TWI809864B - Sensing method and sensing system - Google Patents

Abstract

The disclosure provides a sensing method and a sensing system suitable for detecting the respiratory pattern of an organism. The respiratory pattern of the organism is sensed by a sensing device. The respiratory pattern of the organism is identified. According to the identified respiratory pattern, we can judge whether the organism is abnormal.

Description

Sensing method and sensing system

The disclosure relates to a sensing method and a sensing system, in particular, a sensing method and a sensing system suitable for detecting the breathing pattern of a living body.

In recent years, the elderly care, nursing care, and long-term care services have flourished in response to the global aging trend. The rapid increase in the proportion of the elderly population and the shrinking of the young and middle-aged population have led to a serious imbalance in the care ratio. Relying on technological means to assist care has become an inevitable trend. Numerous technological care solutions are in the ascendant, but services for the elderly focus on human nature. In addition to physical health care, the concept of holistic medical care covering psychology, social interaction, and spirituality is gradually being valued. Coupled with the sudden spread of the epidemic, the isolation policy has caused inconvenience among patients, medical staff and their families, and many regrettable incidents have inevitably occurred. If the current technological care and telemedicine solutions are properly configured and planned, it can provide leapfrog satisfaction among bedridden patients, caregivers and relatives.

However, because the bedridden person will stop breathing due to unexpected situations, the caregiver and relatives will be caught off guard and have the regret of not being able to see the last side. Therefore, a method and system that can detect the breathing state of the bedridden person is extremely needed to be able to Let the caregiver or medical staff judge it and deal with it immediately.

Therefore, this disclosure proposes a sensing method and sensing system suitable for detecting the breathing patterns of living organisms, which can instantly determine whether the breathing patterns of living organisms are abnormal, so that caregivers or medical personnel can judge it and perform real-time monitoring. deal with.

An embodiment of the present disclosure provides a sensing method suitable for detecting the breathing pattern of a living body, including: sensing the breathing pattern of the living body by a sensing device; identifying the breathing pattern of the living body; Breathing patterns, and then judge whether there is any abnormality in the organism.

An embodiment of the present disclosure provides a sensing system suitable for detecting the breathing pattern of a living body, including: a sensing device, which senses and identifies the breathing pattern of a living body, and converts it into a corresponding breathing pattern according to the type of the breathing pattern. Detection signal; processing device, connected to the sensing device, receiving the corresponding detection signal, and analyzing the detection signal, and then judging whether the organism is abnormal

In order to further understand the features and technical content of the disclosure, please refer to the following detailed description and drawings related to the disclosure. However, the provided drawings are only for reference and illustration, and are not intended to limit the disclosure.

FIG. 1 shows a sensing method suitable for detecting the breathing pattern of a living body according to an embodiment of the present disclosure. The sensing method includes: sensing the breathing pattern of the living body by the sensing device (step (S101), then identifying the breathing pattern of the living body (step S102), and further judging the breathing pattern of the living body according to the recognized breathing pattern. Whether there is abnormality in the body (step S103). The aforementioned sensing method also includes when the biological body is abnormal, it can immediately notify the abnormality for processing. In this disclosure, the aforementioned biological body includes humans or animals, and the sensing device is a physiological radar. Physiological radars include non-contact self-injection locking radars or frequency modulated continuous wave radars.

The aforementioned breathing pattern can be judged according to the existing medical common sense, and the breathing pattern can include normal breathing (Eupnea), tachypnea (tachypnea), bradypnea (bradypnea), sleep apnea (sleep apnea), tidal breathing (Cheyne -Stokes), dying breath (Agonal) and other patterns, as shown in Figure 2. Figure 2 shows various breathing patterns. From top to bottom, it represents different waveforms of normal breathing, tachypnea, bradypnea, sleep apnea, tidal breathing, and near-death breathing. Although this disclosure only discloses the six breathing patterns shown in FIG. 2 , it is not limited thereto, and it can also be added to this disclosure based on existing breathing patterns for judgment.

FIG. 3 shows a sensing system suitable for detecting the breathing pattern of a living body according to an embodiment of the present disclosure. The foregoing sensing system includes a sensing device 301 and a processing device 302 . The sensing device 301 senses and identifies the breathing pattern of the living body 303, and converts it into a corresponding detection signal according to the type of the breathing pattern. The processing device 302 is connected to the sensing device 301, receives corresponding detection signals, and analyzes the detection signals to determine whether there is any abnormality in the biological body 303. When the biological body has abnormality, it can immediately notify the abnormality for processing. In this disclosure, the aforementioned organisms include humans or animals, and the sensing device is a physiological radar, which includes a non-contact self-injection locking radar or a frequency-modulated continuous wave radar.

