CN114469061B - Abdominal respiration monitoring system and method - Google Patents
Abdominal respiration monitoring system and method Download PDFInfo
- Publication number
- CN114469061B CN114469061B CN202210068468.9A CN202210068468A CN114469061B CN 114469061 B CN114469061 B CN 114469061B CN 202210068468 A CN202210068468 A CN 202210068468A CN 114469061 B CN114469061 B CN 114469061B
- Authority
- CN
- China
- Prior art keywords
- signal
- respiration
- respiratory
- abdominal
- training
- 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
- 230000029058 respiratory gaseous exchange Effects 0.000 title claims abstract description 184
- 230000003187 abdominal effect Effects 0.000 title claims abstract description 77
- 238000012544 monitoring process Methods 0.000 title claims abstract description 54
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000012549 training Methods 0.000 claims abstract description 67
- 239000011159 matrix material Substances 0.000 claims abstract description 50
- 230000004044 response Effects 0.000 claims abstract description 44
- 238000012216 screening Methods 0.000 claims abstract description 26
- 230000008569 process Effects 0.000 claims abstract description 23
- 238000013145 classification model Methods 0.000 claims abstract description 15
- 230000000241 respiratory effect Effects 0.000 claims description 99
- 238000001514 detection method Methods 0.000 claims description 24
- 238000007781 pre-processing Methods 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 18
- 108010076504 Protein Sorting Signals Proteins 0.000 claims description 12
- 230000011218 segmentation Effects 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 9
- 239000000284 extract Substances 0.000 claims description 6
- 239000003623 enhancer Substances 0.000 claims description 5
- 238000013526 transfer learning Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 3
- 230000036391 respiratory frequency Effects 0.000 claims description 2
- 210000001015 abdomen Anatomy 0.000 abstract description 11
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 230000033001 locomotion Effects 0.000 description 10
- 208000006545 Chronic Obstructive Pulmonary Disease Diseases 0.000 description 6
- 238000002474 experimental method Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000011176 pooling Methods 0.000 description 5
- 230000002685 pulmonary effect Effects 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000007423 decrease Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000003068 static effect Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000003434 inspiratory effect Effects 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000002354 daily effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 239000013589 supplement Substances 0.000 description 2
- 208000000059 Dyspnea Diseases 0.000 description 1
- 206010013975 Dyspnoeas Diseases 0.000 description 1
- 206010028289 Muscle atrophy Diseases 0.000 description 1
- 208000037656 Respiratory Sounds Diseases 0.000 description 1
- 235000011034 Rubus glaucus Nutrition 0.000 description 1
- 244000235659 Rubus idaeus Species 0.000 description 1
- 235000009122 Rubus idaeus Nutrition 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000002238 attenuated effect Effects 0.000 description 1
- 230000037237 body shape Effects 0.000 description 1
- 230000001684 chronic effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000034994 death Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000002526 effect on cardiovascular system Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000005713 exacerbation Effects 0.000 description 1
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 230000003278 mimic effect Effects 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 210000003205 muscle Anatomy 0.000 description 1
- 230000020763 muscle atrophy Effects 0.000 description 1
- 201000000585 muscular atrophy Diseases 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000000414 obstructive effect Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000036387 respiratory rate Effects 0.000 description 1
- 208000023504 respiratory system disease Diseases 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 230000033764 rhythmic process Effects 0.000 description 1
- 238000011272 standard treatment Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009885 systemic effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/08—Detecting, measuring or recording devices for evaluating the respiratory organs
-
- 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
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/08—Detecting, measuring or recording devices for evaluating the respiratory organs
- A61B5/0816—Measuring devices for examining respiratory frequency
-
- 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
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B2505/00—Evaluating, monitoring or diagnosing in the context of a particular type of medical care
- A61B2505/09—Rehabilitation or training
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Surgery (AREA)
- Public Health (AREA)
- Pathology (AREA)
- Veterinary Medicine (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Physiology (AREA)
- Pulmonology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention provides an abdominal respiration monitoring system and method, which relate to the technical field of non-contact monitoring and comprise the following steps: the signal acquisition module is used for continuously transmitting a frequency modulation continuous wave radar signal to the region where the user is in and receiving a corresponding reflected signal in the process of performing abdominal respiration training on the user; the respiration monitoring module is used for extracting a respiration signal reflected by a user from the reflected signal, screening the respiration signal to obtain a corresponding effective respiration signal, extracting the characteristics of the effective respiration signal to obtain a corresponding channel impulse response matrix, inputting the channel impulse response matrix into a respiration classification model obtained by pre-training to process the channel impulse response matrix to obtain the respiration type adopted by the user in the abdominal respiration training process, so as to monitor the effectiveness of the abdominal respiration training of the user. The abdomen type respiration monitoring system has the beneficial effects that abdomen type respiration monitoring is carried out based on the acoustic equipment, the abdomen type respiration monitoring system is easy to deploy and realize, the cost is low, the use is convenient, and the abdomen type respiration monitoring system is suitable for patients to use at home.
Description
Technical Field
The invention relates to the technical field of non-contact monitoring, in particular to an abdominal respiration monitoring system and method.
Background
Chronic Obstructive Pulmonary Disease (COPD) is a progressive systemic respiratory disease that can lead to respiratory inefficiency, cardiovascular complications, muscle atrophy, and ultimately death. COPD is incurable and rehabilitation doctors recommend that patients use Pulmonary Rehabilitation (PR) extensively to prevent COPD exacerbation. Pulmonary Rehabilitation (PR) has proven to be an effective intervention to maintain the health of COPD patients and has become the standard treatment for COPD patients. Pulmonary Rehabilitation (PR) is typically performed under supervision of a rehabilitation team in a clinical setting, however, going to a hospital for rehabilitation training consumes a significant amount of commute time, reducing patient aggressiveness. Furthermore, limited medical resources make many patients with chronic obstructive pulmonary unacceptable for immediate treatment, especially in developing countries. In order to allow patients to get a more timely and effective treatment, the global initiative for chronic obstructive pulmonary disease (GOLD) suggests that patients employ home-based rehabilitation programs, which can help avoid exacerbation of the condition by effectively scheduling the training for pulmonary rehabilitation at home for patients inconvenient to go to a hospital or clinic.
