An Automatic Foot and Shank IMU Synchronization Algorithm: Proof-of-concept.

When using wearable sensors for measurement and analysis of human performance, it is often necessary to integrate and synchronise data from separate sensor systems. This paper describes a synchronization technique between IMUs attached to the shanks and insoles attached at the feet and aims to solve the need to compute the ankle joint angle, which relies on synchronized sensor data. This will additionally enable concurrent analysis using gait kinematic and kinetic features. A proof-of-concept of the algorithm, which relies on cross-correlation of gyroscope sensor data from the shank and foot, to align the sensor systems is demonstrated. The algorithm output is validated against those signals synchronized using manually annotated heel-strike and toe-off ground-truth signal landmarks, identified in both the shank and feet signals using previously published definitions. Results demonstrate that the developed algorithm is capable of synchronizing both sensor systems, based on IMU data from both healthy participants and participants suffering from knee osteoarthritis, with a mean lag time bias of 25.56ms when compared to the ground truth. A proof-of-concept of technique to synchronise IMUs attached to the shanks and insoles attached at the feet is demonstrated and offers an alternative approach to sensor system synchronisation.

Template-Based Recognition of Human Locomotion in IMU Sensor Data Using Dynamic Time Warping.

Increased levels of light, moderate and vigorous physical activity (PA) are positively associated with health benefits. Therefore, sensor-based human activity recognition can identify different types and levels of PA. In this paper, we propose a two-layer locomotion recognition method using dynamic time warping applied to inertial sensor data. Based on a video-validated dataset (ADAPT), which included inertial sensor data recorded at the lower back (L5 position) during an unsupervised task-based free-living protocol, the recognition algorithm was developed, validated and tested. As a first step, we focused on the identification of locomotion activities walking, ascending and descending stairs. These activities are difficult to differentiate due to a high similarity. The results showed that walking could be recognized with a sensitivity of 88% and a specificity of 89%. Specificity for stair climbing was higher compared to walking, but sensitivity was noticeably decreased. In most cases of misclassification, stair climbing was falsely detected as walking, with only 0.2-5% not assigned to any of the chosen types of locomotion. Our results demonstrate a promising approach to recognize and differentiate human locomotion within a variety of daily activities.

Smartphone- and Smartwatch-Based Remote Characterisation of Ambulation in Multiple Sclerosis During the Two-Minute Walk Test.

Leveraging consumer technology such as smartphone and smartwatch devices to objectively assess people with multiple sclerosis (PwMS) remotely could capture unique aspects of disease progression. This study explores the feasibility of assessing PwMS and Healthy Control's (HC) physical function by characterising gait-related features, which can be modelled using machine learning (ML) techniques to correctly distinguish subgroups of PwMS from healthy controls. A total of 97 subjects (24 HC subjects, 52 mildly disabled (PwMSmild, EDSS [0-3]) and 21 moderately disabled (PwMSmod, EDSS [3.5-5.5]) contributed data which was recorded from a Two-Minute Walk Test (2MWT) performed out-of-clinic and daily over a 24-week period. Signal-based features relating to movement were extracted from sensors in smartphone and smartwatch devices. A large number of features (n = 156) showed fair-to-strong (R 0.3) correlations with clinical outcomes. LASSO feature selection was applied to select and rank subsets of features used for dichotomous classification between subject groups, which were compared using Logistic Regression (LR), Support Vector Machines (SVM) and Random Forest (RF) models. Classifications of subject types were compared using data obtained from smartphone, smartwatch and the fusion of features from both devices. Models built on smartphone features alone achieved the highest classification performance, indicating that accurate and remote measurement of the ambulatory characteristics of HC and PwMS can be achieved with only one device. It was observed however that smartphone-based performance was affected by inconsistent placement location (running belt versus pocket). Results show that PwMSmod could be distinguished from HC subjects (Acc. 82.2 ± 2.9%, Sen. 80.1 ± 3.9%, Spec. 87.2 ± 4.2%, F 1 84.3 ± 3.8), and PwMSmild (Acc. 82.3 ± 1.9%, Sen. 71.6 ± 4.2%, Spec. 87.0 ± 3.2%, F 1 75.1 ± 2.2) using an SVM classifier with a Radial Basis Function (RBF). PwMSmild were shown to exhibit HC-like behaviour and were thus less distinguishable from HC (Acc. 66.4 ± 4.5%, Sen. 67.5 ± 5.7%, Spec. 60.3 ± 6.7%, F 1 58.6 ± 5.8). Finally, it was observed that subjects in this study demonstrated low intra- and high inter-subject variability which was representative of subject-specific gait characteristics.

