High-resolution Accelerometer Research Articles

Africa's overlooked top predator: Towards a better understanding of martial eagle feeding ecology in the Maasai Mara, Kenya

Raptors exert top‐down influences on ecosystems via their effects on prey population dynamics and community composition. Most raptors are sympatric with other predators, thus complicating our understanding of their relative influence in these systems. Estimates of kill rates and prey biomass recycling have been used as predation metrics that allow quantitative comparison among species and assessment of the relative role of single species within complex food webs. Few studies have produced findings of kill rates or prey biomass recycling for raptors. We used a supervised machine learning algorithm to behaviourally classify high resolution accelerometer informed GPS locations of tagged adult non‐breeding martial eagles Polemaetus bellicosus in the Maasai Mara region of Kenya to estimate kill rates and prey biomass recycling. Eagle locations classified as feeding were clustered using distance and time thresholds to identify kills and calculate kill rates. Identified kill sites were quickly ground‐truthed to confirm kills and identify prey species. We estimated kill rates for martial eagles at 0.59 kills day‐1 for males and 0.38 kills day‐1 for females, and we estimated biomass recycling per ground‐truthed kill at 1796 g for males and 3860 g for females. From our sample of identified ground‐truthed kills, ‘gamebirds' was the most frequently recorded prey category for male eagles and ‘small ungulates' was the most frequently recorded prey category for female eagles. These results position martial eagles close to sympatric mammalian top predators in trophic pyramids and provide evidence for their classification as a top predator.

Non-contact wearable synchronous measurement method of electrocardiogram and seismocardiogram signals.

Cardiovascular disease is one of the leading threats to human lives and its fatality rate still rises gradually year by year. Driven by the development of advanced information technologies, such as big data, cloud computing, and artificial intelligence, remote/distributed cardiac healthcare is presenting a promising future. The traditional dynamic cardiac health monitoring method based on electrocardiogram (ECG) signals only has obvious deficiencies in comfortableness, informativeness, and accuracy under motion state. Therefore, a non-contact, compact, wearable, synchronous ECG and seismocardiogram (SCG) measuring system, based on a pair of capacitance coupling electrodes with ultra-high input impedance, and a high-resolution accelerometer were developed in this work, which can collect the ECG and SCG signals at the same point simultaneously through the multi-layer cloth. Meanwhile, the driven right leg electrode for ECG measurement is replaced by the AgCl fabric sewn to the outside of the cloth for realizing the total gel-free ECG measurement. Besides, synchronous ECG and SCG signals at multiple points on the chest surface were measured, and the recommended measuring points were given by their amplitude characteristics and the timing sequence correspondence analysis. Finally, the empirical mode decomposition algorithm was used to adaptively filter the motion artifacts within the ECG and SCG signals for measuring performance enhancement under motion states. The results demonstrate that the proposed non-contact, wearable cardiac health monitoring system can effectively collect ECG and SCG synchronously under various measuring situations.

Telemetry and Accelerometer Tracking of Green Toads in an Urban Habitat: Methodological Notes and Preliminary Findings.

Advancements in tracking technologies provide an increasingly important tool in animal monitoring and conservation that can describe animal spatial behavior in native habitats and uncover migratory routes that otherwise may be difficult or impossible to map. In addition, high-resolution accelerometer sensors provide powerful insights into animal activity patterns and can help to identify specific behaviors from accelerometer profiles alone. Previously, such accelerometers were restricted to larger animals due to size and mass constraints. However, recent advances make it possible to use such devices on smaller animals such as the European green toad (Bufotes viridis), the focus of our current study. We deploy custom made tracking devices, that consist of very-high-frequency transmitters and tri-axial accelerometers, to track toads in their native urban environment in Vienna (Austria). A total of nine toads were tracked, ranging from three to nine tracking days per individual during the post-breeding season period. We demonstrate that our devices could reliably monitor toad movement and activity during the observation period. Hence, we confirmed the predominantly nocturnal activity patterns and recorded low overall movement at this urban site. Accelerometer data revealed that toads exhibited brief but intense activity bursts between 10 pm and midnight, resting periods during the night and intermittent activity during the day. Positional tracking alone would have missed the major activity events as they rarely resulted in large positional displacements. This underscores the importance of and value in integrating multiple tracking sensors for studies of movement ecology. Our approach could be adapted for other amphibians or other animals with mass constraints and may become standard monitoring equipment in the near future.

