Angelo Cappello

Evaluation of Accelerometer-Based Fall Detection Algorithms on Real-World Falls

Despite extensive preventive efforts, falls continue to be a major source of morbidity and mortality among elderly. Real-time detection of falls and their urgent communication to a telecare center may enable rapid medical assistance, thus increasing the sense of security of the elderly and reducing some of the negative consequences of falls. Many different approaches have been explored to automatically detect a fall using inertial sensors. Although previously published algorithms report high sensitivity (SE) and high specificity (SP), they have usually been tested on simulated falls performed by healthy volunteers. We recently collected acceleration data during a number of real-world falls among a patient population with a high-fall-risk as part of the SensAction-AAL European project. The aim of the present study is to benchmark the performance of thirteen published fall-detection algorithms when they are applied to the database of 29 real-world falls. To the best of our knowledge, this is the first systematic comparison of fall detection algorithms tested on real-world falls. We found that the SP average of the thirteen algorithms, was (mean±std) 83.0%±30.3% (maximum valueu200a=u200a98%). The SE was considerably lower (SEu200a=u200a57.0%±27.3%, maximum valueu200a=u200a82.8%), much lower than the values obtained on simulated falls. The number of false alarms generated by the algorithms during 1-day monitoring of three representative fallers ranged from 3 to 85. The factors that affect the performance of the published algorithms, when they are applied to the real-world falls, are also discussed. These findings indicate the importance of testing fall-detection algorithms in real-life conditions in order to produce more effective automated alarm systems with higher acceptance. Further, the present results support the idea that a large, shared real-world fall database could, potentially, provide an enhanced understanding of the fall process and the information needed to design and evaluate a high-performance fall detector.

'Outwalk': a protocol for clinical gait analysis based on inertial and magnetic sensors.

A protocol named Outwalk was developed to easily measure the thorax–pelvis and lower-limb 3D kinematics on children with cerebral palsy (CP) and amputees during gait in free-living conditions, by means of an Inertial and Magnetic Measurement System (IMMS). Outwalk defines the anatomical/functional coordinate systems (CS) for each body segment through three steps: (1) positioning the sensing units (SUs) of the IMMS on the subjects’ thorax, pelvis, thighs, shanks and feet, following simple rules; (2) computing the orientation of the mean flexion–extension axis of the knees; (3) measuring the SUs’ orientation while the subject’s body is oriented in a predefined posture, either upright or supine. If the supine posture is chosen, e.g. when spasticity does not allow to maintain the upright posture, hips and knees static flexion angles must be measured through a standard goniometer and input into the equations that define Outwalk anatomical CSs. In order to test for the inter-rater measurement reliability of these angles, a study was carried out involving nine healthy children (7.9xa0±xa02xa0years old) and two physical therapists as raters. Results showed RMS error of 1.4° and 1.8° and a negligible worst-case standard error of measurement of 2.0° and 2.5° for hip and knee angles, respectively. Results were thus smaller than those reported for the same measures when performed through an optoelectronic system with the CAST protocol and support the beginning of clinical trials of Outwalk with children with CP.

First in vivo assessment of “Outwalk”: a novel protocol for clinical gait analysis based on inertial and magnetic sensors

A protocol named “Outwalk” was recently proposed to measure the thorax–pelvis and lower-limb kinematics during gait in free-living conditions, by means of an inertial and magnetic measurement system (IMMS). The aim of this study was to validate Outwalk on four healthy subjects when it is used in combination with a specific IMMS (Xsens Technologies, NL), against a reference protocol (CAST) and measurement system (optoelectronic system; Vicon, Oxford Metrics Group, UK). For this purpose, we developed an original approach based on three tests, which allowed to separately investigate: (1) the consequences on joint kinematics of the differences between protocols (Outwalk vs. CAST), (2) the accuracy of the hardware (Xsens vs. Vicon), and (3) the summation of protocols’ differences and hardware accuracy (Outwalkxa0+xa0Xsens vs. CASTxa0+xa0Vicon). In order to assess joint-angles similarity, the coefficient of multiple correlation (CMC) was used. For test 3, the CMC showed that Outwalkxa0+xa0Xsens and CASTxa0+xa0Vicon kinematics can be interchanged, offset included, for hip, knee and ankle flexion–extension, and hip ab-adduction (CMCxa0>xa00.88). The other joint-angles can be interchanged offset excluded (CMCxa0>xa00.85). Tests 1 and 2 also showed that differences in offset between joint-angles were predominantly induced by differences in the protocols; differences in correlation by both hardware and protocols; differences in range of motion by the Xsens accuracy. Results thus support the commencement of a clinical trial of Outwalk on transtibial amputees.