However, due to poor patient compliance, the effect of performing abdominal breathing training at home is not as good as that guided by a rehabilitation engineer in a hospital. A recent study showed that over 40% of family PR patients were classified as non-compliant. Patient compliance is poor for two reasons: one) abdominal breathing is a long-term training process and does not immediately produce results. Patients are to adhere to training several times per day, each training requiring a high concentration on breath, which can be relatively tedious, making it difficult for the patient to adhere. Without supervision and guidance, patients are easily inert and do not move. Second), early stage in training, abdominal breathing fatigues the patient and dyspnea due to active pulling of the inelastic muscles, resulting in the patient wanting to forgo training.
In order to help the patient practice his abdominal breathing better at home, some telemonitoring techniques have been proposed in the prior art. To cope with the boring breathing movements, some researchers have devised a breath exercise game for breath training, with the user breathing step by step in accordance with the game. To obtain immediate biofeedback, the patient is often required to wear some sensors, such as stretch-based respiration sensors, which can cause discomfort to the patient and possibly reduce the training efficiency of the patient. In addition, breeze detects respiratory phases with respiratory sounds and uses this information to provide biofeedback. But this method does not know the breathing type. With the development of network technology, video-based training programs have become a major trend. The rehabilitation team can conduct video to conduct network conferences with the patient and provide supervision. However, this approach requires that the rehabilitation team and the patient be on-line at the same time, is suitable for guided diagnosis, but is not suitable for long-term monitoring. In practice it is impractical to manually monitor respiratory movements one-to-one as it would consume a significant amount of time for the observer. Meanwhile, the use of cameras in a home environment also presents privacy concerns.
Disclosure of Invention
In view of the problems in the prior art, the present invention provides an abdominal respiration monitoring system, comprising:
The signal acquisition module is used for continuously transmitting a frequency modulation continuous wave radar signal to the region where the user is located and receiving a corresponding reflected signal in the process of performing abdominal respiration training on the user;
The respiration monitoring module is connected with the signal acquisition module, and the respiration monitoring module comprises:
The respiratory signal detection submodule is used for extracting respiratory signals reflected by the user from the reflected signals;
The respiratory signal screening submodule is connected with the respiratory signal detection submodule and is used for screening the respiratory signals to obtain corresponding effective respiratory signals;
And the breath type classification sub-module is connected with the breath phase detection sub-module and is used for extracting features of the effective breath signals to obtain a corresponding channel impulse response matrix, inputting the channel impulse response matrix into a breath classification model obtained through pre-training to obtain the breath type adopted by the user in the abdominal respiration training process, so as to monitor the effectiveness of the abdominal respiration training of the user.
Preferably, the respiratory signal detection submodule includes:
the first detection unit is used for carrying out signal cross-correlation processing on the sent frequency modulation continuous wave radar signal and the received corresponding reflected signal so as to obtain the peak value of each reflected signal;
and the second detection unit is connected with the first detection unit and is used for carrying out signal autocorrelation processing on each peak value to obtain a signal sequence formed by each peak value related to respiration, and the signal sequence is used as the respiration signal reflected by the user.
Preferably, the respiratory signal screening submodule includes:
The clustering unit is used for clustering the corresponding peaks according to the signal frequency of the breathing signal to obtain a plurality of frequency clusters;
The screening unit is connected with the clustering unit and is used for screening the frequency clusters corresponding to the signal frequency meeting the preset respiratory frequency, and taking a signal sequence formed by each peak value in the frequency clusters as the effective respiratory signal.
Preferably, the respiratory signal screening submodule further includes a phase segmentation unit connected with the screening unit, and the phase segmentation unit is used for performing phase segmentation processing according to the effective respiratory signal to obtain and output an inhalation respiratory ratio corresponding to each respiratory cycle contained in the effective respiratory signal.
Preferably, the phase dividing unit includes:
the segmentation subunit is used for carrying out phase segmentation on the effective respiratory signal to obtain the corresponding respiratory cycle;
And the processing subunit is connected with the segmentation subunit and is used for dividing each breathing cycle into an inhalation time and an exhalation time respectively, and further outputting the inhalation-exhalation ratio corresponding to each breathing cycle processed according to the inhalation time and the exhalation time.
Preferably, the frequency-modulated continuous wave radar signal includes a plurality of chirped harmonic bands and a corresponding plurality of blank bands arranged at intervals.
Preferably, the respiration monitoring module further includes a model training submodule connected to the respiration type classification submodule, and the model training submodule includes:
The data acquisition unit is used for acquiring the effective respiratory signals and the corresponding real respiratory types acquired in the process of abdominal respiratory training of a plurality of subjects, and respectively extracting the characteristics of each effective respiratory signal to obtain the corresponding channel impulse response matrix;
The preprocessing unit is connected with the data acquisition unit and is used for preprocessing the channel impulse response matrix to obtain a preprocessing matrix;
the training unit is respectively connected with the data acquisition unit and the preprocessing unit, and is used for taking the preprocessing matrix as input, taking the corresponding real breathing type as output, and training by adopting a transfer learning mode to obtain the breathing classification model.
Preferably, in the preprocessing unit, the channel impulse response matrix is subjected to signal filtering processing to obtain the preprocessing matrix containing dynamic information of the mobile component in a tested environment where the subject is located.
