analysis. Using advanced algorithms such as the SVM model, the system can monitor and identify possible breathing and heart rate abnormalities during sleep in real time. Therefore, this intelligent acquisition and processing system can not only comprehensively acquire sleep data, but also provide important auxiliary information for medical professionals through in-depth analysis and anomaly detection.

[0062] (2) In the signal feature processing module of this system, the use of preprocessing units greatly improves the efficiency of data processing. By eliminating high-frequency noise through low-pass filtering and high-pass filtering, combined with normalization processing, the system effectively reduces interference and improves data accuracy. In terms of abnormality detection, the SVM model is used to detect respiratory abnormalities. Combined with the kernel function algorithm and classification prediction unit, the system can more accurately identify sleep apnea events, providing a reliable basis for the diagnosis of sleep disorders.

[0063] (3) The system uses a comprehensive evaluation module to deeply analyze the output results of the respiratory abnormality coefficient Hxyc and heart rate variability HRV, and generates corresponding diagnostic recommendations. This not only helps medical professionals better understand patients' sleep problems, but also provides patients with personalized treatment plans. For example, when the heart rate is normal but the breathing is abnormal, the system may recommend further respiratory system examinations; and when the heart rate is abnormal and the breathing is abnormal at the same time, the system may propose treatment plans for poor sleep quality or emotional stress, thereby paying comprehensive attention to the patient's health status. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a schematic diagram of a sleep data intelligent collection and processing system flow of the present invention;

[0065] Figure 2 This is a schematic diagram of the steps of a method for intelligently collecting and processing sleep data according to the present invention; DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present invention 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.

[0067] Example 1

[0068] See also Figure 1 , the present invention provides a sleep data intelligent collection and processing system. To achieve the above purpose, the present invention is implemented through the following technical solutions: including a sleep signal detection module, a signal feature processing module, a breathing abnormality algorithm module, a heart rate abnormality algorithm module and a comprehensive evaluation module;

[0069] The sleep signal detection module uses relevant breathing signal detection equipment and heart rate signal detection equipment to collect breathing signals and heart rate signals during real-time sleep;

[0070] The signal feature processing module is used to pre-process the collected respiratory signal and heart rate signal, and then extract the features of the respiratory signal and heart rate signal to obtain the respiratory feature signal and the heart rate feature signal;

[0071] The breathing abnormality algorithm module is used to perform data annotation based on the historical data of the extracted breathing characteristic signal, and use the extracted features and annotated breathing signal data to train the SVM model. For each historical data of the breathing characteristic signal marked as Xi, the kernel function value of the historical data and the real-time collected breathing characteristic signal X is calculated. , and then calculate the weighted sum to obtain the respiratory abnormality coefficient Hxyc, and then use the symbol function Classify the respiratory abnormality coefficient Hxyc output;

[0072] The heart rate abnormality algorithm module triggers the heart rate abnormality algorithm when a sleep apnea event occurs, uses a signal processing algorithm to perform abnormality detection based on the extracted heart rate characteristic signal, calculates and obtains the heart rate variability HRV, and sets the heart rate abnormality detection threshold L to evaluate the heart rate abnormality;

[0073] The comprehensive evaluation module is used to conduct comprehensive analysis and generate corresponding diagnostic recommendations based on the output results of the abnormal breathing coefficient Hxyc and heart rate variability HRV.

[0074] In this embodiment, the system firstly acquires and monitors the respiratory signal and heart rate signal during sleep through the real-time acquisition and monitoring of the sleep signal detection module, and the system can obtain the respiratory signal and heart rate signal during sleep, providing sufficient data support for subsequent analysis. Secondly, the application of the signal feature processing module enables the collected signals to be effectively preprocessed and feature extracted, improving the data quality and information volume, thereby providing a more reliable basis for subsequent algorithm analysis. In addition, the combined use of the abnormal breathing algorithm module and the abnormal heart rate algorithm module can not only timely identify and evaluate the abnormal breathing and heart rate during sleep, but also accurately detect and evaluate abnormalities through algorithm calculation and setting of the abnormal heart rate detection threshold L, thereby providing doctors with more accurate diagnostic information and treatment suggestions. The introduction of the comprehensive evaluation module comprehensively analyzes the results of the abnormal breathing coefficient and heart rate variability, providing doctors with a more complete sleep quality assessment and disorder diagnosis, and promoting the development and clinical practice of the field of sleep medicine.

