CN116138745B - Sleep respiration monitoring method and device integrating millimeter wave radar and blood oxygen data - Google Patents
Sleep respiration monitoring method and device integrating millimeter wave radar and blood oxygen data Download PDFInfo
- Publication number
- CN116138745B CN116138745B CN202310438289.4A CN202310438289A CN116138745B CN 116138745 B CN116138745 B CN 116138745B CN 202310438289 A CN202310438289 A CN 202310438289A CN 116138745 B CN116138745 B CN 116138745B
- Authority
- CN
- China
- Prior art keywords
- time
- data
- characteristic information
- sliding window
- respiratory
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4806—Sleep evaluation
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/0205—Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves
- A61B5/0507—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves using microwaves or terahertz waves
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
- A61B5/14542—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue for measuring blood gases
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4806—Sleep evaluation
- A61B5/4818—Sleep apnoea
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
- A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
- A61B5/6802—Sensor mounted on worn items
- A61B5/681—Wristwatch-type devices
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
Abstract
The invention provides a sleep respiration monitoring method and equipment integrating millimeter wave radar and blood oxygen data, wherein the method comprises the following steps: acquiring radar data provided by millimeter wave radar equipment and blood oxygen data acquired by blood oxygen equipment; determining sleeping time and waking time according to the radar data; extracting features of the blood oxygen data to obtain oxygen reduction feature information of the monitored object in the time from falling asleep to waking in the morning; extracting features of the radar data to obtain respiration feature information, heart rate feature information and body movement feature information of a monitored object in a sleeping time to a waking time; and determining abnormal breathing events from the sleep time to the morning wake time according to the oxygen reduction characteristic information, the breathing characteristic information, the heart rate characteristic information and the body movement characteristic information.
Description
Technical Field
The invention relates to the field of human body data monitoring and analysis, in particular to a sleep respiration monitoring method and equipment integrating millimeter wave radar and blood oxygen data.
Background
Traditional sleep breathing abnormality monitoring mainly uses a multi-lead sleep monitor (PSG) to monitor various physiological signals of a tested person in the sleeping process, and a technician detects breathing abnormality events by analyzing related images of an electroencephalogram, an oximetry chart, a nasal airflow chart and chest and abdomen movements throughout the night.
Conventional multi-lead sleep monitors require the subject to wear a large number of sensors, such as: the nose catheter, the thermistor, the chest and abdomen belt and the electroencephalogram device influence the sleeping comfort degree of the tested person to a certain extent. The chest and abdomen belt presses human tissues, so that a tested person cannot be in a natural sleep state, and the final monitoring result is affected; the lead equipment is complex to operate in the wearing process, needs to be guided by professional staff, is difficult to independently finish wearing, and can influence the monitoring result due to incorrect wearing positions; some activities of the monitored personnel in the sleeping process, such as turning over and body movement, easily cause the displacement of the monitoring equipment and influence the monitoring result; the device is difficult to remove completely during wear, and for some night-time activities, inconvenient effects can be caused. The traditional PSG equipment has high cost, high price and inconvenient use, and is difficult to realize large-scale popularization and application.
Disclosure of Invention
In view of the above, the present invention provides a sleep respiration monitoring method for fusing millimeter wave radar and blood oxygen data, comprising: acquiring radar data provided by millimeter wave radar equipment and blood oxygen data acquired by blood oxygen equipment; determining sleeping time and waking time according to the radar data; extracting features of the blood oxygen data to obtain oxygen reduction feature information of the monitored object in the time from falling asleep to waking in the morning; extracting features of the radar data to obtain respiration feature information, heart rate feature information and body movement feature information of a monitored object in a sleeping time to a waking time; and determining abnormal breathing events from the sleep time to the morning wake time according to the oxygen reduction characteristic information, the breathing characteristic information, the heart rate characteristic information and the body movement characteristic information.
Optionally, the oxygen reduction characteristic information, the breathing characteristic information, the heart rate characteristic information and the body movement characteristic information are respectively information sequences obtained by interception and calculation under various sliding window length and step length settings; the operation of determining a respiratory abnormality event specifically includes: respectively identifying the information sequences under each sliding window setting by using a machine learning model to obtain a respiratory abnormality event sequence under each sliding window setting, wherein the respiratory abnormality event comprises event starting and ending time and confidence degrees of various respiratory abnormality types; and integrating the respiratory abnormality event sequences under each sliding window setting according to the starting and ending time of the event and the confidence degrees of various respiratory abnormality types to obtain a respiratory abnormality event sequence.
Optionally, the operation of integrating the respiratory abnormality event sequence under each sliding window setting to obtain a respiratory abnormality event sequence specifically includes: acquiring respiratory abnormal events with overlapping event start and end time intervals in a respiratory abnormal event sequence under each sliding window setting; and comparing the confidence coefficient of the respiratory abnormal event with the overlapping event starting time interval and event ending time interval, and reserving the respiratory abnormal event with the highest confidence coefficient.