Experimental analysis of the performance of a new shipboard gravity gradient measurement system

The gravity gradient tensor, which has a higher resolution than gravity, is used in a variety of fields, including the discovery of energy resources, auxiliary navigation, and national defense building. Our team has achieved significant advancements in various essential technologies, such as high-resolution accelerometers, and has constructed China’s first self-controllable shipboard gravity gradient measurement system. In the laboratory, accuracy is determined using the mass gravitation technique, static test accuracy of Tuv and Txy is 7.22 E and 3.58 E, while dynamic test accuracy of Tuv and Txy is 9.09 E and 4.16 E. For outfield shipborne test measurement, the internal accord accuracy of Tuv and Txy of the repeat line is 28.2E@750m and 28.8E@750m, and that of the intersection point is 28.2E@750m and 26.8E@750m. The performance of the system is completely validated by dynamic and static testing, laying the groundwork for the practical implementation of gravity gradient technology.

An accelerometer-derived ballistocardiogram method for detecting heart rate in free-ranging marine mammals.

ABSTRACTPhysio-logging methods, which use animal-borne devices to record physiological variables, are entering a new era driven by advances in sensor development. However, existing datasets collected with traditional bio-loggers, such as accelerometers, still contain untapped eco-physiological information. Here, we present a computational method for extracting heart rate from high-resolution accelerometer data using a ballistocardiogram. We validated our method with simultaneous accelerometer–electrocardiogram tag deployments in a controlled setting on a killer whale (Orcinus orca) and demonstrate the predictions correspond with previously observed cardiovascular patterns in a blue whale (Balaenoptera musculus), including the magnitude of apneic bradycardia and increase in heart rate prior to and during ascent. Our ballistocardiogram method may be applied to mine heart rates from previously collected accelerometery data and expand our understanding of comparative cardiovascular physiology.

Measuring the impact: new insights into flood-borne large wood collisions with river structures using an isolated sensor-unit

Large Wood (LW) transported during floods or channelized mass flows poses a high risk for engineered structures, often leading to significant damage or total failure of the impacted structure. To date little is known about impact magnitudes caused by LW collisions. To better control for such interactions, a better understanding of transport dynamics and impact forces is required. The present laboratory study employs state-of-the-art sensor units installed in scaled logs to capture acceleration data from collisions of waterborne LW with 2 in-stream structures—bridge pier and retention structure—each providing different examples of rigid engineered systems. Through precise measurements of acceleration and impact duration (stopping time), the resultant impact forces of LW collisions can be calculated. Here, for the first time, impact forces were quantified in a scaled stream environment based on the inertial frame of the object causing the impact, rather than the more commonly used instrumented structure approach. High-resolution accelerometer measurements were compared to conventional analytical (force balance) approaches. They revealed the need for accurate inertia measurements to appropriately account for prevailing hydraulic flow conditions and the effects of LW interactions in fluvial environments. Although log velocity and stopping time are crucial parameters for assessing LW impact forces, accurate measurements are still elusive due to limitations in available sensing techniques. By presenting proof-of-concept results, this study contributes to an improved understanding of LW impact forces during floods. Based on these encouraging results, we recommend more sensor-based field studies in future, needed for the design of resilient structures.

Application of Advanced Multi-Sensor Non-Destructive Testing System for the Evaluation of Pavements Affected by Transverse Crack-Heaving