A new formulation of the coefficient of multiple correlation to assess the similarity of waveforms measured synchronously by different motion analysis protocols.

Different 3D motion analysis protocols are currently available, but little is known regarding the level of similarity of their outcomes, e.g. whether a joint-angle waveform from one protocol can be interchanged with that measured by another protocol. Similarity assessments are therefore urgent to ease the comparison of results. In this context, a major issue is how to quantify the similarity between waveforms measured synchronously through different protocols, within each of many movement-cycles (e.g. gait-cycle), when the effect of protocols on waveforms similarity is the only of interest. For this purpose we developed a new formulation of the statistical index called coefficient of multiple correlation (CMC). The CMC measures the overall similarity of waveforms taking into account the concurrent effects of differences in offset, correlation, and gain. The within-day CMC originally proposed by Kadaba et al. [7], was firstly reinterpreted in terms of excluded factors. Then, the new formulation was set to assess the inter-protocol similarity, removing the between-gait-cycle variability. An example showing the effectiveness of the new formulation is presented regarding the pelvis-trunk and ankle kinematics.

Double calibration vs. global optimisation: Performance and effectiveness for clinical application

For clinical application the quantification of the actual subject-specific kinematics is necessary. Soft tissue artefact (STA) propagation to joint kinematics can nullify the clinical interpretability of stereophotogrammetric analysis. STA was assessed to be strongly subject- and task-specific. The global optimisation, whose performance was assessed only on simulated data, is at the basis of several of the STA compensation methods proposed in the literature. On the other hand, the double calibration was recently proposed and resulted very effective on experimental data. In the present work, the performance of double calibration and global optimisation in reducing soft tissue artefact propagation to relevant knee joint kinematics was compared by using 3D fluoroscopy as gold standard. The mean RMSE over the repetitions for the double calibration is in the order of 1-2 degrees for joint rotations and 1-3 mm for translation, while for the global optimisation is in the order of 10 degrees and 10-15 mm, respectively. The double calibration should then be preferred for the quantification of the subject specific kinematics.

Hilbert–Huang-Based Tremor Removal to Assess Postural Properties From Accelerometers

Tremor is one of the symptoms of several disorders of the central and peripheral nervous system, such as Parkinsons disease (PD). The impairment of postural control is another symptom of PD. The conventional method of posture analysis uses force plates, but accelerometers can be a valid and reliable alternative. Both these measurement techniques are sensitive to tremor. Tremor affects postural measures and may thus lead to misleading results or interpretations. Linear low-pass filters (LPFs) are commonly employed for tremor removal. In this study, an alternative method, based on Hilbert-Huang transformation (HHT), is proposed. We examined 20 PD subjects, with and without tremor, and 20 control subjects. We compared the effectiveness of LPF and HHT-based filtering on a set of postural parameters extracted from acceleration signals. HHT has the advantage of providing a filter, which with no a priori knowledge, efficiently manages the nonlinear, nonstationary interference due to tremor, and beyond tremor, gives descriptive measures of postural function. Some of the differences found using LPF can instead be ascribed to inefficient noise/tremor suppression. Filter order and cutoff frequency are indeed critical when subjects exhibit a tremorous behavior, in which case LPF parameters should be chosen very carefully.

Effect of sub-optimal neuromotor control on the hip joint load during level walking

Skeletal forces are fundamental information in predicting the risk of bone fracture. The neuromotor control system can drive muscle forces with various task- and health-dependent strategies but current modelling techniques provide a single optimal solution of the muscle load sharing problem. The aim of the present work was to study the variability of the hip load magnitude due to sub-optimal neuromotor control strategies using a subject-specific musculoskeletal model. The model was generated from computed tomography (CT) and dissection data from a single cadaver. Gait kinematics, ground forces and electromyographic (EMG) signals were recorded on a body-matched volunteer. Model results were validated by comparing the traditional optimisation solution with the published hip load measurements and the recorded EMG signals. The solution space of the instantaneous equilibrium problem during the first hip load peak resulted in 10(5) dynamically equivalent configurations of the neuromotor control. The hip load magnitude was computed and expressed in multiples of the body weight (BW). Sensitivity of the hip load boundaries to the uncertainty on the muscle tetanic stress (TMS) was also addressed. The optimal neuromotor control induced a hip load magnitude of 3.3 BW. Sub-optimal neuromotor controls induced a hip load magnitude up to 8.93 BW. Reducing TMS from the maximum to the minimum the lower boundary of the hip load magnitude varied moderately whereas the upper boundary varied considerably from 4.26 to 8.93 BW. Further studies are necessary to assess how far the neuromotor control can degrade from the optimal activation pattern and to understand which sub-optimal controls are clinically plausible. However we can consider the possibility that sub-optimal activations of the muscular system play a role in spontaneous fractures not associated with falls.

