Timed up & go quantification algorithm using IMU and sEMG signal

https://doi.org/10.1016/j.bspc.2022.104309Get rights and content

Highlights

  • TUG test is divided into 6 phases though building algorithm using EMG and IMU signals.

  • Quantitative parameters are also calculated by identifying leg movement and muscle contraction through an inertial sensor and an EMG sensor.

  • ICC is 0.9 or higher between TUG times obtained by algorithm and by video as gold standard.

  • Several parameters were confirmed to show significant difference according to age.

Abstract

The physical function of the elderly is important in an aging society. Among various physical function tests that are available, Timed Up & Go (TUG) is simple, although it includes various phases, and it’s widely used in the medical field. In the TUG test, only the total execution time is manually evaluated. Therefore, during the test, it’s not possible to check the physical function according to each movement in detail. In this study, an algorithm is presented for calculating TUG phase and quantitative parameters by identifying leg movement and muscle contraction through an inertial and an electromyography sensor. TUG test was performed on 127 subjects, and the algorithm was verified using the resulting data. TUG times obtained by Algorithm and by Stopwatch were compared with Video TUG time, which was the accurate value. The time obtained by Algorithm was 0.09 s different from the accurate value, while Stopwatch was 0.37 s different. The statistical analysis confirmed that the algorithm was more accurate by 0.28 s, which was found to be a significant result. TUG phases obtained by Algorithm was compared with TUG phases confirmed by Video. The time difference in all phases was within 0.1 s, and the ICC between two methods was 0.9 or higher. It was confirmed whether quantitative parameters showed significant difference according to age, and several parameters was confirmed to show significant difference. Through TUG phases and quantitative parameters, the deviation is eliminated and precise evaluation becomes possible.

Introduction

As society continues aging, the average human lifespan continues increasing. However, in comparison, the healthy life expectancy is 10 years less, so most of the elderly are living in old age while suffering from various diseases [1]. To solve this problem, there is increasing interest in early diagnosis and the prevention of geriatric diseases. To prevent geriatric disease, it is essential to systematically manage the physical functions of the elderly.

For this purpose, there are various tests that can be used to evaluate the physical function of the elderly, such as Short physical performance battery (SPPB), Timed up & go (TUG), 6 min walk test, hand grip, and step test [2]. Among them, the TUG is one of the most useful evaluation methods in the health care field, because it is temporally and spatially efficient, and because it contains various motion elements. TUG is a test that measures the time it takes for a subject to get up from a chair, walk 3 m, turn, and walk back to the chair to sit down.

It is known that the TUG time significantly increases with age. Bohannon et al. conducted a TUG test on 4,395 subjects and reported significant differences between 8.1 s for 60–69 years of age and 11.3 s for 80–99 years of age. TUG is also related to balance ability [3]. The Berg balance score (BBS) and TUG time, which can be used to objectively evaluate balance ability, show a high negative correlation [4], [5]. Many studies have analyzed the correlation between TUG and falls. For example, Reynaud et al. estimated falls in Chronic obstructive pulmonary disease (COPD) patients based on the results of the TUG test. In the case of patients who had experienced a fall in the previous one year, when the TUG time of 10.9 s was used as a diagnostic criterion, the fall recurrence was predicted with a sensitivity of 100% and a specificity of 97% [6]. Further, Tsuyoshi et al. found that the dual-task cost (DTC) obtained through single TUG and dual task TUG was related to the fall history of the elderly. Older adults with slower single TUG and lower DTC values reported more frequent falls. TUG is also a good predictor of cognitive decline [7]. Green et al. conducted the TUG test on 189 community-dwelling elderly people, and the results showed about 76% accuracy in predicting cognitive decline in cognitively intact participants, and about 89% accuracy in classifying cognitive states [8].

