Search

Inertial measurement units are an efficient tool to estimate the orientation of a rigid body with respect to a global or navigation frame. Thanks to their relatively small scale, these devices are often employed in clinical environments in form of wearable devices. A direct consequence of this large use of inertial sensors has been the development of many sensor fusion techniques for pose estimation in many practical applications. In this paper we study the feasibility of a nonlinear "Unscented" variant of the well-known Kalman Filter for gyroscope/accelerometer sensor fusion in pelvis pose estimation during treadmill walking. In addition, orientation estimation has been obtained without IMU magnetometer data, in order to propose a method suitable also for environments where magnetic disturbances could arise. Pelvis heading (yaw), bank (roll) and attitude (pitch) angles have been evaluated both using the proposed filter and a gold standard optometric system. The root mean square errors obtained using the proposed sensor fusion with respect to the gold standard are below 1 degree for each axis, showing also a significant high correlation (> 0.90). Findings of this study highlight the suitability of a magnetometer-free UKF approach for pose estimation of pelvis during human walking on treadmill, providing information useful also for further estimation of center of mass displacement in the same experimental conditions.

This study focuses on automatic stroke-screening of the arm factor in the FAST (Face, Arm, Speech, and Time) stroke screening method. The study provides a methodology to collect data on specific arm movements, using signals from the gyroscope and accelerometer in mobile devices. Fifty-two subjects were enrolled in this study (20 stroke patients and 32 healthy subjects). Given in the instructions of the application, the patients were asked to perform two arm movements, Curl Up and Raise Up. The two exercises were classified into three parts: curl part, raise part, and stable part. Stroke patients were expected to experience difficulty in performing both exercises efficiently on the same arm. We proposed 20 handcrafted features from these three parts. Our study achieved an average accuracy of 61.7%-74.2% and an average area under the ROC curve (AUC) of 66.2%-81.5% from the combination of both exercises. Compared to the FAST method used by examiners in a previous study (Kapes et al., 2014) that showed with an accuracy of 69%-77% for every age group, our study showed promising results for early stroke identification, giving that our study is based only on the arm factor.

A miniaturized accelerometer attached to the heart has been used for monitoring functional parameters such as early systolic velocity and displacement. Currently, processing of the accelerometer signal for derival of these functional parameters depends on determining start systole by detecting the ECG R-peaks. This study proposes an alternative method using only the accelerometer signal to detect start systole, making additional ECG recordings for this purpose redundant. A signal processing method for automatic detection of start systole by accelerometer alone was developed and compared with detected R-peaks in 15 pigs during 5 different interventions showing a difference of 30 ± 17 ms. Furthermore, the derived early systolic velocity and displacement using only accelerometer measurements correlated well (r 2 =0.91 and 0.82, respectively) with minor differences compared to the current method using ECG R-peaks as time reference. The results show that an accelerometer can be used to monitor cardiac function without the need to measure ECG which can simplify the monitoring system.

Magneto-Inertial technology is a well-established alternative to optical motion capture for human motion analysis applications since it allows prolonged monitoring in free-living conditions. Magneto and Inertial Measurement Units (MIMUs) integrate a triaxial accelerometer, a triaxial gyroscope and a triaxial magnetometer in a single and lightweight device. The orientation of the body to which a MIMU is attached can be obtained by combining its sensor readings within a sensor fusion framework. Despite several sensor fusion implementations have been proposed, no well-established conclusion about the accuracy level achievable with MIMUs has been reached yet. The aim of this preliminary study was to perform a direct comparison among four popular sensor fusion algorithms applied to the recordings of MIMUs rotating at three different rotation rates, with the orientation provided by a stereophotogrammetric system used as a reference. A procedure for suboptimal determination of the parameter filter values was also proposed. The findings highlighted that all filters exhibited reasonable accuracy (rms errors <; 6.4°). Moreover, in accordance with previous studies, every algorithm's accuracy worsened as the rotation rate increased. At the highest rotation rate, the algorithm from Sabatini (2011) showed the best performance with errors smaller than 4.1° rms.

In a gradually aging society, families and hospitals have a growing demand for reliable and unobtrusive physiological signal monitoring for elderly people. However, the existing respiration rate monitoring methods and algorithms are still unsatisfactory. In this work, we introduce a physiological signal acquisition patch which integrates 3-axis accelerator and 3-axis gyroscope to estimate respiration rate, as well as ECG(electrocardiogram) sensor and surface temperature sensor. A complete set of respiration rate estimation algorithms is embedded in our patch, which can be used to identify whether the patch is worn or not, and to recognize, segment, de-noise and reconstruct the respiration signal. In-situ experiments have been conducted to prove the validity of the algorithms described in this paper and the possibilities of estimating respiration rate using a physiological signal acquisition patch. The mean absolute error (MAE) is 0.11(about ±0.7 times in a minute), which is the least among similar studies that acquire respiratory rate from 3-axis accelerators or electrocardiogram.

