Stephen J. Preece

Activity identification using body-mounted sensors--a review of classification techniques.

With the advent of miniaturized sensing technology, which can be body-worn, it is now possible to collect and store data on different aspects of human movement under the conditions of free living. This technology has the potential to be used in automated activity profiling systems which produce a continuous record of activity patterns over extended periods of time. Such activity profiling systems are dependent on classification algorithms which can effectively interpret body-worn sensor data and identify different activities. This article reviews the different techniques which have been used to classify normal activities and/or identify falls from body-worn sensor data. The review is structured according to the different analytical techniques and illustrates the variety of approaches which have previously been applied in this field. Although significant progress has been made in this important area, there is still significant scope for further work, particularly in the application of advanced classification techniques to problems involving many different activities.

A Comparison of Feature Extraction Methods for the Classification of Dynamic Activities From Accelerometer Data

Driven by the demands on healthcare resulting from the shift toward more sedentary lifestyles, considerable effort has been devoted to the monitoring and classification of human activity. In previous studies, various classification schemes and feature extraction methods have been used to identify different activities from a range of different datasets. In this paper, we present a comparison of 14 methods to extract classification features from accelerometer signals. These are based on the wavelet transform and other well-known time- and frequency-domain signal characteristics. To allow an objective comparison between the different features, we used two datasets of activities collected from 20 subjects. The first set comprised three commonly used activities, namely, level walking, stair ascent, and stair descent, and the second a total of eight activities. Furthermore, we compared the classification accuracy for each feature set across different combinations of three different accelerometer placements. The classification analysis has been performed with robust subject-based cross-validation methods using a nearest-neighbor classifier. The findings show that, although the wavelet transform approach can be used to characterize nonstationary signals, it does not perform as accurately as frequency-based features when classifying dynamic activities performed by healthy subjects. Overall, the best feature sets achieved over 95% intersubject classification accuracy.

Monte Carlo modelling of the spectral reflectance of the human eye

The interpretation of in vivo spectral reflectance measurements of the ocular fundus requires an accurate model of radiation transport within the eye. As well as considering the scattering and absorption processes, it is also necessary to account for appropriate histological variation. This variation results in experimentally measured spectra which vary, both with position in the eye, and between individuals. In this paper the results of a Monte Carlo simulation are presented. Three histological variables are considered: the RPE melanin concentration, the choriodal haemoglobin concentration and the choroidal melanin concentration. By considering these three variables, it is possible to generate model spectra which agree well with in vivo experimental measurements of the nasal fundus. The model has implications for the problem of extracting histological parameters from spectral reflectance measurements. These implications are discussed and a novel approach to interpretation of images of the ocular fundus suggested.

Large scale population assessment of physical activity using wrist worn accelerometers: The UK Biobank Study

Background Physical activity has not been objectively measured in prospective cohorts with sufficiently large numbers to reliably detect associations with multiple health outcomes. Technological advances now make this possible. We describe the methods used to collect and analyse accelerometer measured physical activity in over 100,000 participants of the UK Biobank study, and report variation by age, sex, day, time of day, and season. Methods Participants were approached by email to wear a wrist-worn accelerometer for seven days that was posted to them. Physical activity information was extracted from 100Hz raw triaxial acceleration data after calibration, removal of gravity and sensor noise, and identification of wear / non-wear episodes. We report age- and sex-specific wear-time compliance and accelerometer measured physical activity, overall and by hour-of-day, week-weekend day and season. Results 103,712 datasets were received (44.8% response), with a median wear-time of 6.9 days (IQR:6.5–7.0). 96,600 participants (93.3%) provided valid data for physical activity analyses. Vector magnitude, a proxy for overall physical activity, was 7.5% (2.35mg) lower per decade of age (Cohen’s d = 0.9). Women had a higher vector magnitude than men, apart from those aged 45-54yrs. There were major differences in vector magnitude by time of day (d = 0.66). Vector magnitude differences between week and weekend days (d = 0.12 for men, d = 0.09 for women) and between seasons (d = 0.27 for men, d = 0.15 for women) were small. Conclusions It is feasible to collect and analyse objective physical activity data in large studies. The summary measure of overall physical activity is lower in older participants and age-related differences in activity are most prominent in the afternoon and evening. This work lays the foundation for studies of physical activity and its health consequences. Our summary variables are part of the UK Biobank dataset and can be used by researchers as exposures, confounding factors or outcome variables in future analyses.

Spectral filter optimization for the recovery of parameters which describe human skin

The paper presents a method for finding spectral filters that minimize the error associated with histological parameters characterizing normal skin tissue. These parameters can be recovered from digital images of the skin using a physics-based model of skin coloration. The relationship between the image data and histological parameter values is defined as a mapping function from the image space to the parameter space. The accuracy of this function is determined by the choice of optical filters. An optimization criterion for finding the optimal filters is defined by combing methodology from differential geometry with statistical error analysis. It is shown that the magnitude of errors associated with the optimal filters is typically half of that for typical RGB filters on a three-parameter model of human skin coloration. Finally, other medical image applications are identified to which this generic methodology could be applied.