Preferably, the data acquisition unit includes:
An acquisition subunit, configured to acquire the effective respiratory signals and the corresponding real respiratory types acquired during the abdominal respiratory training of the plurality of subjects;
The data enhancer unit is connected with the acquisition subunit and is used for carrying out data enhancement on the effective breathing signal by adopting a preset data window to obtain an enhanced breathing signal;
And the characteristic extraction subunit is connected with the data enhancer unit and is used for respectively carrying out characteristic extraction on each enhanced respiratory signal to obtain the corresponding channel impulse response matrix.
The invention also provides an abdominal respiration monitoring method which is applied to the abdominal respiration monitoring system, and comprises the following steps:
Step S1, continuously transmitting a frequency modulation continuous wave radar signal to the region where a user is located and receiving a corresponding reflected signal by the abdominal respiration monitoring system in the process of performing abdominal respiration training on the user;
Step S2, the abdominal respiration monitoring system extracts a respiration signal reflected by the user from the reflected signal;
Step S3, the abdominal respiration monitoring system screens the respiration signals to obtain corresponding effective respiration signals;
And S4, the abdominal respiration monitoring system performs feature extraction on the effective respiration signals to obtain a corresponding channel impulse response matrix, and inputs the channel impulse response matrix into a respiration classification model obtained through pre-training to obtain the respiration type adopted by the user in the abdominal respiration training process so as to monitor the effectiveness of the abdominal respiration training of the user.
The technical scheme has the following advantages or beneficial effects:
1) The abdomen type respiration monitoring is carried out based on the acoustic equipment, the deployment and the realization are easy, the cost is low, the use is convenient, and the abdomen type respiration monitoring device is suitable for patients to use at home;
2) In the process of performing abdominal respiration training by a user, the respiration type of the patient and the inspiration-to-respiration ratio of the patient can be monitored by continuously sensing and analyzing the change of the surrounding environment by sending a frequency modulation continuous wave radar signal, so that the patient or a remote doctor can monitor the effectiveness of the abdominal respiration training;
3) The breathing signals are segmented through the breathing period to expand data so as to obtain fine-granularity information and the multichannel microphone signals are fused to supplement three-dimensional information, and then a personalized breathing classification model is built through transfer learning, so that the robustness and high performance of the breathing classification model are effectively enhanced.
Drawings
FIG. 1 is a schematic diagram of an abdominal respiration monitoring system according to the preferred embodiment of the present invention;
FIG. 2 is a schematic cross-correlation diagram of different lags at specific times in a preferred embodiment of the present invention;
FIG. 3 (a) is a graph showing the CIR trend of the channel impulse response matrix when the peak represents the maximum chest-to-chest distance of the user in the preferred embodiment of the present invention;
FIG. 3 (b) is a graph showing the CIR trend of the channel impulse response matrix when the peak represents the distance from the chest to the user's chest when the chest is contracted to the minimum in the preferred embodiment of the present invention;
FIG. 3 (c) is a graph showing the CIR trend of the channel impulse response matrix when the peak value is between the maximum inhalation point and the minimum exhalation point in the preferred embodiment of the present invention;
FIG. 4 is a schematic diagram of a model network structure according to a preferred embodiment of the present invention;
fig. 5 is a flow chart of an abdominal respiration monitoring method according to the preferred embodiment of the invention.
Detailed Description
The invention will now be described in detail with reference to the drawings and specific examples. The present invention is not limited to the embodiment, and other embodiments may fall within the scope of the present invention as long as they conform to the gist of the present invention.
In accordance with the foregoing and other problems of the prior art, a preferred embodiment of the present invention provides an abdominal respiration monitoring system, as shown in fig. 1, comprising:
The signal acquisition module 1 is used for continuously sending a frequency modulation continuous wave radar signal to the region where the user is in and receiving a corresponding reflected signal in the process of performing abdominal respiration training on the user;
respiration monitoring module 2, connecting signal acquisition module 1, respiration monitoring module 2 includes:
A respiration signal detection sub-module 21, configured to extract a respiration signal reflected by the user from the reflected signal;
The respiratory signal screening submodule 22 is connected with the respiratory signal detection submodule 21 and is used for screening respiratory signals to obtain corresponding effective respiratory signals;
the breath type classification sub-module 23 is connected with the breath phase detection sub-module 22 and is used for extracting features of the effective breath signals to obtain a corresponding channel impulse response matrix, and inputting the channel impulse response matrix into a breath classification model obtained through pre-training to obtain the breath type adopted by the user in the abdominal respiration training process so as to monitor the effectiveness of the abdominal respiration training of the user.
Specifically, in this embodiment, the signal acquisition module 1 is an acoustic device, and an existing intelligent speaker may be used, where the acoustic device has a speaker capable of sending a frequency modulated continuous wave radar signal (frequency modulated CARRIER WAVE, frequency modulated carrier) and a microphone array capable of receiving a reflected signal. The signal frequency of the frequency modulated continuous wave radar signal is preferably 18-22kHz.
Further, when the user breathes, the breathing types adopted include chest breathing and abdomen breathing, the chest breathing and the abdomen breathing have certain periodicity, and the chest breathing and the abdomen breathing cause different body actions. Based on this, the reflected signal is demodulated by the respiratory signal detection sub-module 21 and the distance interval with the most pronounced periodic variation of power is found as the respiratory signal reflected by the user; subsequently, the effective breathing signal consistent with the breath is obtained through the sound wave shape screening of the breathing signal by the breathing signal screening submodule 22; and the respiratory type classification sub-module 23 is connected with the respiratory phase detection sub-module 22 and is used for extracting features of the effective respiratory signals to obtain advanced features which can represent respiratory types, namely corresponding channel impulse response matrixes, so that the channel impulse response matrixes are input into a respiratory classification model obtained by training in advance to be processed to obtain the respiratory types adopted by the user in the abdominal respiratory training process, and the effectiveness of the abdominal respiratory training of the user is monitored.
