//Logo Image
AuthorsChe-Chang Yang (2011-01-16); recommended: Yeh-Liang Hsu (2011-02-21).
Note: This article is the Chapter 4 of Che-Chang Yang’s doctoral thesis “Development of a Home Telehealth System for Telemonitoring Physical Activity and Mobility of the Elderly”.

Chapter 4. Algorithm design for real-time physical activity identification

This chapter presents the algorithm design for real-time physical activity identification. Combined with the wearable motion detector (WMD), several still postures and basic dynamic movements can be identified in real-time. Fall detection and manual emergency response is also incorporated in this algorithm. The algorithm is processed in real-time in the WMD and the identified activity events can be instantly transmitted to a distributed data server (DDS) via a ZigBee wireless protocol. The preliminary performance evaluation of the algorithm is conducted, and a demonstration of continuous remote monitoring using the WMD is also presented in this chapter.

4.1 Introduction

Physical activity can be regarded as any bodily movement or posture produced by skeletal muscles and resulting in energy expenditure [Capspersen et al., 1985]. Aging and various aspects of behavioral variables directly affect one’s physical activity level [Pennathur et al., 2003]. For elderly people, changes in physical activity are highly related to the transition from relatively independent living to ill and declined functional health status. This transition process is often subtle and cannot be easily perceived by care-givers or even the elderly themselves [Celler et al., 1995]. Therefore, it is important to provide objective and accurate assessment approaches to one’s long-term physical activity level in a free-living environment (e.g., home) for evaluation of quality of life and overall functional health status [Noury et al., 2002; Lyons et al., 2005].

The difficulties in assessing physical activity are primarily due to the subtle and complex nature of body movement, which requires accurate and reliable measuring techniques. A majority of assessment methods in the past relied on questionnaires or observation by regular visits, which may lead to inconsistent assessment results and inability to be applied in large-scale studies [Meijer et al., 1991]. Advances in sensors, microprocessors and wireless communication technologies have been the driving factors to facilitate continuous tele-monitoring of physical activity. Accelerometers have been regarded as appropriate motion sensors for the use of human movement measurement. These accelerometry-based systems can provide significant information on the details of physical activities, and further quantify the subject’s mobility level at acceptable cost [Bouten et al., 1997].

Most traditional motion sensor-based monitoring systems capitalize on off-line data processing techniques to analyze recorded human movements [Najafi et al., 2002, 2003]. In these systems, portable data loggers with sensor units are commonly used to record the continuous data of human movement. The recorded data is then uploaded to a personal computer (PC) and is analyzed by complex signal processing techniques, such as fast Fourier transform (FFT), wavelet transform or other frequency-domain processing methods. However, these off-line systems are unable to provide real-time identification of human movements, let alone tele-monitoring of physical activities.

Real-time activity identification and automatic movement classification have been developed recently. Mathie et al [2003] proposed a generic framework for automated human movement classification using accelerometry data. The movement classification within that framework does not require complex signal processing, making it applicable to embedded systems and portable devices. Karantonis et al [2006] also presented a real-time system for physical activity classification in home environments. All the movements are classified in the real-time processing of a wearable system. However, a PC-based program is still required to identify walking. The advantages of such approaches are the capability of providing not only real-time monitoring data but also the potential of instant and timely response to emergent events.

This chapter presents the algorithm design for real-time physical activity identification. Combined with the waist-mounted wearable motion detector (WMD), 3 still postures (sitting still, standing still, and lying still) and 5 dynamic movements (two sit-stand transitions, two lie-sit transitions and walking) Fall detection and manual emergency response is also incorporated in this algorithm. The algorithm is processed in real-time in the WMD and the identified activity events can be instantly transmitted to a distributed data server (DDS) via a ZigBee wireless protocol. The preliminary performance evaluation of the algorithm is conducted, and a demonstration of continuous remote monitoring using the WMD is also presented in this chapter.

4.2 Acceleration and tilt sensing

When the accelerometer output is sampled via an analog-to-digital converter (ADC), the output signal can be re-converted to analog voltage according to Equation (4-1). V is the analog voltage (V), ADC is the digital values of the ADC readings, and Vdd is the supply power for the ADC. The term  stands for the sensitivity of the ADC (n is the number of bits). The microcontroller in the WMD offers 12 ADC channels with 10-bit resolution and 3.3V operation voltage that correspond to a sensitivity of .

