Non-invasive and Unrestrained Monitoring of Human Respiratory System by Sensorized Environment
from observed data, and 3) a presenting part for letting them utilize understood information.
The human respiratory system is very complex. Therefore, it requires monitoring with many sensors such as pressure sensors to monitor the chest and abdomenfs movement, thereto for airflow around the nose and mouth, a contact-type of microphone for snoring, oximeters for oxygen saturation, and mercury sensors for posture(Fig. 1 (B)). Since these sensors need to be attached directly to the person, they impose much physiological or mental burden on him or her.
Figure 1: Environment Sensorization
Indeed, the conventional contact-type sensors can monitor physiological values certainly as long as they are used adequately. Actually, however, they very often fail to monitor continuously. For example, in 48 percent of the cases (21 of 43 cases) in the clinical study the authors conducted, the monitoring system failed to measure the physiological values continuously enough to diagnose disease. This suggests that even in a hospital where more priority is given to accurate monitoring than to comfortable monitoring, non-invasive and unrestrained monitoring is required to minimize this failure. The developed living space, SELF (Sensorized Environment for LiFe) is sensorized as shown in Fig. 1. In this study, sensorization meansmaking the room itself a sensor for inputting human daily behavior by embedding sensors into the room invisibly to keep the roomfs appearance natural and maintain its original function. The figure explains main components of the typical conventional computers are transformed to the room. For example, a keyboard, which is a kind of touch sensor, becomes a bed-shaped touch sensor. A microphone is embedded into a lighting fixture. A display is embedded into a washstand. Figure 1 (C) shows the photograph of the constructed bed room. SELF support daily healthcare at home as follows: 1) SELF observes a person using the pressure sensor bed and the ceiling dome microphone when he or she goes to bed and sleeps, 2) SELF reports useful information to the person using the washstand display when he or she goes to the washstand typically after waking up or before going to bed.
This section describes a pressure sensor bed and a ceiling dome microphone as an example of the sensorized environment. Sensors are embedded in both systems invisibly.
Pressure sensor bed
The pressure sensor bed consists of a pressure distribution sensor array, a controller, and a bed. The pressure distribution sensor has 210 Force Sensing Resistors (FSRs) which are set at 7[cm] intervals. An FSR is a thin film sensor made from piezoresistive polymer. The sampling frequency of the pressure image is 20 [Hz]. The measuring range of each pressure sensor is 0 to 1[kg]. The pressure sensor bed is used for monitoring the breath curve, oxygen desaturation(*1) frequency, posture, and body movement.
Based on the fact that Cheyne-Stokes-like breathing occur at high probability in oxygen desaturation, the pressure sensor bed can estimate oxygen desaturation frequency by measuring Cheyne-Stokes-like breathing(*2) frequency. Figure 2 comparesthe frequency of Cheyne-Stokes-like breathing and oxygen desaturation 4%(*3) at an interval of 10 minutes with 3 patients with disease of different seriousness.
Figure 2. Comparison of histogram of Oxygen desaturation 4% (OD4) detected by conventional sensor and Cheyne-Stokes-like breathing detected by pressure sensor bed
(*1) Oxygen saturation (SpO2) expresses the percentage of oxyhemoglobin molecules to all hemoglobin molecules and is almost 100 % in healthy subjects. Oxygen desaturation means the percentage falls for some reasons. Doctors find oxygen saturation monitoring important physiologically to judge whether respiration is normal.
(*2) It is a kind of periodic respiration.
The authors defined Cheyne-Stokes-like breathing as the breath with respiration
effort without ventilation accompanying gradual increase and gradual decrease,
or sudden increase and gradual decrease.
(*3) Oxygen desaturation 4% means oxygen saturation falls 4%. From Fig. 2, a high correlation was confirmed between the Cheyne-Stokes index monitored by the pressure sensor bed and oxygen desaturation 4% index.
Ceiling dome microphone
The authors invented the ceiling dome microphone which consists of a ceiling dome, a lighting fixture, and a omnidirectional microphone. This device has two functions: indirect lighting and gathering sound. The ceiling dome is used to reflect both light and sound. A microphone is set at the focal point of the reflector. The diameter of the dome is 900[mm]. The device enables detection not only snoring sounds but also normal breathing sounds with high sensitivity while keeping the roomfs appearance natural. It is positioned above the bed. The gain obtained by the ceiling dome is maintained at more than 20[dB] for high frequency sounds of more than 6[kHz]. Since the frequency of normal breathing sounds ranges from 5 to 15 [kHz], this device can detect breathing sounds, i.e., air flow at the mouth and nose.
Figure 3 shows an example of breathing sounds detection using the ceiling dome microphone. Roomfs background noise includes such noises as those of an air conditioner and a computer. The figure shows that both the inhalation and exhalation component of the breath cycle are detected quite clearly.
Figure 3. Detection of normal breathing sounds by ceiling dome microphone
UNDERSTANDING INTERNAL STATUS BASED ON A MODEL OF HUMAN FUNCTIONS
Human respiratory system
Figure 4 shows physiology and physics of human respiration, and a method for monitoring conditions of the human respiratory system. The human respiratory system deeply relates to brain activity, the circulatory system, respiratory organs such as lung, respiratory muscles such as diaphragm, peripheral organs such as nasal cavity and oral cavity, human posture and so forth as shown in Fig. 4 a). To monitor conditions of the human respiratory system, a model of the respiratory system for estimating the conditions from sensor data is necessary. This section describes the developed model of the human respiratory system and a method of estimating the conditions using the model.