[0075] Compared with existing technical means, the improvement of this sleep monitoring system lies in its multi-module design and the accuracy of algorithm processing. Traditional sleep monitoring often relies on a single or limited number of signal acquisition devices, while this system improves the comprehensiveness and accuracy of monitoring by integrating the detection of respiratory signals and heart rate signals. In addition, the algorithm model and data processing method used are more intelligent and refined, which can better cope with the complex characteristics and changing patterns of sleep signals, and improve the accuracy and efficiency of abnormality detection and evaluation.

[0076] Example 2

[0077] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the sleep signal detection module includes a breathing signal detection unit and a heart rate signal detection unit;

[0078] The respiratory signal detection unit collects the respiratory signals during sleep in real time by using the polysomnography (PSG) system. The respiratory signals include nasal airflow, oral airflow and blood oxygen saturation.

[0079] The heart rate signal detection unit collects the heart rate signal during sleep by using the heart rate sensor. The heart rate signal includes the number of heartbeat intervals N and the number of heartbeats per minute R.

[0080] The signal feature processing module includes a preprocessing unit and a feature extraction unit;

[0081] The preprocessing unit is used to perform preprocessing based on the collected respiratory signal and heart rate signal. The specific preprocessing scheme is as follows: low-pass filtering and high-pass filtering are used to eliminate high-frequency noise from the respiratory signal and heart rate signal, and the interference caused by body movement factors is removed. The continuous respiratory signal data is then divided into windows of fixed length, and finally the collected respiratory signal and heart rate signal are normalized;

[0082] The feature extraction unit includes a breathing feature extraction unit and a heart rate feature extraction unit;

[0083] The respiratory feature extraction unit is used to extract time domain features from the preprocessed respiratory signal by analyzing and counting the time series data, and the time domain features include the average respiratory frequency Pl, the respiratory time Hj, the number of blood oxygen saturation decreases Xj, and the blood oxygen saturation decrease amplitude Xf;

[0084] The calculation method is as follows:

[0085] The average respiratory rate Pl is calculated by recording the nasal and oral airflows of the respiratory signal, identifying the respiratory cycle, one inhalation and one exhalation constitute one cycle, calculating the number of breaths per unit time, and then taking the average value;

[0086] The breathing time Hj records the breathing signal, identifies the start and end of each breathing cycle, and calculates the time difference between the start time points of two adjacent breathing cycles;

[0087] The number of blood oxygen saturation decreases Xj and the amplitude of blood oxygen saturation decrease Xf are determined by recording the blood oxygen saturation signal, determining the baseline blood oxygen saturation value and the decrease threshold, identifying each blood oxygen saturation decrease event, and recording the start time and end time of the decrease to calculate the amplitude of each decrease;

[0088] The heart rate feature extraction unit is used to extract features from the preprocessed heart rate signal, and perform time domain analysis to obtain the intervals between two adjacent heart beats using the heart beat interval number N and the heart beats per minute R. , and then by averaging The interval uses the average value algorithm to obtain the average heartbeat interval .