In this disclosure, the processing device 302 applies time-domain or frequency-domain analysis to calculate the characteristic value of the breathing pattern, and judges whether the organism is abnormal according to the characteristic value. Fig. 4 shows the flow chart of the method for applying time-domain or frequency-domain analysis to calculate the respiratory pattern. The foregoing method steps include: performing time-domain analysis and processing on the detection signal (step S401) and applying short-time-distance Fourier transform to the detection signal. Computed to produce the computed signal. The calculated signal is subjected to spatial data extraction of one-dimensional data by a one-dimensional convolutional layer (step S402 ). Next, the calculated signal is processed by long-short-term memory processing (step S403). The aforementioned long-short-term memory processing is a special Recurrent Neural Network (RNN), mainly to solve the problem of gradient disappearance and gradient explosion in the long sequence training process, and the Recurrent Neural Network (RNN), is A neural network for processing sequence data. Compared with the general neural network, it can handle the data of sequence change. Secondly, apply the technique of preventing overfitting (DropOut) (step S404), the main purpose is to prevent overfitting, the primary parameter rate is the ratio of turning off the hidden layer nodes, and randomly turn off the connection between the hidden layer nodes and the input neurons, Do not update the weights, resulting in multiple results, and then compare and remove extreme values, so as to avoid the phenomenon of overfitting. Then apply deep neural network processing (Deep Neural Network, DNN) (step S405), the calculated signal has a neural network with many hidden layers. The first layer is the input layer, the last layer is the output layer, and the layers in between are all hidden layers. Layers are fully connected, and finally, the type of breathing pattern can be determined (step S406).

The aforementioned breathing pattern can be judged according to the existing medical common sense. The aforementioned breathing pattern includes normal breathing (Eupnea), tachypnea (tachypnea), bradypnea (bradypnea), sleep apnea (sleep apnea), tidal breathing (Cheyne -Stokes), dying breath (Agonal) and other patterns, as shown in Figure 2. Although this disclosure only discloses the six breathing patterns shown in FIG. 2 , it is not limited thereto, and it can also be added to this disclosure based on existing breathing patterns for judgment.

The aforementioned physiological radar may include a frequency modulated continuous wave radar or a non-contact self-injection locking radar. Figure 5A shows a block diagram of a simple frequency modulated continuous wave radar. As shown in FIG. 5A , the frequency modulated continuous wave radar includes a radar transmitter 51 , a radar receiver 52 and a CPU 53 , and an antenna 54 and an antenna 55 . The radar transmitter 51 continuously transmits multiple signals S1 to the living body H through the antenna 54, and then the radar receiver 52 receives the multiple signals S2 reflected back by the living body H through the antenna 55 and performs multiple signal differences and multiple signal differences. and the sum value, and then couple the multiple signal difference values with the multiple signal sum values to generate a coupled signal, which is then input to the central processing unit 53 for processing. The radar transmitter 51 can be a signal synthesizer, the radar receiver 52 can be a mixer, a low-pass filter and an analog-to-digital converter, and the central processing unit 53 can be a microprocessor, a graphics processor, a digital signal processor, etc. wait. FIG. 5B shows a block diagram of a simple non-contact self-injection locking physiological signal radar sensor. As shown in Figure 5B, the self-injection-locked integrated circuit 61 transmits a signal to the living body H through the antenna 62, and the signal reflected by the living body H passes through the antenna 63 back to the self-injection-locked integrated circuit 61, and then the self-injection-locked integrated circuit 61 61 generates a frequency modulated and amplitude modulated radio frequency signal S3. Although this disclosure only discloses the aforementioned frequency-modulated continuous wave radar sensor in FIG. 5A and the non-contact self-injection-locked physiological signal radar sensor in FIG. 5B , it is not limited thereto. In the present disclosure, the variable continuous wave radar sensor and the non-contact self-injection locking physiological signal radar sensor are added or subtracted for sensing.

The detection of breathing patterns disclosed in this disclosure can be combined with the existing physiological sensing care system to provide a bridge for bedridden people to communicate with the outside world with a hidden video intercom screen. Due to the need for privacy protection, the video intercom screen is normally closed. When doctors, nurses and family members need video communication, it can be launched from the remote activation screen to wake up patients and elders for two-way communication. Through the video screen, professional medical staff can remotely check the patient's condition in real time and give professional advice when necessary. The video intercom screen and color photography lens (such as RGB lens) provide remote photoplethysmography (Remote Photoplethysmography; rPPG) non-contact heartbeat and blood oxygen analysis, allowing professional medical staff to observe the patient's physiological state during consultation. The video screen can provide relatives with long-distance conversations. The usual emotional connection and the dying conversation at important moments can solve the problem of being unable to arrive in time due to isolation or residential transportation. The video intercom screen can also serve as a source of external information and provide real-time content services. Apply for a patent for bedside automated consultation and digital content delivery services provided by this hidden video intercom system for bedside care.

