A novel motion monitoring system for activities of daily living

Capability to perform activities of daily living (ADLs) is a major factor in quality of life (QOL). While it can be difficult for the elderly, disabled, or patients with chronic diseases to deal with ADLs, they need to spend a great deal of money on healthcare and assistive technologies to keep a good QOL. The situation can be improved if a real-time ADLs monitoring and recognition system is available to provide health information to physicians, pharmacists, or caregivers to offer timely diagnosis, prescription, or emergency reaction. We have developed a wireless wearable motion monitoring system that is suitable for monitoring ADLs involving limbs. The system consists of six Bluetooth low energy (BLE) transponders that are small and light enough to be mounted on all limbs. Each transponder, called SensorTag (by Texas Instruments), is equipped with a tri-axial accelerometer, a triaxial magnetometer, and a tri-axial gyroscope for motion monitoring. Each SensorTag can be linked to a smartphone for long-term outdoor monitoring. A graphic user interface is created to acquire signals from BLE receivers, display the signals in real-time, process data, and store for off-line analysis. This system was tested in three scenarios, and signals were analyzed off-line using a quaternion-based motion reconstruction algorithm. First, a SensorTag was examined against a marker-based motion capture system in a linear motion test. Second, a SensorTag was worn on a subject’s wrist to monitor food-intake trajectory. Finally, six SensorTags were worn on wrist, knee, and ankle joints of left and right hands to monitor gait on a straight path. Results showed various error rates in different scenarios, however, the error rates are within an acceptable range, and more importantly the patterns of the motions are reproducible.


Motivation for monitoring activities of daily living
Activities of daily living (ADLs) are generally categorized in to two main categories [1].The first type is the basic ac-tivities of daily living (BADLs), which includes self-care tasks such as bathing, dressing, toileting, brushing teeth, eating and functional mobility.The other type is the instrumental activities of daily living (IADLs), which allows people to keep an independent lifestyle with additional services, such as cooking, driving, using telephone or computer, shopping, keeping track of finances and managing medication [2,3].The ability to perform ADLs is a major factor in determining one's quality of life (QOL) [4, 5, and 6].While it can be very difficult for elderly, disabled or chronic disease patients to deal with ADLs, they also need to spend so much on healthcare and assistive technologies to keep a reasonable QOL.
The problem becomes more serious as the aging population is growing.According to the report by the Population Division (a contribution to the 2002 World Assembly on Ageing and its follow-up), the number of aging people in the world has been increasing every year since 1950 [7].Muscle strength, balance and body function declines with aging, and the possibility of neurodegenerative diseases such as Parkinson's disease (PD) increases [8].PD patients, for example, have typical symptoms like rest tremor, bradykinesia, hypokinesia or rigidity [9].Most tremors happen at hands or fingers resulting in failed attempts to perform ADLs like holding forks.Apart from the difficulties in executing ADLs, the therapy relies on the PD patient's symptoms [10,11].Different dosage should be prescribed according to the symptom reports, which can be extracted from the motor features report in diaries.However, the report can only cover a short period of time.As a result, researchers have developed various systems to monitor these symptoms and ADLs.A summary of the state-of-the-art of the monitoring systems, which can capture the motion, is explained in the following section.