Mobile phones offer the possibility to monitor and track health parameters. Our aim was to test the feasibility and accuracy of measuring beat-to-beat heart rate using smartphone accelerometers by recording the vibrations generated by the heart during its function and transmitted to the chest wall, i.e. the so-called seismocardiographic signal (SCG)., Methods: 9 healthy male volunteers were studied in supine (SUP) and in standing (ST) posture. A smartphone (iPhone6, Apple) was positioned on the thorax (POS1) to acquire SCG signal. While supine, a second smartphone was positioned on the navel (POS2). The SCG signal was recorded for 3 minutes during spontaneous respiration, synchronous with 3-leads ECG. Using a fully automated algorithm based on amplitude thresholding after rectification, the characteristic peak of the SCG signal (IVC) was detected and used to compute beat-to-beat heart duration, to be compared with the corresponding RR intervals extracted from the ECG., Results: A 100% feasibility of the approach resulted for POS1 in SUP, while 89% in POS2, and 78% for POS1 in ST. In supine, for each smartphones' position, the automated algorithm correctly identified the cardiac beats with >98% accuracy. Linear correlation (r2) with RR was very high (>0.98) in each posture and position, with no bias and narrow limits of agreement., Conclusions: The obtained results proved the feasibility of the proposed approach and the robustness of the applied algorithm in measuring the beat-to-beat heart rate from smartphone-derived SCG, with high accuracy compared to conventional ECG-derived measure.

Elderly fall detection based on accelerometers is an active research area. Nowadays authors are addressing specific problems such as failure rates and energy consumption, but in most cases their strategies do not conciliate these objectives. In this paper we propose a double threshold based methodology with two novel detection features, a product between the sum vector magnitude and the signal magnitude area, and a normalization of the signal magnitude area over five 1 s windows. The methodology was validated using the public Mobifall dataset, and one developed for this work. It achieved 99 % of accuracy with Mobifall, and 97 % with the self-developed dataset. This methodology is based on an activity by activity analysis performed for determining which activities are prone to fail, as an alternative way of reducing detection failures.

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.

Stress testing is used to measure the performance of the heart in an elevated stress state, in order to monitor or diagnose certain heart problems. Many measurements can be used to determine the performance of the heart, with the Tei index being the measurement of interest in this work. The Tei index has been used as a reliable method to evaluate systolic and diastolic performance, as it overcomes some limitations of the classical echocardiographic indices. It is calculated based on the time intervals derived from echocardiography. This paper presents an exploratory study, which uses an accelerometer to record mechanical events occurring in each cardiac cycle, also known as the seismocardiogram (SCG). From timing measurements corresponding to various events in the heart, a metric for myocardial performance is calculated based on the Tei index. The use of SCG in addition to ECG has the potential to provide further insights about the heart during stress testing, since the SCG quantifies mechanical actions of the heart.

Sufficient range of motion of the knee joint is necessary for performing many activities of daily living. Ambulatory monitoring of knee function can provide valuable information about progression of diseases like knee osteoarthritis and recovery after surgical interventions like total knee arthroplasty. In this paper, we describe a skin-mounted, conformal, accelerometer-based system for measuring knee angle and range of motion that does not require a skilled operator to apply devices. We establish the accuracy of this technique with respect to clinical gold standard goniometric measurements on a dataset collected from normative subjects during the performance of repeated bouts of knee flexion and extension tests. Results show that knee angle and range of motion estimates are highly correlated with goniometer measurements, and track differences in knee angle and range of motion to within 1%. These results demonstrate the ability of this system to characterize knee angle and range of motion, enabling future longitudinal monitoring of knee motion in naturalistic environments.

In this paper, a method for the assessment of the Unified Parkinson Disease Rating scale (UPDRS) related to tremor is presented. The method described consists of hand resting and posture state detection, tremor detection and tremor quantification based on accelerometer and gyroscope readings from a wrist worn sensor. The initial results on PD patient recordings on home environment indicate the feasibility of the proposed method in monitoring UPDRS tremor in patient home environment.

Energy expenditure (EE) estimation from accelerometer-based wearable sensors is important to generate accurate assessment of physical activity (PA) in individuals. Approaches hitherto have mainly focused on using accelerometer data and features extracted from these data to learn a regression model to predict EE directly. In this paper, we propose a novel framework for EE estimation based on statistical estimation theory. Given a test sequence of accelerometer data, the probability distribution on the PA classes is estimated by a classifier and these predictions are used to estimate EE. Experimental evaluation, performed on a large dataset of 152 subjects and 12 activity classes, demonstrates that EE can be estimated accurately using our framework.

Monitoring fetal wellbeing is key in modern obstetrics. While fetal movement is routinely used as a proxy to fetal wellbeing, accurate, noninvasive, long-term monitoring of fetal movement is challenging. A few accelerometer-based systems have been developed in the past few years, to tackle common issues in ultrasound measurement and enable remote, self-administrated monitoring of fetal movement during pregnancy. However, many questions remain unanswered to date on the optimal setup in terms of body-worn accelerometers as well as signal processing and machine learning techniques used to detect fetal movement. In this paper, we systematically analyze the trade-offs between sensor number and positioning, the presence of reference accelerometers outside of the abdominal area and provide guidelines on dealing with class imbalance. Using a dataset of 15 measurements collected employing 6 three-axial accelerometers we show that including a reference accelerometer on the back of the participant consistently improves fetal movement detection performance regardless of the number of sensors utilized. We also show that two accelerometers plus a reference accelerometer are sufficient for optimal results.