                                                                                  (4-1)

As shown in Equation (4-2), the KXPA4-2050 output is related to its supply voltage (Vdd) and the acceleration (Acc) along the sensitive axis. Note that Vout and Vdd are in the unit of volt (V) and Acce in g (9.81m/s2). 0.66 is its sensitivity in per gravity (V/mg). Figure 4.1 depicts the coordinate definition of acceleration of KXPA4-2050. Because the WMD is powered by regulated DC3.3V, the output is 1.65V in the absence of any acceleration along the sensitive axis. Note that in Figure 4.1 the arrow along each axis indicates the positive accelerated direction. More clearly, the measured acceleration can be positive or negative according to Table 4.1 which shows the output of each axis of KXPA4-2050 in different orientations with respect to the Earth’s ground that has gravity (1g) downward. Referred to Equation (4-2), an output of 2.31V is obtained in the axis that senses 1g gravity.

                                                                           (4-2)

Figure 4.1 The tri-axial coordinate definition of KXPA4-2050 (The arrow of each axis indicates the directions that are positively accelerated)

Table 4.1 KXPA4 outs with respect to the x-y-z orientations

Orientation

Outputs (Vout)

x-axis

1.65

2.31

1.65

0.99

1.65

1.65

y-axis

2.31

1.65

0.99

1.65

1.65

1.65

z-axis

1.65

1.65

1.65

1.65

2.31

0.99

The accelerations along each axis can be computed with Equation (4-2). The tilt angles of the axes with respect to the horizon can be further calculated from the measured acceleration according to Equation (4-3). , , and  are the tilt angles of x-, y-, and z-axis relative to the horizon. Because of the sinusoidal relationship between tilt angle and acceleration, the maximal sensitivity in tilt sensing is approximately 0.017452g per degree of tilt, and the minimal sensitivity is 0.0001523g per degree of tilt. Note that tilt sensing using Equation (4-3) is mostly valid and accurate when the accelerometer is positioned still.

                                                                                 (4-3)

When the accelerometer rotates freely in motion, the accuracy of tilt sensing based on Equation (4-3) decreases. Consequently, the tilt angles can be more accurately calculated according to Equation (4-4). The maximal sensitivity (0.0 17452g per degree of tilt) of one axis can be maintained at any tilt orientation when combing other two axes.

                                                               (4-4)

4.3 Structure of the algorithm

The hardware design of the wearable motion detector is presented in the Chapter 3. The algorithm design for real-time physical activity identification is further described in this chapter. Figure 4.1 shows the flowchart of the algorithm that is implemented in the PIC microcontroller of the WMD. This algorithm can discriminate human postures (lying or upright) and identify several basic physical activities (sit-stand postural transitions, bed movement and walking). Fall detection is also provided in this algorithm. Due to the limited signal processing and memory capacity of the PIC microcontroller, it is crucial to reduce memory (RAM) use of the PIC microcontroller and to reduce time delay or interruption during operation. The sampling rate fs is selected 60Hz, being as low as the level that significant activity characteristics can be reserved. The clock of the PIC microcontroller is set 40MHz to maximize signal processing speed and to reduce the operation delay. The sampled data is in discrete-time sequences and the applied signal processing techniques are processed in time domain.

4.3.1 General process of the algorithm

As described earlier, the algorithm is designed to identify 3 still postures (lying still, sitting still, and standing still), 4 postural transitions (sit-to-stand, stand-to-sit, sit-to-lie and lie-to-sit), and walking. Detection of possible falls is also featured in this algorithm. In the case where there is no definite result determined by the algorithm, the event will be recorded as an uncertain movement or an uncertain posture.

Figure 4.2 the main algorithm flowchart for the wearable motion detector

When the WMD is turned on, the algorithm starts to determine the status of the emergency button. If the button is pressed and held for over 5 seconds (D1=Yes), an emergency signal or a help call is raised, and the signal will immediately be sent. If the button is not triggered (D1=No), a quick primary sampling (P1) process is activated to collect acceleration data for 0.5 second. When the primary sampling process is completed, the next decision D2 determines whether the status is active (movement exists) or not by investigating the signal magnitude area (SMA) of the 0.5-second a data from the process P1 according to Equation (4-5). By empirical trial, a threshold of 1.15g is given to the decision D2. If the computed SMA from the data is below that threshold (D2=No), the status is temporarily deemed still, and the 0.5-second data will be lead to the following sections for identifying still postures. If the sampled data is equal to or above the threshold, the status is temporarily deemed active, and the secondary sampling (P2) is immediately activated to read the accelerometer signals for 2 seconds. The 0.5-second data collected in the primary sampling process and the 2-second data collected in the secondary sampling process is then concatenated together to obtained a data of 2.5 seconds in length. The use of such a dual-stage (P1, P2) data sampling strategy ensures that a dynamic movement can be captured within the sampling interval.