U-turn speed is a valid and reliable smartphone-based measure of multiple sclerosis-related gait and balance impairment.

BACKGROUND: People living with multiple sclerosis (MS) experience impairments in gait and mobility, that are not fully captured with manually timed walking tests or rating scales administered during periodic clinical visits. We have developed a smartphone-based assessment of ambulation performance, the 5 U-Turn Test (5UTT), a quantitative self-administered test of U-turn ability while walking, for people with MS (PwMS). RESEARCH QUESTION: What is the test-retest reliability and concurrent validity of U-turn speed, an unsupervised self-assessment of gait and balance impairment, measured using a body-worn smartphone during the 5UTT? METHODS: 76 PwMS and 25 healthy controls (HCs) participated in a cross-sectional non-randomised interventional feasibility study. The 5UTT was self-administered daily and the median U-turn speed, measured during a 14-day session, was compared against existing validated in-clinic measures of MS-related disability. RESULTS: U-turn speed, measured during a 14-day session from the 5UTT, demonstrated good-to-excellent test-retest reliability in PwMS alone and combined with HCs (intraclass correlation coefficient [ICC] = 0.87 [95 % CI: 0.80-0.92]) and moderate-to-excellent reliability in HCs alone (ICC = 0.88 [95 % CI: 0.69-0.96]). U-turn speed was significantly correlated with in-clinic measures of walking speed, physical fatigue, ambulation impairment, overall MS-related disability and patients' self-perception of quality of life, at baseline, Week 12 and Week 24. The minimal detectable change of the U-turn speed from the 5UTT was low (19.42 %) in PwMS and indicates a good precision of this measurement tool when compared with conventional in-clinic measures of walking performance. SIGNIFICANCE: The frequent self-assessment of turn speed, as an outcome measure from a smartphone-based U-turn test, may represent an ecologically valid digital solution to remotely and reliably monitor gait and balance impairment in a home environment during MS clinical trials and practice.

Gait Characteristics Harvested During a Smartphone-Based Self-Administered 2-Minute Walk Test in People with Multiple Sclerosis: Test-Retest Reliability and Minimum Detectable Change.

The measurement of gait characteristics during a self-administered 2-minute walk test (2MWT), in persons with multiple sclerosis (PwMS), using a single body-worn device, has the potential to provide high-density longitudinal information on disease progression, beyond what is currently measured in the clinician-administered 2MWT. The purpose of this study is to determine the test-retest reliability, standard error of measurement (SEM) and minimum detectable change (MDC) of features calculated on gait characteristics, harvested during a self-administered 2MWT in a home environment, in 51 PwMS and 11 healthy control (HC) subjects over 24 weeks, using a single waist-worn inertial sensor-based smartphone. Excellent, or good to excellent test-retest reliability were observed in 58 of the 92 temporal, spatial and spatiotemporal gait features in PwMS. However, these were less reliable for HCs. Low SEM% and MDC% values were observed for most of the distribution measures for all gait characteristics for PwMS and HCs. This study demonstrates the inter-session test-retest reliability and provides an indication of clinically important change estimates, for interpreting the outcomes of gait characteristics measured using a body-worn smartphone, during a self-administered 2MWT. This system thus provides a reliable measure of gait characteristics in PwMS, supporting its application for the longitudinal assessment of gait deficits in this population.

Performance Evaluation of State of the Art Systems for Physical Activity Classification of Older Subjects Using Inertial Sensors in a Real Life Scenario: A Benchmark Study.