The present disclosure has been described by the above-mentioned related embodiments, but the above-mentioned embodiments are only examples for implementing the present disclosure. It must be pointed out that the disclosed embodiments do not limit the scope of this disclosure. On the contrary, modifications and equivalent arrangements included in the spirit and scope of the patent claims are included in the scope of the present disclosure.

S101: step S102: step S103: step 301: Sensing device 302: processing device 303: Biological organisms S401: step S402: step S403: step S404: step S405: step S406: step 51:Radar Transmitter 52:Radar receiver 53: CPU 54: Antenna 55: Antenna 61: Self-injection locking IC 62: Antenna 63: Antenna S1~S3: signal

FIG. 1 shows a sensing method suitable for detecting the breathing pattern of a living body according to an embodiment of the present disclosure.

Figure 2 shows various breathing patterns.

FIG. 3 shows a sensing system suitable for detecting the breathing pattern of a living body according to an embodiment of the present disclosure.

Fig. 4 shows a flowchart of a method for calculating breathing patterns using time-domain or frequency-domain analysis.

Figure 5A shows a block diagram of a simple frequency modulated continuous wave radar.

FIG. 5B shows a block diagram of a simple non-contact self-injection locking physiological signal radar sensor.

S101: step

S102: step

S103: step

Claims (14)

A sensing method, suitable for detecting the breathing pattern of a living body, comprising: sensing a breathing pattern of a living body by a sensing device, and converting a corresponding detection signal according to the breathing pattern; by A processing device is connected to the sensing device, receives the corresponding detection signal, and performs time-domain analysis and processing on the detection signal to generate a calculated signal, and performs a calculation on the calculated signal through a one-dimensional convolution layer. The extraction of the spatial data of the three-dimensional data, the calculated signal is processed by long and short-term memory processing, and a type of the breathing pattern is judged; and the biological type is judged according to the type of the breathing pattern judged Whether the body is abnormal. The sensing method according to claim 1, further includes: when the organism has an abnormality, the medical staff can be immediately notified of the abnormality for processing. The sensing method according to claim 1, wherein the breathing patterns include normal breathing, tachypnea, bradypnea, sleep apnea, tidal breathing, and near-death breathing. The sensing method as claimed in claim 1, wherein the organism comprises a human or an animal. The sensing method according to claim 1, wherein the sensing device is a physiological radar, and the physiological radar includes a non-contact self-injection locking radar or a frequency modulated continuous wave radar. The sensing method as claimed in claim 1, wherein after the calculated signal is processed by long-short-term memory processing, it further includes applying an over-fitting prevention technique. The sensing method according to claim 6, further comprising applying deep neural network processing after applying the overfitting prevention technique. A sensing system, suitable for detecting the breathing pattern of a living body, comprising: a sensing device, which senses a breathing pattern of an organism, and converts a corresponding detection signal according to the breathing pattern; and a processing device, connected to the sensing device, receives the corresponding detection signal, and performs time-domain analysis and processing on the detection signal to generate a calculated signal, and performs one-dimensional processing on the calculated signal by a one-dimensional convolution layer The extraction of the spatial data of the data, the calculated signal is processed by long-term and short-term memory processing, and a type of the breathing pattern is judged, and then it is judged whether the organism is There are exceptions. The sensing system as claimed in claim 8, wherein the breathing patterns include normal breathing, tachypnea, bradypnea, sleep apnea, tidal breathing, and near-death breathing. The sensing system according to claim 8, wherein the sensing device is a physiological radar, and the physiological radar includes a non-contact self-injection locking radar or a frequency modulated continuous wave radar. The sensing system as claimed in claim 8, wherein the processing device applies time-domain or frequency-domain analysis to calculate the characteristic value of the breathing pattern, and judges whether the living body is abnormal according to the characteristic value. The sensing system as claimed in claim 8, wherein the organism comprises a human or an animal. The sensing system as claimed in claim 8, wherein after the calculated signal is processed by long-short-term memory processing, it further includes applying an over-fitting prevention technique. The sensing system as claimed in claim 13, further comprising applying deep neural network processing after applying the anti-overfitting technique.