Heart activity, or at least heart rate variability, is associated with body position. Our previous studies have confirmed that impedance pneumography may be used to record respiratory function, but the calibration coefficients for this method depend on position. Data were collected from 24 students (12 male, 12 female), who alternated positions between lying (on front, back, and right side), sitting and standing. Signals from an attached iPhone's internal sensors (accelerometer, gyroscope, magnetometer) were recorded and attitude relative to gravity was calculated. The signals were subsequently segmented and marked. Five algorithms were trained and cross-validated for different sensor combinations. Without differentiation of sitting and standing, 100% accuracy was achieved using all algorithms. The classifier best differentiating these two states was based on random forests, with overall accuracy of 90%. Simple methods based on a proposed hybrid classifier were tested for online measurement without the need for signal segmentation, with 99% accuracy. The prospect of the algorithms' use in long-term studies (particularly cardiorespiratory monitoring) was assessed.

The ability to monitor arterial blood pressure continuously with unobtrusive body worn sensors may provide a unique and potentially valuable assessment of a patient's cardiovascular health. Pulse wave velocity (PWV) offers an attractive method to continuously monitoring blood pressure. However, PWV technologies based on timing measurements between the ECG and a distal PPG suffer from inaccuracies on mobile patients due to the confounding influence of pre-ejection period (PEP). In this paper, we presented a wearable, continuous blood pressure monitor (ViSi Mobile) that can measure and track changes in PEP. PEP is determined from precordial vibrations captured by an accelerometer coupled to the patient's sternum. The performance of the PEP measurements was evaluated on test subjects with postural change and patient activity. Results showed potential to improve cNIBP accuracy in active patients.

Any abnormal hypersynchronus activity of neurons can be characterized as an epileptic seizure (ES). A broad class of non-epileptic seizures is comprised of Psychogenic non-epileptic seizures (PNES). PNES are paroxysmal events, which mimics epileptic seizures and pose a diagnostic challenge with epileptic seizures due to their clinical similarities. The diagnosis of PNES is done using video-electroencephalography (VEM) monitoring. VEM being a resource intensive process calls for alternative methods for detection of PNES. There is now an emerging interest in the use of accelerometer based devices for the detection of seizures. In this work, we present an algorithm based on Gaussian mixture model (GMM's) for the identification of PNES, ES and normal movements using a wrist-worn accelerometer device. Features in time, frequency and wavelet domain are extracted from the norm of accelerometry signal. All events are then classified into three classes i.e normal, PNES and ES using a parametric estimate of the multivariate normal probability density function. An algorithm based on GMM's allows us to accurately model the non-epileptic and epileptic movements, thus enhancing the overall predictive accuracy of the system. The new algorithm was tested on data collected from 16 patients and showed an overall detection accuracy of 91% with 25 false alarms.

Manual palpation is still the gold standard for assessment of pulse presence during cardiopulmonary resuscitation (CPR) for professional rescuers. However, this method is unreliable, time-consuming and subjective. Therefore, reliable, quick and objectified assessment of pulse presence in cardiac arrest situations to assist professional rescuers is still an unmet need. Accelerometers may present a promising sensor modality as pulse palpation technology for which pulse detection at the carotid artery has been demonstrated to be feasible. This study extends previous work by presenting an algorithm for automatic, accelerometer-based pulse presence detection at the carotid site during CPR. We show that accelerometers might be helpful in automated detection of pulse presence during CPR.

In this study, we developed a sleep posture estimation algorithm using 3-axis accelerometer signals measured from a patch-type sensor. Firstly, we inspected the characteristics of accelerometer signals for different sleep postures. Based on the results, we established decision rules to estimate 5 postures containing supine, left, right lateral, prone postures, and non-sleep postures such as sitting and standing. The algorithm was tested by the data from thirteen subjects during night time PSG. As a result, the algorithm estimated sleep postures with an average agreement of 99.16%, and cohen's kappa of 0.98 compared with reference sleep postures determined by position sensor and video recording. The proposed method with the device could be used as supportive purpose in routine PSG study and out-of-hospital environment.

Freezing of gait (FOG) is a common motor impairment to suffer an inability to walk, experienced by Parkinson's disease (PD) patients. FOG interferes with daily activities and increases fall risk, which can cause severe health problems. We propose a novel smartphone-based system to detect FOG symptoms in an unconstrained way. The feasibility of single device to sense gait characteristic was tested on the various body positions such as ankle, trouser pocket, waist and chest pocket. Using measured data from accelerometer and gyroscope in the smartphone, machine learning algorithm was applied to classify freezing episodes from normal walking. The performance of AdaBoost.M1 classifier showed the best sensitivity of 86% at the waist, 84% and 81% in the trouser pocket and at the ankle respectively, which is comparable to the results of previous studies.