Figure 4. Physiology, physics and measuring method of human respiratory system
Physiology of human respiratory system
Respiration keeps oxygen and carbon dioxide in body fluids within a constant range. Tidal air is controlled by respiratory muscles such as the diaphragm and intercostals based on the oxygen and carbon dioxide concentration detected by some internal sensors. The sensors exist in carotid body of the common carotid artery and central chemosensitive area of the medulla oblongata. The control of the human respiratory system is shown in Fig. 4 b).
Non-invasive and unrestrained monitoring of human respiratory system
The followings are physiological values that the developed human model can estimate from sensor data. Physics of the human respiratory system and the methods for monitoring these physical values are outlined in Fig. 4.
- Posture, body movement
Posture is recognized by analyzing pressure sensor signals based on a posture model. Body movement is calculated by detecting changes in pressure sensor output.
- Breath curve
Breath curve is calculated by the "same phase sum method" which is a method for appropriately summing the control offset considering the phase difference in the output change in the pressure sensor.
- Oxygen desaturation in blood
Oxygen desaturation is detected by detecting Cheyne-Stokes-like breathing from the calculated breath curve. The principle is that Cheyne-Stokes-like breathing occurs with high probability when oxygen desaturation appears. This method can detect only the frequency of oxygen desaturation, not absolute values of oxygen desaturation.
- Airflow at mouth and nose
Airflow at the mouth and nose is detected using the ceiling dome microphone.
Evaluation of estimation function of SELF
The authors conducted experiments for a real patient suffering from Sleep Apnea Syndrome. Figure 5 compares physiological values measured by a conventional system and that estimated by the developed system. Apnea estimation by our system is done using both a detecting function of snoring sounds and a monitoring function of the amplitude of breath curve. If the amplitude is less than a certain threshold and that there are no snoring/breathing sounds, the system finds obstructive apnea. Obstructive apnea is a typical apnea such that the patient cannot inhale the air despite of breath effort and that the concentration of the oxygen in blood falls. Oxygen desaturation estimation is done by analyzing the change of the amplitude of breath curve. The figure shows the system can estimate apnea correctly in 8 of 10 cases. The figure also shows the system can estimate ogygen desaturation in 9 of 10 cases.
Figure 5. Comparison monitoring and estimating function of SELF with conventional one
This section reports a washstand display as an examples of an information
providing furniture. The washstand tends to be used every morning and night,
which means if the washstand has a function of providing information as well
as mirroring the inhabitant, it is able to convey periodically. According
to questionnaire survey conducted by us, women spend at the washstand from
30 minutes to 1 hour a day. Since there are enough time to read information
from the washstand, the washstand is one of the most suitable place for providing
a person with useful information. Figure 6 shows the developed washstand display.
It consists of a liquid crystal display (LCD) having a touch sensor and a
vision system. The LCD is used for displaying not only a personfs face but
also some health information analyzed by SELF.
Figure 7 shows the image displayed on the LCD. The health information is displayed around the face. This report consists of weekly and daily information on the changes in health condition and is expressed with bar graphs and text. Changes to be notified to the person are determined based on the average condition of the person and of other people. For the basis of information on the average condition of people, we used a medical textbook.
Figure 6. Constructed washstand display
Figure 7. Example of output of washstand display
This report describes a sensorized environment for non-invasive and unrestrained monitoring the human respiratory system.The system consists of 1) a sensorized environment for robustly and naturally observing inhabitants, 2) a human model for understanding conditions of the human respiratory system, and 3) a presenting environment for letting them utilize analyzed information. As examples of the sensing and presenting environment, this report reports the daily living space (SELF) which has a ceiling dome microphone, a pressure sensor bed, and a washstand display. The experiments conducted for a real patient suffering from Sleep Apnea Syndrome proved the developed system can estimate apnea, oxygen desaturation frequency using sensory data from the pressure sensor bed and the ceiling dome microphone.
- Y. Nishida, T. Hori, "Non-invasive and Unrestrained Monitoring of Human Respiratory System by Sensorized Environment, " Proc. of the First IEEE International Conference on Sensors (Sensor 2002), pp. 62.4(1)-(6), June 2002
- T. Kuga, M. Takayama, T. Ishii, Y. Nishida, "Respiration monitoring of sleep apnea syndrome using a pressure sensor bed," ûôÈ, Vol. 13, No. 2, pp.1-11, 2001
- Y. Nishida, T. Hori, "Sensorized Living Space with a Model of Human Functions," Proc. of Digital Human Modeling Workshop at IEEE International Conference on Intelligent Robots and Systems (IROS2001), 2001
- Y. Nishida, T. Hori, T. Suehiro, S. Hirai, "Sensorized Environment for Self-communication Based on Observation of Daily Human Behavior," Proc. of 2000 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS2000), pp.1364-1372, 2000
- Y. Nishida, T. Hori, T. Suehiro, S. Hirai, "Monitoring of Breath Sound
under Daily Environment by Ceiling Dome Microphone," Proc. of 2000 IEEE
International Conference on Systems, Man, and Cybernetics (SMC2000), pp.1822-1829,