Pavement tenting, also referred to as crack-heaving, is a distress condition that primarily affects bituminous roads constructed in cold climates. This type of distress spreads over long stretches of roadways and can drastically affect drivers’ safety and comfort. The phenomenon occurs in freezing winter temperatures offering a limited and dire time window for testing. This paper discusses using an integrated multi-sensor non-destructive testing methodology to evaluate and characterize pavements affected by tenting. A survey van equipped with high-definition video and thermal cameras, LIDAR laser scanner, high-resolution accelerometer, and ground-penetrating radar (GPR) technologies was used to assess several roads suspected of tenting. The plurality of measuring devices and the data fusion and synchronization capabilities proved useful in revealing important pavement tenting characteristics that would have been otherwise overlooked. The data analysis led to the development of test parameters, derived from longitudinal profile measurements, that captured reasonably well the intensity and frequency of the tented cracks. The parameters were successfully employed to characterize the tested roads and determine the extent of critically affected segments. The study also showed the potential of GPR measurements to investigate underneath moisture conditions contributing to the formation of the tented cracks. Finally, the findings and tools developed in this study were discussed and compared with observations of local engineers who have extensive experience and insight on the subject matter. The knowledge and recommendations gathered in this final effort were also synthesized and incorporated into the paper.

Behaviour Classification on Giraffes (Giraffa camelopardalis) Using Machine Learning Algorithms on Triaxial Acceleration Data of Two Commonly Used GPS Devices and Its Possible Application for Their Management and Conservation.

Averting today’s loss of biodiversity and ecosystem services can be achieved through conservation efforts, especially of keystone species. Giraffes (Giraffa camelopardalis) play an important role in sustaining Africa’s ecosystems, but are ‘vulnerable’ according to the IUCN Red List since 2016. Monitoring an animal’s behavior in the wild helps to develop and assess their conservation management. One mechanism for remote tracking of wildlife behavior is to attach accelerometers to animals to record their body movement. We tested two different commercially available high-resolution accelerometers, e-obs and Africa Wildlife Tracking (AWT), attached to the top of the heads of three captive giraffes and analyzed the accuracy of automatic behavior classifications, focused on the Random Forests algorithm. For both accelerometers, behaviors of lower variety in head and neck movements could be better predicted (i.e., feeding above eye level, mean prediction accuracy e-obs/AWT: 97.6%/99.7%; drinking: 96.7%/97.0%) than those with a higher variety of body postures (such as standing: 90.7–91.0%/75.2–76.7%; rumination: 89.6–91.6%/53.5–86.5%). Nonetheless both devices come with limitations and especially the AWT needs technological adaptations before applying it on animals in the wild. Nevertheless, looking at the prediction results, both are promising accelerometers for behavioral classification of giraffes. Therefore, these devices when applied to free-ranging animals, in combination with GPS tracking, can contribute greatly to the conservation of giraffes.

Sleep classification from wrist-worn accelerometer data using random forests

Accurate and low-cost sleep measurement tools are needed in both clinical and epidemiological research. To this end, wearable accelerometers are widely used as they are both low in price and provide reasonably accurate estimates of movement. Techniques to classify sleep from the high-resolution accelerometer data primarily rely on heuristic algorithms. In this paper, we explore the potential of detecting sleep using Random forests. Models were trained using data from three different studies where 134 adult participants (70 with sleep disorder and 64 good healthy sleepers) wore an accelerometer on their wrist during a one-night polysomnography recording in the clinic. The Random forests were able to distinguish sleep-wake states with an F1 score of 73.93% on a previously unseen test set of 24 participants. Detecting when the accelerometer is not worn was also successful using machine learning (hbox {F1-score} > 93.31%), and when combined with our sleep detection models on day-time data provide a sleep estimate that is correlated with self-reported habitual nap behaviour (hbox {r}=.60). These Random forest models have been made open-source to aid further research. In line with literature, sleep stage classification turned out to be difficult using only accelerometer data.

Cross-Coupling Coefficient Estimation of a Nano-gAccelerometer by Continuous Rotation Modulation on a Tilted Rate Table

Nano- $g$ accelerometers are widely used in space exploration and measurement of the earth’s gravitational field. It is essential to precisely evaluate error effects at high orders such as cross-coupling for applications in a dynamic environment. Nevertheless, it remains challenging to meet the precision requirements using conventional calibration measures. In this article, we propose a method to separate the cross-coupling coefficients of a linear single-axis accelerometer by mounting it on a steadily rotating rate table that is tilted at a fixed deviation angle with respect to the horizontal plane. The gravity component is periodically modulated along the input axis per revolution. Simultaneously, a series of centripetal acceleration is applied along the cross axis in sequence while adjusting the rotation frequency of the rate table by steps. Thus, the cross-coupling coefficient can be separated by its dependence both on the modulated gravity acceleration and the centripetal acceleration. In comparison to the static multipoint angular rotation test on a tilted dividing head, the proposed dynamic modulation method demonstrates improved robustness against corruption from bias drift, with an improved uncertainty. This method to separate the cross-coupling coefficient is suitable for testing high-resolution accelerometers, without requiring high bias stability or sensitive response sustaining at ultralow frequency.