                         (4-5)

The 2.5-second data is regarded as one of the dynamic movements to be identified. The process P3 applies median filtering (n=3) and moving average (window length=3) to the data. The median filtering has the effect of eliminating signal spikes which is not directly related to the movement and the moving-average operation reduces signal fluctuations while preserving the overall signal profiles. Identical to the decision D2, the decision D3 again computes the SMA of the 2.5-second concatenated data with Equation (4-5) to determine whether dynamic movement possibly exists. This decision can determine a still status with minor or subtle movements that could be seen active in the decision D2. After process in P3, the 2.5-second data representing a dynamic movement is fed to the operations among the decision D5 to D10 which identify sit-stand/ lie-sit posture transitions and possible sign of a fall.

4.3.2 Identifying dynamic activities

The symbolic representation of signal sequence which functions as a high-pass filter is commonly used in many operations in the algorithm to register apparent changes or trends in a data sequences. This process maps a data sequence P=[,,,…, ] into a binary sequence C=[,,,…, ] according to Equation (4-6). The entry ci in a binary sequence is assigned “1” if the difference si between two consecutive data points  and  exceeds a specific threshold sthr. Otherwise, ci is assigned “0”. Figure 4.3 shows an example of a trunk tilt data during a sit-to-stand transition and its corresponding symbolic representation (binary sequence). Note that the binary sequence is expressed as a histogram to highlight the entries .

, i=1,2,3,…,149                                                     (4-6)

Figure 4.3 An example of a trunk tilt data and its binary sequence

Before entering the procedure of dynamic movement identification, it is important to discriminate the trunk position at the end of state. Therefore, the decision D4 screens the data segment from 2.0s to 2.5s to determine the trunk orientation according to Figure 4.4. An upright posture here is defined as the trunk tilt within the range of 60°forward and backward with reference to vertical.

Figure 4.4 Illustration of the trunk tilt orientation

(1)  Identifying sit-stand postural transitions

The vertical acceleration is used to identify sit-stand postural transitions because it is more sensitive to external motions through empirical trials and observations. Through observation, the acceleration patterns of sit-stand postural transitions can be characterized by R1: peak order, R2: peak distance and R3: peak values.

Figure 4.5 and 4.6 show examples of the vertical acceleration patterns during sit-to-stand and stand-to-sit postural transitions. In Figure 4.5, it can be observed that an upper acceleration peak PUpper appears first and is followed by a lower acceleration peak PLower for a sit-to-stand transition. A reverse order of PUpper and PLower for a stand-to-sit transition is also observed in Figure 4.6. The peak distance Dpeak is the interval (in seconds) between the upper acceleration peak and the lower acceleration peak. Peak values are the accelerations at the both peaks.

Figure 4.5 An example of vertical acceleration pattern during sit-to-stand transition at a subject’s normal speed

Figure 4.6 An example of vertical acceleration pattern during stand-to-sit transition at a subject’s normal speed

Figure 4.7 and Figure 4.8 show examples of the vertical acceleration patterns during slow, normal and fast sit-stand transitions of the same subject. Faster transitions can be found to produce shorter peak distances, and slow transitions produces longer peak distances for both sit-to-stand and stand-to-sit transitions. In addition, the magnitudes of the upper acceleration peak and the lower acceleration peak in faster transitions are higher. Slow transitions have the peaks of lower magnitudes.

Figure 4.7 An example of vertical acceleration pattern during slow, normal and fast stand-to-sit transitions

Figure 4.8 An example of vertical acceleration pattern during slow, normal and fast stand-to-sit transitions

Coherent properties of peak distance (R2) and peak values (R3) were also observed in an extensive test with 15 recruited ostensibly healthy subjects of various ages (11 males and 4 females, 5 young subjects under 35yr, 5 middle-aged subjects between 35 to 65yr, and 5 elder subjects over 65yr). Table 4.1 shows the average peak distances of sit-stand transitions from all the subjects. These values also reveal the evident tendency mentioned above.