Optionally, extracting features from the blood oxygen data specifically includes: identifying an oxygen reduction event from the blood oxygen data, the oxygen reduction event being that the blood oxygen data is below a preset blood oxygen threshold; intercepting the blood oxygen data by utilizing sliding windows with different scales and step sizes, and judging whether the blood oxygen data in each sliding window and within a certain time range behind the sliding window contain the oxygen reduction event or not; for a sliding window with blood oxygen reduction in a defined analysis duration range, determining a difference value between a blood oxygen reduction minimum value and a reduction start value and a time difference between a time point corresponding to the blood oxygen reduction minimum value and a starting point of the sliding window; and forming oxygen reduction characteristic information corresponding to the sliding window by using the zone bit of whether the blood oxygen reduction event occurs or not, and the difference value and the time difference.
Optionally, the radar data comprise distance-dimension respiration doppler data, namely the movement speed of chest and abdomen positions at different distances along the radar sight direction, and the respiration characteristic information comprises chest and abdomen movement characteristic information; the method for acquiring the chest and abdomen movement characteristic information of the monitored object in the sleeping time to the morning waking time specifically comprises the following steps: intercepting the distance dimension respiratory Doppler data by utilizing sliding windows with different scales and step sizes; and according to the distance dimension respiratory Doppler data in each sliding window, dividing the chest and abdomen boundaries according to the distance, respectively summing the chest and abdomen distance dimension respiratory Doppler data along the distance dimension, and solving the correlation coefficient of the chest and abdomen sum vector and the abdomen sum vector after chest and abdomen sum respectively, wherein the correlation coefficient is used as chest and abdomen motion characteristic information corresponding to the sliding window.
Optionally, the radar data comprises distance-dimensional respiratory doppler data and the respiratory characteristic information comprises respiratory rate variation characteristic information.
Optionally, obtaining the respiration rate variation characteristic information of the monitored subject from the time of falling asleep to the time of waking in the morning specifically includes: summing the distance dimension respiration Doppler data along the distance dimension to obtain a respiration rate curve; intercepting the respiratory rate curve by utilizing sliding windows and detection windows with different scales and step sizes, wherein the length of the detection window is longer than that of the sliding window, and expanding the detection window by taking the sliding window as a center to obtain the detection window; and calculating the ratio of the respiration rate in the sliding window to the respiration rate outside the sliding window in the detection window, and taking the ratio as the corresponding respiration rate change characteristic information in the window.
Optionally, the radar data comprises a heart rate curve; obtaining heart rate characteristic information of a monitored object in a sleeping time to a waking time, wherein the heart rate characteristic information comprises the following specific steps of: intercepting the heart rate curve by utilizing sliding windows with different scales and step sizes, calculating the time difference between a time point corresponding to the maximum peak value of the heart rate rise and the end point of the sliding window, and calculating the difference between the peak value and the average value of the heart rate in a certain time before the peak value as the heart rate rise value characteristic corresponding to the sliding window; and utilizing the heart rate rising value characteristic and the time difference to form heart rate rising characteristic information corresponding to the sliding window.
Optionally, the radar data comprise body motion power data, namely echo intensities of human body parts at different distances; the method for obtaining the body movement characteristic information of the monitored object in the sleeping time to the morning waking time specifically comprises the following steps: adding the body movement power data along the distance dimension to obtain a power adding curve along the time dimension, and smoothing the curve obtained by adding the power along the time dimension; and intercepting a power addition curve by utilizing sliding windows with different scales and step sizes, calculating the power ratio of two adjacent seconds in the time dimension aiming at the data in each sliding window, and determining the maximum value of the power ratio in the sliding window as the body movement characteristic information corresponding to the sliding window.
Correspondingly, the invention also provides sleep respiration monitoring equipment, which comprises the following components: a processor and a memory coupled to the processor; the memory stores instructions executable by the processor, and the instructions are executed by the processor, so that the processor executes the sleep respiration monitoring method integrating millimeter wave radar and blood oxygen data.
According to the sleep respiration monitoring method provided by the embodiment of the invention, the millimeter wave radar and the hand belt type blood oxygen monitoring equipment are utilized to acquire all the characteristics required by sleep and sleep respiration abnormality analysis, firstly, the sleeping time and the morning wake time of a tested person can be identified according to radar data so as to intercept data in a sleeping state, and then respiratory abnormality events can be identified from four dimensions of blood oxygen, respiration, heart rate and body movement, so that a more accurate identification result can be obtained. The scheme is convenient to monitor, the using method is simple, and privacy is not violated. The radar and blood oxygen equipment has small influence on sleeping quality of people, can not cause damage to human bodies, has stable data transmission and convenient checking result, can not only monitor people for a long time sleeping and abnormal respiration, but also assist doctors in primary screening of abnormal respiration diseases.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of a sleep monitoring system;
FIG. 2 is a flow chart of a sleep respiration monitoring method in an embodiment of the present invention;
FIG. 3 is an example graph of visual statistics of respiratory anomaly events;
FIG. 4 is a schematic diagram of recognition results of a machine learning model for two sliding window outputs according to an embodiment of the present invention;
fig. 5 is a schematic diagram of capturing respiratory characteristic information according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made apparent and fully in view of the accompanying drawings, in which some, but not all embodiments of the invention are shown. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention. The technical features of the different embodiments of the invention described below may be combined with one another as long as they do not conflict with one another.