In a preferred embodiment of the present invention, the fm continuous wave radar signal includes a plurality of chirped harmonic bands and a corresponding plurality of blank bands spaced apart.
Specifically, in this embodiment, the frequency of the chirped harmonic band varies linearly with time, and the signal f=f 0 +b/Tt is transmitted at time T, where f 0 represents the initial frequency, B represents the bandwidth of the chirped harmonic band, and T represents the duration of the chirped harmonic band. Preferably, f 0 =18 khz, b=4 khz, t=0.05 s to fit the hardware configuration of most commercial intelligent speakers. The corresponding microphone array preferably supports a sampling frequency of 48kHz. It is further preferred that a blank band is added after each chirped harmonic band to effectively avoid signal overlap interference, and more preferred that the chirped harmonic band and the blank band have the same band length.
In a preferred embodiment of the present invention, the respiratory signal detection sub-module 21 includes:
A first detecting unit 211, configured to perform signal cross-correlation processing on the transmitted fm continuous wave radar signal and the received corresponding reflected signal, so as to obtain a peak value of each reflected signal;
The second detecting unit 212 is connected to the first detecting unit 211, and is configured to perform signal autocorrelation processing on each peak to obtain a signal sequence formed by each peak related to respiration, as a respiration signal reflected by the user.
Specifically, in the present embodiment, by measuring the propagation delay τ of the reflected signal from the speaker transmitting the fm continuous wave radar signal to the microphone array receiving, the distance d from the reflecting surface to the intelligent speaker can be calculated by the following formula:
d＝c*τ/2
Where c represents the propagation velocity of the fm continuous wave radar signal.
In order to measure the propagation delay τ, the transmitted fm continuous wave radar signal may be cross-correlated with the received reflected signal, a technique that measures the similarity of the two sequences, from which the peak value of each reflected signal may be obtained.
Delay is measured by cross-correlation, which is defined as follows:
Where X corr (N) represents the cross-correlation value of the similarity between the transmitted fm continuous wave radar signal X and the N sample shifted version of the received reflected signal y, N represents the sample length of the fm continuous wave radar signal X and the received reflected signal y, and N represents the number of right-shifted samples.
Because the propagation delay of the received reflected signal should be only amplitude attenuated compared to the transmitted fm continuous wave radar signal, a number of sets Lags of peaks may be selected as the number of shifted samples of the received reflected signal to the original transmitted fm continuous wave radar signal. This means that the received signals from the different paths are reflections of the transmitted signal x in a Lags sample delay with a hysteresis value lag. Based on this, the propagation delay is calculated as follows:
Where F s denotes the sampling frequency of the microphone array.
Based on this, the distance d from the reflecting surface to the intelligent speaker is:
as shown in fig. 2, for a cross-correlation diagram of different lags at a particular time, the cross-correlation value varies with the lag value lag, with each peak of the lag representing a reflected signal. For example, when the cross correlation value reaches a peak at lag=102, substituting the distance formula described above, it means that at this point in time, there is a strong signal with a reflection distance of 36 cm. Similarly, other peaks represent other reflection points from different distances.
After obtaining the peaks characterizing the distance by the cross-correlation process, it is necessary to find further reflection paths that can be used to detect respiration. For a transmitted frequency modulated continuous wave radar signal, the reflected signal may be obtained from several different distances. The reflected signal of the continuously transmitted frequency modulated continuous wave radar signal is calculated in the time domain. Information can then be obtained as to whether or not there is a particular reflective surface at a particular time. From a time domain perspective, for a particular distance, if it is exactly the reflection point of the chest or abdomen, the corresponding reflection signal will vary periodically with respiration. Thus, autocorrelation may be performed on all Lags peaks to find the respiratory signal with the highest periodicity caused by respiration, i.e. resulting in a user reflection.
In a preferred embodiment of the present invention, respiratory signal screening sub-module 22 includes:
a clustering unit 221, configured to cluster each corresponding peak value according to the signal frequency of the respiratory signal to obtain a plurality of frequency clusters;
The screening unit 222 is connected to the clustering unit 221, and is configured to screen a frequency cluster with a signal frequency corresponding to a preset respiration frequency, and take a signal sequence formed by each peak in the frequency cluster as an effective respiration signal.
In particular, in this embodiment, in order to detect the respiratory phase, it is necessary to know the chest or abdomen movement direction of the user. Some work detects chest movement directly by tracking the point of direct reflection. In practice, however, the results are easily affected because the point of direct reflection may change with slight movements of the body and the reflection path with good periodicity may not be a direct reflection path. To make the system robust, we use the breathing properties to determine the inspiration phase and expiration phase. Depending on the reflection point, there are three modes for the CIR variation of the channel impulse response matrix corresponding to the respiration signal. If the peak happens to represent a state where the user inhales to a maximum or exhales to a minimum, the signal changes to a sine wave. As shown in fig. 3 (a), the peak represents the distance to the chest when the chest of the user expands to its maximum. From this point on, the power amplitude at this distance decreases with exhalation and then increases with inhalation. As shown in fig. 3 (b), when the peak is the distance from the chest of the user when the chest is contracted to a minimum, the channel impulse response matrix CIR decreases when the user inhales and increases when the user exhales. However, when the peak is between the maximum inhalation point and the minimum exhalation point, the situation is different. As shown in fig. 3 (c), the channel impulse response matrix CIR decreases when the user inhales, and then increases when the user breathes between the maximum inhalation point and the minimum exhalation point. The CIR of the channel impulse response matrix decreases again when the user continuously exhales, and increases again when he starts inhaling. It can be found that the frequency of the sine wave generated by breathing in the third case is twice the frequency during the same breathing time.