An Inverse Method for the Recovery of Tissue Parameters from Colour Images

The interpretation of colour images is presented as an inverse problem in which a mapping is sought between image colour vectors and the physiological parameters characterizing a tissue. To ensure the necessary one-to-one correspondence between the image colours and the parameters, the mapping must be unique. This can be established through testing the sign of the determinant of the Jacobian matrix, a multi-dimensional equivalent of a discrete derivative, over the space of all parameter values. Furthermore, an optimisation procedure is employed to find the set of filters for image capture which generate image vectors minimizing the mapping error. This methodology applied to interpretation of skin images shows that the standard RGB system of filters provides for a unique mapping between image values and parameters characterizing the normal skin. It is further shown that an optimal set of filters reduces the error of quantification by a factor of 2, on average.

Variation in Pelvic Morphology May Prevent the Identification of Anterior Pelvic Tilt

Abstract Pelvic tilt is often quantified using the angle between the horizontal and a line connecting the anterior superior iliac spine (ASIS) and the posterior superior iliac spine (PSIS). Although this angle is determined by the balance of muscular and ligamentous forces acting between the pelvis and adjacent segments, it could also be influenced by variations in pelvic morphology. The primary objective of this anatomical study was to establish how such variation may affect the ASIS-PSIS measure of pelvic tilt. In addition, we also investigated how variability in pelvic landmarks may influence measures of innominate rotational asymmetry and measures of pelvic height. Thirty cadaver pelves were used for the study. Each specimen was positioned in a fixed anatomical reference position and the angle between the ASIS and PSIS measured bilaterally. In addition, side-to-side differences in the height of the innominate bone were recorded. The study found a range of values for the ASIS-PSIS of 0–23 degrees, with a mean of 13 and standard deviation of 5 degrees. Asymmetry of pelvic landmarks resulted in side-to-side differences of up to 11 degrees in ASIS-PSIS tilt and 16 millimeters in innominate height. These results suggest that variations in pelvic morphology may significantly influence measures of pelvic tilt and innominate rotational asymmetry.

Automatic identification of gait events using an instrumented sock

BackgroundTextile-based transducers are an emerging technology in which piezo-resistive properties of materials are used to measure an applied strain. By incorporating these sensors into a sock, this technology offers the potential to detect critical events during the stance phase of the gait cycle. This could prove useful in several applications, such as functional electrical stimulation (FES) systems to assist gait.MethodsWe investigated the output of a knitted resistive strain sensor during walking and sought to determine the degree of similarity between the sensor output and the ankle angle in the sagittal plane. In addition, we investigated whether it would be possible to predict three key gait events, heel strike, heel lift and toe off, with a relatively straight-forward algorithm. This worked by predicting gait events to occur at fixed time offsets from specific peaks in the sensor signal.ResultsOur results showed that, for all subjects, the sensor output exhibited the same general characteristics as the ankle joint angle. However, there were large between-subjects differences in the degree of similarity between the two curves. Despite this variability, it was possible to accurately predict gait events using a simple algorithm. This algorithm displayed high levels of trial-to-trial repeatability.ConclusionsThis study demonstrates the potential of using textile-based transducers in future devices that provide active gait assistance.

Effect of rocker shoe design features on forefoot plantar pressures in people with and without diabetes

BACKGROUND There is no consensus on the precise rocker shoe outsole design that will optimally reduce plantar pressure in people with diabetes. This study aimed to understand how peak plantar pressure is influenced by systematically varying three design features which characterise a curved rocker shoe: apex angle, apex position and rocker angle. METHODS A total of 12 different rocker shoe designs, spanning a range of each of the three design features, were tested in 24 people with diabetes and 24 healthy participants. Each subject also wore a flexible control shoe. Peak plantar pressure, in four anatomical regions, was recorded for each of the 13 shoes during walking at a controlled speed. FINDINGS There were a number of significant main effects for each of the three design features, however, the precise effect of each feature varied between the different regions. The results demonstrated maximum pressure reduction in the 2nd-4th metatarsal regions (39%) but that lower rocker angles (<20°) and anterior apex positions (>60% shoe length) should be avoided for this region. The effect of apex angle was most pronounced in the 1st metatarsophalangeal region with a clear decrease in pressure as the apex angle was increased to 100°. INTERPRETATION We suggest that an outsole design with a 95° apex angle, apex position at 60% of shoe length and 20° rocker angle may achieve an optimal balance for offloading different regions of the forefoot. However, future studies incorporating additional design feature combinations, on high risk patients, are required to make definitive recommendations.

A comparison of kinematic algorithms to estimate gait events during overground running

The gait cycle is frequently divided into two distinct phases, stance and swing, which can be accurately determined from ground reaction force data. In the absence of such data, kinematic algorithms can be used to estimate footstrike and toe-off. The performance of previously published algorithms is not consistent between studies. Furthermore, previous algorithms have not been tested at higher running speeds nor used to estimate ground contact times. Therefore the purpose of this study was to both develop a new, custom-designed, event detection algorithm and compare its performance with four previously tested algorithms at higher running speeds. Kinematic and force data were collected on twenty runners during overground running at 5.6m/s. The five algorithms were then implemented and estimated times for footstrike, toe-off and contact time were compared to ground reaction force data. There were large differences in the performance of each algorithm. The custom-designed algorithm provided the most accurate estimation of footstrike (True Error 1.2 ± 17.1 ms) and contact time (True Error 3.5 ± 18.2 ms). Compared to the other tested algorithms, the custom-designed algorithm provided an accurate estimation of footstrike and toe-off across different footstrike patterns. The custom-designed algorithm provides a simple but effective method to accurately estimate footstrike, toe-off and contact time from kinematic data.