Although the motion sequences of TUG are complex, only total TUG time is considered in the traditional TUG. The traditional method, which only measures total TUG time, doesn’t check various motions in detail [9]. It is difficult for an expert to measure the TUG phase in detail, and to calculate quantitative parameters directly. Therefore it is necessary to extract TUG phase and quantitative parameters of the TUG using the instrumented TUG. The needs for automated TUG tests with additional quantitative parameters of TUG have been recognized by various researchers in the field [10], [11]. Sprint et al. summarized the feasibility of automated assessment for each disease using wearable sensors [10]. In studies of hemiplegia, cadence, RMS and CV of IMU signals among TUG quantitative parameters can be used to significantly differentiate between independent and supervised hemiplegic participants. The studies found that Alzheimer’s disease patients have lower stride length and gait regularity than healthy group. Herman et al. reported that obtaining quantitative parameters using objective metrology such as iTUG may have clinical significance in the early stages of Parkinson’s disease (PD) [11]. The PD group took a longer TUG phase such as sit-to-stand, stand-to-sit duration than the tremor dominant (TD) group. And The PD group had larger gait parameters such as walking duration and number of steps, and the amplitude of the rotational angular velocity during turn than the TD group. Green et al. reported that detailed times and some of quantified TUG parameters (e.g., TUG phase times, range mid-swing point, cadence, etc.) included essential information that was not confirmed by the berg balance scale (BBS) or the total TUG time [5]. In previous studies, there were only parameters related to movement using IMU sensor. Muscle contraction could not be evaluated when movement is evaluated. So We used an instrumented TUG including a sEMG sensor. Parameters related to actual activity and muscle quality were calculated to evaluate the muscles as well. The instrumented TUG of this study is proposed to extract TUG phase time, gait parameters, inertia and EMG-related parameters by using IMU sensor and surface EMG.

Toshiyo et al. classified the TUG test into several phases for the quantitative analysis of fall risk [12]. IMU sensors were attached to the waist and femur, and phases were divided into Sit–stand, Walk1, Turn1, Walk2, Turn2, and Stand–sit using the angular velocity threshold. Palmerini et al. divided TUG movements obtained by wearing a single accelerometer on the waist into five types (sit-to-walk, gait, turning, gait, and walk-to-sit) [13]. Greene et al. subdivided the TUG time using a kinematic sensor in front of each shank and extracted gait parameters and angular velocity parameters using the motion information from the sensors. Of the 44 parameters, 29 were used to distinguish the presence or absence of a fall history with 77.3% sensitivity and 75.9% specificity [5].

In this study, the TUG time was subdivided and quantitatively confirmed using an inertial sensor and a surface electromyography sensor. The inertial sensor detects movement, while the EMG sensor detects muscle contraction. Leg movement and muscle contraction are important clues for distinguishing movements during the TUG test. We also intend to verify the usefulness of TUG segmentation time and quantitative parameters obtained through clinical trials on the elderly [14], [15], [16]. As people get older, physical function declines with age [17]. In addition, older people with diseases to physical function decline more rapidly [18]. We have conducted a study to find the elderly who have lower physical function than their age. This is a necessary discovery for the prevention of aging and disease, and many studies are being conducted to make such a discovery in medical and bioengineering [19], [20], [21], [22]. Ko et al. estimated the age of dynamic balance ability (DBA), one of the physical functions, through the TUG test using the IMU. And the DBA of the elderly in the local community was managed to prevent diseases and accidents caused by aging [22]. They expected that by estimating the age of DBA through the TUG test, it would be a study that could contribute a lot to quality of elderly life. Furthermore, it would be possible to find the patient type. And it is expected that it will also be able to find the patient type. Obtaining the quantitative parameters of TUG is a basic step for physical function evaluation. Based on this, it is necessary to find parameters that can well represent the difference between the healthy group and the specific patient group. As a first step, we identified some parameters that can be distinguished according to physical functions of an age group. This is to check whether there is a possibility that the quantitative parameters we extracted can be distinguished according to physical function. Vervoort et al. noted that 28 of the 72 iTUG parameters were able to differentiate between the young group aged 18 ∼ 45 years and the elderly group aged 46 ∼ 75 years old [23]. He also reported that this technique has become the basis for the classification of old adults aging normally and those aging with pathologies.