Based on the method, the peak values can be clustered according to the signal frequency of the respiration signals, preferably, a K-Means clustering algorithm is adopted for clustering to obtain a plurality of frequency clusters, further, the frequency clusters corresponding to the preset respiration frequency are obtained through screening, and the signal sequence formed by the peak values in the frequency clusters is used as an effective respiration signal.
In the preferred embodiment of the present invention, the respiratory signal screening sub-module 22 further includes a phase dividing unit 223 connected to the screening unit 222, and configured to perform phase dividing processing according to the effective respiratory signal to obtain and output an inspiratory respiratory ratio corresponding to each respiratory cycle included in the effective respiratory signal.
Specifically, when performing abdominal respiration monitoring, besides the respiratory type of the user, the inspiratory respiratory ratio corresponding to each respiratory cycle of the user needs to be acquired, so as to further accurately monitor the effectiveness of abdominal respiration training of the user. Based on this, in this embodiment, the peak value of the highest autocorrelation is selected to determine the phase based on the frequency cluster corresponding to the respiration frequency whose signal frequency corresponds to the preset respiration frequency. Next, the peaks and troughs of the sine wave are found and used for phase segmentation. In a preferred embodiment of the present invention, the phase dividing unit 223 includes:
a dividing subunit 2231, configured to phase-divide the effective respiratory signal to obtain a corresponding respiratory period;
The processing subunit 2232 is connected to the dividing subunit 2231, and is configured to divide each respiratory cycle into an inhalation time and an exhalation time, and further output an inhalation-exhalation ratio corresponding to each respiratory cycle processed according to the inhalation time and the exhalation time.
Specifically, in the present embodiment, the longer stage is set as exhalation and the shorter stage is set as inhalation by utilizing the characteristic that the breathing time is always longer than the inhalation time. As the inspiration time and expiration time are detected, the inspiration-expiration ratio may then be calculated by the following formula:
wherein I/e_ratio represents the inspiration-to-Expiration ratio, inspiration _time represents the inspiration time, and expiration_time represents the Expiration time.
Although respiration has been detected by the above process, it is still difficult to distinguish the type of respiration with the human eye. Based on the above, the technical scheme adopts a breath classification model to divide breath types. In the preferred embodiment of the present invention, the respiration monitoring module 2 further comprises a model training sub-module 24, connected to the respiration type classification sub-module 23, the model training sub-module 24 comprising:
The data obtaining unit 241 is configured to obtain effective respiratory signals and corresponding real respiratory types acquired during the abdominal respiratory training of the multiple subjects, and extract features of each effective respiratory signal to obtain a corresponding channel impulse response matrix;
A preprocessing unit 242, connected to the data acquisition unit 241, for preprocessing the channel impulse response matrix to obtain a preprocessing matrix;
The training unit 243 is respectively connected to the data obtaining unit 241 and the preprocessing unit 242, and is configured to take the preprocessing matrix as input, take the corresponding real breathing type as output, and train to obtain a breathing classification model by adopting a migration learning mode.
In a preferred embodiment of the present invention, the preprocessing unit 242 performs signal filtering processing on the channel impulse response matrix to obtain an over-preprocessing matrix containing dynamic information of the mobile device in a tested environment where the subject is located.
Specifically, in this embodiment, since the channel impulse response matrix CIR is a combination of static information (i.e., sofas, walls, and human torso) and dynamic information (e.g., the body movements of the passing person and the user themselves) in the test environment. However, only respiratory induced body movements help us classify the respiratory type. It is therefore necessary to eliminate those unwanted signals.
Further specifically, first, it is necessary to filter the reflected signal at a long distance. This can eliminate the effects of static and distant object movement. In the channel impulse response matrix CIR, due toTherefore, the tuning (tune) detection range can be performed by only modifying the maximum lag value. In real life we find that most users tend to place acoustic devices, such as smart speakers, in front of them at a location of about 50cm to 1 m. Therefore, the sensing range is set to be 30cm to 1m empirically, so that in the daily use environment, most users do not need to move the intelligent sound box, and the interference of remote objects is eliminated. We convert this to a distance of the lag number, i.e. 84 to 280.
Next, we cancel the static component and only focus on moving objects. Note that the CIR of the channel impulse response matrix measures the superposition of components of the same distance, whereas the CIR of the channel impulse response matrix of the dynamic object cannot be measured directly. To remove the effect of static objects, we calculate the channel impulse response matrix CIR difference on the time axis as:
dynamic_CIR(t)＝CIR(t)-CIR(t-1)
after this step, only the mobile component is retained.
In a preferred embodiment of the present invention, the data acquisition unit 241 includes:
an acquisition subunit 2411, configured to acquire valid respiratory signals and corresponding real respiratory types acquired during abdominal respiration training of a plurality of subjects;
A data enhancer unit 2412, connected to the acquisition subunit 2411, configured to perform data enhancement on the effective respiratory signal by using a preset data window to obtain an enhanced respiratory signal;
The feature extraction subunit 2413 is connected to the data enhancement subunit 2412, and is configured to perform feature extraction on each enhanced respiratory signal to obtain a corresponding channel impulse response matrix.
In particular, in this embodiment, training the breath classification model requires a large amount of data, otherwise the model is easily overfitted. However, manually collecting data for all scenes is expensive and difficult to implement. To make the model more robust, we extend the training set by data enhancement without overfitting. Because the system is used every day, the data enhancement strategy can be designed according to different factors without extreme scenarios.