Table 4.1 Average peak distances of sit-stand transitions in different subject groups

 

Young

Middle-aged

Elder

Sit-to-stand

Slow

0.82s

1.06s

 

Normal

0.73s

0.69s

0.70s

Fast

0.56s

0.53s

 

Stand-to-sit

Slow

0.97s

1.16s

 

Normal

0.83s

0.88s

0.98s

Fast

0.56s

0.69s

 

The distribution of peak distances in sit-stand transitions of all samples collected in the test is shown in Figure 4.9. For the sit-to-stand case, most peak distances (95 out of 98) distribute over the range between 0.3s to 1.3s. For the stand-to-sit case, most peak distances (95 out of 101) distribute over the range between 0.4s to 1.5s. As a result, the upper and lower bound of the peak distances, and , can be assigned according to the results in this figure. In this algorithm, the parameters for sit-to-stand transition are s and s. For stand-to sit case, they are s and s.

Figure 4.9 Distribution of peak distance in sit-stand transitions among all the subjects

Regarding the peak values (R3), Table 4.2 shows the average peak values of sit-stand transitions in different subject groups, and Figure 4.10 and Figure 4.11 further show the distributions of the peak values from all samples collected in the test. For the sit-to-stand case, most upper acceleration peak values (88.7%) are greater than 0.1g, and all lower acceleration peak values are less than 0g. For the stand-to-sit case, most upper acceleration peak values (76%) are greater than 0.1g, and most lower acceleration peak values (93%) are less than -0.1g. Therefore, the threshold parameters can be selected for the upper and lower acceleration peaks. For the sit-to-stand case, the thresholds g and g are assigned. For the stand-to-sit case, the thresholds g and g are assigned.

Table 4.2 Average peak values of sit-stand transitions in different subject groups

 

Young

Middle-aged

Elder

Sit-to-stand

Slow

Max. av (g)

0.16

0.14

 

Min. av (g)

-0.15

-0.20

Normal

Max. av (g)

0.24

0.26

0.24

Min. av (g)

-0.20

-0.28

-0.28

Fast

Max. av (g)

0.48

0.43

 

Min. av (g)

-0.49

-0.41

Stand-to-sit

Slow

Max. av (g)

0.18

0.17

 

Min. av (g)

-0.14

-0.18

Normal

Max. av (g)

0.28

0.19

0.37

Min. av (g)

-0.23

-0.27

-0.30

Fast

Max. av (g)

0.49

0.32

 

Min. av (g)

-0.41

-0.38

Figure 4.10 Distribution of positive vertical acceleration peaks of sit-stand transitions in different subject groups

Figure 4.11 Distribution of negative vertical acceleration peaks of sit-stand transitions in different subject groups

As shown in the main algorithm flowchart in Figure 4.2, the following decisions D5 and D6 identify sit-stand transitions from a dynamic movement with an upright trunk posture at the end has been recognized (D4=Yes). Figure 4.11 further illustrates the decomposed procedure of the decisions D5 and D6 which embodies several required thresholds and parameters for distinguishing sit-stand transitions. The first step (R1) is to check the type of peak orders to The following two steps (R2 and R3) check the peak distance and the peak amplitudes (PUpper and PLower) to further confirm whether the dynamic movement is indeed a sit-to-stand or stand-to-sit transition. The threshold values used in this identification process were experimentally obtained in the test with 15 subjects mentioned above. R1, R2 or R3 must be all satisfied to identify sit-to-stand or stand-to-sit transition. Otherwise the decision D5 and D6 fail to identify the dynamic movement and the result of a uncertain movement is given.

 

Figure 4.12 The decomposed flowchart of sit-stand transition identification

(2)  Identifying walking movement

Figure 4.13 shows an example of the vertical acceleration pattern during walking. As shown in this figure, a walking movement simplified as a patterns with fast and repeated oscillating changes in the vertical acceleration. Previous studies revealed that the frequencies of normal level walking are between 0.7Hz to 4Hz with the peak acceleration values between 0.4g to -0.3g [Sekine et al., 2000; Mathie et al., 2003]. Those parameters were adopted in this identification process. The decision D8 also determines a walking movement if the decision D7 fails to recognize a stand-to-sit transition (D7=No).