Fig. 1 shows a sleep monitoring system based on millimeter wave radar, which comprises a user side, a cloud computing side and a display side, wherein the user side specifically comprises millimeter wave radar equipment and blood oxygen equipment, the millimeter wave radar equipment collects echo signals in a scene, and the original data are preprocessed through a related algorithm of digital signal processing, so that the obtained data comprise, but are not limited to, one-dimensional range profile, noise power, breathing data, heart rate data and the like; blood oxygen equipment (wristwatch type) collects blood oxygen information of a person; the cloud computing end is used for receiving and analyzing the data of the user end, generating a result required by the sleep monitoring report by analyzing the analyzed data, storing the result into the database, synchronizing the result to the network background, transmitting the data to the display end by the background, finally displaying a health report required by the monitoring personnel, and viewing or downloading the health report on a webpage.
The invention provides a sleep breathing abnormality monitoring method based on millimeter wave radar data, which is executed by the cloud computing terminal in the embodiment, and can be executed by other electronic equipment such as a computer or a mobile terminal in other embodiments, and the method comprises the following operations as shown in fig. 2:
s1, radar data provided by millimeter wave radar equipment and blood oxygen data acquired by blood oxygen equipment are acquired. The millimeter wave radar device may preprocess the echo signal to obtain a series of radar data, specifically, continuous data in the whole monitoring period, that is, data of the tested person changing with time in the monitoring period, such as one-dimensional range profile, respiration data, heart rate data, body movement data, and the like. Similarly, the blood oxygen data is also continuous data changing along with time, the blood oxygen acquisition equipment can send the blood oxygen data to the radar equipment, the radar equipment sends the blood oxygen data and the radar data obtained through preprocessing to the cloud end through a network, the data is converted into files which can be identified by an algorithm through analysis and packaging at the cloud end, and the files are stored in the server. The network terminal calls the algorithm terminal through a timing program and transmits a file path, and the algorithm terminal reads the data in the server through the file path and processes the data.
S2, determining sleeping time and morning wake time according to the radar data. The data obtained in the monitoring period, that is, all the data obtained when the radar device and the blood oxygen collecting device are in the working state, but the monitoring period is not equal to the sleeping period of the monitored person, so that the time when the person enters the sleeping state and wakes up needs to be identified according to the radar data, and the purpose is to intercept the data in the sleeping period.
Depending on the specific type of radar data in different embodiments, different methods may be used to obtain the fall asleep time and the wake-up time. For example, the time of the person lying down for the first time and rising up for the last time in the monitoring period can be identified as the falling asleep and morning wake time based on the one-dimensional range profile data only. It is of course possible to further accurately identify in combination with other data, such as comprehensively analysing the status of the person on the basis of breathing status, body movement data, heart rate data etc., which data can be used to analyse whether the person has entered a relatively stable state, thereby determining when the person to be monitored has entered sleep after lying down and has been awake before getting up.
And S3, extracting features of the blood oxygen data to obtain oxygen reduction feature information of the tested person in the time from falling asleep to waking in the morning. Specifically, firstly, the blood oxygen data in the sleep state can be intercepted, and analysis is performed on continuous blood oxygen data, for example, the time point and the time period when the blood oxygen is lower than the threshold value, the corresponding blood oxygen value, the lowest value, the blood oxygen change rate and the like can be determined as the oxygen reduction characteristic value. The time when the oxygen reduction does not occur has corresponding characteristic information, such as a flag bit, which indicates that the blood oxygen is normal at the time, and the time when the oxygen reduction occurs has a flag bit, which indicates that the blood oxygen is abnormal at the time, and the values are the same, so the oxygen reduction characteristic information is still regarded as a time-varying information flow.
And S4, carrying out feature extraction on the radar data to obtain the respiration feature information, the heart rate feature information and the body movement feature information of the monitored object in the sleeping time to the morning waking time. Similar to the oxygen-reduced feature information, the three feature information obtained in this step all belong to a time-varying information stream. By way of example, the respiration characteristic information may be a respiration rate or the like that varies with time, the heart rate characteristic information may be a heart rate or a rate of change in the heart rate or the like that varies with time, and the body movement characteristic information may be a body movement amplitude or the like that varies with time (such as turning over in sleep, limb movements). The characteristic information can be obtained in various ways according to the specific type of radar data in different embodiments, and can be obtained by adopting different methods based on power data provided by radar equipment or other data (such as Doppler, body motion power) obtained by performing mathematical transformation on the power data, and the like.
S5, determining abnormal breathing events from the sleep time to the morning waking time according to the oxygen reduction characteristic information, the breathing characteristic information, the heart rate characteristic information and the body movement characteristic information. The recognition can be realized by adopting a machine learning model, training is carried out on the model by adopting a supervised, unsupervised or semi-supervised training method in advance, the characteristic information is used as input data of the model, the model recognizes and intercepts four characteristic information for a period of time, and whether the model belongs to abnormal breathing and an abnormal type result in the period of time are output. With respect to the types of abnormalities, specific may include obstruction, central, mixed breathing abnormalities, and hypopnea. And identifying the characteristic information in the sleep state one by one in a time period to obtain all respiratory abnormal events occurring in the whole sleep state, wherein each respiratory abnormal event comprises whether the respiratory abnormal event belongs to abnormality, specific event type and corresponding starting and ending time.