Gait analysis is an important diagnostic measure to investigate the pattern of walking. Traditional gait analysis is generally carried out in a gait lab, with equipped force and body tracking sensors, which needs a trained medical professional to interpret the results. This procedure is tedious, expensive, and unreliable and makes it difficult to track the progress across multiple visits. In this paper, we present a smart insole called FreeWalker, which provides quantitative gait analysis outside the confinement of traditional lab, at low- cost. The insole consists of eight pressure sensors and two motion tracking sensors, i.e. 3-axis accelerometer and 3-axis gyroscope. This enables measurement of under-foot pressure distribution and motion sequences in real-time. The insole is enabled with onboard SD card as well as wireless data transmission, which help in continuous gait-cycle analysis. The data is then sent to a gateway, for analysis and interpretation of data, using a user interface where gait features are graphically displayed. We also present validation result of a subject's left foot, who was asked to perform a specific task. Experiment results show that we could achieve a data-sampling rate of over 1 KHz, transmitting data up to a distance of 20 meter and maintain a battery life of around 24 hours. Taking advantage of these features, FreeWalker can be used in various applications, like medical diagnosis, rehabilitation, sports and entertainment.

Falls are a major health risk that diminish the quality of life among elderly people. Apart from falls themselves, most dramatic consequences are usually related with long lying periods that can cause serious side effects. These findings call for pervasive long-term fall detection systems able to automatically detect falls. In this paper, we propose an effective fall detection algorithm for mobile platforms. Using data retrieved from wearable sensors, such as Inertial Measurements Units (IMUs) and/or SmartPhones (SPs), our algorithm is able to detect falls using features extracted from accelerometer and gyroscope. While mostly of the mobile-based solutions for fall management deal only with accelerometer data, in the proposed approach we combine the instantaneous acceleration magnitude vector with changes of the user's heading in a Threshold Based Algorithm (TBA). In such a way, we were able to handle falls detection with minimal computational load, increasing the overall system accuracy with respect to traditional fall management methods. Experimental results show the strong detection performance of the proposed solution in discriminating between falls and typical Activities of Daily Living (ADLs) presenting fall-like acceleration patterns.

The magnetic distortions in indoor environment affects the accuracy of yaw angle estimation using magnetometer. Thus, the accuracy of indoor localization based on inertial-magnetic sensors will be affected as well. To address this issue, this paper proposes a magnetometer-free solution for indoor human localization and yaw angle estimation. The proposed algorithm fuses a wearable inertial sensor consisting of MEMS-based accelerometer and gyroscope with a portable ultra-wideband (UWB) localization system in a cascaded two-step filter consisting of a tilt Kalman filter and a localization Kalman filter. By benchmarking against an optical motion capture system, the experimental results show that the proposed algorithm can accurately track position and velocity as well as the yaw angle without using magnetometer.

Patients with concussions, strokes and neuromuscular disease such as Parkinson's disease, often have difficulties in keeping balance and suffer from abnormal gaits. Gait assessment conducted by a physician or therapist in clinics is standard clinical practice for assessing such injuries. However, this approach is subjective, leading to potential problems of unrepeatability, poor sensitivity and unreliability. To conduct the assessment in an objective way, a computer-based gait assessment system is designed and presented in this paper. The system performs assessments on dynamic balance and gaits by analyzing the skeleton frames of a subject captured by the Microsoft Kinect RGB-D sensor. Results show that the proposed system effectively scores subjects.

Optimum Experimental Design (OED) is an information gathering technique used to estimate parameters, which aims to minimize the variance of parameter estimation and prediction. In this paper, we further investigate an OED for MEMS accelerometer calibration of the 9-parameter auto-calibration model. Based on a linearized 9-parameter accelerometer model, we show the proposed OED is both G-optimal and rotatable, which are the desired properties for the calibration of wearable sensors for which only simple calibration devices are available. The experimental design is carried out with a newly developed wearable health monitoring device and desired experimental results have been achieved.

Tremor is a common symptom shared in both Parkinson's disease (PD) and Essential tremor (ET) subjects. The differential diagnosis of PD and ET tremor is important since the treatment depends on specific medication. A novel feature was developed based on a hypothesis stating that the tremor of PD subject has a larger fluctuation while performing resting task than action task. Tremor signal was collected using a gyroscope sensor attached to subject's finger. The angular velocity signal was analyzed by transforming a one-dimensional to two-dimensional signal based on relation of different units of time-delay. The tremor fluctuation was defined as the area of 95% confidence ellipse covering the two-dimensional signal. Experimenting with 32 PD and 20 ET subjects, a ratio of fluctuation of resting to kinetic task can be a sensitive feature to discriminate PD from ET with 100% accuracy.