 

Figure 4.13. An example of the vertical acceleration patterns during walking

(3)  Identifying lie-sit postural transitions

Tilt data and its end posture state (trunk orientation) are required to identify and distinguish lie-sit type postural transitions. In the binary sequence of the trunk tilt data, a lie-to-sit transition produces positive trunk tilt increments and ends with upright posture. On the contrary, a sit-to-lie transition produces negative trunk tilt increments and ends with lying posture. Decision D8 identifies whether a lie-to-sit transition if a walking movement is not recognized in D7. For a dynamic movement (D2=Yes) with recognized lying end state (D4=No), the decision D10 is used to identify sit-to-lie transition if the sign of fall is not recognized (D9=No).

4.3.3 Identifying possible falls and still postures

A fall can be intuitively regarded as a movement accompanied by unusually high acceleration peaks in a short time interval, which is followed by a period of lying still posture. Therefore, fall detection in this algorithm is designed with two identification stages. The first stage is to determine whether there is a sign of fall in a dynamic movement. A sign of fall is defined as the existence of a SMA greater than 1.8g [Karantonis et al., 2006].   

For the first identification stage, a dynamic movement (D2=Yes, D3=Yes), with a non-upright (D4=No) posture is processed in the decision D9 to determine whether a sign of fall exists. The sign of fall is defined to indicate an occurrence of fall with an existence of peak acceleration greater than Vf=±1.5g. For the second identification stage, the algorithm starts to count the time duration of lying still posture in the following identification loop. The decision D11 examines the trunk tilt data to determine a lying posture. Therefore, a dynamic movement with a previous sign of fall can be raised to indicate a possible fall if a prolonged TD=20s period of lying still expires (D12=Yes).

Identifying and distinguishing sitting still or standing still posture requires not only the trunk tilt orientation data, but also the information of previously identified postural transitions or walking movement. If a still posture is not a lying posture (D11=No) and its previous movement is either a sit-to-stand transition or walking movement, this still posture can be recognized as standing still (D15=Yes). On the other hand, a still posture is sitting still if the previous movement is either a stand-to-sit or lie-to-sit transition (D14=Yes).

4.4 Performance evaluation of the algorithm

4.4.1 Basic performance test: sensitivity and specificity

In order to evaluate the performance of the device in identifying still postures and dynamic activities, 10 subjects were recruited for the laboratory-based test. Sensitivity and specificity for identifying postural transitions (sit-stand transitions, lie-sit transitions) and walking were recorded. In the sensitivity test, the subjects were asked to perform separate test items according to the facilitator’s instruction. The test items include the reciprocal postural transitions (sit-stand transitions and sit-lie transitions) and continuous walking movements. The subjects repeated each postural transition 20 times. In addition, the sensitivity of walking was obtained from 50s of continuous walking steps performed by each subject.

For the specificity test, 5 test items were arranged into phase A (sit-to-stand and sit-to-lie), phase B (stand-to-sit and walking) and phase C (lie-to-sit). According to the algorithm, when a subject is sitting still, his possible posture transition from that state is either sit-to-stand transition or sit-to-lie transition. Therefore in phase A, the subjects were initially in the sitting still posture and they can perform any movement other than the sit-to-stand or sit-to-lie transitions. Similarly in phase B, the subjects were initially standing still, and then performed any movements other than stand-to-sit transition or walking movement. In phase C, the subjects performed any movements other than lie-to-sit transition as they were initially lying. Each subject performed a total of 50 movements Note that during the test the subjects were allowed to vary their movement speeds or time intervals arbitrarily, and they were informed to avoid performing ambiguous movements. For example, in phase A, the subjects were asked to avoid performing a movement similar to sit-to-lie transition even though they subjectively considered it a free movement.

Table 4.3 shows the sensitivity and specificity obtained from 200 and 500 data samples respectively. Note that the evaluation did not include sitting still or standing still postures because both still postures are associated with the results of previously identified postural transitions or walking.