[0089] In this embodiment, the system uses a polysomnography (PSG) system and a heart rate sensor to realize the real-time acquisition of respiratory signals and heart rate signals during sleep by dividing the sleep signal detection module into a respiratory signal detection unit and a heart rate signal detection unit. This design not only provides comprehensive sleep signal data, but also ensures the accuracy and continuity of the data, laying a solid foundation for subsequent analysis and diagnosis. Secondly, the application of the preprocessing unit and the feature extraction unit in the signal feature processing module effectively eliminates the noise and interference in the signal through a series of signal processing methods, and improves the quality and reliability of the data. The feature extraction unit obtains the key features of the respiratory signal and the heart rate signal through time domain analysis and statistical methods, providing important information support for subsequent algorithm analysis and diagnosis. Therefore, this modularly designed sleep monitoring system can not only comprehensively and accurately collect sleep signals, but also effectively extract key features, providing strong support for sleep quality assessment and disorder diagnosis. At the same time, the respiratory signal detection unit and the heart rate signal detection unit in the sleep signal detection module, as well as the preprocessing unit and the feature extraction unit in the signal feature processing module, adopt a series of sophisticated processing methods and algorithms. These methods can not only eliminate noise and interference in the signal, but also extract key features in the signal, as well as heart rate intervals and heartbeat times. The acquisition and analysis of these features provide an important basis for the identification and evaluation of sleep abnormalities. Through this refined signal processing and feature extraction, the sleep monitoring system can more accurately capture abnormal conditions during sleep.

[0090] Example 3

[0091] This embodiment is explained in Example 2. Please refer to Figure 1 ,Specifically: the breathing abnormality algorithm module includes a kernel function algorithm unit and a SVM model algorithm unit;

[0092] The kernel function algorithm unit is used to use the sleep apnea events manually annotated by doctors as the annotations of training data, which will be annotated as "apnea" and "normal breathing" category labels. After the historical respiratory signal features are extracted, the data is marked as the historical respiratory feature vector Xi. The historical respiratory feature vector Xi={Pl i , Hj i , Xj i , Xf i}, the respiratory feature vector collected in real time is marked as the respiratory feature vector X, and the radial basis function kernel is constructed based on the historical respiratory feature vector Xi and the respiratory feature vector X. ;

[0093] Radial Basis Function Kernel Obtained by the following formula;

[0094] ;

[0095] Where exp represents the exponential function, Represents the parameters of the radial basis function kernel, which controls the width of the kernel function.

[0096] The SVM model algorithm unit includes a classification prediction unit and a classification decision unit;

[0097] The classification prediction unit is used to train the SVM model using the extracted respiratory signal features and the annotated labels, evaluate the model performance using the cross-validation technique, and adjust the SVM model parameters, including the penalty parameter C and the kernel function parameter γ to optimize the SVM model, and then extract the same features from the respiratory feature vector X, use the trained SVM model to construct the respiratory abnormality algorithm formula, output the respiratory abnormality coefficient Hxyc, and analyze the "apnea" or "normal breathing" category;

[0098] The respiratory abnormality coefficient Hxyc is obtained by the following algorithm formula;

[0099] ;

[0100] In the formula, Represents the Lagrange multiplier, which is a non-negative number obtained through the training process, indicating the importance of each training sample in the final model. corresponds to the support vectors, which are points on or between the boundaries of the model, represents the category label of the i-th sample in the training data set. For binary classification problems, the label is usually -1 or 1; b represents the bias term, a constant obtained through the training process, which is used to adjust the position of the decision boundary.

[0101] Classification decision unit is used to use symbolic function The formula classifies the output results of the output breathing abnormality coefficient Hxyc;

[0102] Symbolic Function Obtained by the following formula;

[0103] ;

[0104] In the formula, Represents the category label predicted by the SVM model, and sign represents the sign function used to determine the predicted category.

[0105] The respiratory abnormality coefficient Hxyc calculated by the SVM model output and the symbolic function is applied Determine the prediction category. The specific prediction category method is as follows;

[0106] When the respiratory abnormality coefficient Hxyc>0, it means =1, indicating that a sleep apnea event occurred in the respiratory signal during the predicted time period, which is considered as abnormal breathing. In this case, further abnormal heart rate detection is initiated;

[0107] When the respiratory abnormality coefficient Hxyc < 0, it means =-1, indicating that there is no sleep apnea event in the respiratory signal during the predicted time period, and the breathing is judged to be normal, that is, the respiratory characteristics during this time period are normal and do not meet the apnea standard.