Current state-of-the-art of motion monitoring technologies and shortcomings
Various technologies have been used for human body monitoring.A summary of these technologies with their pros and cons is shown in Table 1.Human motion monitoring with markers has been used for human body and body part motion analysis since Johansson's (1973) moving light display psychological experiment [12].Although the marker-based optical motion capture (mo-cap) systems provide very accurate motion data, they are expensive to setup and the monitoring space is limited [13].With the development of Microelectromechanical systems (MEMS) devices, inertial sensors are used in some simple applications like video gaming consoles and smartphones for orientation computation [14].Wii MotionPlus is a typical remote controller with more accurate motion sensing abilities, but it can be easily deceived because it only monitors the player's hand motion, and the player can sit on a couch and pretend to play tennis or golf.In other words the console does not provide whole body motion monitoring.The advantages of inertial measurement unit (IMU) sensing devices are that they are easy to setup, small to wear, and they are cheaper than the optical ca -mera system.However, their information is not as accurate as mo-cap systems.The sensor data drifts as time progresses and there is a lack of a reference system to fix the error [15].
In IMUs, gyroscope measures the angular rate for orientation [16].However, it drifts over time and needs the direction of gravity (accelerometer) or earth magnetic field (magnetometer) as reference for revision [17, 18, and 19].This type of sensor that can measure magnetic, angular rate, and Table 1.Comparison of marker-based optical motion capture (mo-cap) system, depth sensing camera or scanner and inertial sensors mo-cap system.gravity on three axes are often referred to as MARG, and have been used for advanced monitoring of aircraft attitudes, including roll, pitch, and yaw [20].Based on MARG, a quaternion algorithm is developed for the drift compensation that makes it possible to achieve indoor navigation where GPS cannot work [21,22].
Depth camera based systems have also been used for human motion monitoring.The depth camera measures the points' distance of the captured scene, and its price and data accuracy have become attractive with the invention of Microsoft Kinect.Unlike the Wii MotionPlus console, Kinect monitors the whole human body and it requires players to move the same way instructed by the gaming role.However, depth camera still has occlusion problem, and can only measure from a limited distance.
Researches showed that the human body movements can be estimated with muscle activities using Electromyography (EMG) signal alone or a combination of EMG and IMU to obtain more details [23,24].Myo (Thalmic Labs Inc.), which provides 8 EMG pods and a 9-axis IMU in an armband, has many applications in motion sensing.Myo has been used in gesture control experiments.Although it has so many advantages, it requires eight nodes for each hand that makes it impractical for users.
For ADLs monitoring, to track trajectory of one's motions regardless of indoor or outdoor, a portable or wearable device is the best choice.In this paper, we have developed a wearable motion monitoring system using the MARG sensors to track the trajectory of the limbs' motions.A graphic (3)

Function
The system wa ar displaceme ight path.The owing.
. Monitorin this experim sorTag, and t nter on a linea ture lab.The systems wer line comparis two systems.h 5.18 cm erro 74 cm that cau k of orientatio ar displaceme m the start to 81 cm, respec .T ation (15), yie 1.99% for the     A novel motion n monitoring system for acti ivities of daily y living   data collection.The accuracy of the system was tested in different scenarios, and compared to a camera-based mo-cap system or a measurable criterion.
Linear displacement tests demonstrated our system is less accurate than the camera-based mo-cap systems; however, the error rate is still under 8%.While the proposed system is much cheaper than the camera-based mo-cap systems, it is also easier to set it up, especially in the outdoor scenarios.This is valuable to monitor activity of subjects while they are not limited to indoor.
For the food in-take test, we mimicked the process and reconstructed the food in-take motion pattern.The error in X and Y axes of 40 tests on each hand are in acceptable range (less than 20 cm), while the Z axis error reached 30 cm due to the use of low quality gyroscope.
In the walking test, hands moved back and forth, and the OEA calculates the relative acceleration, which leads to a different result from the knee and ankle movements.The knee and ankle joints' movements are almost in a straight line, hence, the motion reconstruction results are more reliable and the error rates of ankle joints in these tests are less than 13%.
Based on the results, there are three ways to improve the system in the future.First, utilize a better gyroscope in the SensorTag to avoid drifting problem.Second, develop a customized OEA for the ADLs' monitoring that has improved orientation and displacement calculation.Finally, a more practical smartphone application could be developed to connect six or more SensorTags simultaneously, and save data in higher sampling rates.
on Pervasive Health and Technology 12 2016 -03 2017 | Volume 3 | Issue 9 | e3A novel motion n monitoring system for acti ivities of daily y living X.

Figure 7
Figure 7. and "W" a

Figure 8
Figure 8. T eration to shown.