In this study, we present an algorithm that automatically calculates the TUG phase with IMU and sEMG sensors, and we check whether there is a statistically significant difference in quantitative parameters, including the TUG phase, according to age. Through this, it will be suggested that the TUG assessment equipment composed of IMU and sEMG and the algorithm based on it can be a basic system for more precise physical function assessment than the presently used method. This paper begins by presenting the measurement system and measurement method, and the driving idea of the algorithm to calculate the TUG phase. In addition, TUG quantitative parameters are presented, and information on clinical trials to verify them is presented as well. In the result section, the time calculated by the algorithm is compared with the time calculated by the physiotherapist’s stopwatch and video analysis. It is confirmed whether there was a significant difference in the TUG quantitative parameters according to age. The discussion summarizes these results and presents the conclusion of this paper.

Section snippets

Method

For algorithm development in this study, the research stage depicted in Fig. 1 was carried out. First, a pilot test was performed to construct a quantitative algorithm. An algorithm that can calculate TUG phase and TUG parameters is constructed using the data obtained from the pilot test. To validate the TUG quantitative algorithm, we conduct a clinical trial in the elderly. First, TUG time comparison among the algorithms is performed, as are stopwatch measurements and video measurements to

TUG phase separated by TUG quantification algorithm

Through the algorithm proposed in this paper, TUG time could be subdivided into reaction time, stand-up time, go time, turn time, return time, and sit-down time. These times were calculated by identifying the motions that distinguish each phase from the signal. Fig. 5 shows the division of phases in the gyro ML graph that is used to judge gait. The reaction time was expressed as the point from where the buzzer sounded to where the movement and contraction started. Stand-up time is from this

Discussion

Early diagnosis and prevention of geriatric diseases is an essential aspect of increasing the healthy lifespan among the aging population. To this end, it is necessary to systematically manage the physical functions of the elderly. In this study, we tried to quantify the TUG test, which is a typically used method for evaluating the physical functions of the elderly. A TUG clinical trial was conducted using a device that included an IMU and an electromyogram, along with an app to operate the

CRediT authorship contribution statement

Jun-Woo Lee: Methodology, Writing – original draft. Dong-Jun Park: Methodology, Validation. Min-Kyu Kim: Data curation, Validation. Myung-Jun Shin: Conceptualization. Jong-Hwan Park: Resources. Byeong-Ju Lee: Investigation, Data curation. Eun-Lee Lee: Investigation. Joon-Soo Jeong: Investigation. Se-Jin Ahn: Supervision.

Declaration of Competing Interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Sejin Ahn reports financial support was provided by National Research Foundation of Korea.

Acknowledgments

This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education (NRF-2020R1l1A3064576) and Convergence Medical Institute of Technology R&D project(CMIT2021-06), Pusan National University Hospital.

References (34)

  • B.R. Greene et al.

    Assessment of cognitive decline through quantitative analysis of the timed up and go test

    IEEE Trans. Biomed. Eng.

    (2011)
  • S. Reinfelder

    Timed Up-and-Go phase segmentation in Parkinson's disease patients using unobtrusive inertial sensors

    2015 37th Annual international Conference of the IEEE Engineering in Medicine and Biology Society (EMBC)

    (2015)
  • G. Sprint et al.

    Toward automating clinical assessments: a survey of the timed up and go

    IEEE Rev. Biomed. Eng.

    (2015)
  • T. Herman

    Identifying axial and cognitive correlates in patients with Parkinson’s disease motor subtype using the instrumented Timed Up and Go

    Exp. Brain Res.

    (2014)
  • Y. Higashi

    Quantitative evaluation of movement using the timed up-and-go test

    IEEE Eng. Med. Biol. Mag.

    (2008)
  • Palmerini, Luca, et al. Quantification of motor impairment in Parkinson's disease using an instrumented timed up and go...
  • V. Lummel et al.

    Intra-rater, inter-rater and test-retest reliability of an instrumented timed up and go (iTUG) test in patients with Parkinson’s disease

    PLoS ONE

    (2016)
  • View full text