The popularity of using wearable inertial sensors for physical activity classification has dramatically increased in the last decade due to their versatility, low form factor, and low power requirements. Consequently, various systems have been developed to automatically classify daily life activities. However, the scope and implementation of such systems is limited to laboratory-based investigations. Furthermore, these systems are not directly comparable, due to the large diversity in their design (e.g., number of sensors, placement of sensors, data collection environments, data processing techniques, features set, classifiers, cross-validation methods). Hence, the aim of this study is to propose a fair and unbiased benchmark for the field-based validation of three existing systems, highlighting the gap between laboratory and real-life conditions. For this purpose, three representative state-of-the-art systems are chosen and implemented to classify the physical activities of twenty older subjects (76.4 ± 5.6 years). The performance in classifying four basic activities of daily life (sitting, standing, walking, and lying) is analyzed in controlled and free living conditions. To observe the performance of laboratory-based systems in field-based conditions, we trained the activity classification systems using data recorded in a laboratory environment and tested them in real-life conditions in the field. The findings show that the performance of all systems trained with data in the laboratory setting highly deteriorates when tested in real-life conditions, thus highlighting the need to train and test the classification systems in the real-life setting. Moreover, we tested the sensitivity of chosen systems to window size (from 1 s to 10 s) suggesting that overall accuracy decreases with increasing window size. Finally, to evaluate the impact of the number of sensors on the performance, chosen systems are modified considering only the sensing unit worn at the lower back. The results, similarly to the multi-sensor setup, indicate substantial degradation of the performance when laboratory-trained systems are tested in the real-life setting. This degradation is higher than in the multi-sensor setup. Still, the performance provided by the single-sensor approach, when trained and tested with real data, can be acceptable (with an accuracy above 80%).

Video analysis validation of a real-time physical activity detection algorithm based on a single waist mounted tri-axial accelerometer sensor.

We have validated a real-time activity classification algorithm based on monitoring by a body worn system which is potentially suitable for low-power applications on a relatively computationally lightweight processing unit. The algorithm output was validated using annotation data generated from video recordings of 20 elderly volunteers performing both a semi-structured protocol and a free-living protocol. The algorithm correctly identified sitting 75.1% of the time, standing 68.8% of the time, lying 50.9% of the time, and walking and other upright locomotion 82.7% of the time. This is one of the most detailed validations of a body worn sensor algorithm to date and offers an insight into the challenges of developing a real-time physical activity classification algorithm for a tri-axial accelerometer based sensor worn at the waist.

Fall detection algorithms for real-world falls harvested from lumbar sensors in the elderly population: a machine learning approach.

Automatic fall detection will promote independent living and reduce the consequences of falls in the elderly by ensuring people can confidently live safely at home for linger. In laboratory studies inertial sensor technology has been shown capable of distinguishing falls from normal activities. However less than 7% of fall-detection algorithm studies have used fall data recorded from elderly people in real life. The FARSEEING project has compiled a database of real life falls from elderly people, to gain new knowledge about fall events and to develop fall detection algorithms to combat the problems associated with falls. We have extracted 12 different kinematic, temporal and kinetic related features from a data-set of 89 real-world falls and 368 activities of daily living. Using the extracted features we applied machine learning techniques and produced a selection of algorithms based on different feature combinations. The best algorithm employs 10 different features and produced a sensitivity of 0.88 and a specificity of 0.87 in classifying falls correctly. This algorithm can be used distinguish real-world falls from normal activities of daily living in a sensor consisting of a tri-axial accelerometer and tri-axial gyroscope located at L5.

The measurement of sedentary patterns and behaviors using the activPAL™ Professional physical activity monitor.

Epidemiological studies have associated the negative effects of sedentary time and sedentary patterns on health indices. However, these studies have used methodologies that do not directly measure the sedentary state. Recent technological developments in the area of motion sensors have incorporated inclinometers, which can measure the inclination of the body directly, without relying on self-report or count thresholds. This paper aims to provide a detailed description of methodologies used to examine a range of relevant variables, including sedentary levels and patterns from an inclinometer-based motion sensor. The activPAL Professional physical activity logger provides an output which can be interpreted and used without the need for further processing and additional variables were derived using a custom designed MATLAB® computer program. The methodologies described have been implemented on a sample of 44 adolescent females, and the results of a range of daily physical activity and sedentary variables are described and presented. The results provide a range of objectively measured and objectively processed variables, including total time spent sitting/lying, standing and stepping, number and duration of daily sedentary bouts and both bed hours and non-bed hours, which may be of interest when making association between physical activity, sedentary behaviors and health indices.

Embedded fall and activity monitoring for a wearable ambient assisted living solution for older adults.

With the rapidly increasing over 60 and over 80 age groups in society, greater emphasis will be put on technology to detect emergency situations, such as falls, in order to promote independent living. This paper describes the development and deployment of fall-detection, activity classification and energy expenditure algorithms, deployed in a tele-monitoring system. These algorithms were successfully tested in an end-user trial involving 9 elderly volunteers using the system for 28 days.

Activity recognition using dynamic multiple sensor fusion in body sensor networks.