Fig. 3 is an example of plotting respiratory anomaly events as visual statistics, where the abscissa indicates the start time of event occurrence, the ordinate indicates the event duration, and the vertical line segment indicates that some respiratory anomaly event occurred at the corresponding time, OSA is obstructive sleep apnea, CSA is central sleep apnea, MSA is mixed sleep apnea, HPY is hypopnea.
On the basis of obtaining the respiratory abnormal event, some indexes, such as average duration, longest duration and the like of various abnormal events in the table above in fig. 2, can be further counted and calculated, and the information can be displayed to a user through a display end.
According to the sleep respiration monitoring method provided by the embodiment of the invention, the millimeter wave radar and the hand belt type blood oxygen monitoring equipment are utilized to acquire all the characteristics required by sleep and sleep respiration abnormality analysis, firstly, the sleeping time and the morning wake time of a tested person can be identified according to radar data so as to intercept data in a sleeping state, and then respiratory abnormality events can be identified from four dimensions of blood oxygen, respiration, heart rate and body movement, so that a more accurate identification result can be obtained. The scheme is convenient to monitor, the using method is simple, and privacy is not violated. The radar and blood oxygen equipment has small influence on sleeping quality of people, can not cause damage to human bodies, has stable data transmission and convenient checking result, can not only monitor people for a long time sleeping and abnormal respiration, but also assist doctors in primary screening of abnormal respiration diseases.
For the embodiment that needs to distinguish specific abnormal types, in order to make the accuracy of the recognition result of the model higher, in one embodiment, the oxygen reduction feature information, the respiration feature information, the heart rate feature information and the body movement feature information are respectively obtained by intercepting and calculating the information sequences under various sliding window lengths and stepping settings.
Specifically, taking two sliding window lengths and steps as examples, specifically, the first sliding window length and step are adopted to intercept and analyze blood oxygen data and radar data, and the characteristic information of the ith sliding window of the first sliding window and step can be expressed as { O } 1i 、B 1i 、H 1i 、M 1i } t1i Wherein O is 1i Representing oxygen reduction characteristic information, B 1i Representing respiratory characteristic information, H 1i Representing heart rate characteristic information, M 1i Representing body movement characteristic information; intercepting and analyzing blood oxygen data and radar data using a second sliding window length and step, the characteristic information of the ith sliding window of the second sliding window and step length may be expressed as { O } 2i 、B 2i 、H 2i 、M 2i } t2i . Because the sliding window is used for detecting whether the abnormal breathing event occurs in the period of time, the extracted characteristic information is a specific or same-dimension characteristic vector, and the length of the characteristic vector finally fed into the model is not influenced by the sliding window and the stepping, namely { O } 1i 、B 1i 、H 1i 、M 1i } t1i And { O 2i 、B 2i 、H 2i 、M 2i } t2i Is the same length.
Further, in step S5: and respectively identifying the characteristic information sequences corresponding to the different sliding windows and the steps by using a machine learning model to obtain a corresponding breathing abnormal event sequence, wherein the breathing abnormal event comprises time information and confidence degrees of various breathing abnormal types. The machine learning model may be specifically an SVM model, and the recognition results are output for sliding window data of various lengths and steps, respectively. Taking fig. 4 as an example, the ordinate represents the confidence of an abnormal event, the abscissa represents the time, and the two lines represent the results for the two sliding window data outputs, respectively. Wherein event 1 and event 4 are the recognition results for the first type of sliding window data, event 2 and event 3 are the recognition results for the second type of sliding window data, these events may be of different anomaly types, the magnitude of the segment representing the confidence of this anomaly type, e.g., event 1 is OSA, confidence is 0.8; event 4 is CSA with a confidence level of 0.9; event 2 is CSA with a confidence level of 0.6; event 3 is an MSA confidence of 0.95. Other positions with confidence of 0 indicate no abnormal event.
And integrating each abnormal breathing event sequence according to the time information and the confidence degree of the abnormal breathing type to obtain one abnormal breathing event sequence. In the example shown in fig. 4, there are time-overlapping portions of two recognition results, and there are various methods of integrating the recognition results of such a state. In a preferred embodiment, respiratory anomaly events in which time information in each respiratory anomaly event sequence overlaps can be obtained; the confidence of the overlapped respiratory abnormal events exists in the comparison time information, and the respiratory abnormal event with the highest confidence is reserved. All remain for other respiratory abnormalities where no overlap exists.
Still taking fig. 4 as an example, the result of the integration according to the preferred embodiment described above is that event 1 is OSA, event 2 is MSA, and original event 2 and original event 4 are discarded due to low confidence.
The following embodiments provide alternative ways of extracting the above-described characteristic information.
