//Logo Image
AuthorsChe-Chang Yang (2011-01-17); recommended: Yeh-Liang Hsu (2011-02-21).
Note: This article is the Chapter 7 of Che-Chang Yang’s doctoral thesis “Development of a Home Telehealth System for Telemonitoring Physical Activity and Mobility of the Elderly”.

Chapter 7. Monitoring and assessment of activities of daily living in the home environment

This chapter presents the monitoring and assessment of activities of daily living (ADL) in the home environment. The research background is briefly introduced first. The design of the monitoring devices, including the human movement detector and the appliance usage detector is then described. A selection of simple ADL estimates (the ADL features) is defined and demonstrated for quantifying and assessing ADL performances using 5-month ADL data of an elder person living alone at home. The characteristics of the ADL features in analyzing ADL data are compared, and the technical aspects regarding the monitoring system are also discussed.

7.1 Introduction to home ADL monitoring

Activities of daily living refer to several daily tasks which are required for personal self-care and independent living, such as eating, dressing or bathing [Katz et al., 1963]. Lawton et al. also defined the instrumental ADLs (I-ADLs) as the activities that require interaction with objects or instruments for self-care or communications with people, such as the use of telephone or basic home appliances [Lawton et al., 1969]. The ability of performing ADL directly reflects individual’s living independency. Ageing process is an expected cause of limitations in performance of daily activities that changes from advanced or moderate ADLs to a lower, basic ADLs (B-ADLs) level [Pennathur et al., 2003].

ADLs have been widely used in clinical and research fields to evaluate the level of disability, or functional status of elder people. Katz et al. [Katz et al., 1970, Katz and Akpom, 1976] defined a scale for ADL, and it has been used by geriatrics to evaluate the dependence level of the elderly [Katz, 1983]. Six major activities were selected: bathing, dressing, toileting, transferring, continence, and feeding. An I-ADL scale was also developed by Lawton et al. [1969], and it includes the evaluation in the ability of telephone use, food preparation, housekeeping, laundry, the ability for shopping, responsibility for own medication, and the ability to handle finance. Currently the Barthel Index (BI) has been clinically adopted to assess functional ability. The Barthel Index investigates performance level of 10 ADL items to evaluate the living independency of elderly or disabled people.

Traditional ADL assessments for the elderly usually rely on subjective judgments by clinical or specialized personnel. Long-term ADL profiles of the elderly acquired at home environment can provide additional comprehensive information related to the living behaviors, and thus their functional ability can be better determined. Technologies have the potentials to assist ADL measurements in an unobtrusive way without disturbing the daily life of the elderly. In 1993, the research group in the University of New South Wales (UNSW) initiated the development of a cost-effective remote monitoring approach to identifying changes of health status from simple activity measures collected by a number of sensors [Celler et al., 1994, 1995]. In their proposed systems, a telemedicine platform was used to enable effective, automatic and continuous data transmission and collection, as well as to prompt timely response and care intervention.

Advances in sensors and telecommunication technologies have prompted the concept and development of “smart homes” that are achieved by implementing diverse sensors and intelligent algorithms to monitor the human-environment interactions. For healthcare purposes, such smart homes or home monitoring systems can collect health-related data to build a profile of health and functional status of people [Noury et al., 2003]. In such smart homes sophisticated instruments and sensors may be used. However, home ADL monitoring systems usually consist of a range of simple and low-cost sensors distributed in a home environment [Ogawa et al., 2002]. The passive infrared (PIR) sensor is the most common sensor for detecting human occupancy or for determining active movements within a sensitive range of specific space. Mechanical, magnetic or photoelectrical switches are also the sensors that can be used to detect transfers between rooms [Suzuki et al., 2006]. Energy expenditure estimation using PIRs was also studied though the preliminary outcome is not acceptable [Kaushik et al., 2006].

Sensors can be also used to monitor activities related to I-ADLs. The ADLife developed by Tunstall [http://www.tunstall.co.uk] is a tele-care solution for the elderly people. The system can detect door/electrical appliance usages (e.g., oven, refrigerators) to provide detailed information on the elder persons’ I-ADLs within their homes. A study of monitoring of home appliances usage of elder people living alone showed that daily and nocturnal activities can be differentiated and several daily activities can be calculated [Franco et al., 2008]. The approach to detecting ADLs from power line impulses and classifying the usage of electrical appliances were also presented [Berenguer et al., 2008].

This chapter presents the research on monitoring and assessment of activities of daily living (ADL) in the home environment. Though various types of sensors have been used to monitor home ADLs of specific interests, the use of single PIR was demonstrated to identify daily living patterns [Suzuki et al., 2006]. Mobility changes can be observed and identified from a multi-sensor monitoring system utilizing only PIRs at home [Chan et al., 2005]. In addition, home activities related to the usage of home appliances can be monitored electrically from power line. Therefore, in this study only the PIRs and the AC current transformers (CTs) are selected as the ADL sensors because the two types of sensors are deemed adequate for monitoring home activities related to daily living rhythm of elderly people. This approach not only reduces diverse hardware modalities but also simplifies the complexity in instrumentation and data analysis.