Further specifically, the first factor that requires attention is the respiratory rate. To classify the type of breath, a sequence of ordered consecutive spatial frames is used as input. The input length and the selection of frames are significant. In order to balance computational efficiency and accuracy, it is empirically preferred that the fixed data window be 5, which corresponds to 50 frames of samples. When the breathing cycle is less than 5, some additional data needs to be supplemented. There are two ways to supplement: first, motion compensated frame interpolation may be used to provide longer signals. Secondly, adding adjacent respiratory data into the respiratory cycle, and simulating the segmentation of inaccurate data. Also, when the respiration time is greater than the window size, only a portion of the signal is used as input by extracting frames of the original respiration signal or cutting the respiration data into window sizes. For robustness of the data, a random factor α may be set to represent the chance of interpolation and β for compression. Because the pattern of breath classification is most pronounced when inspiration reaches a maximum, the inspiratory peak may be set as the window center. Since the characteristics of breathing are different at different stages, to ensure that this information is not lost, it is preferable to randomly move the window to both sides by a function:
GetWindow(peak):return[peak-random*l,peak+(1-random)*l]
where range e (0, 1), peak is the period of time that the subject just completed inhaling, and l is the manually set window length. In addition, by window selection, the respiratory cycles can be treated equally regardless of whether they are longer or shorter than the window size.
Another factor to consider is the distance of the patient from the intelligent speaker. We observe that different distances of motion can be simulated by introducing drift in the tap index, which can be achieved by translating the channel impulse response matrix CIR onto the spatial axis. A random factor y may be set to determine when to increase the random distance on the y-axis. In addition, the influence of different angles at which the subject is located on the intelligent speaker is considered. Theoretically, for an omni-directional microphone, the received signal is a superposition of all equidistant signals in all directions in space. Thus, for a single microphone, the relative angle of the user does not affect the result. Taking into account the inherent relative positions of the multiple microphones, spatial translational invariance of the CNN can be exploited to efficiently extract advanced features for classification. Thus, it is not necessary to augment the data from the point of arrival of the signal. It was verified in experiments that the variation of the angle of arrival does not affect the performance of the model.
Since CNN is suitable for spatially continuous data, and can also be applied to time series model with stable structure, and meets the requirement of system, CNN is selected as model network. As shown in fig. 4, the model network structure includes a first convolution layer 100, a first pooling layer 101, a second convolution layer 102, a second pooling layer 103, a third convolution layer 104, a third pooling layer 105, and a full connection layer 106, which are connected in sequence. Specifically, the model takes as input the channel impulse response matrix CIR from 6 microphones. The size of each input is D x T x N, where D is the detection range represented by the peak, T is the sequence of frames in the time domain, and N is the number of microphones. The design of the three convolution layers aims at extracting advanced features from the channel impulse response matrix CIR. Convolution kernels are 3 x 3 in size and are widely used for a variety of tasks. For the three convolutional layers, the kernels are set to 16, 32 and 64, respectively. These parameters are set empirically. Following each convolution layer, batch normalization is further applied. The activation function is a ReLU. The features are then fed to a pooling layer (pooling) to reduce the number of parameters to learn.
CNNs only handle a single channel individually. But as previously described each distance is a superposition of multiple reflected signals. The overlapping signals need to be further processed to avoid overfitting. Because both the perceptron and the beamforming have the same mathematical expression, y= wX, a data-driven approach can be used instead of beamforming. According to the general approximation theorem, a function can be approximated arbitrarily if a feedforward neural network has a linear output layer and at least one hidden layer. Because the function that needs to be fitted is linear and the amount of data is small, only one extra hidden layer is added to fit. This step automatically extracts spatial information that may be helpful in place of beamforming for implementation to achieve spatial selectivity. Therefore, it is called automatic beamforming.
While the generic model is promising and environmentally unchanged, it is still desirable that the model be better able to serve users. Because of the different breathing rhythms, diaphragm intensities and body shapes of each patient, training data to avoid small samples cannot be fully generalized to new users. Therefore, in the technical scheme, the tag data of a small number of new users are personalized by using the transfer learning technology. The user can independently perform abdominal breathing exercise only through training of a rehabilitation team. Thus, a small amount of data may be collected at this stage. The network is then frozen except for the last layer and the last layer is retrained using the obtained data. In this case, we personalize the solution to a new user with a custom function selection.
In a preferred embodiment of the present invention, the solution is implemented on SEEED RESPEAKER-Mic circular array kit (Philips notebook USB speaker), SPA20 and Raspberry Pi4B development boards. The devices are tied together and placed on a desktop to mimic the hardware configuration of most commercial intelligent speakers on the market. The loudspeaker continuously generates and emits FMCW CHRIP at 18-22 kHz. The microphone records a signal with a reflected sample rate set to 48 kHz. The recorded signals are sent to a notebook computer for further analysis. The signals are preferably processed and classified by Matlab and Python, and the breath types are classified by the processed CIR matrix. Preferably, the framework used for deep learning is Pytorch. Using Vernier GoThe respiratory belt serves as baseline detection data for the respiratory phase (FDA-certified devices). Two cameras are placed in front of and to the sides of the volunteer to record the type of breathing of the user. In the case where the participant feels relaxed, he or she will be asked to take a three minute abdominal or chest breath. The whole experimental process is completed under supervision.
The experimental verification was performed by ten volunteers (five men and five women). Before experimental verification, the volunteers are ensured to fully grasp the breathing skills, and accidents do not occur in the experiment. Before the experiment, all volunteers watch the video of the abdominal respiration training teaching and read the abdominal respiration related guidance. Each volunteer was assured of having the ability to breathe on his own and be examined by a therapist prior to performing the experiment. Experiments were performed in three environments in daily life.
The experimental verification result shows that the median error rate in the breath detection of the technical scheme is 0.2BPM, the I/E ratio is accurately deduced, and the average absolute percentage error is less than 5.6%. The recall rate of the abdominal respiration is 96% and the detection accuracy is 95%.