In one embodiment, regarding the extraction of blood oxygen characteristic information, the following manner may be specifically adopted:
s31, identifying an oxygen reduction event from the blood oxygen data, wherein the oxygen reduction event is that the blood oxygen data falls below a preset blood oxygen threshold value, and the duration of the whole reduction process reaches a certain length (for example, 10 seconds). Specifically, on the sleep data during the whole night sleep, a blood oxygen reduction threshold may be set, and in the time dimension, when the blood oxygen is reduced below the threshold and the whole reduction process lasts for more than 10 seconds, an oxygen reduction event is considered to occur, the starting time of the oxygen reduction event and the difference value between the blood oxygen corresponding to the starting time and the reduced minimum blood oxygen are recorded, so that the oxygen reduction event in the whole night sleep time is recorded regularly.
S32, intercepting blood oxygen data by utilizing sliding windows with different scales and step sizes, and judging whether the blood oxygen data in each sliding window and within a certain time range after the sliding window contain the oxygen reduction event or not, wherein the length of a certain time after the sliding window ends is a preset value, such as a plurality of seconds. And (3) a certain time from the beginning of the sliding window to the end of the sliding window is called analysis duration, if the analysis duration contains an oxygen reduction event, the oxygen reduction zone bit is set to be 1, otherwise, the oxygen reduction zone bit is set to be 0. And determining the difference value between the lowest blood oxygen reduction value and the reduction starting value and the time difference between the time point corresponding to the lowest blood oxygen reduction value and the starting point of the sliding window for the sliding window in which blood oxygen reduction occurs in the delimited analysis duration range.
S33, forming oxygen reduction characteristic information corresponding to the sliding window by using the zone bit of whether blood oxygen reduction occurs or not and the difference value and the time difference. For example, a flag bit of 1 indicates that an oxygen reduction event occurs in the analysis duration, a flag bit of 0 indicates that an oxygen reduction event does not occur in the analysis duration, and oxygen reduction characteristic information corresponding to one sliding window is [ oxygen reduction flag bit, blood oxygen difference value, time difference ]; a sliding window with a flag bit of 0, a blood oxygen difference value of 0 and a time difference of 0 can be regarded as a feature vector.
According to the above preferred embodiment, when a respiratory abnormality is identified, the oxygen reduction feature information input to the machine learning model is a feature vector for a sliding window, which has an oxygen reduction flag, a blood oxygen minimum value, and a time difference of corresponding sliding window data, whereby accuracy of the respiratory abnormality identification result can be improved.
In one embodiment, the radar data comprises distance-dimension breathing Doppler data, namely the movement speed of chest and abdomen positions at different distances along the radar sight line direction; the respiratory characteristic information includes chest and abdomen movement characteristic information, and regarding obtaining the chest and abdomen movement characteristic information, the following manner may be adopted in particular:
S41A, intercepting distance dimension respiratory Doppler data by utilizing sliding windows with different scales and step sizes.
S42A, dividing chest and abdomen boundaries according to the distance dimension respiratory Doppler data in each sliding window, respectively summing the chest and abdomen distance dimension respiratory Doppler data along the distance dimension, and obtaining the correlation coefficients of chest summation feature vectors and abdomen summation feature vectors after chest and abdomen summation feature vectors as chest and abdomen movement feature information. Specifically, the chest and abdomen position is determined by the distance, radar equipment is placed on the head top of a tested person, the human body is perpendicularly irradiated, the chest position is 30-50 cm, the abdomen position is 50-70 cm, and the data of the chest and the abdomen can be obtained by intercepting radar signals of the two distances.
And dividing chest and abdomen boundaries according to the distance in different sliding windows, respectively summing the chest and abdomen distance dimension respiratory Doppler data along the distance dimension to respectively obtain chest characteristic vectors and abdomen characteristic vectors with the lengths of the sliding windows, and solving correlation coefficients of the chest characteristic vectors and the abdomen characteristic vectors to serve as chest and abdomen motion characteristic information corresponding to the sliding windows. For example, the correlation coefficient solution formula is
Wherein X is a characteristic vector obtained by summing respiratory Doppler along a distance dimension in a chest distance dimension in a sliding window, Y is a characteristic vector obtained by summing the power along the distance dimension in an abdomen distance dimension in the sliding window, cov is covariance, var is variance, and r (X, Y) is a correlation coefficient.
In one embodiment, the radar data includes distance-dimensional respiratory Doppler data that is used to extract respiratory characteristic information, including in this embodiment respiratory rate variation characteristic information. The breathing rate change characteristic information can be extracted in the following way:
S41B, summing the distance-dimensional respiratory Doppler data along the distance dimension to obtain a summed curve, and taking the curve as a respiratory rate curve.
S42B, utilizing sliding windows and detection windows with different scales and step sizes to intercept a respiratory rate curve, wherein the length of the detection window is larger than that of the sliding window, and expanding the detection window by taking the sliding window as the center. As shown in fig. 5, the abscissa represents time, and the ordinate represents respiratory rate, where a represents a sliding window, and B1 and B2 are extended to both sides, respectively, to obtain a detection window.