Energetic assessment of trunk postural modifications induced by a wearable audio-biofeedback system

This paper investigates the trunk postural modifications induced by a wearable device which assesses the trunk sway and provides biofeedback information through sonification of trunk kinematics. The device is based on an inertial wearable sensing unit including three mono-axial accelerometers and three rate gyroscopes embedded and mounted orthogonally. The biofeedback device was tested on nine healthy subjects during quiet stance in different conditions of sensory limitation eyes closed on solid surface, eyes open on foam cushion surface, eyes closed on foam cushion surface. Five trials were performed for each condition; the order of the trials was randomized. The results reported in this paper show how subjects reduced their rotational kinetic energy by using the biofeedback information and how this reduction was related to the limitation of sensory information.

Non-linear re-calibration of force platforms

Force platforms (FPs) are used in human movement analysis to measure the ground reaction force and the center of pressure (COP), and calculate derived kinetic and energetic quantities. We propose a re-calibration method that compensates for the FP non-linearity induced by top plate bending under loading. The method develops a previous solution that was proposed for a linear re-calibration and proved suitable for both local and global error compensation (Cedraro et al., 2008). The new method was experimentally tested on 4 commercial FPs by estimating the non-linear re-calibration matrix in a first training trial and by using it to assess the three force components and the COP in a validation trial, comparing the new method to the previously proposed solution for global, linear re-calibration. The average COP accuracy (mm) in the training trial was (mean±std): 2.3±1.4, 2.6±1.5, 11.8±4.3, 14.0±2.5 for the 4 FPs before re-calibration, and 0.7±0.4, 0.6±0.2, 0.5±0.2, 2.3±1.3 after non-linear re-calibration. In the validation trial, for one of the 4 tested FPs, mean errors for the three force components (N) and COP (mm) were: 3.6±2.3 (F(X)), 3.0±0.7 (F(Y)), 5.0±2.5 (F(Z)), 1.2±0.68 (COP) after linear re-calibration, and 2.5±0.7 (F(X)), 2.6±0.5 (F(Y)), 3.9±1.2 (F(Z)), 0.6±0.3 (COP) after non-linear re-calibration. The proposed global, non-linear method performed equally well as the local, linear re-calibration method, proving well-suited to compensate for the mild non-linear behavior of FP with the advantage of estimating a single re-calibration matrix.

A portable system for in-situ re-calibration of force platforms: Experimental validation

A system for the in-situ re-calibration of six-component force platforms is presented. The system consists of a device, a data-acquisition procedure and an algorithm. The device, simple and lightweight, is composed of a high-precision, 3-D load cell, loaded through a triangular stage, and precisely positioned on the force platform under re-calibration by means of a template. The data-acquisition procedure lasts about 1h and requires up to 13 measurements consisting of manual positioning the load cell on the force platform, and then having the operator exerting loads on both load cell and force platform by his/her body movement. As a result, the procedure makes use of loads in the same range of posture and gait tests. The algorithm estimates the local or global six-by-six re-calibration matrix of the force platform through a least-squares optimization, and is presented in detail in a separate paper [Cedraro A, Cappello A, Chiari L. A portable system for in-situ re-calibration of force platforms: Theoretical validation. Gait Posture 2008;28:488-94]. The system was validated on four commercial force platforms (Amti OR6, Bertec 4060-08, Bertec 4080-10, and Kistler 9286A). The average accuracy in the measurement of the center of pressure were 2.3+/-1.4mm, 2.6+/-1.5mm, 11.8+/-4.3mm, 14.0+/-2.5mm before re-calibration, 1.1+/-0.6mm, 1.8+/-1.1mm, 1.0+/-0.6mm, 3.2+/-1.1mm after global re-calibration, and 0.7+/-0.4mm, 0.8+/-0.5mm, 0.5+/-0.3mm, 2.0+/-1.2mm after local re-calibration (results presented in random order). The force platform re-calibration influenced the value, sign, and timing of net joint moments, estimated during a gait task through an inverse dynamics approach.