[0108] In this embodiment, the design of the breathing abnormality algorithm module makes full use of the combination of the kernel function algorithm unit and the SVM model algorithm unit to provide an efficient and reliable method for the identification and classification of sleep apnea events. First, the kernel function algorithm unit constructs a radial basis function kernel by training the sleep apnea events manually annotated by doctors. , the historical breathing feature vector Xi and the real-time collected breathing feature vector X are effectively mapped into feature space, laying the foundation for the subsequent SVM model training. Secondly, the SVM model algorithm unit uses the combination of the classification prediction unit and the classification decision unit to classify the breathing features collected in real time using the trained SVM model, output the breathing abnormality coefficient Hxyc, and classify and judge the abnormality coefficient through the symbolic function. This design can not only quickly and accurately identify abnormal breathing events during sleep, but also provide effective starting conditions for subsequent abnormal heart rate detection, thereby improving the comprehensiveness and accuracy of the sleep monitoring system.

[0109] Example 4

[0110] This embodiment is explained in Example 3, please refer to Figure 1 ,Specifically: the abnormal heart rate algorithm module includes a heart rate abnormal algorithm unit and a heart rate abnormal evaluation unit;

[0111] The heart rate abnormality algorithm unit is triggered by monitoring the apnea event, and uses the signal processing algorithm to detect heart rate abnormality based on the extracted real-time heart rate features, and calculates and obtains the heart rate variability HRV;

[0112] Heart rate variability HRV is obtained through the following algorithm formula;

[0113]

[0114] The heart rate abnormality evaluation unit is used to compare and evaluate the preset heart rate abnormality detection threshold L with the result of the heart rate variability HRV output, and analyze the heart rate situation during apnea. The specific evaluation results are as follows;

[0115] When the heart rate variability HRV>heart rate abnormality detection threshold L, it is judged as heart rate abnormality;

[0116] When heart rate variability HRV = heart rate abnormality detection threshold L, the heart rate is judged to be normal;

[0117] When the heart rate variability HRV is less than the heart rate abnormality detection threshold L, it is determined that the heart rate is abnormal.

[0118] In this embodiment, the design of the heart rate abnormality algorithm module aims to achieve effective detection and evaluation of heart rate variability during sleep through the heart rate abnormality algorithm unit and the heart rate abnormality evaluation unit, providing important support for the comprehensive evaluation of sleep quality. First, the heart rate abnormality algorithm unit extracts heart rate features in real time by monitoring the triggering of apnea events and uses a signal processing algorithm to detect heart rate abnormalities, thereby obtaining heart rate variability HRV. This process can not only capture the changes in heart rate during sleep in a timely manner, but also provide a data basis for subsequent abnormality evaluation. Secondly, the heart rate abnormality evaluation unit compares and evaluates the preset heart rate abnormality detection threshold L with the real-time collected heart rate variability HRV to specifically determine whether the heart rate during sleep is normal or abnormal.

[0119] Example 5

[0120] This embodiment is explained in Example 1, please refer to Figure 1 Specifically: the comprehensive evaluation module is used to comprehensively analyze the output results of the respiratory abnormality coefficient Hxyc and the heart rate variability HRV, and generate relevant diagnostic suggestions. The specific comprehensive analysis content is as follows;

[0121] When the heart rate is normal but the breathing is abnormal, the individual is assessed to have respiratory problems. Although the heart rate is normal, the abnormal breathing requires further formal examination of the respiratory system to rule out potential respiratory problems.

[0122] When the heart rate is normal but breathing is abnormal, the patient is at potential health risk. A high heart rate may indicate emotional stress and insufficient physical activity, while abnormal breathing may indicate poor sleep quality or potential respiratory problems. Further medical examination is required at this time.