The scope of our work focuses on investigating the potential use of the built-in accelerometer of the smartphones for the recording of the respiratory activity and deriving the respiratory rate. Five healthy subjects performed an inspiratory load protocol. The excursion of the right chest was recorded using the built-in triaxial accelerometer of a smartphone along the x, y and z axes and with an external uniaxial accelerometer. Simultaneously, the respiratory airflow and the inspiratory mouth pressure were recorded, as reference respiratory signals. The chest acceleration signal recorded in the z axis with the smartphone was denoised using a scheme based on the ensemble empirical mode decomposition, a noise data assisted method which decomposes nonstationary and nonlinear signals into intrinsic mode functions. To distinguish noisy oscillatory modes from the relevant modes we use the detrended fluctuation analysis. We reported a very strong correlation between the acceleration of the z axis of the smartphone and the reference accelerometer across the inspiratory load protocol (from 0.80 to 0.97). Furthermore, the evaluation of the respiratory rate showed a very strong correlation (0.98). A good agreement was observed between the respiratory rate estimated with the chest acceleration signal from the z axis of the smartphone and with the respiratory airflow signal: Bland-Altman limits of agreement between -1.44 and 1.46 breaths per minute with a mean bias of -0.01 breaths per minute. This preliminary study provides a valuable insight into the use of the smartphone and its built-in accelerometer for respiratory monitoring.

This study presents the design and implementation of a wearable wireless device, connected to a smart phone, which monitors and prevents repetitive ankle sprain due to chronic ankle instability (CAI). The device prevents this common foot injury by electrical stimulation of the peroneal muscles using surface electrodes which causes dorsiflexion of the foot. This is done after measuring ankle kinematics using inertial motion sensors and predicting ankle sprain. The prototype implemented here has a fast response time of 7 msec which enables prevention of ankle sprain before ligament damage occurs. Wireless communication between the components of the device, in addition to their small size, low cost and low power consumption, makes it unobtrusive, easy to wear and not hinder normal activities. The device connects via Bluetooth to an android smart phone application for continuous data logging and reporting to keep track of the incidences of possible ankle sprain and correction. This is a significant feature of this device since it enables monitoring of patients with CAI and quantifying progression of the condition or improvement in the case of treatment.

In this paper, we present an activity classification-based algorithm for the automatic detection of Levodopa Induced Dyskinesia in Parkinson's Disease (PD) patients. Two PD patients experiencing motor fluctuations related to chronic Levodopa therapy performed a protocol of simple daily life activities on at least two different occasions. A Random Forest classifier was able to classify the performed activities by the patients with an overall accuracy of 86%. Based on the detected activity, a K Nearest Neighbor classifier detected the presence of dyskinesia with accuracy ranging from 75% to 88%.

Health issues for elderly people may lead to different injuries obtained during simple activities of daily living (ADL). Potentially the most dangerous are unintentional falls that may be critical or even lethal to some patients due to the heavy injury risk. Many fall detection systems are proposed but only recently such health care systems became available. Nevertheless sensor design, accuracy as well as energy consumption efficiency can be improved. In this paper we present a single 3-axial accelerometer energy-efficient sensor system. Power saving is achieved by selective event processing triggered by fall detection procedure. The results in our simulations show 100% accuracy when the threshold parameters are chosen correctly. Estimated energy consumption seems to extend battery life significantly.

Activity monitor systems are increasing used recently. They are important for athletes and casual users to manage physical activity during daily exercises. In this paper, we use a triaxial accelerometer to design and implement an intelligent belt system, which can detect the user's step and flapping motion. In our system, a wearable intelligent belt is worn on the user's waist to collect activity acceleration signals. We present a step detection algorithm to detect real-time human step, which has high accuracy and low complexity. In our system, an Android App is developed to manage the intelligent belt. We also propose a protocol, which can guarantee data transmission between smartphones and wearable belt effectively and efficiently. In addition, when users flap the belt in emergency, the smartphone will receive alarm signal sending by the belt, and then notifies the emergency contact person, which can be really helpful for users in danger. Our experiment results show our system can detect physical activities with high accuracy (overall accuracy of our algorithm is above 95%) and has an effective alarm subsystem, which is significant for the practical use.

Upper limb coordination is necessary for the performance of activities of daily living (ADLs). This coordination is impaired in individuals suffering from motor deficits. The evolution of inter- and intra-limb coordination patterns may provide insight into mechanisms of recovery. In this paper, we investigate the ability of inertial sensors to capture measures of limb coordination in non-disabled individuals during the performance of ADL inspired tasks. We evaluate limb coordination as measured by time and frequency domain features extracted from inertial sensors for a subset of upper limb tasks, and evaluate the relative sensitivity of these measures to different task types.

Essential tremor (ET) is a highly prevalent movement disorder. Patients with ET exhibit a complex progressive and disabling tremor, and medical management often fails. Deep brain stimulation (DBS) has been successfully applied to this disorder, however there has been no quantifiable way to measure tremor severity or treatment efficacy in this patient population. The quantified amelioration of kinetic tremor via DBS is herein demonstrated through the application of a smartphone (iPhone) as a wireless accelerometer platform. The recorded acceleration signal can be obtained at a setting of the subject's convenience and conveyed by wireless transmission through the Internet for post-processing anywhere in the world. Further post-processing of the acceleration signal can be classified through a machine learning application, such as the support vector machine. Preliminary application of deep brain stimulation with a smartphone for acquisition of a feature set and machine learning for classification has been successfully applied. The support vector machine achieved 100% classification between deep brain stimulation in `on' and `off' mode based on the recording of an accelerometer signal through a smartphone as a wireless accelerometer platform.