Multiple sensor fusion is a main research direction for activity recognition. However, there are two challenges in those systems: the energy consumption due to the wireless transmission and the classifier design because of the dynamic feature vector. This paper proposes a multi-sensor fusion framework, which consists of the sensor selection module and the hierarchical classifier. The sensor selection module adopts the convex optimization to select the sensor subset in real time. The hierarchical classifier combines the Decision Tree classifier with the Naïve Bayes classifier. The dataset collected from 8 subjects, who performed 8 scenario activities, was used to evaluate the proposed system. The results show that the proposed system can obviously reduce the energy consumption while guaranteeing the recognition accuracy.

Ubiquitous ambient assisted living solution to promote safer independent living in older adults suffering from co-morbidity.

This paper describes the development, deployment and trial results from 9 volunteers using the eCAALYX system. The eCAALYX system is an ambient assisted living telemonitoring system aimed at older adults suffering with co-morbidity. Described is a raw account of the challenges that exist and results in bringing a Telemedicine system from laboratory to real-world implementation and results for usability, functionality and reliability.

Cross-sectional analysis of levels and patterns of objectively measured sedentary time in adolescent females.

BACKGROUND: Adolescent females have been highlighted as a particularly sedentary population and the possible negative effects of a sedentary lifestyle are being uncovered. However, much of the past sedentary research is based on self-report or uses indirect methods to quantity sedentary time. Total time spent sedentary and the possible intricate sedentary patterns of adolescent females have not been described using objective and direct measure of body inclination. The objectives of this article are to examine the sedentary levels and patterns of a group of adolescent females using the ActivPAL™ and to highlight possible differences in sedentary levels and patterns across the week and within the school day. A full methodological description of how the data was analyzed is also presented. METHODS: One hundred and eleven adolescent females, age 15-18 yrs, were recruited from urban and rural areas in the Republic of Ireland. Participants wore an ActivPAL physical activity monitor for a 7.5 day period. The ActivPAL directly reports total time spent sitting/lying every 15 seconds and accumulation (frequency and duration) of sedentary activity was examined using a customized MATLAB(®) computer software programme. RESULTS: While no significant difference was found in the total time spent sitting/lying over the full 24 hour day between weekday and weekend day (18.8 vs. 18.9 hours; p = .911), significantly more sedentary bouts of 1 to 5 minutes and 21 to 40 minutes in duration were accumulated on weekdays compared to weekend days (p < .001). The mean length of each sedentary bout was also longer (9.8 vs. 8.8 minutes; p < .001). When school hours (9 am-3 pm) and after school hours (4 pm-10 pm) were compared, there was no difference in total time spent sedentary (3.9 hours; p = .796) but the pattern of accumulation of the sedentary time differed. There were a greater number of bouts of > 20 minutes duration during school hours than after school hours (4.7 vs. 3.5 bouts; p < .001) while after school time consisted of shorter bouts < 20 minutes. CONCLUSIONS: School is highlighted as a particularly sedentary setting for adolescent females. Interventions to decrease sedentary time at school and the use of wearable devices which distinguish posture should be encouraged when examining sedentary patterns and behaviors in this population.

Fall algorithm development using kinematic parameters measured from simulated falls performed in a quasi-realistic environment using accelerometry.

This study aims to determine the optimal temporal, angular and acceleration parameters and thresholds for an accelerometer based, chest-worn, fall detection algorithm. In total, 10 healthy male subjects performed 14 different fall types, 3 times by each. The falls were performed onto in a quasi-realistic environment consisting of mats of a minimum thickness. Optimum parameters for; t(falling): time-to-fall, θ(max): max-angle, t(θmax) : max-angle-time, t(RTStanding) : Return-to-standing-time and t(lying) : lying-time were determined using a data set consisting of a total of 420 falls.

A system for activity recognition using multi-sensor fusion.

This paper proposes a system for activity recognition using multi-sensor fusion. In this system, four sensors are attached to the waist, chest, thigh, and side of the body. In the study we present two solutions for factors that affect the activity recognition accuracy: the calibration drift and the sensor orientation changing. The datasets used to evaluate this system were collected from 8 subjects who were asked to perform 8 scripted normal activities of daily living (ADL), three times each. The Naïve Bayes classifier using multi-sensor fusion is adopted and achieves 70.88%-97.66% recognition accuracies for 1-4 sensors.

Optimum gravity vector and vertical acceleration estimation using a tri-axial accelerometer for falls and normal activities.

