Machine Learning Techniques for Automated Tremor Detection in the Presence of External Stressors

In this study tremor data of 25 subjects (Senile tremor = 5, Alcohol induced tremor = 9, Healthy individuals = 11) were collected using a wearable device consisting of five Inertial Measuring Units (IMUs) and an embedded optical sensor. The subjects were made to draw the Archimedes spiral under the influence of external stressors. Features were extracted from measured acceleration data and also from an optical sensor. Using the selected features few supervised machined learning algorithms were explored for automatic classification of tremor. Performance matrix used to evaluate the classifier was accuracy, recall, and precision. It is observed that the algorithms are able to accurately classify healthy, senile tremor and alcohol induced tremor.

Daily Activity Recognition and Tremor Quantification from Accelerometer Data for Patients with Essential Tremor Using Stacked Denoising Autoencoders

Human activity recognition (HAR) has received more and more attention, which is able to play an important role in many fields, such as healthcare and intelligent home. Thus, we have discussed an application of activity recognition in the healthcare field in this paper. Essential tremor (ET) is a common neurological disorder that can make people with this disease rise involuntary tremor. Nowadays, the disease is easy to be misdiagnosed as other diseases. We have combined the essential tremor and activity recognition to recognize ET patients’ activities and evaluate the degree of ET for providing an auxiliary analysis toward disease diagnosis by utilizing stacked denoising autoencoder (SDAE) model. Meanwhile, it is difficult for model to learn enough useful features due to the small behavior dataset from ET patients. Thus, resampling techniques are proposed to alleviate small sample size and imbalanced samples problems. In our experiment, 20 patients with ET and 5 healthy people have been chosen to collect their acceleration data for activity recognition. The experimental results show the significant result on ET patients activity recognition and the SDAE model has achieved an overall accuracy of 93.33%. What’s more, this model is also used to evaluate the degree of ET and has achieved the accuracy of 95.74%. According to a set of experiments, the model we used is able to acquire significant performance on ET patients activity recognition and degree of tremor assessment.

Detailed analysis and comparison of different activity metrics

Actigraphic measurements are an important part of research in different disciplines, yet the procedure of determining activity values is unexpectedly not standardized in the literature. Although the measured raw acceleration signal can be diversely processed, and then the activity values can be calculated by different activity calculation methods, the documentations of them are generally incomplete or vary by manufacturer. These numerous activity metrics may require different types of preprocessing of the acceleration signal. For example, digital filtering of the acceleration signals can have various parameters; moreover, both the filter and the activity metrics can also be applied per axis or on the magnitudes of the acceleration vector. Level crossing-based activity metrics also depend on threshold level values, yet the determination of their exact values is unclear as well. Due to the serious inconsistency of determining activity values, we created a detailed and comprehensive comparison of the different available activity calculation procedures because, up to the present, it was lacking in the literature. We assessed the different methods by analysing the triaxial acceleration signals measured during a 10-day movement of 42 subjects. We calculated 148 different activity signals for each subject’s movement using the combinations of various types of preprocessing and 7 different activity metrics applied on both axial and magnitude data. We determined the strength of the linear relationship between the metrics by correlation analysis, while we also examined the effects of the preprocessing steps. Moreover, we established that the standard deviation of the data series can be used as an appropriate, adaptive and generalized threshold level for the level intersection-based metrics. On the basis of these results, our work also serves as a general guide on how to proceed if one wants to determine activity from the raw acceleration data. All of the analysed raw acceleration signals are also publicly available.

The Dilemma of Analyzing Physical Activity and Sedentary Behavior with Wrist Accelerometer Data: Challenges and Opportunities

Physical behaviors (e.g., physical activity and sedentary behavior) have been the focus among many researchers in the biomedical and behavioral science fields. The recent shift from hip- to wrist-worn accelerometers in these fields has signaled the need to develop novel approaches to process raw acceleration data of physical activity and sedentary behavior. However, there is currently no consensus regarding the best practices for analyzing wrist-worn accelerometer data to accurately predict individuals’ energy expenditure and the times spent in different intensities of free-living physical activity and sedentary behavior. To this end, accurately analyzing and interpreting wrist-worn accelerometer data has become a major challenge facing many clinicians and researchers. In response, this paper attempts to review different methodologies for analyzing wrist-worn accelerometer data and offer cutting edge, yet appropriate analysis plans for wrist-worn accelerometer data in the assessment of physical behavior. In this paper, we first discuss the fundamentals of wrist-worn accelerometer data, followed by various methods of processing these data (e.g., cut points, steps per minute, machine learning), and then we discuss the opportunities, challenges, and directions for future studies in this area of inquiry. This is the most comprehensive review paper to date regarding the analysis and interpretation of free-living physical activity data derived from wrist-worn accelerometers, aiming to help establish a blueprint for processing wrist-derived accelerometer data.