We have proposed a novel solution to a fundamental problem encountered in implementing non-ingestion based medical adherence monitoring systems, namely, how to reliably identify pill medication intake. We show how wireless wearable devices with tri-axial accelerometer can be used to detect and classify hand gestures of users during solid-phase medication intake. Two devices were worn on the wrists of each user. Users were asked to perform two activities in the way that is natural and most comfortable to them: (1) taking empty gelatin capsules with water, and (2) drinking water and wiping mouth. 25 users participated in this study. The signals obtained from the devices were filtered and the patterns were identified using dynamic time warping algorithm. Using hand gesture signals, we achieved 84.17 percent true positive rate and 13.33 percent false alarm rate, thus demonstrating that the hand gestures could be used to effectively identify pill taking activity.

Remote monitoring of in-home activity and body posture provides useful physiological parameters for understanding patient status and for making clinical judgment in patients living with heart failure. The Corventis PiiX is a mobile chest-adherent device that provides continuously recorded accelerometer data from patients in an in-home setting. In particular, a 3-axis accelerometer enables quantification of full body activity and body posture changes. In this paper, the prevalence of sedentary activity with subject age, the population-wide distribution of upright body posture and resting body posture, and a diagnostic for reduced activity levels and resting body posture is presented with retrospective analysis on a large nonrandomized dataset.

Falls are a common and serious problem faced by older populations. There is a growing interest in estimating the risk of falling for older people using body-worn sensors and simple movement tasks, allowing appropriate fall prevention programs to be administered in a timely manner to the high risk population. This study investigated the capability and validity of using a waist-mounted triaxial accelerometer (TA) and a directed routine (DR) that includes three movement tasks to discriminate between fallers and non-fallers and between multiple fallers and non-multiple fallers. Data were collected from 98 subjects who were stratified into two separate groups, one for model training and the other for model validation. Logistic regression models were constructed using the TA features from the entire DR and from each single DR task, and were validated using unseen data. The best models were obtained using features from the alternate step test to classify between fallers and non-fallers with κ = 0.34-0.41, sensitivity = 68%-71% and specificity = 63%-73%. However, the overall validation performances were poor. The study emphasizes the importance of independent validation in fall prediction studies.

Numerous recent studies have aimed to improve myoelectric control of prostheses. However, the majority of these studies is characterized by two problems that could be easily fulfilled with recent resources supplied by the scientific literature. First, the majority of these studies use only intact subjects, with the unproved assumption that the results apply equally to amputees. Second, usually only electromyography data are used, despite other sensors (e.g., accelerometers) being easy to include into a real life prosthesis control system. In this paper we analyze the mentioned problems by the classification of 40 hand movements in 5 amputated and 40 intact subjects, using both sEMG and accelerometry data and applying several different state of the art methods. The datasets come from the NinaPro database, which supplies publicly available sEMG data to develop and test machine learning algorithms for prosthetics. The number of subjects can seem small at first sight, but it is not considering the literature of the field (which has to face the difficulty of recruiting trans-radial hand amputated subjects). Our results indicate that the maximum average classification accuracy for amputated subjects is 61.14%, which is just 15.86% less than intact subjects, and they show that intact subjects results can be used as proxy measure for amputated subjects. Finally, our comparison shows that accelerometry as a modality is less affected by amputation than electromyography, suggesting that real life prosthetics performance may easily be improved by inclusion of accelerometers.

Wrist-worn accelerometers are becoming more prevalent as a means to assess use of the impaired upper extremity in daily life after stroke. However, wrist accelerometry does not measure joint movements of the hand, which are integral to functional use of the upper extremity. In this study, we used a custom-built, non-obtrusive device called the manumeter to measure both arm use (via wrist accelerometry) and hand use (via finger magnetometry) of a group of unimpaired subjects while they performed twelve motor tasks at three intensities. We also gave the devices to four stroke subjects and asked them to wear them for six hours a day for one month. From the in-lab testing we found that arm use was a strong predictor of hand use for individual tasks, but that the slope of the relationship varied by up to a factor of ~12 depending on the task being performed. Consistent with this, in the daily use data collected from stroke subjects we found a broad spread in the relationship between arm and hand use. These results suggest that analyzing the spread of the relationship between daily hand and arm use will give more insight into upper extremity recovery than wrist accelerometry or finger magnetometry alone, because the spread reflects the nature of the daily tasks performed as well as the amount of upper extremity use.

Current generation smartphone can be seen as one of the most ubiquitous device for physical activity recognition. In this paper we proposed a physical activity recognizer to provide u-healthcare services in a cost effective manner by utilizing cloud computing infrastructure. Our model is comprised on embedded triaxial accelerometer of the smartphone to sense the body movements and a cloud server to store and process the sensory data for numerous kind of services. We compute the time and frequency domain features over the raw signals and evaluate different machine learning algorithms to identify an accurate activity recognition model for four kinds of physical activities (i.e., walking, running, cycling and hopping). During our experiments we found Support Vector Machine (SVM) algorithm outperforms for the aforementioned physical activities as compared to its counterparts. Furthermore, we also explain how smartphone application and cloud server communicate with each other.