UNLABELLED: This study aims to determine an optimum estimate for the gravitational vector and vertical acceleration profiles using a body-worn tri-axial accelerometer during falls and normal activities of daily living (ADL), validated using a camera based motion analysis system. Five young healthy subjects performed a number of simulated falls and normal ADL while trunk kinematics were measured by both an optical motion analysis system and a tri-axial accelerometer. Through low-pass filtering of the trunk tri-axial accelerometer signal between 1 Hz and 2.7 Hz using a 1(st) order or higher, Butterworth IIR filter, accurate gravity vector profile can be obtained using the method described here. RESULTS: A high mean correlation (≥ 0.83: Coefficient of Multiple Correlations) and low mean percentage error (≤ 2.06 m/s(2)) were found between the vertical acceleration profile generated from the tri-axial accelerometer based sensor to those from the optical motion capture system. This proposed system enables optimum gravity vector and vertical acceleration profiles to be measured from the trunk during falls and normal ADL.

Assessment of waist-worn tri-axial accelerometer based fall-detection algorithms using continuous unsupervised activities.

This study aims to evaluate a variety of existing and novel fall detection algorithms, for a waist mounted accelerometer based system. Algorithms were tested against a comprehensive data-set recorded from 10 young healthy subjects performing 240 falls and 120 activities of daily living and 10 elderly healthy subjects performing 240 scripted and 52.4 hours of continuous unscripted normal activities.

The application of inertial and magnetic sensors to the monitoring of calf muscle pump activity.

On long distance journeys passengers at high risk from deep vein thrombosis (DVT) are recommended to exercise on a regular basis to contract the calf muscle pump and encourage venous return. If a passenger fails to complete an exercise program that induces active contraction of the calf muscle pump they will remain at increased risk of DVT. This paper presents a novel inertial and magnetic sensor-based technique for monitoring calf muscle pump activity. The technique could be implemented into a system for monitoring the level of calf muscle pump activity in persons with limited mobility. Such a system could be used to provide a reminder to the user that there is a need to exercise should they have forgotten to exercise, failed to exercise sufficiently or exercised incorrectly. The proposed technique was evaluated by comparison with calf muscle pump activity measured using an electromyography (EMG) sensor. Results show that the technique can be used to monitor calf muscle pump activity over a wide range of leg exercises.

Fall-detection through vertical velocity thresholding using a tri-axial accelerometer characterized using an optical motion-capture system.

Falls in the elderly population are a major problem for today's society. The immediate automatic detection of such events would help reduce the associated consequences of falls. This paper describes the development of an accurate, accelerometer-based fall detection system to distinguish between Activities of Daily Living (ADL) and falls. It has previously been shown that falls can be distinguished from normal ADL through vertical velocity thresholding using an optical motion capture system. In this study however accurate vertical velocity profiles of the trunk were generated by simple signal processing of the signals from a tri-axial accelerometer (TA). By recording simulated falls onto crash mats and ADL performed by 5 young healthy subjects, using both a single chest mounted TA and using an optical motion capture system, the accuracy of the vertical velocity profiles was assessed. Data analysis was performed using MATLAB to determine the peak velocities recorded and RMS error during four different fall and six ADL types. Results show high correlations and low percentage errors between the vertical velocity profiles generated by the TA to those recorded using the optical motion capture system. In addition, through thresholding of the vertical velocity profiles generated using the TA at -1.3m/s, falls can be distinguished from normal ADL with 100% sensitivity and specificity.

The design and development of a long-term fall detection system incorporated into a custom vest for the elderly.

A fall detection system and algorithm, incorporated into a custom designed garment has been developed. The developed fall detection system uses a tri-axial accelerometer, microcontroller, battery and Bluetooth module. This sensor is attached to a custom designed vest, designed to be worn by the elderly person under clothing. The fall detection algorithm was developed and incorporates both impact and posture detection capability. The vest and fall algorithm was tested on young healthy subjects performing normal activities of daily living (ADL) and falls onto crash mats, while wearing the best and sensor. Results show that falls can de distinguished from normal activities with a sensitivity >90% and a specificity of >99%, from a total data set of 264 falls and 165 normal ADL. By incorporating the fall-detection sensor into a custom designed garment it is anticipated that greater compliance when wearing a fall-detection system can be achieved and will help reduce the incidence of the long-lie, when falls occur in the elderly population. However further long-term testing using elderly subjects is required to validate the systems performance.