Temporal-Sound based User Interface for Smart Home

We propose a gesture-based interface to control a smart home. Our system replaces existing physical controls with our temporal sound commands using accelerometer. In our preliminary experiments, we recorded the sounds generated by six different gestures (knocking the desk, mouse clicking, and clapping) and converted them into spectrogram images. Classification learning was performed on these images using a CNN. Due to the difference between the microphones used, the classification results are not successful for most of the data. We then recorded acceleration values, instead of sounds, using a smart watch. 5 types of motions were performed in our experiments to execute activity classification on these acceleration data using a machine learning library named Core ML provided by Apple Inc.. These results still have much room to be improved.

CARL: a running recognition algorithm for free-living accelerometer data

Wearable accelerometers hold great promise for physical activity epidemiology and sports biomechanists. However, identifying and extracting data from specific physical activities, such as running, remains challenging. Objective: To develop and validate an algorithm to identify bouts of running in raw, free-living accelerometer data from devices worn at the wrist or torso (waist, hip, chest). Approach: The CARL (continuous amplitude running logistic) classifier identifies acceleration data with amplitude and frequency characteristics consistent with running. The CARL classifier was trained on data from 31 adults wearing accelerometers on the waist and wrist, then validated on free-living data from 30 new, unseen subjects plus 166 subjects from previously-published datasets using different devices, wear locations, and sample frequencies. Main Results: On free-living data, the CARL classifier achieved mean accuracy (F1 score) of 0.984 (95% confidence interval 0.962-0.996) for data from the waist and 0.994 (95% CI 0.991-0.996) for data from the wrist. In previously-published datasets, the CARL classifier identified running with mean accuracy (F1 score) of 0.861 (95% CI 0.836-0.884) for data from the chest, 0.911 (95% CI 0.884-0.937) for data from the hip, 0.916 (95% CI 0.877-0.948) for data from the waist, and 0.870 (95% CI 0.834-0.903) for data from the wrist. Misclassification primarily occurred during activities with similar torso acceleration profiles to running, such as rope jumping and elliptical machine use. Significance: The CARL classifier can accurately identify bouts of running as short as three seconds in free-living accelerometry data. An open-source implementation of the CARL classifier is available at <<GITHUBURL>>.

Assessment of Perception on Disaster Proneness of Lembang Fault in District of Cisarua, West Java Indonesia

The Lembang Fault is a major fault located at the northern Bandung. This fault has a high disaster risk, including ground shaking, surface rupture, and possible landslides or liquefaction. This fault can cause earthquakes of 6.5-7 magnitude, making 8 million people in four Regencies and Cities around West Bandung Regency, Cimahi City, Bandung City and Bandung Regency exposed to major disaster risk. This research focuses on assessing the Perception of Disaster Proneness of the Lembang Fault in the District of Cisarua, West Java, Indonesia. This research was conducted using a case study and deductive-qualitative approach. In addition, this research was carried out by combining engineering and social research methodologies. The survey location point is determined based on hazard data (Peak Ground Acceleration data), vulnerability data (covering building density, slope, curvature, soil character, distance from faults, etc.) and population density data. This study indicates that the public’s perception of the disaster in the Lembang Fault is very subjective. How they act is based on experience or based on their beliefs. Therefore, an essential part of this research is assessing and measuring the community’s perception of the Lembang Fault towards disasters that may arise. The government must make serious efforts to convey that the disaster in the Lembang fault is much bigger and can happen at any time. Therefore, building resilient communities that genuinely understand the dangers of living in disaster-prone areas is essential.