The invention also provides an abdominal respiration monitoring method which is applied to the abdominal respiration monitoring system, as shown in fig. 5, and comprises the following steps:
Step S1, continuously transmitting a frequency modulation continuous wave radar signal to the region where a user is located and receiving a corresponding reflected signal by the abdominal respiration monitoring system in the process of performing abdominal respiration training by the user;
step S2, the abdominal respiration monitoring system extracts a respiration signal reflected by a user from the reflected signal;
step S3, screening the respiratory signals by the abdominal respiration monitoring system to obtain corresponding effective respiratory signals;
and S4, performing feature extraction on the effective respiratory signals by the abdominal respiration monitoring system to obtain a corresponding channel impulse response matrix, inputting the channel impulse response matrix into a respiratory classification model obtained through pre-training, and processing to obtain the respiratory type adopted by the user in the abdominal respiration training process so as to monitor the effectiveness of the abdominal respiration training of the user.
The foregoing description is only illustrative of the preferred embodiments of the present invention and is not to be construed as limiting the scope of the invention, and it will be appreciated by those skilled in the art that equivalent substitutions and obvious variations may be made using the description and drawings, and are intended to be included within the scope of the present invention.
Claims (8)
1. An abdominal respiration monitoring system, comprising:
The signal acquisition module is used for continuously transmitting a frequency modulation continuous wave radar signal to the region where the user is located and receiving a corresponding reflected signal in the process of performing abdominal respiration training on the user;
The respiration monitoring module is connected with the signal acquisition module, and the respiration monitoring module comprises:
The respiratory signal detection submodule is used for extracting respiratory signals reflected by the user from the reflected signals;
The respiratory signal screening submodule is connected with the respiratory signal detection submodule and is used for screening the respiratory signals to obtain corresponding effective respiratory signals;
The breath type classification submodule is connected with the breath signal screening submodule and is used for extracting features of the effective breath signals to obtain a corresponding channel impulse response matrix, inputting the channel impulse response matrix into a breath classification model obtained through pre-training to obtain the breath type adopted by the user in the abdominal respiration training process, so as to monitor the effectiveness of the abdominal respiration training of the user;
The respiratory signal detection submodule includes:
the first detection unit is used for carrying out signal cross-correlation processing on the sent frequency modulation continuous wave radar signal and the received corresponding reflected signal so as to obtain the peak value of each reflected signal;
the second detection unit is connected with the first detection unit and is used for carrying out signal autocorrelation processing on each peak value to obtain a signal sequence formed by each peak value related to respiration, and the signal sequence is used as the respiration signal reflected by the user;
the respiratory signal screening submodule comprises:
The clustering unit is used for clustering the corresponding peaks according to the signal frequency of the breathing signal to obtain a plurality of frequency clusters;
The screening unit is connected with the clustering unit and is used for screening the frequency clusters corresponding to the signal frequency meeting the preset respiratory frequency, and taking a signal sequence formed by each peak value in the frequency clusters as the effective respiratory signal.
2. The abdominal respiration monitoring system according to claim 1, wherein the respiration signal screening submodule further comprises a phase dividing unit connected to the screening unit, and configured to perform phase dividing processing according to the effective respiration signal to obtain and output an inhalation respiration ratio corresponding to each respiration period included in the effective respiration signal.
3. The abdominal respiration monitoring system according to claim 2, wherein the phase dividing unit comprises:
the segmentation subunit is used for carrying out phase segmentation on the effective respiratory signal to obtain the corresponding respiratory cycle;
a processing subunit connected to the dividing subunit for dividing each of the respiratory cycles into an inhalation time and an inhalation time, respectively
And the expiration time is further output according to the inspiration-expiration ratio corresponding to each breathing cycle processed by the inspiration time and the expiration time.
4. The abdominal respiration monitoring system of claim 1, wherein the fm continuous wave radar signal comprises a plurality of chirped harmonic bands and a corresponding plurality of blank bands disposed in spaced relation.
5. The abdominal respiration monitoring system of claim 1, wherein the respiration monitoring module further comprises a model training submodule coupled to the respiration type classification submodule, the model training submodule comprising:
The data acquisition unit is used for acquiring the effective respiratory signals and the corresponding real respiratory types acquired in the process of abdominal respiratory training of a plurality of subjects, and respectively extracting the characteristics of each effective respiratory signal to obtain the corresponding channel impulse response matrix;
The preprocessing unit is connected with the data acquisition unit and is used for preprocessing the channel impulse response matrix to obtain a preprocessing matrix;
the training unit is respectively connected with the data acquisition unit and the preprocessing unit, and is used for taking the preprocessing matrix as input, taking the corresponding real breathing type as output, and training by adopting a transfer learning mode to obtain the breathing classification model.
6. The system according to claim 5, wherein the preprocessing unit performs signal filtering processing on the channel impulse response matrix to obtain the preprocessing matrix containing dynamic information of moving components in a subject environment of the subject.
7. The abdominal respiration monitoring system of claim 6, wherein the data acquisition unit comprises:
An acquisition subunit, configured to acquire the effective respiratory signals and the corresponding real respiratory types acquired during the abdominal respiratory training of the plurality of subjects;
The data enhancer unit is connected with the acquisition subunit and is used for carrying out data enhancement on the effective breathing signal by adopting a preset data window to obtain an enhanced breathing signal;
And the characteristic extraction subunit is connected with the data enhancer unit and is used for respectively carrying out characteristic extraction on each enhanced respiratory signal to obtain the corresponding channel impulse response matrix.