Smartphone applications have been demonstrated for their capacity to measure gait in functionally autonomous environments beyond the limitations of a traditional gait laboratory. A software application enables the iPhone to function as a wireless accelerometer platform. The recorded acceleration of gait can be transmitted wirelessly as an email attachment through Internet connectivity. The objective of the research was to demonstrate the capacity of the smartphone to quantify gait features of a trans-tibial prosthesis. The iPhone a standard smartphone was mounted to the carbon fiber blade of the prosthesis through an adapter developed by a 3D printer. The application demonstrated considerable accuracy and reliability for the quantification of gait characteristics.

A feature of healthy gait is a clearly defined heel strike upon initial contact of the foot with the ground. However, a common consequence of ageing is deterioration of the heel first nature of gait towards a shuffling gait (flat foot at contact). Physiotherapy can be effective in correcting this but is costly and labour intensive. Gait rehabilitation could be accelerated with home exercise, guided by a biofeedback device that distinguishes between heel first and shuffling gait. This paper describes an algorithm that distinguishes between heel-to-toe gait and shuffling gait on the basis of angular velocity of the foot, using an inertial measurement unit. Measurements were made of normal and abnormal gait and used to develop an algorithm that distinguishes between good and bad steps. Results demonstrate very good algorithm performance, with a classification accuracy at the accuracy-optimal threshold of 92.7% when compared with physiotherapist labels. The sensitivity and specificity at this threshold are 84.4% and 97.5% respectively. These performance metrics suggest that this algorithm is usable in a biofeedback device.

Two algorithms for evaluating postural transitions (PTs) in cohorts of 40 healthy younger and 40 older adults are described and evaluated. The time of sit-to-stand (SiSt) and stand-to-sit (StSi) transitions and their duration were measured with two tri-axial accelerometers, one on the chest and one on the lower back. Each algorithm was optimized for these sensor placements. The first algorithm for sensor placement on the chest used a scalar product and vertical velocity estimates. The second algorithm for sensor placement on the lower back used a vector magnitude and a discrete wavelet transform. Both algorithms performed excellently in PT classification for younger and older adults (>86%). However, the chest based sensor and algorithm were better for estimating transition duration (TD) with ICCs to video analysis ranging from 0.678 to 0.969.

Recently, inertial sensors have been used to measure human movements for the purpose of rehabilitation. In evaluation of motor function for rehabilitation, reference data measured from healthy subjects with inertial sensors is needed. In addition, since lower limbs perform 3-dimensional movements during gait, it is needed to analyze 3-dimensional gait movements. The plot of 3-dimensional vector locus can be useful to understand 3-dimensional movements. The purposes of this paper were to show usefulness of the vector locus in understanding circumduction during gait, and to test an evaluation parameter for 3-dimensional movements during walking as reference data. Gait of 12 healthy subjects were measured, and then vector loci were plotted in the 3-dimensional space. Shape of the vector locus was evaluated by the width of each component as one of the reference data. It was suggested that the vector loci plotted in 3-dimensioinal space or those projected on horizontal plane were useful to understand circumduction during gait. It was also suggested that the width of vector locus was useful to evaluate differences of movement between subjects.

The aim of this study was to propose and comparatively evaluate four methods for assessing stride-by-stride changes of direction of progression, during straight walking using measurements of a magnetic and inertial unit placed above the malleolus. The four methods were evaluated by comparing their estimate of the gait changes of direction of progression with that obtained from an instrumented gait mat used as a gold standard. The methods were applied to the data obtained from the gait of both healthy subjects and patients with Huntington Disease, the latter characterized by a jerky swing phase. The results showed that the errors associated to the best estimates of the gait direction changes were about 10% of its range of variability for the healthy subjects and increased to about 30% for the patients, both walking at comfortable speed when the range of variability is the largest. Additional testing on gait at various radius of curvature should be carried out to fully validate the MIMU-based estimates.

Gait is a sensitive biomarker of decline in both cognitive and physical function. Therefore, the collection of gait data is an important feature of clinical assessments. Accelerometer-based body worn sensors are quickly becoming the preferred tool for assessing gait because they are small, useable in a wide variety of settings, offer more continuous spatio-temporal analysis and are inexpensive when compared with traditional gait assessment methodologies. The purpose of this study was to determine the validity and within test reliability of a low cost body worn movement sensor with associated algorithms to assess gait in a large group of older and younger healthy adults. We collected gait data over intermittent walks on an instrumented walkway for a within trial validation and also used the same accelerometer derived gait data for a within test reliability analysis. ICCs for validation and reliability were >0.756 and >0.965, respectively.