High-resolution micro-grating accelerometer based on a gram-scale proof mass.

Micro-grating accelerometer detecting small displacement by an optical system can be widely applied in inertial navigation and seismic monitoring. We proposed a micro-grating accelerometer prototype with a proof mass of gram-scale to decrease the thermal mechanical noise, which is the fundamental limit of a high-resolution accelerometer. The theoretical model for the contrast ratio of a micro-grating accelerometer is established based on Gaussian beam theory, and the adjustment method based on a scanning slit beam profiler improves the contrast ratio of 0th order effectively. Compared to our former prototype, experiment results indicate the noise floor is decreased from 0.9 mg/√Hz to 137 ng/√Hz, and the bias stability is decreased from 0.35 mg to 3.1 µg.

Quantification of Motor Function in Huntington Disease Patients Using Wearable Sensor Devices

Previous studies have demonstrated the feasibility and promise of wearable sensors as objective measures of motor impairment in Parkinson disease and essential tremor. However, there are few published studies that have examined such an application in Huntington disease (HD). This report provides an evaluation of the potential to objectively quantify chorea in HD patients using wearable sensor data. Data were derived from a substudy of the phase 2 Open-PRIDE-HD study, where 17 patients were screened and 15 patients enrolled in the substudy and ultimately 10 patients provided sufficient wearable sensor data. The substudy was designed to provide high-resolution data to inform design of predictive algorithms for chorea quantification. During the entire course of the 6-month study, in addition to chorea ratings from 18 in-clinic assessments, 890 home assessments, and 1,388 responses to daily reminders, 33,000 h of high-resolution accelerometer data were captured continuously from wearable smartwatches and smartphones. Despite its limited sample size, our study demonstrates that arm chorea can be characterized using accelerometer data during static assessments. Nonetheless, the small sample size limits the generalizability of the model. The sensor-based model can quantify the chorea level with high correlation to the chorea severity reported by both clinicians and patients. In addition, our analysis shows that the chorea digital signature varies between patients. This work suggests that digital wearable sensors have the potential to support clinical development of medications in patients with movement disorders, such as chorea. However, additional data would be needed from a larger number of HD patients with a full range of chorea severity (none to severe) with and without intervention to validate this potentially predictive technology.

Pull-in dynamics of overdamped microbeams

We study the dynamics of MEMS microbeams undergoing electrostatic pull-in. At DC voltages close to the pull-in voltage, experiments and numerical simulations have reported ‘bottleneck’ behaviour in which the transient dynamics slow down considerably. This slowing down is highly sensitive to external forces, and so has widespread potential for applications that use pull-in time as a sensing mechanism, including high-resolution accelerometers and pressure sensors. Previously, the bottleneck phenomenon has only been understood using lumped mass-spring models that do not account for effects such as variable residual stress and different boundary conditions. We extend these studies to incorporate the beam geometry, developing an asymptotic method to analyse the pull-in dynamics. We attribute bottleneck behaviour to critical slowing down near the pull-in transition, and we obtain a simple expression for the pull-in time in terms of the beam parameters and external damping coefficient. This expression is found to agree well with previous experiments and numerical simulations that incorporate more realistic models of squeeze film damping, and so provides a useful design rule for sensing applications. We also consider the accuracy of a single-mode approximation of the microbeam equations—an approach that is commonly used to make analytical progress, without systematic investigation of its accuracy. By comparing to our bottleneck analysis, we identify the factors that control the error of this approach, and we demonstrate that this error can indeed be a very small.