[0123] In this embodiment, the design of the comprehensive evaluation module adds a layer of intelligent diagnostic capability to the sleep monitoring system. By comprehensively analyzing the results of the abnormal breathing coefficient Hxyc and the heart rate variability HRV output, it provides doctors with more targeted diagnostic suggestions, further improving the clinical application value of the system. First, when the heart rate is normal and the breathing is abnormal, the system evaluates that the individual may have respiratory problems. Even if the heart rate is normal, the abnormal breathing may indicate potential sleep quality problems or respiratory abnormalities. In this case, the comprehensive evaluation module recommends further prompting for formal examinations to rule out potential respiratory problems and provide individuals with a more comprehensive health assessment. Secondly, when the heart rate is normal and the breathing is abnormal, the system believes that the patient has potential health risks. A high heart rate may indicate emotional stress or insufficient physical activity, while abnormal breathing may indicate poor sleep quality or potential respiratory problems. In this case, the comprehensive evaluation module recommends further medical examinations to further evaluate the patient's health status and provide corresponding treatment and management suggestions for the individual. The design of this comprehensive evaluation module can not only provide a more comprehensive health assessment, but also provide doctors with more targeted diagnostic suggestions, which brings significant benefits to the clinical application of sleep monitoring systems.

[0124] Example 6

[0125] See also Figure 1 and Figure 2 , a method for intelligently collecting and processing sleep data, comprising the following steps:

[0126] S1. First, according to the signal type to be collected, select a sleep monitoring device and sensor with corresponding functions to record the changes of breathing signals and heart rate signals in real time during sleep;

[0127] S2. After the signal is collected, the raw data is preprocessed to eliminate possible noise and interference, and then the characteristic signal and the heart rate characteristic signal are extracted from the preprocessed breathing signal and heart rate signal for subsequent abnormality detection and analysis;

[0128] S3, extract features from the preprocessed respiratory signal, manually label the sleep apnea event category labels through doctors, and construct the radial basis function kernel , and then further calculate and analyze the respiratory abnormality coefficient Hxyc, and use the symbolic function The formula classifies the output results of the output breathing abnormality coefficient Hxyc, and predicts and classifies the abnormalities produced by breathing during sleep;

[0129] S4. When a breathing pause event is detected, a heart rate abnormality detection is further performed. Based on the features extracted from the pre-processed heart rate signal, a signal processing algorithm is used to perform heart rate abnormality detection calculation to obtain the heart rate variability HRV, and then the degree of the heart rate variability HRV is evaluated to evaluate the abnormal and normal conditions of the heart rate;

[0130] S5. Finally, a comprehensive analysis is performed based on the results of abnormal detection of respiratory signals and heart rate signals, and relevant diagnostic suggestions are provided.