The system diagram of the proposed monitoring system in this research described in Chapter 1 is repeated in Figure 7.1. This system is based on the decentralized home-telehealth system (DHTS) developed by Gerontechnology Research Center in Yuan Ze University [Hus et al., 2007]. As described above, there are two kinds of home ADL detectors in this system, the human movement detectors using PIRs and the appliance usage detectors using CTs. The human movement detector is a device for detecting active human movements within a sensitive range. The appliance usage detectors detect the home appliance usages (on/off status) by sensing its delivered AC power. All the detected ADL signals are transmitted to a DDS via the ZigBee RF wireless protocol.

Figure 7.1 The system diagram of the proposed monitoring system

The design of the monitoring instruments, including the human movement detector and the appliance usage detector is then described in Section 7.2 to 7.4. In addition to demonstrating the DHTS capabilities of facilitating home ADL monitoring of the elderly, it is also expected that this system can provide caregivers the access to the profiles of daily living of the elderly living alone at home. In Section 7.5, a selection of simple ADL estimates (the ADL features) is defined for quantifying and assessing ADL performances, and demonstrated using the 5-months ADL data of an elder person living alone at home. The characteristics of the ADL features in analyzing ADL data are compared, and the technical aspects regarding the monitoring system are also discussed in Section 7.6 of this chapter.

7.2 Design of the human movement detector

The human movement detectors sense the changes of human infrared intensity which represent occurrences of active movements within a sensible range. In other words, an inactively immobilized subject at one location will not trigger the human movement detectors. The human movement detector also measures room temperature and humidity. Once activities are detected by the human movement detector, it transmits the signals (activity count and temperature/humidity) to the DDS via a ZigBee 2.4GHz wireless protocol. If there is no activity, the detector reports the temperature and humidity every 1 minute.

7.2.1 Schematic design

Figure 7.2 shows the system block diagram of the human movement detector. The detailed schematic design of this system is shown in Figure 7.3 to Figure 7.5. All the components inside this detector can be powered by DC3.3V, except the passive infrared (PIR) sensor module (KC7783R) requiring at least DC4V. Therefore this detector is DC5V-powered from an AC-DC power adaptor. A voltage regulator (LM3940, National Semiconductor) is used to convert DC5V input power to regulated DC3.3V power. The DC5V input power is also bypassed to the PIR sensor module.

The PIR sensor module is used to detect the existence of human movement. The module uses a master PIR control chip (KC778B, COMedia Ltd.) for signal amplification and logic control. The PIR sensor is covered by a Fresnel lens that the incident IR can be uniformly distributed onto the sensor. When human movement exists within the sensitive range of the PIR sensor module, the module asserts high (as its supply power) in its module output; otherwise the output is zero (GND).

The processing core of this detector is the PIC microcontroller (PIC18LF6722, Microchip Co.). It is in a 64-pin plastic thin quad flatpack (TQFP) package and provides basic and advanced functions in signal input/output, A/D conversion, data communication and processing. The output of the PIR sensor module is connected to a digital input channel (PortB, RB5) of the PIC microcontroller. Note that an optional channel of serial SCK (Port RB1) and DATA (Port RB2) is reserved for connecting a digital humidity and temperature sensor (SHT7x series, Sensirion AG) when the environmental parameters are required.

A watchdog (WD) timer (MAX6369, MAXXIM) can also be optionally activated if external timing is required. In this design, the WD timer is a sleep mode switch of the PIC microcontroller. The pin RA4 of the PIC microcontroller normally outputs high (as VDD) and low (zero as GND) periodically per time interval while functioning. The watchdog input (WDI) of the WD Timer reads the status (high or low) of the pin RA4 to determine whether the PIC microcontroller is in run mode or in sleep mode. The WD timer starts timing when the PIC microcontroller is in sleep mode of which the pin RA4 stops changing its alternating high/low status. If the watchdog timeout (WDt) expires, the WD timer asserts a low output at its watchdog output (WDO) which is connected to the pin RB0 of the PIC microcontroller. The low status at pin RB0 activates the PIC microcontroller to return to run mode. The WDt can range from 1s to 180s. The setting can be configured with a 3-bit logic selector.

The PIC microcontroller offers two serial RS-232 channels, in which COM-A is for communicating with PCs or any other compatible devices, and COM-B is internally connected to the ZigBee RF Module (XBee Series 2 OEM RF module, Digi International) for wireless data communication and wireless sensor network (WSN). The ZigBee RF module is compatible with IEEE 802.15.4 “ZigBee” protocol operating in 2.4GHz with data rate up to 250kbps. This ZigBee RF module features the capabilities of point-to-point, and multi-mesh wireless networking, with the maximum data transmission range of 40m (indoor) and 120m (outdoor) between each node. The ZigBee RF module is in sleep mode, it can be “waken” by a low status (GND) at the pin RB3 of the PIC microcontroller. The ZigBee RF module is in sleep mode when the pin RB3 is high.