Using pseudo electrostatic spring constant to improve the resolution of micro-machined accelerometer

High-resolution accelerometers have commonly used an electrostatic force-feedback sigma-delta closed-loop. In order to improve the resolution of the accelerometers, high-order noise shaping filters has been used. However, it is difficult to trade off between the stability and the noise shaping ability to design high-order sigma-delta closed-loop, especially with the high-Q acceleration sensors. In this paper, a pseudo electrostatic spring constant is proposed based on the electrostatic force-feedback sigma-delta closed-loop accelerometer. It uses to adjust the resonant frequency of the system transfer function to achieve a lower noise level in the signal band, and this adjustment is outside the closed-loop without influence on the stability of the closed-loop system. The concept of the pseudo electrostatic spring constant, the system analysis and simulation are presented. The integrated measurement system is implemented on the PCB. The system noise floor can achieve about 160ng/√Hz after the adjustment by using the optimal pseudo electrostatic spring constant. The experimental results confirm the theoretical analysis and simulation results.

How Elephant Seals (Mirounga leonina) Adjust Their Fine Scale Horizontal Movement and Diving Behaviour in Relation to Prey Encounter Rate.

Understanding the diving behaviour of diving predators in relation to concomitant prey distribution could have major practical applications in conservation biology by allowing the assessment of how changes in fine scale prey distribution impact foraging efficiency and ultimately population dynamics. The southern elephant seal (Mirounga leonina, hereafter SES), the largest phocid, is a major predator of the southern ocean feeding on myctophids and cephalopods. Because of its large size it can carry bio-loggers with minimal disturbance. Moreover, it has great diving abilities and a wide foraging habitat. Thus, the SES is a well suited model species to study predator diving behaviour and the distribution of ecologically important prey species in the Southern Ocean. In this study, we examined how SESs adjust their diving behaviour and horizontal movements in response to fine scale prey encounter densities using high resolution accelerometers, magnetometers, pressure sensors and GPS loggers. When high prey encounter rates were encountered, animals responded by (1) diving and returning to the surface with steeper angles, reducing the duration of transit dive phases (thus improving dive efficiency), and (2) exhibiting more horizontally and vertically sinuous bottom phases. In these cases, the distance travelled horizontally at the surface was reduced. This behaviour is likely to counteract horizontal displacement from water currents, as they try to remain within favourable prey patches. The prey encounter rate at the bottom of dives decreased with increasing diving depth, suggesting a combined effect of decreased accessibility and prey density with increasing depth. Prey encounter rate also decreased when the bottom phases of dives were spread across larger vertical extents of the water column. This result suggests that the vertical aggregation of prey can regulate prey density, and as a consequence impact the foraging success of SESs. To our knowledge, this is one of only a handful of studies showing how the vertical distributions and structure of prey fields influence the prey encounter rates of a diving predator.

Sleep Assessment in Large Cohort Studies with High-Resolution Accelerometers.

Accelerometry can be a practical replacement for polysomnography in large observational studies. This review discusses the need for sleep characterization in large observational studies, exemplified by the practices of the ongoing German National Cohort study. After brief descriptions of the physical principles and state-of-the-art accelerometer devices and an overview of public data analysis algorithms for sleep-wake differentiation, we demonstrate that the spectral properties of acceleration data provide additional features that can be exploited. This leads to a periodogram-based sleep detection algorithm. Finally, we address issues of data handling and quality assurance in large cohort studies.

Post-tagging behaviour and habitat use in shortnose sturgeon measured with high-frequency accelerometer and PSATs

External tagging of fish using pop-up satellite tags (PSATs) can cause trauma and stress associated with capture, handling, tagging injury and tag placement that impedes body function and mobility, and these can affect the behaviour of the fish. We measured short-term (days) post-release behavioural response of estuarine shortnose sturgeon to tagging with PSATs and high-frequency accelerometers. We developed a secure, low-noise attachment method for high-resolution (50 Hz) accelerometer PSATs in shortnose sturgeon. The archived time series of acceleration was used to determine sturgeon post-tagging activity, estimate swimming speed, erratic behaviour and behavioural clusters in relation to ambient light level and temperature, depth and tidal flow. Short-term (hours) post-release response was characterized by resting periods on the river bottom and decreased swimming activity that was associated with individual-specific modulations in the swimming gait and high-energy burst acceleration movements. Locomotion routines suggested a relation to tidal flow, yet the short-term time series did not allow for routine movements to be classified. The approach used provides a useful method of revealing behavioural modifications during the post-release recovery period of PSAT-tagged fish. We discuss how short-term (acute) effects on behaviour and potential longer-term (chronic) effects on survival are especially relevant in tagging studies.