Table 4.3 Evaluation of sensitivity and specificity of the algorithm in identifying still postures and dynamic movements

Posture/activity

Sensitivity (%)

Specificity (%)

Sit-to-stand

92.2

91.5

stand-to-sit

95.6

88.5

Sit-to-lie

92.2

99.5

Lie-to-sit

95.6

88.0

Walking

98.9

99.5

4.4.2 Demonstration of continuous activity tele-monitoring at home

Implementing the algorithm, the wearable motion detector was used to demonstrate the possibility of continuous tele-monitoring of physical activity at home. Figure 4.14 shows the activity chronograph of a subject at home. This figure contains 6.75-hour data in the 15-hour monitoring period. The subject did not wear the WMD continuously in some situations, such as taking a shower, going outside, etc., therefore the vertical time axis is not continuous. In this figure, lines of various lengths represent respective events (event 1 to 9) identified by the system. For example, the subject was lying from 4:00 to 9:00, and mostly sitting from 9:00 to 12:00.

The recorded activity data can be classified as in Figure 4.15 by using a computer program showing the numbers and percentages of each dynamic activity or still posture within the entire monitoring period. The subject also made a report to register his actual activities during this period, and the monitored data reveals moderate correlation with the report. While the chronograph in Figure 4.14 shows the pattern of the subject’s daily activity, the quantitative data in Figure 4.15 can be used to form indices of the mobility level of the subject.

Event number 1: lying still, no.2: sitting still, no.3: standing still, no.4: walking,

no.5: sit-to-stand, no.6: stand-to-sit, no.7: lie-to-sit, no.8: sit-to-lie, no.9: possible fall: 9

Figure 4.14 Example of activity chronograph of the continuously recorded data in the ambulatory monitoring at home

dded

Figure 4.15 The results of the recorded activities

4.5 Discussion

To identify basic daily physical activities, an algorithm designed for the wearable motion detector is presented in this chapter. A miniature tri-axial capacitive accelerometer is used as the motion sensor due to its superiority of accurate acceleration/tilt sensing, fast response and low power consumption. With accelerometry measurement, this algorithm can identify human postures, walking and postural transitions. The potential capability of fall detection is also presented. High identification accuracy is achieved in the performance evaluation, though actual human movements in daily living are more complex than what are considered and characterized in the algorithm.

During the performance evaluation and implementation of the system in the home environment, the following limitations regarding real-time processing and tele-monitoring of physical activities were also observed:

(1)     Tri-axial accelerometer

An advanced capacitive tri-axial accelerometer is used to measure both the acceleration and trunk tilt of the human body. For such a gravity-responsive accelerometer, the most precise tilt sensing can be maintained when the accelerometer is at static, or under constant acceleration. Degradation of tilt sensing accuracy in changing acceleration magnitude is inevitable [Elble, 2005]. Therefore, the acceleration data acquired from the wearable motion detector cannot be used to precisely interpret the complex nature of human motions. In spite of this constraint, tilt sensing and acceleration measurement using one tri-axial accelerometer is still valid for most daily physical activities because the resulting sensor outputs still preserve apparent characteristics accelerometirc patterns.

(2)     The wearable motion detector

The wearable motion detector is carried at the waist/pant belt for ease of attachment and convenience. However, it might limit some movements in lying posture and therefore further influences the wearer’s comfort. In addition, the subject could not wear the device in some situations, such as taking a shower, going outside, etc. This has been an important usability issue for a fall detector in general, as these are often the situations where users are at great risk of falling at home (getting out of the shower, getting out of bed to go to the bathroom at night, etc). In addition, reduction in the size and weight of the device is also an important design issue in improving usability.

Moreover, the power capacity of batteries is an important design consideration for the practicality of wearable devices. To extend the time for use, a battery cartridge of larger power capacity can be selected. Proper selection of sampling rate in the microprocessor is also important in order to maintain optimal power longevity. For example, when the subject is at static state, the microprocessor should be set at a lower sampling rate because higher signal resolution is not necessary.

(3)     Real-time identification

The algorithm developed in this chapter has the ability to provide real-time status of physical activity. However, because each physical activity must be identified in continuous intervals with discrete short samples, real-time systems usually cannot provide the identification accuracy higher than that of the off-time analysis systems [Karantonis, et al., 2006]. The limitation concerning the identification accuracy of both off-line and real-time systems is the complex nature of human movement. People usually perform mixed and combined movements in their normal activities of living, even the diverse characteristics of one activity item can be observed from the same person. Advances in signal processing techniques such as self-adaptive parameter tuning can be further employed to enhance the identification accuracy.

