Mitsuhiro Ogawa

Long term remote behavioral monitoring of elderly by using sensors installed in ordinary houses

In maintaining the health of elderly people, it can be useful to monitor their health status through their daily routines in their own home. This paper reports on the remote monitoring of the daily routine behavior of elderly patients in their domestic houses. We attempted to monitor the daily behavior of two elderly ladies who live in Mizusawa, Japan. A 74-year-old woman, who lived alone, was monitored for about two years, and another 72-year-old woman who lived alone was monitored for about a year. Several sensors were installed, including infrared sensors to detect human movement, magnetic switches to detect the opening and closing of doors, wattmeters embedded in wall sockets to detect the use of household appliances, a flame detector to detect the use of a cooking stove and a CO, sensor to detect the presence of a subject in a room by monitoring the carbon dioxide expired. An industrial networking system was introduced into each house to combine the sensors. The sensor outputs were recorded on a personal computer located in each house. The data were automatically transferred daily to another site via the Internet using CATV. With the, sensors, a network and data system, the monitoring was fully automatic and did not require the placement of any sensors on the subjects (e.g. electrodes and cuff sensors) or any operations by subjects. Information on several daily behavior patterns, such as the number of door openings, the length of sleep, absences from the house, use of a cooking stove, and time spent watching television were clearly identified either from a single sensor output, or by combining several sensor outputs. Examination of the data allowed some daily behavior, such as worship (in the form of Japanese Buddhism) and the tending of planters to be evaluated. Such monitoring can contribute to the maintenance of health.

Time-Course Action Analysis of Daily Life Investigations in the Welfare Techno House in Mizusawa

In this study, we have examined the possibility of non-intrusive monitoring at home. We describe the experimental results of the relation between an individuals recorded behavior and the actions detected by the sensors during the course of the subjects daily life. Infrared sensors and the opening of doors and windows and the use of appliances were used. In this paper, it is assumed that the subjects were sleeping when there were no responses by the infrared sensors or when only the bedroom sensor was responding. The sensors were able to detect excretion by using the outputs of the infrared sensor and the water valve of the lavatory, even if a subject did not record it in the time-course of daily life during the investigation. Our findings show that the actions of subjects who stayed at Welfare Techno House (WTH) in Mizusawa could be estimated from outside the house, especially for sleeping and excretion.

Fully automated biosignal acquisition in daily routine through 1 month

With the goal of keeping people healthy and active, long-term at-home measurement of physiological function during daily life is valuable, and is attempted in this study. As a first trial of at-home measurements, a fully automated physiological data acquisition system was developed. This system contains unconstrained physiological parameter monitors: an ECG monitor for use during bathing, a bed temperature monitor, and a weight monitor located in the toilet seat. It also contains a computer with data acquisition programs and a data format rule. Unconstrained physiological parameter monitors and fully automatic measurement procedures provide data acquisition non-invasively and without the subjects awareness. This system has been proved by an experiment involving measurement over one month of a healthy male subject (aged 29 years) in laboratory-based facilities. The system functioned for one month of data acquisition with no operator involved. The system may be used for home health monitoring of patients and also of healthy subjects.

Rhythm of daily living and detection of atypical days for elderly people living alone as determined with a monitoring system

We have developed a system for monitoring the health of elderly people living at home. Infrared and other sensor outputs are collected using a monitoring program installed on a personal computer (PC) in the home at a sampling rate of 1 Hz. Once each day, the data are transferred to a server through the Internet using a cable television (TV) connection. An elderly subject was monitored for a 12-day baseline period and completed a daily questionnaire about her activities. This enabled us to identify the rhythm of daily living (sleeping, 23:00–04:59; getting up/breakfast, 05:00–08:59; indoor activities/going out, 09:00–16:59; and dinner/going to bed, 17:00–22:59) and the average outputs from the sensors in the rooms. The subject was then monitored for a further six months. By identifying sensor output counts outside the limits of mean ±3SD, we were able to detect atypical days. During the six-month monitoring period, 29 atypical days were detected. We suggest that the monitoring system may be effective in tele-rehabilitation.

Assessment of bed temperature monitoring for detecting body movement during sleep: comparison with simultaneous video image recording and actigraphy

This study was designed to assess changes in bed temperature as an index of body movement. Changes in the temperature of the bed or mattress were used to estimate motility and number of awakenings, and total time in bed. Simultaneous temperature records, video image recordings and actigraphic data were obtained from 12 young adults for three consecutive nights. Significant agreement of time in bed and body movement among the three measurements was found. Intra-individual variability was low for time in bed. In a second study, 10 healthy elderly subjects used a temperature monitor at home for four consecutive nights. The time in bed was 8.1 h, a little longer than for the younger adults, but not markedly different. These studies show the potential usefulness of bed temperature monitoring as an index of sleep behaviour. Although the precise sleep index could not be observed, behaviour could be evaluated simply.