(Invited) A 1-mG MEMS Sensor

MEMS (microelectromechanical systems) technology has contributed substantially to the miniaturization of inertial sensors, such as accelerometers and gyroscopes [1]. Nowadays, MEMS inertial sensors have been widely used in a variety of application in our daily life. In recent years, accurate sensing of low acceleration, specifically below 1 G ( G = 9.8 m/s2, gravitational acceleration), has been one of the major interests for MEMS accelerometer development; those highly-sensitive accelerometers are to be used for possible applications such as healthcare and medical use, navigation systems, and micro-gravity monitoring in space [2]. According to the background, we have proposed a design approach of sub-1G MEMS accelerometer with a gold proof mass [3]. The high-density of gold has enabled us to suppress the thermo-mechanical noise, as it is called Brownian noise (BN ), when compared with typical proof mass material such as silicon; as the magnitude of BN is inversely proportional to the proof mass [4], the increase of proof-mass density would be an effective method to have low BN with a minimum proof-mass footprint. In this paper, we present a 1mG MEMS sensor that was designed to be an extension of our previous work. A proof-of-concept device has been developed with a proof mass made of electroplated gold. The target BN was set to be below 1 μG/Hz1/2, which has potential of acceleration sensing with a resolution of 1 mG or even lower. The proof mass was designed to be 2.4 mm × 2.1 mm in area with the thickness of 12 μm, which was nearly four times larger than that of our previous sensor. To reduce the metal warpage, the proof mass was segmented into a number of sub-blocks that were cross-linked by the additional metal layer. The device was fabricated by the post-CMOS (complementary metal-oxide semiconductor) process [6] utilizing gold electroplating; the sensor can be implemented on to CMOS-LSI (large-scale integration). Thus, MEMS structures and sensing circuitry could be integrated as a single chip with a minimum footprint and parasitic elements. Fig. 1(a) shows a top view of the fabricated MEMS device. The proof mass consists of the array of 18 × 21 sub-blocks. Each proof mass corner is suspended by micromechanical spring. A close-up SEM (scanning electron microscope image) is shown in Fig. 1(b). The stopper is used to limit the displacement of the proof mass, which would prevent the self-destruction of the MEMS structure at the input of excess acceleration. The fundamental characteristics of the MEMS device were experimentally obtained by the frequency response, where we measured capacitance and phase as a function of signal frequency; the actual BN was estimated to be 0.16 μG/Hz1/2 [5] that was sufficiently lower than the target value of 1 μG/Hz1/2. Also, we measured the capacitance change between the proof mass and the fixed counter electrode as a function of input acceleration. The experimental results indicate that the MEMS device was capable of capacitive sensing input acceleration below 1G. The evaluation results show that the proposed sensor design could further improve the sensing resolution of MEMS accelerometers, which would be required for possible future applications. The presentation will cover the design, fabrication process, and measurement results of the proof-of-concept device. [1] N. Yazdi, F. Ayazi, and K. Najafi, “Micromachined Inertial Sensors”, Proc. IEEE, Vol. 86, pp.1640-1659, 1998. [2] G. Krishnan, C. U. Kshirsagar, G. K. Ananthasuresh, and N. Bhat, “Micromachined High-Resolution Accelerometers,” J. Indian Inst. Sci., vol. 87, pp. 333-361, 2007. [3] D. Yamane et al., “Design of sub-1g microelectromechanical systems accelerometers” Appl. Phys. Lett., vol. 104, issue 7, pp. 074102, 2014. [4] M. Lemkin, and B. E. Boser, “A Three-Axis Micromachined Accelerometer with a CMOS Position-Sense Interface and Digital Offset-Trim Electronis,” IEEE J. Solid-State Circuits, vol. 34, pp. 456-468, 1999. [5] D. Yamane, T. Konishi, T. Matsushima, H. Toshiyoshi, K. Machida, and K. Masu, "A Sub-1G Capacitive Sensor for Integrated CMOS-MEMS Accelerometers," in Proc. 7th Asia-Pacific Conference on Transducers and Micro/Nano Technologies (APCOT2014), EXCO, Daegu, Korea. June 29 - July 2, 2014, P2-55. [6] K. Machida, T. Konishi, D. Yamane, H. Toshiyoshi, and K. Masu, “Integrated CMOS-MEMS Technology and Its Applications,” ECS Trans., vol. 61, pp. 21-39, 2014. Figure 1