Estimating gait frequency is an important component in the detection and diagnosis of various medical conditions. Smartphone-based kinematic sensors offer a window of opportunity in free-living gait frequency estimation. The main issue with smartphone-based gait frequency estimation algorithms is how to adjust for variations in orientation and location of the phone on the human body. While numerous algorithms have been implemented to account for these differences, little work has been done in comparing these algorithms. In this study, we compare various position independent algorithms to determine which are more suited to robust gait frequency estimation. Using sensor data collected from volunteers walking with a smartphone, we examine the effect of using three different time series with the magnitude, weighted sum, and closest vertical component algorithms described in the paper. We also test two different methods of extracting step frequency: time domain peak counting and spectral analysis. The results show that the choice of time series does not significantly affect the accuracy of frequency measurements. Furthermore, both time domain and spectral approaches show comparable results. However, time domain approaches are sensitive to false-positives while spectral approaches require a minimum set of repetitive measurements. Our study suggests a hybrid approach where both time-domain and spectral approaches be used together to complement each other's shortcomings.

Humans are living longer but morbidity has also increased; threatening to create a serious global burden. Our approach is to monitor gait for early warning signs of morbidity. Here we present highlights from a series of experiments into gait as a potential biomarker for Parkinson's disease (PD), ageing and fall risk. Using body-worn accelerometers, we developed several novel camera-less methods to analyze head and pelvis movements while walking. Signal processing algorithms were developed to extract gait parameters that represented the principal components of vigor, head jerk, lateral harmonic stability, and oscillation range. The new gait parameters were compared to accidental falls, mental state and co-morbidities. We observed: 1) People with PD had significantly larger and uncontrolled anterioposterior (AP) oscillations of the head; 2) Older people walked with more lateral head jerk; and, 3) the combination of vigorous and harmonically stable gait was demonstrated by non-fallers. Our findings agree with research from other groups; changes in human gait reflect changes to well-being. We observed; different aspects of gait reflected different functional outcomes. The new gait parameters therefore may be complementary to existing methods and may have potential as biomarkers for specific disorders. However, further research is required to validate our observations, and establish clinical utility.

Dual-task gait allows assessment of impaired executive function and mobility control in older individuals, which are risk factors of falls. This study investigated gait changes in older individuals due to the addition of a cognitive load, using wearable pressure-sensing insole and tri-axial accelerometer measures. These wearable sensors can be applied at the point-of-care. Eleven elderly (65 years or older) individuals walked 7.62 m with and without a verbal fluency cognitive load task while wearing FScan 3000E pressure-sensing insoles in both shoes and a Gulf Coast X16-1C tri-axial accelerometer at the pelvis. Plantar-pressure derived parameters included center of force (CoF) path and temporal measures. Acceleration derived measures were descriptive statistics, Fast Fourier Transform quartile, ratio of even-to-odd harmonics, and maximum Lyapunov exponent. Stride time, stance time, and swing time all significantly increased during dual-task compared to single-task walking. Minimum, mean, and median CoF stance velocity; cadence; and vertical, anterior-posterior, and medial-lateral harmonic ratio all significantly decreased during dual-task walking. Wearable plantar pressure-sensing insole and lower back accelerometer derived-measures can identify gait differences between single-task and dual-task walking in older individuals and could be used in point-of-care environments to assess for deficits in executive function and mobility impairments.

Improving community mobility is a common goal for persons with stroke. Measuring daily physical activity is helpful to determine the effectiveness of rehabilitation interventions. In our previous studies, a novel wearable shoe-based sensor system (SmartShoe) was shown to be capable of accurately classify three major postures and activities (sitting, standing, and walking) from individuals with stroke by using Artificial Neural Network (ANN). In this study, we utilized decision tree algorithms to develop individual and group activity classification models for stroke patients. The data was acquired from 12 participants with stroke. For 3-class classification, the average accuracy was 99.1% with individual models and 91.5% with group models. Further, we extended the activities into 8 classes: sitting, standing, walking, cycling, stairs-up, stairs-down, wheel-chair-push, and wheel-chair-propel. The classification accuracy for individual models was 97.9%, and for group model was 80.2%, demonstrating feasibility of multi-class activity recognition by SmartShoe in stroke patients.

The seismocardiogram (SCG) signal traditionally measured using a chest-mounted accelerometer contains low-frequency (0-100 Hz) cardiac vibrations that can be used to derive diagnostically relevant information about cardiovascular and cardiopulmonary health. This work is aimed at investigating the effects of respiration on the frequency domain characteristics of SCG signals measured from 18 healthy subjects. Toward this end, the 0-100 Hz SCG signal bandwidth of interest was sub-divided into 5 Hz and 10 Hz frequency bins to compare the spectral energy in corresponding frequency bins of the SCG signal measured during three key conditions of respiration--inspiration, expiration, and apnea. Statistically significant differences were observed between the power in ensemble averaged inspiratory and expiratory SCG beats and between ensemble averaged inspiratory and apneaic beats across the 18 subjects for multiple frequency bins in the 10-40 Hz frequency range. Accordingly, the spectral analysis methods described in this paper could provide complementary and improved classification of respiratory modulations in the SCG signal over and above time-domain SCG analysis methods.

This paper presents a prototype of a wireless, point-of-care medical device to measure head impacts in contact or collision sports. The device is currently capable of measuring linear acceleration, time, and the duration of impact. The location of the impact can also be recorded by scaling the prototype design to multiple devices. An experimental apparatus was built to simulate head impacts and to verify the data from the device. Preliminary results show that the biomechanical measures from the device are sufficiently accurate.