[0131] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. A sleep data intelligent collection and processing system, characterized in that: It includes a sleep signal detection module, a signal feature processing module, a breathing abnormality algorithm module, a heart rate abnormality algorithm module and a comprehensive evaluation module; The sleep signal detection module collects the breathing signal and heart rate signal during sleep in real time by using relevant breathing signal detection equipment and heart rate signal detection equipment; The sleep signal detection module includes a breathing signal detection unit and a heart rate signal detection unit; The respiratory signal detection unit collects the respiratory signal during sleep in real time by using a polysomnography (PSG) system, wherein the respiratory signal includes nasal airflow, oral airflow and blood oxygen saturation; The heart rate signal detection unit collects the heart rate signal during sleep by using the heart rate sensor, wherein the heart rate signal includes the number of heartbeat intervals N and the number of heartbeats per minute R; The signal feature processing module is used to pre-process the collected breathing signal and heart rate signal, and then extract the features of the breathing signal and the heart rate signal to obtain the breathing feature signal and the heart rate feature signal; The signal feature processing module includes a preprocessing unit and a feature extraction unit; The preprocessing unit is used to perform preprocessing based on the collected respiratory signal and heart rate signal. The specific preprocessing scheme is as follows: use low-pass filtering and high-pass filtering to eliminate high-frequency noise from the respiratory signal and heart rate signal, and remove interference caused by body movement factors, then divide the continuous respiratory signal data into windows of fixed length, and finally perform normalization processing on the collected respiratory signal and heart rate signal; The feature extraction unit includes a breathing feature extraction unit and a heart rate feature extraction unit; The respiratory feature extraction unit is used to extract time domain features from the pre-processed respiratory signal by analyzing and counting the time series data, and the time domain features include the average respiratory frequency Pl, the respiratory time Hj, the number of blood oxygen saturation decreases Xj, and the blood oxygen saturation decrease amplitude Xf; The heart rate feature extraction unit is used to extract features from the preprocessed heart rate signal, and obtain the intervals between two adjacent heart beats by performing time domain analysis based on the heart beat interval number N and the heart beats per minute number R. , and then through the two adjacent heartbeat intervals Calculate the average value to obtain the average heartbeat interval ; The breathing abnormality algorithm module is used to perform data annotation based on the historical data of the extracted breathing characteristic signal, use the extracted features and the annotated breathing signal data to train the SVM model, mark each historical data of the breathing characteristic signal as Xi, generate the historical breathing characteristic signal Xi, and calculate the kernel function value between the historical breathing characteristic signal Xi and the real-time collected breathing characteristic signal X. , and then calculate the weighted sum to obtain the respiratory abnormality coefficient Hxyc, and then use the category label Classify the respiratory abnormality coefficient Hxyc output; The heart rate abnormality algorithm module triggers the heart rate abnormality algorithm when a sleep apnea event occurs, uses a signal processing algorithm to perform abnormality detection based on the extracted heart rate characteristic signal, calculates and obtains the heart rate variability HRV, and sets a heart rate abnormality detection threshold L to perform heart rate abnormality assessment; The comprehensive evaluation module is used to perform comprehensive analysis and generate corresponding diagnosis recommendations based on the output results of the abnormal breathing coefficient Hxyc and the heart rate variability HRV.

2. The sleep data intelligent collection and processing system according to claim 1, characterized in that: The abnormal breathing algorithm module includes a kernel function algorithm unit and a SVM model algorithm unit; The kernel function algorithm unit is used to use the sleep apnea events manually labeled by doctors as the labels of training data, which are labeled as "apnea" and "normal breathing" category labels, and the data after the historical respiratory feature signal is extracted is labeled as the historical respiratory feature signal Xi, wherein the historical respiratory feature signal Xi={Pl i , Hj i , Xj i , Xf i }, the respiratory feature vector collected in real time is marked as the respiratory feature vector X, and the radial basis function kernel is constructed based on the historical respiratory feature signal Xi and the respiratory feature vector X. ; The radial basis function kernel Obtained by the following formula; ； Where exp represents the exponential function, Represents the parameters of the radial basis function kernel, which controls the width of the kernel function.

3. The sleep data intelligent collection and processing system according to claim 2, characterized in that: The SVM model algorithm unit includes a classification prediction unit and a classification decision unit; The classification prediction unit is used to train the SVM model using the extracted respiratory feature signal and the annotated label, evaluate the model performance using the cross-validation technique, and adjust the SVM model parameters, including the penalty parameter C and the kernel function parameter γ to optimize the SVM model, and then extract the same features from the respiratory feature vector X, use the trained SVM model to construct the respiratory abnormality algorithm formula, output the respiratory abnormality coefficient Hxyc, and analyze the "apnea" or "normal breathing" category; The abnormal breathing coefficient Hxyc is obtained by the following algorithm formula; ； In the formula, represents the Lagrange multiplier, represents the category label of the i-th sample in the training data set, and b represents the bias term.

4. The sleep data intelligent collection and processing system according to claim 3, characterized in that: The classification decision unit is used to use the category label The formula classifies the output results of the output breathing abnormality coefficient Hxyc; The category label Obtained by the following formula; ； In the formula, It represents the category label predicted by the SVM model, and Sign represents the symbol function used to determine the predicted category.

5. The sleep data intelligent collection and processing system according to claim 4, characterized in that: The respiratory abnormality coefficient Hxyc calculated by the SVM model output is then labeled Determine the prediction category. The specific prediction category method is as follows; When the respiratory abnormality coefficient Hxyc>0, it means =1, indicating that a sleep apnea event occurred in the respiratory signal during the predicted time period, which is considered as abnormal breathing. In this case, further abnormal heart rate detection is initiated; When the respiratory abnormality coefficient Hxyc < 0, it means =-1, indicating that there is no sleep apnea event in the respiratory signal during the predicted time period, and the breathing is judged to be normal, that is, the respiratory characteristics during this time period are normal and do not meet the apnea standard.

6. The sleep data intelligent collection and processing system according to claim 5, characterized in that: The abnormal heart rate algorithm module includes an abnormal heart rate algorithm unit and an abnormal heart rate evaluation unit; The heart rate abnormality algorithm unit is triggered by monitoring the apnea event, and performs heart rate abnormality detection by using a signal processing algorithm based on the extracted real-time heart rate features, and calculates and obtains the heart rate variability HRV; The heart rate variability HRV is obtained by the following algorithm formula; ； The heart rate abnormality evaluation unit is used to compare and evaluate the preset heart rate abnormality detection threshold L with the result of the heart rate variability HRV output, and analyze the heart rate situation during apnea. The specific evaluation results are as follows; When the heart rate variability HRV>heart rate abnormality detection threshold L, it is judged as heart rate abnormality; When heart rate variability HRV = heart rate abnormality detection threshold L, the heart rate is judged to be normal; When the heart rate variability HRV is less than the heart rate abnormality detection threshold L, it is determined that the heart rate is abnormal.

7. The sleep data intelligent collection and processing system according to claim 6, characterized in that: The comprehensive evaluation module is used to comprehensively analyze the results of the abnormal breathing coefficient Hxyc and the heart rate variability HRV output, and generate relevant diagnostic suggestions. The specific comprehensive analysis content is as follows; When the heart rate is normal but the breathing is abnormal, the individual is assessed to have respiratory problems. Although the heart rate is normal, the abnormal breathing requires further formal examination of the respiratory system to rule out potential respiratory problems. When the heart rate is normal but breathing is abnormal, the patient is at potential health risk. A high heart rate may indicate emotional stress and insufficient physical activity, while abnormal breathing may indicate poor sleep quality or potential respiratory problems. Further medical examination is required at this time.

8. A method for intelligently collecting and processing sleep data, comprising the intelligent collection and processing system for sleep data according to any one of claims 1 to 7, characterized in that: The following steps are involved: S1. First, according to the signal type to be collected, select a sleep monitoring device and sensor with corresponding functions to record the changes of breathing signals and heart rate signals in real time during sleep; S2. After the signal is collected, the raw data is preprocessed to eliminate possible noise and interference, and then the characteristic signal and the heart rate characteristic signal are extracted from the preprocessed breathing signal and heart rate signal for subsequent abnormality detection and analysis; S3, extract features from the preprocessed respiratory signal, manually label the sleep apnea event category labels through doctors, and construct the radial basis function kernel , and then further calculate and analyze the respiratory abnormality coefficient Hxyc, and use the category label The formula classifies the output results of the output breathing abnormality coefficient Hxyc, and predicts and classifies the abnormalities produced by breathing during sleep; S4. When a breathing pause event is detected, a heart rate abnormality detection is further performed. Based on the features extracted from the pre-processed heart rate signal, a signal processing algorithm is used to perform heart rate abnormality detection calculation to obtain the heart rate variability HRV, and then the degree of the heart rate variability HRV is evaluated to evaluate the abnormal and normal conditions of the heart rate; S5. Finally, a comprehensive analysis is performed based on the results of abnormal detection of respiratory signals and heart rate signals, and relevant diagnostic suggestions are provided.

Images