S43B, calculating the ratio of the respiration rate in the sliding window to the respiration rate outside the sliding window in the detection window as the respiration rate change characteristic information. The respiratory rate change characteristic information feature=fA/fB1+B2, fA represents the respiratory rate in the sliding window, and fB1+B2 represents the respiratory rate outside the sliding window in the detection window. Thus, the characteristic information of the breathing rate change with time can be obtained. In a preferred embodiment, both embodiments are employed simultaneously, i.e. the breathing characteristic information comprises thoracoabdominal movement characteristic information and breathing rate change characteristic information. The chest and abdomen movement characteristics reflect the movement relationship between the chest and abdomen through a correlation coefficient, and when the chest and abdomen movement directions are opposite, the correlation coefficient is negative, and when the chest and abdomen movement directions are the same, the correlation coefficient is positive. When abnormal respiration occurs, the direction and the correlation of the chest and abdomen movement can be reflected by a correlation coefficient, which is often accompanied with the contradictory movement of the chest and abdomen. The change in respiratory rate reflects the condition of the human breath, and when respiratory anomalies occur, there is typically a condition of a significant decrease in respiratory rate, even zero. Therefore, any one of the two types of breathing characteristic information can be used as an effective basis for identifying breathing abnormality, and when the two types of characteristic information are adopted at the same time, the accuracy of the breathing abnormality event identification result can be further improved.
In one embodiment, the radar data includes heart rate curves, which data is used to extract heart rate characteristic information. The heart rate characteristic information can be extracted in the following way:
and S41C, finding all peaks of heart rate rise from the heart rate curve. The heart rate curve may be smoothed using a hamming window, and the heart rate rising peak and subsequent analysis may be identified based on the smoothed curve.
S42C, intercepting the heart rate curve by utilizing sliding windows with different scales and step sizes, determining the maximum value in the peak value of the heart rate rise and the time difference between the corresponding time point and the end point of the sliding window, and calculating the difference between the maximum peak value and the average value of the heart rate in a certain time before the peak value as the heart rate rise value characteristic corresponding to the sliding window;
S43C, utilizing the heart rate rising value and the time difference to form heart rate rising characteristic information corresponding to the sliding window. The [ heart rate rise difference, time difference ] is obtained and is a two-dimensional feature vector.
By counting the data of heart rate variation when abnormal breathing events occur, the heart rate is generally increased, and the difference between the starting time of the increase of the heart rate and the ending time of the abnormal breathing events is approximately within a certain time range.
According to the above preferred embodiment, when a respiratory abnormality is identified, the heart rate characteristic information input to the machine learning model is a characteristic vector for a sliding window in which a heart rate rise difference value and time information are present, whereby accuracy of the respiratory abnormality identification result can be improved.
In one embodiment, the radar data includes body motion power data, i.e. echo intensities of human body parts at different distances, for extracting body motion characteristic information. The body movement characteristic information can be extracted in the following way:
and S41D, adding the body movement power data along the distance dimension to obtain a curve along the time dimension, filtering the curve along the time dimension by using median filtering, and smoothing the filtered result by Hamming smoothing.
S42D, intercepting a power addition curve by utilizing sliding windows with different scales and step sizes, calculating the power ratio of two adjacent seconds in the time dimension aiming at the data in each sliding window, and determining the maximum value of the power ratio in the sliding window as body movement characteristic information corresponding to the sliding window.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. While still being apparent from variations or modifications that may be made by those skilled in the art are within the scope of the invention.
Claims (9)
1. A sleep respiration monitoring method integrating millimeter wave radar and blood oxygen data is characterized by comprising the following steps:
acquiring radar data provided by millimeter wave radar equipment and blood oxygen data acquired by blood oxygen equipment;
determining sleeping time and waking time according to the radar data;
extracting features of the blood oxygen data to obtain oxygen reduction feature information of the monitored object in the time from falling asleep to waking in the morning;
extracting features of the radar data to obtain respiration feature information, heart rate feature information and body movement feature information of a monitored object in a sleeping time to a waking time;
determining abnormal breathing events from sleep time to morning wake time according to the oxygen reduction characteristic information, the breathing characteristic information, the heart rate characteristic information and the body movement characteristic information, wherein the oxygen reduction characteristic information, the breathing characteristic information, the heart rate characteristic information and the body movement characteristic information are respectively intercepted and calculated information sequences under various sliding window length and step length settings; the operation of determining a respiratory abnormality event specifically includes:
respectively identifying the information sequences under each sliding window setting by using a machine learning model to obtain a respiratory abnormality event sequence under each sliding window setting, wherein the respiratory abnormality event comprises event starting and ending time and confidence degrees of various respiratory abnormality types;
and integrating the respiratory abnormality event sequences under each sliding window setting according to the starting and ending time of the event and the confidence degrees of various respiratory abnormality types to obtain a respiratory abnormality event sequence.
2. The method of claim 1, wherein the step of integrating the sequence of respiratory abnormalities for each sliding window setting to obtain a sequence of respiratory abnormalities comprises:
acquiring respiratory abnormal events with overlapping event start and end time intervals in a respiratory abnormal event sequence under each sliding window setting;
and comparing the confidence coefficient of the respiratory abnormal event with the overlapping event starting time interval and event ending time interval, and reserving the respiratory abnormal event with the highest confidence coefficient.