Figure 7.2 System block diagram of the human movement detector

Figure 7.3 Schematic of the human movement detector (1)

Figure 7.4 Schematic of the human movement detector (2)

Figure 7.5 Schematic of the human movement detector (3)

7.2.2 PCB design, space layout and finished product

Figure 7.6 shows the PCB design and overall space layout of the human movement detector which uses a universal case (RH3135) of the dimension 120×70×25(mm). Note that if this device serves as a temperature/humidity monitor without using the PIR sensor module, a battery pack of two AA or three AAA batteries can be stored inside this box, occupying the position of the PIR sensor module. The space for the SMA 2.4GHz antenna of the ZigBee RF module is also reserved. Figure 7.7 shows the finished products of the PCB and the device assembly.

Figure 7.6 The PCB design and space layout of the human movement detector

DSC04244-1DSC04249-1

Figure 7.7 The PCB and the product

7.3 Design of the appliance usage detector

The appliance usage detector is an intermediate device between a main AC outlet and the appliance to be monitored. It can detect the use of single or multiple main-powered home appliances. The appliance usage detector senses the AC current consumed by the appliances. A specific threshold of AC current is given to determine the on/off status of the appliances, indicating whether the appliances are in use or not. This detector can periodically transmit the count signals to the DDS when the appliances being monitored are in use.

7.3.1 Schematic design

Figure 7.8 shows the system block diagram of the appliance usage detector, and Figure 7.9 to Figure 7.11 show the schematics of the detector. An AC-DC switching power supply unit (A5-110915B) is used to provide the circuit with DC5V from any AC power input from 100-240V/50-60Hz. This AC-DC power supply has a PCB-mount package so that the module is small and can be integrated onto a PCB. A voltage regulator (LM3940) outputs a regulated DC3.3V power from DC5V power. A replaceable fuse is used to protect the whole circuit against power overload, electrical shock, short circuit or any electrical failure that might damage the detector.

The schematic design is similar to that of the human infrared detector, except that the sensor is replaced by a CT (current transducer/transformer) that can be soldered on a PCB. The CT (CTL-6-P-4-H, U_RD) is a small transformer to sense the AC current in the power line bypassed from the input power. The CT output voltage is small and therefore an amplifier circuit based on LM358 OPA is used. The amplified CT output is then directed to the pin RA0 of the PIC microcontroller for 10bit A/D conversion. Note that the load resistor R4 coupled to the CT output (Figure 7.9) determines the sensitivity of the CT raw output. The larger resistance that resistor is used, the more sensitive the CT outputs. But a more linear response at the CT output can be obtained when the resistance of the resistor is kept low. Typically a 10Ω resistor is used. The OPA output level can be adjusted with the resistor R5 and R6.

Figure 7.8 The block diagram of the appliance usage detector

Figure 7.9 The schematic of the appliance usage detector (1)

Figure 7.10 The schematic of the appliance usage detector (2)

Figure 7.11 The schematic of the appliance usage detector (3)

7.3.2 PCB design, space layout and finished product

Figure 7.12 shows the PCB design and space layout of the appliance usage detector. The dimension of the housing is 125×85×55(mm). This detector has an IEC type AC inlet for connecting a power cord to mains power line. There is also an IEC type AC socket on the top side of the detector for connecting the power cord to the appliances to be monitored. Figure 7.13 shows the PCB and the finished product of the electricity detector.

Figure 7.12 The PCB design and space layout of the appliance usage detector

DSC042322-2DSC04251-1

Figure 7.13 The PCB and finished product of the appliance usage detector

7.4 Operation of the ADL detectors

Figure 7.14 illustrates the flowchart of the common program code for both the human movement detector and appliance usage detector. The detector type (human movement detector or appliance usage detector) must be assigned in the program for correct functioning. If the detector type is assigned as the human movement detector, the program will ignore the functions associated with the appliance usage detector, and vice versa.

For a human movement detector, the PIC microcontroller reads the output of the PIR sensor module twice (P2) to determine whether active human movement exists. The first reading is registered in S1, followed by a 1-second delay. Then the second reading is registered in S2. Decision D3 determines whether human movement exists according to the lookup table in Table 7.1. Note that a “0” stands for no movement (off) and “1” stands for a movement presence (on) for the status S1 and S2.

If both S1 and S2 are zeros, the output is 0x30 that indicates no movement. In the other cases the outputs are 0x31 showing movement exists. If human movement exists (the output is 0x31), the PIC microcontroller will then retrieve the temperature and humidity readings from the SHT75 sensor in Process P3. The character 0x31 with the temperature and humidity readings will be transmitted wirelessly via the ZigBee RF module to the DDS according to the data protocol described in Table 7.2, and the system returns to the beginning to repeat the procedure and function. In the absence of any human movement, the human movement detector does not transmit data but will report the temperature and humidity readings every one minute interval (as shown in Decision D3, D4 and Process P3). The readings with the movement count 0x30 are also transmitted via the ZigBee RF module to the DDS. The sampling interval of the human movement detector is about 6 seconds. Therefore, if a human movement detector is continuously triggered (on), there will be 100 ADL counts for a 10-minute epoch for data recording.