8. A method of monitoring abdominal respiration as claimed in any one of claims 1 to 7, comprising:
Step S1, continuously transmitting a frequency modulation continuous wave radar signal to the region where a user is located and receiving a corresponding reflected signal by the abdominal respiration monitoring system in the process of performing abdominal respiration training on the user;
Step S2, the abdominal respiration monitoring system extracts a respiration signal reflected by the user from the reflected signal;
Step S3, the abdominal respiration monitoring system screens the respiration signals to obtain corresponding effective respiration signals;
step S4, the abdominal respiration monitoring system performs feature extraction on the effective respiration signals to obtain corresponding channel impulse response matrixes, and inputs the channel impulse response matrixes into a respiration classification model obtained through pre-training to obtain respiration types adopted by the user in the abdominal respiration training process, so as to monitor the effectiveness of the abdominal respiration training of the user;
The step S2 includes:
Step S21, the abdominal respiration monitoring system carries out signal cross-correlation processing on the sent frequency modulation continuous wave radar signal and the received corresponding reflected signal so as to obtain peak values of the reflected signals;
Step S22, the abdominal respiration monitoring system performs signal autocorrelation processing on each peak value to obtain a signal sequence formed by each peak value related to respiration, and the signal sequence is used as the respiration signal reflected by the user;
the step S3 includes:
step S31, the abdominal respiration monitoring system clusters corresponding peaks according to the signal frequency of the respiration signal to obtain a plurality of frequency clusters;
Step S32, the abdominal respiration monitoring system screens out the frequency cluster corresponding to the signal frequency conforming to the preset respiration frequency, and takes a signal sequence formed by each peak value in the frequency cluster as the effective respiration signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210068468.9A CN114469061B (en) | 2022-01-20 | 2022-01-20 | Abdominal respiration monitoring system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210068468.9A CN114469061B (en) | 2022-01-20 | 2022-01-20 | Abdominal respiration monitoring system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114469061A CN114469061A (en) | 2022-05-13 |
| CN114469061B true CN114469061B (en) | 2024-06-25 |
Family
ID=81472517
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210068468.9A Active CN114469061B (en) | 2022-01-20 | 2022-01-20 | Abdominal respiration monitoring system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114469061B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118415620A (en) * | 2024-04-17 | 2024-08-02 | 北京邮电大学 | Multi-degree-of-freedom lung function monitoring method, device and storage medium based on millimeter waves |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108523892A (en) * | 2018-05-09 | 2018-09-14 | 珠海乐维电子科技有限公司 | A kind of abdominal respiration detecting system and method |
| CN109640793A (en) * | 2016-08-18 | 2019-04-16 | 皇家飞利浦有限公司 | For determining the sensing system and method for type of respiration |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102222940B1 (en) * | 2019-06-04 | 2021-03-05 | 영남대학교 산학협력단 | Respiration training apparatus |
| CN112957687A (en) * | 2021-04-25 | 2021-06-15 | 中国科学院心理研究所 | Training system is breathed to abdominal type |
| CN113616188B (en) * | 2021-08-11 | 2022-10-04 | 燕山大学 | Respiration monitoring method based on frequency modulated continuous waves of inaudible sound |
-
2022
- 2022-01-20 CN CN202210068468.9A patent/CN114469061B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109640793A (en) * | 2016-08-18 | 2019-04-16 | 皇家飞利浦有限公司 | For determining the sensing system and method for type of respiration |
| CN108523892A (en) * | 2018-05-09 | 2018-09-14 | 珠海乐维电子科技有限公司 | A kind of abdominal respiration detecting system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114469061A (en) | 2022-05-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Leng et al. | The electronic stethoscope | |
| US20210113099A1 (en) | Wireless medical sensors and methods | |
| Wang et al. | Using smart speakers to contactlessly monitor heart rhythms | |
| Zhao et al. | PPG-based finger-level gesture recognition leveraging wearables | |
| Fang et al. | Headscan: A wearable system for radio-based sensing of head and mouth-related activities | |
| US20220007964A1 (en) | Apparatus and method for detection of breathing abnormalities | |
| Wan et al. | RespTracker: Multi-user room-scale respiration tracking with commercial acoustic devices | |
| Fioranelli et al. | Contactless radar sensing for health monitoring | |
| US20170209115A1 (en) | Method and system of separating and locating a plurality of acoustic signal sources in a human body | |
| US20220378377A1 (en) | Augmented artificial intelligence system and methods for physiological data processing | |
| CN110520935A (en) | Learn sleep stage from radio signal | |
| Ali et al. | An end-to-end deep learning framework for real-time denoising of heart sounds for cardiac disease detection in unseen noise | |
| CN114469061B (en) | Abdominal respiration monitoring system and method | |
| Aleksic et al. | State of the art of audio-and video-based solutions for AAL | |
| CN112450900B (en) | Non-contact heartbeat detection method based on intelligent sound box | |
| Wang et al. | MultiResp: robust respiration monitoring for multiple users using acoustic signal | |
| Bahache et al. | An inclusive survey of contactless wireless sensing: A technology used for remotely monitoring vital signs has the potential to combating covid-19 | |
| Zakrzewski | Methods for doppler radar monitoring of physiological signals | |
| Jiokeng et al. | Handrate: heart rate monitoring while simply holding a smartphone | |
| Guo et al. | Emergency semantic feature vector extraction from WiFi signals for in-home monitoring of elderly | |
| Fan et al. | A Contactless Breathing Pattern Recognition System Using Deep Learning and WiFi Signal | |
| Pham et al. | Noncontact detection of cardiopulmonary activities of trapped humans in rescue relief events | |
| Wang et al. | Exploiting Passive Beamforming of Smart Speakers to Monitor Human Heartbeat in Real Time | |
| WO2020214865A1 (en) | Systems and methods for contactless motion tracking | |
| Wu et al. | Ubi-Asthma: Toward Ubiquitous Asthma Detection Using the Smartwatch |
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 |