The wearable motion detector with the algorithm has been technically proven to be feasible for continuous tele-monitoring of physical activity at home. This real-time identification approach can provide sufficient information in evaluating the subject’s activities of daily living (ADL) pattern and the physical mobility level. Connected with a DDS, the wearable motion detector enables real-time telemonitoring of basic daily activities at home. Rehabilitation, elderly care and personal health management are the potential applications of this wearable systes.

References

Bouten C. V., Koekkoek K. T. M., Verduin M., Kodde R., and Janssen J. D., 1997. “A Triaxial Accelerometer and Portable Data Processing Unit for the Assessment of Daily Physical Activity,” IEEE Trans. Biomed. Eng., vol. 44, No. 3, pp.136-147.

Capspersen C. J., Powell K. E., and Christenson G. M., 1985. “ Physical Activity, Exercise and Physical Fitness: Definitions and Distinctions for Health Related Research,” Public Health Rep., vol. 110, pp. 126-131.

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

Elble J. R., 2005. “Gravitational Artifact in Accelerometric Measurements of Tremor,” Clinical Neurophysiology, vol. 116, pp. 1638-1643.

Karantonis D. M., Narayanan M. R., Mathie M., Lovell N. H., and Celler B.G., 2006. “Implementation of a Real-Time Human Movement Classifier Using a Triaxial Accelerometer for Ambulatory Monitoring,” IEEE Trans. Info. Tech Biomed., vol. 10, No. 1,pp. 156-167.

Kemp B., Janssen A., and Kamp B.V.D., 1998. “Body Position can be Monitored in 3D Using Miniature Accelerometers and Earth-magnetic Field sensors,” Electroencephalography and clinical Neurophysiology, vol. 109, pp. 484-488.

Lyons G. M., Culhane K. M., Hilton D., Grace P. A., and Lyons, D., 2005. “A Description of an Accelerometer-based Mobility Monitoring Technique,” Medical Engineering & Physics, Vol. 27, pp. 497-504.

Mathie M. J., Coster, A. C. F., Lovell, N. H. Celler, B. G., 2003. “Detection of daily physical activities using a triaxial accelerometer,” Medical & Biological Engineering & Computing, vol. 41, pp. 296-301.

Mathie M. J. 2003. “Monitoring and interpreting human movement patterns using a triaxial accelerometer,” Ph.D. thesis, University of New South Wales, Sydney, Australia,

Meijer G.A.L., Westerterp K.R., Verhoeven F.M.H., Koper H.B.M., and Hoor, F. T., 1991. “Methods to Assess Physical Activity with Special Reference to Motion Sensors and Accelerometers, IEEE Trans. On Biomed. Eng., vol. 38, No.3, pp. 221-229.

Najafi B., and Aminian K., 2002. “Measurement of Stand-Sit and Sit-Stand Transitions Using a Miniature Gyroscope and Its Application in Fall Risk Evaluation in the Elderly,“ IEEE Trans. Biomed. Eng., vol. 49, pp. 843-851.

Najafi B., Aminian K., Paraschiv-Ionescu A., Loew F., Bula C. J., and Robert P., 2003. “Ambulatory system for Human Motion Analysis Using a Kinematic Sensor: Monitoring of Daily Physical Activity in the Elderly,” IEEE Trans. On Biomed. Eng., vol. 50, no. 6, pp.711-723.

Noury N., Virone G., and Creuzet T., 2002. “The Health Integrated Smart Home Information System(HIS2): Rules Based System for The Localization of a Human,” 2nd Annual International IEEE-EMBS Special Topic Conference on Microtechnologies in Medicine & Biology, pp. 318-321.

Nyan M. N., Tay F. E. H., Seah K. H. W., and Sitoh Y. Y., 2006. “Classification of a Gait Patterns in the Time-frequency Domain,” Journal of Biomechanics, Vol. 39, pp. 2647-2656.

Pennathur A., Magham R., Contreras L R., and 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.

Scanaill C. N., Carew S., Barralon P., Noury N., Lyons D., and Lyons G.M., 2006. “A Review of Approaches to Mobility Telemonitoring of the Elderly in Their Living Environment,” Annals of biomed. Eng., Vol. 34, No.4, pp. 547-563.

Sekine M., Tamura T., Togawa T., Fukui Y., 2000. “Classification of waist-acceleration signals in a continuous walking record,” Medical Eng. and Phy. Vol. 22, pp. 285-291.