Table 7.1 The lookup table for PIR status to determine movement

PIR status

S1

S2

Output

0

0

0x30

1

0

0x31

0

1

1

1

For the appliance usage detector, the PIC microcontroller reads the CT output (P1) and then determines the on/off status of the appliances. A voltage threshold is required here to distinguish whether the appliance is switched in on/off /standby mode. This threshold should be manually selected in the program according to the operational and electrical characteristics of the appliances. If the CT output exceeds the threshold, the appliance is deemed switched on (in use). The count 0x31 will be transmitted to the DDS via the ZigBee RF module and then the procedure repeats continuously in every 6-second interval, which is identical to that in the human movement detector. In other words, if an appliance is continuously in use, the DDS will receive 100 ADL counts per 10-minute epoch.

Figure 7.14 The common program flowchart of the home ADL detectors

In addition, both detectors use common format and protocol in wireless data transmission in Process P4. Table 7.2 shows the data protocol with an example of data sequence. A complete data sequence contains 11 ASCII characters. The initial (SN.1) and end characters (SN.11) are always “0x71” and “0xFF”, respectively. Both characters are used as flags to indicate a complete and valid data sequence. When DDS receives and screens a data stream, the characters between the initial and end addresses are identified. Except the flag characters, the first character is a register for sensor ID, which may range from “0x01” to “0xFF”. Each ADL detector has unique sensor ID in a wireless sensor network. The following 6 characters (SN.3 to SN.8) are temperature and humidity readings. For example, in temperature reading characters “2 5 8” mean 25.8°C, and “5 5 7” in the humidity reading characters mean 55.7%RH. The character SN.9 and SN.10 are for ADL register. SN.9 is assigned for human movements and SN.10 for appliance use. In SN.9, 0x30 accounts for “none movement (off)”, and 0x31 accounts for “movement exists (on)”. In SN.10, 0x30 indicates “off”, and 0x31 indicates “on” of the appliance.

Table 7.2 The data format and protocol in wireless data transmission

SN

Address

Descriptions

1

0x71

Initial character

2

0x01

Sensor ID, from 0x01, 0x02, …

3

2

Temperature value [1]

4

5

Temperature value [2]

5

8

Temperature value [3]

6

5

Humidity value [1]

7

5

Humidity value [2]

8

7

Humidity value [3]

9

0x31

Movement count. ON: 0x31; OFF: 0x30

10

0x30

Appliance use count. ON: 0x31; OFF: 0x30

11

0xFF

End character

7.5 Quantifying home ADLs with simple estimates

The following sections demonstrate the use of simple estimates to assess and quantify ADL data. The sample ADL data was collected by installing the monitoring system described in the previous sections in a home environment where an elder person lives alone (female, 74yr). The monitoring period started from the end of April 2010 and ended at the end of December 2010. Note that the duration of effective ADL data collection is about 5 months because of some interrupts during the monitoring period.

7.5.1 Instruments setup

In this research, 5 ADL detectors were installed to collect ADL data, including 4 human movement detectors to monitor human motions, and 1 appliance usage detector to monitor TV usage. The 4 human movement detectors were installed in the kitchen, bedroom, and bathroom and by the doorway. Figure 7.15 shows the attachment of the human movement detectors. The human movement data collected at these 4 locations may reflect most of the subject’s ADLs at home. The appliance usage detector interconnects the TV and the AC wall outlet to sense the AC current in a power line. Thus the on/off status of the home appliance usage can be detected. TV usage is monitored because watching TV is the major activity of the elder subject at home. All the ADL detectors were mains-powered and the collected data was transmitted to a DDS via the 2.4GHz ZigBee wireless protocol. The elder subject was not aware of the operation of the detectors, and thus the system setup did not interfere with the subject’s daily living routines.

Figure 7.15 The setup of the ADL detectors in the subject’s residence

7.5.2 Activity profile monitored on a single day

Figure 7.16 shows the activity data collected in the kitchen on two different dates (May 27 and June 14 in 2010). The data logging interval (epoch) is 10 minutes. Therefore there are 144 epochs from 0:00 to 23:59 per day. The activity counts (shown in the vertical axis) at each epoch (shown in the horizontal axis) are shown in a chronological order. In Figure 7.16, the differences between the activities data of two different dates can be easily observed. For example, the profile from May 27 has a peak activity count of 47 at 18:00 and intensive activities centered in the evening, while the profile from June 14 has a peak activity count of 20, measured in the morning.

Figure 7.16 The preview of activity data collected in the kitchen at two different dates

Figure 7.17 and Figure 7.18 show two different data of TV usage collected on May 14 and July 12 in 2010, respectively. The difference between the two TV usage data can also be easily observed. For example, the TV usage on May 14 mainly centers on two periods of 7:50 to 13:50, and 16:10 to 21:00, and that is apparently different from the profile on July 12 whose TV usages were measured in four time periods, and one period from 2:10 o 2:50 in the early morning may be of particular attention.

Figure 7.17 The TV usage data collected on May 14

Figure 7.18 The TV usage data collected on July 12

7.5.3 Activity profile monitored on a long-term basis

General activity profiles can also be observed on a long-term basis. Figure 7.19 shows the average profile of daily activities in the kitchen for the 5-months period. This profile generally shows the daily living rhythm in terms of the measured activity counts. In the data very few activities were detected during midnight to 6:00 in the morning, which corresponds to the period of sleep of the subject. This figure also shows that the subject had more activities during around 6:00 to 9:00 in the morning and the activities decreases gradually in the afternoon. In the evening the activities increase slightly, and then decrease in the night time. The elderly subject usually has lunch between 10:00 to 11:00am, and dinner between 15:00 to 16:00pm, which is a lot earlier than most people. The observable trend of ADL profile obtained by the monitoring system is consistent with the rhythm of daily living of the subject living alone at home.