The use of an unsupervised learning approach for characterizing latent behaviors in accelerometer data.

The recent increase in data accuracy from high resolution accelerometers offers substantial potential for improved understanding and prediction of animal movements. However, current approaches used for analysing these multivariable datasets typically require existing knowledge of the behaviors of the animals to inform the behavioral classification process. These methods are thus not well-suited for the many cases where limited knowledge of the different behaviors performed exist. Here, we introduce the use of an unsupervised learning algorithm. To illustrate the method's capability we analyse data collected using a combination of GPS and Accelerometers on two seabird species: razorbills (Alca torda) and common guillemots (Uria aalge). We applied the unsupervised learning algorithm Expectation Maximization to characterize latent behavioral states both above and below water at both individual and group level. The application of this flexible approach yielded significant new insights into the foraging strategies of the two study species, both above and below the surface of the water. In addition to general behavioral modes such as flying, floating, as well as descending and ascending phases within the water column, this approach allowed an exploration of previously unstudied and important behaviors such as searching and prey chasing/capture events. We propose that this unsupervised learning approach provides an ideal tool for the systematic analysis of such complex multivariable movement data that are increasingly being obtained with accelerometer tags across species. In particular, we recommend its application in cases where we have limited current knowledge of the behaviors performed and existing supervised learning approaches may have limited utility.

Relationship Of Stride Characteristics To Trunk Acceleration And Metabolic Cost In Collegiate Runners

There has been recent interest regarding the importance of gait characteristics as determinants of running performance. Within homogenous groups of elite distance runners, factors other than VO2max contribute to the variation in running performance. Running economy is one such factor and clarifying the gait characteristics that influence running economy is important to understand running performance. PURPOSE: This study quantified stride characteristics, trunk acceleration, metabolic cost of running, and their interaction in collegiate runners. METHODS: Subjects consented to procedures approved by EMU CHHS Human Subjects Committee. High-resolution accelerometer (HRA, G-Link ADXL 210, Microstrain Inc., VT), full-body optical motion-capture (Vicon MX, Vicon,, CO), and indirect calorimetry (Oxycon Mobile, CareFusion, CA) measurements were collected from seven male NCAA DI collegiate distance runners during treadmill running at speeds of 3.33, 3.89, and 4.44 m·s-1. Root mean square (RMS) accelerations were calculated for the mediolateral (ML), anterior-posterior (AP), vertical (VT) axis and resultant Euclidian scalar (RES). Ratio of axial accelerations (Xra= Xrms/RES) were also determined for VTra, MLra and APra. The kinematic stride characteristics of stance time, stance time (percent gait cycle), stride length, and stride rate were determined from optical motion capture. RESULTS: Stride rate, stance time, and stride length were different across speeds (p < 0.05). Linear stepwise regressions showed that VO2, AP acceleration, and RES acceleration were influenced by stride length (R2 = 0.86, 0.28, and 0.22, respectively, p < 0.05). VT acceleration was related to stance time percent (R2 = 0.243, p < 0.05) and VTra was related to stride length and stance time (R2 = 0.65, p < 0.05), while APra was related to stride rate (R2 = 0.21, p < 0.05). CONCLUSION: We have previously shown that trained runners exhibit higher VT and VTra than untrained and that VTra and APra influence metabolic cost of running. Therefore, these differences may partially be explained by differences in stride length, stance time percent, and stride rate. Of these, stride length and stance time percent appear to be most important.