The effect of horseshoes and surfaces on horse and jockey centre of mass displacements at gallop

Horseshoes influence how horses’ hooves interact with different ground surfaces, during the impact, loading and push-off phases of a stride cycle. Consequently, they impact on the biomechanics of horses’ proximal limb segments and upper body. By implication, different shoe and surface combinations could drive changes in the magnitude and stability of movement patterns in horse-jockey dyads. This study aimed to quantify centre of mass (COM) displacements in horse-jockey dyads galloping on turf and artificial tracks in four shoeing conditions: 1) aluminium; 2) barefoot; 3) GluShu; and 4) steel. Thirteen retired racehorses and two jockeys at the British Racing School were recruited for this intervention study. Tri-axial acceleration data were collected close to the COM for the horse (girth) and jockey (kidney-belt), using iPhones (Apple Inc.) equipped with an iOS app (SensorLog, sample rate = 50 Hz). Shoe-surface combinations were tested in a randomized order and horse-jockey pairings remained constant. Tri-axial acceleration data from gallop runs were filtered using bandpass Butterworth filters with cut-off frequencies of 15 Hz and 1 Hz, then integrated for displacement using Matlab. Peak displacement was assessed in both directions (positive ‘maxima’, negative ‘minima’) along the cranio-caudal (CC, positive = forwards), medio-lateral (ML, positive = right) and dorso-ventral (DV, positive = up) axes for all strides with frequency ≥2 Hz (mean = 2.06 Hz). Linear mixed-models determined whether surfaces, shoes or shoe-surface interactions (fixed factors) significantly affected the displacement patterns observed, with day, run and horse-jockey pairs included as random factors; significance was set at p<0.05. Data indicated that surface-type significantly affected peak COM displacements in all directions for the horse (p<0.0005) and for all directions (p≤0.008) but forwards in the jockey. The largest differences were observed in the DV-axis, with an additional 5.7 mm and 2.5 mm of downwards displacement for the horse and jockey, respectively, on the artificial surface. Shoeing condition significantly affected all displacement parameters except ML-axis minima for the horse (p≤0.007), and all displacement parameters for the jockey (p<0.0005). Absolute differences were again largest vertically, with notable similarities amongst displacements from barefoot and aluminium trials compared to GluShu and steel. Shoe-surface interactions affected all but CC-axis minima for the jockey (p≤0.002), but only the ML-axis minima and maxima and DV-axis maxima for the horse (p≤0.008). The results support the idea that hoof-surface interface interventions can significantly affect horse and jockey upper-body displacements. Greater sink of hooves on impact, combined with increased push-off during the propulsive phase, could explain the higher vertical displacements on the artificial track. Variations in distal limb mass associated with shoe-type may drive compensatory COM displacements to minimize the energetic cost of movement. The artificial surface and steel shoes provoked the least CC-axis movement of the jockey, so may promote greatest stability. However, differences between horse and jockey mean displacements indicated DV-axis and CC-axis offsets with compensatory increases and decreases, suggesting the dyad might operate within displacement limits to maintain stability. Further work is needed to relate COM displacements to hoof kinematics and to determine whether there is an optimum configuration of COM displacement to optimise performance and minimise injury.

Identification of Soft Fall based on Falling State Occurrences

Fall detection techniques for helping the elderly were developed based on identifying falling states using simulated falls. However, some real-life falling states were left undetected, which led to this work on analysing falling states. The aim was to find the differences between active daily living and soft falls where falling states were undetected. This is the first consideration to be based on the threshold-based algorithms using the acceleration data stored in an activity database. This study addresses soft falls in addition to the general falls based on two falling states. Despite the number of false alarms being higher rising from 18.5% to 56.5%, the sensitivity was increased from 52% to 92.5% for general falls, and from 56% to 86% for soft falls. Our experimental results show the importance of state occurrence for soft fall detection, and will be used to build a learning model for soft fall detection.

Integrating acceleration signal processing and image segmentation for condition assessment of asphalt roads

A proactive road maintenance system enables agencies to better allocate resources to manage their road networks. An inventory of the roads’ conditions is an essential component of such maintenance program. This research project proposes a hybrid system to asses the condition of the asphalt roads, which uses a dashboard-mounted smartphone to simultaneously collect the acceleration response of a vehicle and the video footage of the road surface while driving. The system analyzes acceleration data for anomalous events that could indicate a defect. Then the computer vision module of the system applies semantic segmentation in the corresponding frame to the detected anomaly to identify defects. This system demonstrated 84% recall and 88% precision rates in detection of anomalies in two road segments. Despite these promising results, the system can only detect the defects that are passed over and it could miss some defects with small acceleration responses, such as traverse cracks.