 

Figure 7.19 The general profile of daily activities in the kitchen

Figure 7.20 shows the general profile of daily TV usage of the subject at home from the 5-month monitoring period. The intensive TV usage is around 17:00 to 20:00 in the evening of which the usage rate is above 90%. Low usage rate during 0:00 to 3:00 is also observed.

 

Figure 7.20 The average profile of daily TV usage

The differences in daily living rhythms between weekdays and weekends/holidays can also be observed from the ADL profiles. Figure 7.21 shows the average profiles of activities in the kitchen on weekdays and holidays. Figure 7.22 shows the average profile of TV usage on weekdays and on holidays. Similar trend of activity profiles in the kitchen on weekdays and holidays in Figure 7.21 is observed, while Figure 7.22 shows distinct variance between the TV usage profiles on weekdays and holidays. This is mainly because the subject turns on the TV to receive stock market information (from 9:00 to 13:30) everyday during weekdays. The subject turns on the TV late and less often in the morning during holidays. Watching TV news and other entertainment programs is the major activity for the subject in the evening on either weekdays or weekends.

Figure 7.21 The average profiles of activities in the kitchen on weekdays and holidays

Figure 7.22 The average profiles of TV usage on weekdays and holidays

7.5.4 Abnormal activities identified from the ADL data

In addition to ADL profiles, abnormal activities can also be identified from the ADL data. For example, Figure 7.23 shows abnormal TV usage from 2:00 to 2:40 detected on July 12. It was later confirmed that the subject was watching TV in the early morning of that day because she could not fall asleep. On April 29, the subject fell accidentally in her bedroom in the early morning and could not get up, and the system also detected the abnormal activities at the bedside from 3:10 to 4:00am on that day as shown in Figure 7.23.

Figure 7.23 The one-day activity data with abnormal activities in the bedroom

7.6 Quantitative ADL data analysis

In the previous section, distinct activity profiles collected from different dates were previewed. Intuitively, obviously distinct ADL profiles which represent different activity rhythms can be differentiated by observation. However, it is also important to examine the ADL data in terms of quantitative measures so as to analyze ADL data objectively and quantitatively.

7.6.1 The selection of simple estimates for ADL data analysis

A selection of simple estimates (the ADL features) is presented to quantify ADLs in terms of the activity intensity (or frequency) and the regularity (or pattern) of activity occurrence. In Figure 7.24, the activity data in the kitchen collected on May 27 (referring to Figure 7.16) is used as an example to demonstrate the simple estimates for ADL data discussed below. The ADL data in Figure 7.24 can be re-arranged by ranking the epochs in the order of activity counts, as shown in Figure 7.25.

 

Figure 7.24 An example of one-day activity data arranged in a chronological order

Figure 7.25 Total activity data re-arranged in a frequency order

(1)  Feature 1: active time ratio

People may perform their daily activities at varied time periods in a day. The first focus of interest is to see how frequently the individual performs activities over a daily scale. As shown in Figure 7.26,  is the activity count at the ith epoch in the rank profile. The epochs which have activity counts () are called “active epochs”, and the other epochs without activity counts () are called “inactive epochs”. The total 144 epochs can be divided into the “active period (Tact)” and “inactive period (Tina)”. The active period corresponds to the amount of active epochs, and the inactive period the amount of the remaining inactive epochs.

The “active time ratio” is a measure that indicates how frequently the subject performs activities over the entire day. The active time ratio  is defined as the ratio of the active epochs (Tact) to overall epochs (Tact+Tina), as expressed in Equation (7-1). For example, Tact in Figure 7.26 is 47 epochs, and the total number of epochs is 144. Therefore  is 32.6%. A high active time ratio indicates a more frequent activity profile, regardless of the activity intensity in the active epochs.

                                                                              (7-1)

 

Figure 7.26 The active period and the inactive period in a ranked ADL profile

(2)  Feature 2: activity rate (or usage rate for appliance)

In addition to the frequency of ADL during the day defined by the active time ratio, it is also important to have the estimates regarding the intensity of activities. The “activity rate” and “daily activity rate” are defined for the purpose. The activity rate measures the intensity of activity over the active epochs, while the daily activity rate measures the intensity of activity over the entire day.

The activity rate (, or the usage rate for appliance uses) is calculated according to Equation (7-2), which is the total activity counts divided by the maximum amount of activity count during the active epochs . Note that in Equation (7-2), if there is no activity in a day, both the sum of  and  are zero. Therefore the activity rate is zero.

                                                                 (7-2)

For the same example shown in Figure 7.24, the total activity count is 528 from 47 active epochs. As a result, the activity rate is 11.2%, which means that the subject has activities in 11.2% of the time of an active epoch in average. A higher activity rate indicates that the intensity of the activity is higher during the active epochs.

(3)  Feature 3: daily activity rate (or daily usage rate for appliance usage)

Similar to Feature 2, the daily activity rate (or daily usage rate for appliance uses)  generally shows the intensity of the activities the subject performs in a whole day.

                                                                   (7-3)

For the sample data in Figure 7.24, the average activity rate is 3.67%, which is an indicator of the intensity of activity of the day.

(4)  Feature 4: coefficient of variation of daily activities (or coefficient of variance of daily appliance usage)

People may perform activities of varied intensity at different time periods of the day, which results in varied activity counts. In addition to the frequency and intensity of activities, the variance of the activity counts distributing over a whole day () defined in Equation (7-4) can provide an estimate of how uniformly the subject performs activities in a day.

                                                    (7-4)

For the sample data in Figure 7.24, the coefficient of variance is 2.28. A lower coefficient of variance indicates less deviation of activity intensity.

(5)  Feature 5: correlation coefficient of activity profile (or appliance usage profile for appliance use)

It has been reported that elderly people tend to have stable lifestyle [Franco et al., 2008]. Hence, the similarity of the ADL data of a day and the long-term ADL profile can be an important estimate of whether the subject follows his/her own regular ADL rhythm (trend). The correlation coefficient r is a common statistical measure of the interdependence of two or more variables. Therefore, the correlation coefficient is used to compare a daily profile to the long-term average profile. The correlation coefficient ranges from -1 to 1, and a higher value in the correlation coefficient of two profiles means a more similar or correlated trend between the two profiles.

This feature investigates the activity data in terms of its regularity relative to a long-term average pattern in the chronological order, regardless of their base intensities. For example, the correlation coefficient of the one-day activity profile in Figure 7.24 and the 5-monhs average profile as shown in Figure 7.19 is 0.17. The correlation coefficient of TV usage profile between May 14 (Figure 7.17) and the 5-month average (Figure 7.20) is 0.82. A higher correlation coefficient here means a more regular activity rhythm with respect to the long-term average rhythm.

7.6.2 Investigation of ADL features for analyzing different ADLs

This section describes the used of the ADL features introduced in the previous section to quantify the ADL data. Referring to Figure 7.16 which shows two activity data collected on different dates, Table 7.3 compares their ADL features. The subject had more intensive and frequent activities on May 27 than that on June 14 because the active time ratio, average activity rate and activity rate on May 27 are higher. However, the ADL rhythm on June 14 was more regular than that on May 27 as the correlation coefficient of activity profile on June 14 is higher,

Table 7.3 Comparison of two activity data from Figure 7.16

ADL Feature

May 27

Jun14

Daily activity rate

3.70%

1.3%

Active time ratio

0.33

0.22

Activity rate

11.2%

6.0%

Coefficient of variance of daily activity

2.28

2.56

Correlation coefficient of activity profile

0.17

0.47

Table 7.4 compares the ADL features computed from the average activity profiles in the kitchen, bedroom, doorway and the bathroom shown as Figure 7.28. The activities in the kitchen have greater values in the daily activity rate, active time ratio and activity rate. These ADL features indicate a relatively greater activity frequency and intensity in the kitchen among the other 3 locations at home. The smallest coefficient of variance of daily activities and highest correlation coefficient of activity profile are also found in the kitchen activities, indicating a more regular ADL rhythm among the others. On the contrary, the ADLs by the doorway show the least correlation coefficient of activity profile and highest coefficient of variance of daily activities among the other three locations. These features show that the activities by the doorway are least regular. As the subject performed relatively intensive ADLs in the kitchen, this may imply that the subject still preserves a good performance level in the I-ADLs [Lawton et al, 1969]. 

Table 7.4 The comparison of the ADL features computed from the average ADL profile in the four locations at home

ADL features

Kitchen

bedroom

doorway

bathroom

Daily activity rate

2.0%

0.1%

0.2%

0.9%

Active time ratio

0.19

0.06

0.08

0.10

Activity rate

10.9%

1.7%

1.9%

8.8%

Coefficient of variance of daily activities

2.90

4.78

4.79

4.73

Correlation coefficient of activity profile

0.35

0.20

0.19

0.25

Figure 7.28 The average activity profiles in (a) kitchen, (b) bedroom, (c) doorway, and (d) bathroom of the elderly subject

The ADL features can also quantify the ADLs regarding appliance usages, such as the TV usage in this study. Referring to Figure 7.22, Table 7.5 compares the ADL features of the daily TV usage on weekdays and holidays. Compared with the weekday profile, the holiday profile is lower in the daily usage rate, the active time ratio and the usage rate. This shows the subject uses TV more on weekdays. The greater coefficient of variance of daily appliance usage and the lower correlation coefficient of appliance usage profile from the holiday profile also indicate a less regular TV usage than weekdays. By visiting the subject, it has been confirmed that the subject usually watches TV news and programs to help handle finances (stock market and transactions) in the morning during weekdays. This is considered as a major cause of the different performance in the ADL features and the significant changes observed from the two profiles. The intensive TV usage for handling finances also implies a good I-ADL performance of the elder subject [Lawton et al, 1969].

Table 7.5 Comparison of the average profiles of daily TV usage on weekdays and holidays

ADL features

Weekdays

Holidays

Daily usage rate

55.9%

42.3%

Active time ratio

0.57

0.44

Usage rate

98.8%

95.4%

Coefficient of variance of daily appliance usage

0.91

1.21

Correlation coefficient of appliance usage profile

0.81

0.64

According to the TV usage data on July 12 as shown in Figure 7.18, the elder subject watched TV during 2:10 to 3:00 in the early morning. It is an occasional event because the subject rarely watches TV during that time. Therefore, it is of particular interest to investigate whether such occasional TV usage will result in changes of performance in the ADL features. Table 7.6 compares the TV usages on July 12 and May 14. The TV usage on May 14 was similar to the average by comparing its profiles and ADL features with the average profile, For the TV usage on July 12, it has the active time ratio the same as that on May14, and its most ADL features are similar to the TV usage on May 14. However, the correlation coefficient of appliance usage profile is 0.59 of TV usage on July 12 is much lower than that on May 14 or the average levels. In this comparison, the occasional or unusual TV usages result in determinable changes of performance of the ADL features.

Table 7.6 Comparison of the profiles of TV usage on two different dates

ADL features

May 14

July 12

Daily usage rate

46.0%

44.8%

Active time ratio

0.47

0.47

Usage rate

99.0%

96.2%

Coefficient of variance of daily appliance usage

1.09

1.11

Correlation coefficient of appliance usage profile

0.82

0.59

7.7 Discussion

In this chapter an ADL monitoring system based on the Decentralized Home-Telehealth System (DHTS) was presented. In this system, two kinds of ADL detectors were developed. One is the human movement detector utilizing PIRs, and the other the appliance usage detector using CTs. A selection of ADL features were demonstrated to give quantitative estimates regarding the ADL characteristics by analyzing the 5-month ADL data collected from an elder person living alone at home. .

During the monitoring session the system setup did not obstruct the subject’s daily living and the subject was not aware of the operation of the instruments. This system did not cause any discomfort, displeasure, or other major issues to the elder subject. There was no malfunction reported from all the ADL detectors. There was no loss of data owing to the use of ZigBee 2.4GHz WSN. These user feedbacks and results show that this system is reliable and it should be suitable for use in home environments.

Instead of detecting the human occupancy in a specific space, the human movement detector senses the occurrence of apparent human movements within its sensitive range of the selected locations at home, and thus its data can better reflect mobility of the subject under monitoring. The selected ADL features in this chapter can show different ADL characteristics. The daily activity rate, activity rate and active time ratio show the ADL characteristics in terms of the frequency and intensity of activity occurrences. The coefficient of variance of daily activities (or appliance usage) and correlation coefficient of activity profile (or appliance usage profile) were used to estimate the regularity and pattern of ADLs. These ADL features can quantify and discriminate different ADL characteristics through the demonstration of ADL data analysis. With the use of the ADL features, the ADL data of the elder subject shows more activities doing daily routines in the kitchen than the other 3 locations (bedroom, door, and bathroom). The relatively intensive ADLs in the kitchen may imply that the elder subject still maintains adequate functional status in performing I-ADLs. The ADLs by the doorway show the lowest correlation coefficient and the highest coefficient of variance, which indicates a less regular ADL.

The appliance usage detector was used to monitor the TV usage because watching TV is the major activity of the elder subject at home. Very high daily usage rate of 55.9% and correlation coefficient of 0.81 were found in the TV usage profile monitored in the 5-month period. The appliance usage detector can be connected to any home appliances which meet the power requirement.

Unusual activities in the nighttime have been detected by the system, too. The ability of detecting unusual activities and automatic and timely emergency response is important and thus should be implemented onto ADL monitoring systems to enhance the usability of traditional ADL monitoring system.

Appendix: Bill of material lists of the home ADL sensors

Table A7.1 Human movement detector

Designator

Part/item

General footprint

Note

U1

LM3940

TO-220

 

U2

PIC18LF6722

 

64-pin TQFP

U3

MAX6369

SSOP-8

 

U4

XBee Series 2 OEM RF module

 

2.0-pitch pin

J1

DC connector

 

 

J2

Connector

CON3

Single-inlet 2.54-pitch (F)

J3

CON6

J4

CON3

J5

 

Single-inlet 1.27-pitch (F)

BT1

CR2032 socket

 

 

D1

LED

0805

 

D2

 

Y1

XTAL

HC-49

10MHz

R1, R2

Resistor

0805

 

R3, R5

2.2kΩ

R4

100Ω

R6, R10

10kΩ

R7

2kΩ

R9

20kΩ

C2

Capacitor

0805

33uF

C3, C7, C8

0.22uF

C4

0.01uF

C5, C6

22pF

S1, S4-S6

Switch

CON2

2.54-pitch pin

S2, S3

 

 

Table A7.2 Appliance usage detector

Designator

Part/item

General footprint

Note

U1

LM2940

TO-220

 

U2

LM3940

TO-220

 

U3

PIC18LF6722

 

64-pin TQFP

U4

LM358

DIP8

 

U5

XBee Series 2 OEM RF module

 

2.0-pitch pin

J1

AC inlet

 

IEC type

J2

Switching power supply

 

 

J4

Connector

CON6

Single-inline pin socket (F)

J5

CON3

RS-232

J6

AC outlet

 

IEC type

F1

Fuse

 

6×30mm, 250V 10A(T)

Y1

XTAL

HC-49

10MHz

S1

Switch

CON2

 

S2

 

 

S3

 

D1

LED

0805

SMD

D2

SMD

CT1

CT

 

CTL-6-P-4-H

R1, R3

Resistor

0805

2.2kΩ

R2

0805

100Ω

R4

AXIAL0.3

10Ω

R5

10kΩ

R6

270kΩ

R7

0805

2kΩ

C1

Capacitor

0805

0.47uF

C2, C3

33uF

C4, C5

22pF

C6

0.1uF, or 0.22uF

C7

0.01uF

C8, C9

0.22uF

C10

RB.2/.4

470uF

References

Berenguer M., Giordani M., Giraud F., Noury N., “Automatic detection of activities of daily living from detecting and classifying electrical events on the residential power line,” Proceedings of the 10th IEEE International Conference on e-Health Networking, Applications and Service, Singapore, 7-9 July, 2008.

Celler, B. G., Hesketh, T., Earnshaw, W., Ilsar, E., 1994. “An instrumentation system for the remote monitoring of changes in functional status of the elderly at home,” Proceedings of the 16th Annual International Conference of the IEEE EMBS, Vol. 2, pp. 908-909.

Celler B.G., Lovell, B.G., Earnshaw W. A., Ilsar E. D., Betbeder-Matibet L., 1994. “Development of an integrated system for remote monitoring of health status of the elderly at home,” Biomedical Engineering, Application, Basis, Communications, Vol. 6:6, pp. 919-926.

Celler, B.G., Earnshaw, W., Ilsar, E.D., Betbeder-Matibet, L., Harris, M.F., Clark, R., Hesketh, T., Lovell, N. H., 1995. “Remote monitoring of health status if the elderly at home. A multidisciplinary project on aging at the University of New South Wales,” International Journal of Bio-medical Computing, Vol. 40, pp. 147-155.

Chan M., Campo E., Estève D., 2005. “Assessment of activity of elderly people using a home monitoring system,” International Journal of Rehabilitation Research, Vol. 28, pp.69-76.

Franco G.C., Gallay F., Berenguer M., Christine M., Couturier P., 2008. “Non-invasive monitoring of the activities of daily living of elderly people at home- a pilot study of the usage od domestic appliances,” Journal of Telemedicine and Telecare, Vol. 14, pp. 231-235.

Hsu, Y. L., Yang, C. C., Tsai, T. C., Cheng, C. M., Wu, C. H., “Development of a decentralized home telehealth monitoring system”, Telemedicine and e-Health, Vol. 13, No.1, pp. 69-78, February 2007.

Katz, S., Ford, A. B., Moskowitz, R. W., Jackson, B. A., Jaffe, M. W., 1963. “Studies of illness in the aged: The index of ADL, a standardized measure of biological and psychosocial function”, Journal of the American Medical Association, vol. 185, pp. 914-919.

Katz, S., Downs, T. D., Cash, H. R., Grotz, R. C., 1970. “Progress in development of  the index of ADL”, The Gerontologist, pp. 20-30.

Katz S., 1983. “Assessment self-maintenance: Activities of daily living, mobility and instrumental activities of daily living”, Journal of the American Geriatrics Society, vol. 31, no.12, pp. 721-727.

Katz, S., Akpom, C., 1976. “A measure of primary sociobiological functions”, International Journal of Health Service, vol. 6, no. 3, pp. 493-508.

Kaushik A., Celler B.G., 2006. “Use of infrared sensors for extimation of energy expenditure by elderly people living alone at home,” Proceedings of the 28th IEEE EMBS Annual International Conference, NY, USA.

Lawton M. P., Brody, E. M., 1969. “Assessment of older people: Self-maintaining and instrumental activities of daily living”, The Gerontologist, vol. 9, no. 3, pp. 179-186.

Noury N. Virone G., Barralon P., Ye J., Rialle V. and Demongeot J., 2003. “New trends in health smart homes,” Proceedings of 5th International Workshop on Enterprise Networking and Computing in Healthcare Industry. pp. 118–127.

Ogawa M., Togawa T., “Monitoring daily activities and behaviors at home by using brief sensors,” 1st Annual International IEEE-EMBS Special Topic Conference on Microtechnologies in Medicine & Biology, Oct. 12-14, Lyon, France, 2002.

Pennauthor A., Magham R., Contreras L. R., Dowling W., 2003. “Daily living activities in older adults: Part I- a review of physical activity and dietary intake assessment methods”, International Journal of Industrial Ergonomics, Vol. 32, pp. 389-404.

Suzuki R., Ogawa M., Otake S., Izutsu T., Tobimatsu Y., Iwaya T., Izumi S-I., 2006. “Rhythm of daily living and detection of atypical days for elderly people living alone as determined with a monitoring system,” Journal of Telemedicine and Telecare, Vol. 12, pp. 208-214.

Suzuki R., Otake S., Izutsu T., Yoshida M., Iwaya T.,   2006. “Monitoring daily living activities of elderly people in a nursing home using an infrared motion detection system,” Telemedicine and e-health, Vol. 12, No. 2, pp. 146-155.