Remote monitoring of daily activities and behaviors at home

In maintaining the health of people, both elderly and younger, it can be useful to monitor their health status through their daily routines in their own home. This paper reports on the remote monitoring of the daily routine behaviors in an ordinary house. We attempted to monitor the daily behaviors of a subject, mainly in the kitchen and dining room. Several sensors were installed, including infrared sensors to detect human movement, magnetic switches to detect the opening and closing of doors, a carbon dioxide sensor to detect presence of the subject, and temperature sensors at the kitchen sink to detect cooking. A 31-year-old man who lived alone was monitored for about three months. The output of sensors was recorded on a personal computer and the data were transferred to another site by the Internet. Monitoring was performed fully automatically. As a result, daily habits could be clearly identified. Such monitoring can contribute to the maintenance of health.

Fully automated ECG data acquisition in a bathtub

In apply to home health monitoring, the authors attempted ECG recording in a bathtub. For fully automated data acquisition, a personal identification technique was introduced in which differences in ECG waveforms among individuals were used. The wavelet transform was applied to reduce the ECG data, and a neural network was introduced to create a discrimination algorithm by learning. In a preliminary study, it was shown that more than 91% of ECG data could be identified correctly among a family of five members.

Wearable humidity sensor with porous membrane by soft-MEMS techniques

An electric conductimetric sensor (thickness: 80 /spl mu/m) constructed in a sandwich configuration with a hydrophilic poly-tetrafluoroethylene membrane placed between two gold deposited layers was evaluated for use as a moisture sensor. The humidity level was measured by electrical conductivity of the device using the multifrequency LCR-meter at frequencies ranging from 100 Hz to 100 kHz, the device was calibrated at 100 Hz against the moisture air over the range of 30-85 % RH, which includes normal humidity level in the atmosphere and physiologic air such as breath and sweating. The response sensitivity of the conductimetric device was extremely high (i.e. less than 1 sec. for conductivity shift between humid air of 80 % RH and dried air of -60 /spl deg/C dew point) even for recovery to dried air. The sensor performance was reproducible over multiple measurements, showing the highly reproducibility with a coefficient of variation of 1.77 % (n = 5).

An attempt to provide a model of consciousness

A neural model based on the local coding hypothesis was proposed-a group of neurons, called cognitive neurons, and a barrier, called the exclusive connection, and it was shown that consciousness and related mental processes can be explained consistently.

Multivariate Calibration Models to Estimate Non-invasively Blood Glucose Levels Based on A Novel Optical Technique Named Pulse Glucometry

A novel optical technique for the non-invasive in vivo blood glucose concentration (BGL) measurement, named “Pulse glucometry”, was combined with four types of regression analyses, primary component regression (PCR), partial least squares regression (PLS), artificial neural network (ANN) and support vector machines (SVMs) Regression (SVMsR), as well as with two discriminant analyses using ANN and SVMs. The total transmitted radiation intensity (Iλ) and the cardiac-related pulsatile changes superimposed on Iλ in human adult fingertips were measured over the wavelength range from 900 to 1700 nm using a very fast spectrophotometer with sampling speed of 100 spectra/s, obtaining a differential optical density (ΔODλ) related to the blood component in the finger tissues. Subsequently, a calibration model using paired data of a family of ΔODλs and the corresponding measured BGL values was constructed with the four regression methods. Two discriminant analyses also attempted to distinguish BGL levels according to such two criterions of diabetic screening as (1) a case of fasting blood sugar and (2) a case of 2-hour postprandial blood sugar. Using 183 paired data sets, in which measured BGL values ranged from 89.0-219 mg/dl (4.94-12.2 mmol/l), we found that the calibration models based on ANN and SVMsR could provide a good BGL estimation and that all of the measured and estimated BGL data points fell within the clinically acceptable regions analyzed by the Clarke error grid. The two discriminant analyses also showed over 80% accuracy for the case (1) and over 65% accuracy for the case (2). These results provide a preliminary evidence that Pulse glucometry with ANN and SVMsR can be applied effectively to estimate BGL levels non-invasively and a discriminant type calibration using ANN and SVMs appears useful and promising for the screening of BGL levels.