3. The method according to claim 1, characterized in that the feature extraction is performed on the blood oxygen data, in particular comprising:
identifying an oxygen reduction event from the blood oxygen data, the oxygen reduction event being that the blood oxygen data is below a preset blood oxygen threshold;
intercepting the blood oxygen data by utilizing sliding windows with different scales and step sizes, and judging whether the blood oxygen data in each sliding window and within a certain time range behind the sliding window contain the oxygen reduction event or not;
for a sliding window with blood oxygen reduction in a defined analysis duration range, determining a difference value between a blood oxygen reduction minimum value and a reduction start value and a time difference between a time point corresponding to the blood oxygen reduction minimum value and a starting point of the sliding window;
and forming oxygen reduction characteristic information corresponding to the sliding window by using the zone bit of whether the blood oxygen reduction event occurs or not, and the difference value and the time difference.
4. The method according to claim 1, wherein the radar data comprises distance-dimensional respiratory doppler data, namely the movement speed of chest and abdomen positions at different distances along the radar sight line direction, and the respiratory characteristic information comprises chest and abdomen movement characteristic information;
the method for acquiring the chest and abdomen movement characteristic information of the monitored object in the sleeping time to the morning waking time specifically comprises the following steps:
intercepting the distance dimension respiratory Doppler data by utilizing sliding windows with different scales and step sizes;
and according to the distance dimension respiratory Doppler data in each sliding window, dividing the chest and abdomen boundaries according to the distance, respectively summing the chest and abdomen distance dimension respiratory Doppler data along the distance dimension, and solving the correlation coefficient of the chest and abdomen sum vector and the abdomen sum vector after chest and abdomen sum respectively, wherein the correlation coefficient is used as chest and abdomen motion characteristic information corresponding to the sliding window.
5. The method of claim 1 or 4, wherein the radar data comprises distance-dimensional respiratory doppler data and the respiratory characteristic information comprises respiratory rate variation characteristic information.
6. The method according to claim 5, wherein obtaining the respiration rate variation characteristic information of the subject from the time of falling asleep to the time of waking in the morning, specifically comprises:
summing the distance dimension respiration Doppler data along the distance dimension to obtain a respiration rate curve;
intercepting the respiratory rate curve by utilizing sliding windows and detection windows with different scales and step sizes, wherein the length of the detection window is longer than that of the sliding window, and expanding the detection window by taking the sliding window as a center to obtain the detection window;
and calculating the ratio of the respiration rate in the sliding window to the respiration rate outside the sliding window in the detection window, and taking the ratio as the corresponding respiration rate change characteristic information in the window.
7. The method of claim 1, wherein the radar data comprises a heart rate curve;
obtaining heart rate characteristic information of a monitored object in a sleeping time to a waking time, wherein the heart rate characteristic information comprises the following specific steps of:
the heart rate curve is truncated using sliding windows of different dimensions and step sizes,
calculating the time difference between a time point corresponding to the maximum peak value of the heart rate rise and the end point of the sliding window, and calculating the difference between the peak value and the average value of the heart rate in a certain time before the peak value as a heart rate rise value characteristic corresponding to the sliding window;
and utilizing the heart rate rising value characteristic and the time difference to form heart rate rising characteristic information corresponding to the sliding window.
8. The method according to claim 1, wherein the radar data comprises body motion power data, namely echo intensities of human body parts at different distances;
the method for obtaining the body movement characteristic information of the monitored object in the sleeping time to the morning waking time specifically comprises the following steps:
adding the body movement power data along the distance dimension to obtain a power adding curve along the time dimension, and smoothing the curve obtained by adding the power along the time dimension;
and intercepting a power addition curve by utilizing sliding windows with different scales and step sizes, calculating the power ratio of two adjacent seconds in the time dimension aiming at the data in each sliding window, and determining the maximum value of the power ratio in the sliding window as the body movement characteristic information corresponding to the sliding window.
9. A sleep respiration monitoring device, comprising: a processor and a memory coupled to the processor; wherein the memory stores instructions executable by the processor to cause the processor to perform the sleep respiration monitoring method of fusing millimeter wave radar and blood oxygen data of any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310438289.4A CN116138745B (en) | 2023-04-23 | 2023-04-23 | Sleep respiration monitoring method and device integrating millimeter wave radar and blood oxygen data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310438289.4A CN116138745B (en) | 2023-04-23 | 2023-04-23 | Sleep respiration monitoring method and device integrating millimeter wave radar and blood oxygen data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116138745A CN116138745A (en) | 2023-05-23 |
| CN116138745B true CN116138745B (en) | 2023-06-27 |
Family
ID=86339295
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310438289.4A Active CN116138745B (en) | 2023-04-23 | 2023-04-23 | Sleep respiration monitoring method and device integrating millimeter wave radar and blood oxygen data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116138745B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116269298B (en) * | 2023-02-21 | 2023-11-10 | 郑州大学 | Non-contact sleep respiration monitoring method and system based on millimeter wave radar |
| CN117530666B (en) * | 2024-01-03 | 2024-04-05 | 北京清雷科技有限公司 | Breathing abnormality recognition model training method, breathing abnormality recognition method and equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108283489A (en) * | 2017-12-16 | 2018-07-17 | 湖南明康中锦医疗科技发展有限公司 | Sleep-respiratory system and method |
| CN109480787A (en) * | 2018-12-29 | 2019-03-19 | 中国科学院合肥物质科学研究院 | A kind of contactless sleep monitor equipment and sleep stage method based on ULTRA-WIDEBAND RADAR |
| CN110013235A (en) * | 2019-03-29 | 2019-07-16 | 张恒运 | A kind of smart home sleeping apparatus and system |
| CN112716474A (en) * | 2021-01-20 | 2021-04-30 | 复旦大学 | Non-contact sleep state monitoring method and system based on biological microwave radar |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8454528B2 (en) * | 2008-04-03 | 2013-06-04 | Kai Medical, Inc. | Non-contact physiologic motion sensors and methods for use |
| WO2011143631A2 (en) * | 2010-05-14 | 2011-11-17 | Kai Medical, Inc. | Systems and methods for non-contact multiparameter vital signs monitoring, apnea therapy, sway cancellation, patient identification, and subject monitoring sensors |
| JP2014217453A (en) * | 2013-05-02 | 2014-11-20 | 斎藤 光正 | Human body abnormality detection device by stationary wave radar and utilization method thereof |
| CN109009755A (en) * | 2018-09-05 | 2018-12-18 | 成都江雪医疗器械有限公司 | A kind of intelligence control system of sleep apnea Electronic Stryker frame |
| US20230293049A1 (en) * | 2020-07-29 | 2023-09-21 | Cornell University | Systems and methods for monitoring respiration of an individual |
| US20220386944A1 (en) * | 2021-06-04 | 2022-12-08 | Apple Inc. | Sleep staging using machine learning |
-
2023
- 2023-04-23 CN CN202310438289.4A patent/CN116138745B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108283489A (en) * | 2017-12-16 | 2018-07-17 | 湖南明康中锦医疗科技发展有限公司 | Sleep-respiratory system and method |
| CN109480787A (en) * | 2018-12-29 | 2019-03-19 | 中国科学院合肥物质科学研究院 | A kind of contactless sleep monitor equipment and sleep stage method based on ULTRA-WIDEBAND RADAR |
| CN110013235A (en) * | 2019-03-29 | 2019-07-16 | 张恒运 | A kind of smart home sleeping apparatus and system |
| CN112716474A (en) * | 2021-01-20 | 2021-04-30 | 复旦大学 | Non-contact sleep state monitoring method and system based on biological microwave radar |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116138745A (en) | 2023-05-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116138745B (en) | Sleep respiration monitoring method and device integrating millimeter wave radar and blood oxygen data | |
| US10966666B2 (en) | Machine learnt model to detect REM sleep periods using a spectral analysis of heart rate and motion | |
| EP3229676B1 (en) | Method and apparatus for physiological monitoring | |
| US7578793B2 (en) | Sleep staging based on cardio-respiratory signals | |
| JP6077138B2 (en) | Detection of sleep apnea using respiratory signals | |
| CN108388912A (en) | Sleep stage method based on multisensor feature optimization algorithm | |
| US20200107775A1 (en) | Methods and Systems for Monitoring Sleep Apnea | |
| CN105266787A (en) | Non-contact type heart rate detection method and system | |
| CN112998701A (en) | Vital sign detection and identity recognition system and method based on millimeter wave radar | |
| Papini et al. | Photoplethysmography beat detection and pulse morphology quality assessment for signal reliability estimation | |
| US20210256836A1 (en) | System and method for processing multiple signals | |
| CN110866498B (en) | Heart rate monitoring method | |
| CN106413533A (en) | Device, system and method for detecting apnoea of a subject | |
| CA2847412A1 (en) | System and methods for estimating respiratory airflow | |
| CN112363139A (en) | Human body breathing time length detection method and device based on amplitude characteristics and storage medium | |
| Geertsema et al. | Automated non-contact detection of central apneas using video | |
| Rescio et al. | Ambient and wearable system for workers’ stress evaluation | |
| KR101996027B1 (en) | Method and system for extracting Heart Information of Frequency domain by using pupil size variation | |
| CN114176564A (en) | Method for extracting respiratory state based on radar signal | |
| Ankitha et al. | Literature review on sleep APNEA analysis by machine learning algorithms using ECG signals | |
| Rolink et al. | Improving sleep/wake classification with recurrence quantification analysis features | |
| CN113907742A (en) | Sleep respiration data monitoring method and device | |
| CN114758410A (en) | Parameter detection system for identifying human body safety and health conditions | |
| Carter et al. | Deep Learning-Enabled Sleep Staging From Vital Signs and Activity Measured Using a Near-Infrared Video Camera | |
| US20200175685A1 (en) | Method to Derive a Person's Vital Signs from an Adjusted Parameter |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |