Fabio Bagalà

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 value = 98%). The SE was considerably lower (SE = 57.0%±27.3%, maximum value = 82.8%), 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.

Proposal for a multiphase fall model based on real-world fall recordings with body-fixed sensors

Falls are by far the leading cause of fractures and accidents in the home environment. The current Cochrane reviews and other systematic reviews report on more than 200 intervention studies about fall prevention. A recent meta-analysis has summarized the most important risk factors of accidental falls. However, falls and fall-related injuries remain a major challenge. One novel approach to recognize, analyze, and work better toward preventing falls could be the differentiation of the fall event into separate phases. This might aid in reconsidering ways to design preventive efforts and diagnostic approaches. From a conceptual point of view, falls can be separated into a pre-fall phase, a falling phase, an impact phase, a resting phase, and a recovery phase. Patient and external observers are often unable to give detailed comments concerning these phases. With new technological developments, it is now at least partly possible to examine the phases of falls separately and to generate new hypotheses.The article describes the practicality and the limitations of this approach using body-fixed sensor technology. The features of the different phases are outlined with selected real-world fall signals.ZusammenfassungStürze sind die mit Abstand häufigsten Ursachen von Frakturen und häuslichen Verletzungen im Alter. In den Cochrane Reviews und anderen systematischen Analysen wurden mehr als 200 randomisierte Interventionsstudien zur Sturzprävention erfasst. Eine neue Metaanalyse liegt für die Risikofaktoren von Stürzen vor. Dennoch bleiben Stürze und sturzbedingte Verletzungen eine große Herausforderung. Ein neuer Ansatz zur Erkennung, Analyse und Prävention von Stürzen ist es, Stürze in Abschnitte aufzuteilen. Dies könnte bei der Erstellung diagnostischer und präventiver Ansätze helfen. Phänomenologisch ist offenkundig, dass es eine Vorphase, Fallphase, Aufprallphase, Ruhephase und mögliche Erholungsphase gibt. Patienten und Fremdbeobachter sind allerdings nicht in der Lage, hierzu exakte Angaben zu machen. Durch technologische Neuentwicklungen ist es nunmehr möglich, diese Abschnitte zumindest teilweise zu beurteilen und daraus erste Hypothesen abzuleiten.Der Artikel beschreibt dabei die Praktikabilität und Beschränkungen der Verwendung von am Körper getragenen Sensoren. Die Sturzphasen werden anhand von Fallbeispielen verdeutlicht.

A wavelet-based approach to fall detection

Falls among older people are a widely documented public health problem. Automatic fall detection has recently gained huge importance because it could allow for the immediate communication of falls to medical assistance. The aim of this work is to present a novel wavelet-based approach to fall detection, focusing on the impact phase and using a dataset of real-world falls. Since recorded falls result in a non-stationary signal, a wavelet transform was chosen to examine fall patterns. The idea is to consider the average fall pattern as the “prototype fall”.In order to detect falls, every acceleration signal can be compared to this prototype through wavelet analysis. The similarity of the recorded signal with the prototype fall is a feature that can be used in order to determine the difference between falls and daily activities. The discriminative ability of this feature is evaluated on real-world data. It outperforms other features that are commonly used in fall detection studies, with an Area Under the Curve of 0.918. This result suggests that the proposed wavelet-based feature is promising and future studies could use this feature (in combination with others considering different fall phases) in order to improve the performance of fall detection algorithms.

Quantitative Description of the Lie-to-Sit-to-Stand-to-Walk Transfer by a Single Body-Fixed Sensor

Sufficient capacity and quality of performance of complex movement patterns during daily activity, such as standing up from a bed, is a prerequisite for independent living and also may be an indicator of fall risk. Until now, the transfer from lying-to-sit-to-stand-to-walk (LSSW) was investigated by functional testing, subjective rating or for activity classification of subtasks. The aim of this study was to use a single body-fixed inertial sensor to describe the complex movement of the LSSW transfer. Fifteen older patients of a geriatric rehabilitation clinic (median age 81 years) and ten young, healthy persons (median age 37 years) were instructed to stand up from bed in a continuous movement and to start walking. Data acquisition was performed using an inertial measurement unit worn on the lower back. Parameters extracted from the sensor outputs were able to correctly classify the subjects into a correct group with sensitivity and specificity between 90% and 100%. ICCs of the descriptive parameters ranged between 0.85 and 0.95 in the cohort of older patients. The different strategies adopted to transfer from lying to standing up were estimated through an extended Kalman filter. The results obtained in this study suggest the usability of the instrumented LSSW test in clinical settings.

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.

Calibrated 2D Angular Kinematics by Single-Axis Accelerometers: From Inverted Pendulum to

A new method for the estimation of multi-link angular kinematics in the sagittal plane, using one single-axis accelerometer (SAA) per segment, is presented in this paper. A preliminary calibration, using SAAs and a reference system (encoder or stereo-photogrammetry), allows the estimation of sensors position and orientation and segment lengths. These parameters are then used to predict the chain kinematics using the SAAs only. To evaluate the method, the algorithm is first tested on a mechanical arm equipped with a reference encoder. A general method for estimating the kinematics of an N-link chain is also provided. Finally, a three-link biomechanical model is applied to a human subject to estimate the joint angles during squat tasks; a stereo-photogrammetric system is used for validation. The results are very close to the reference values. Mean descriptive (predictive) root mean squared error (RMSE) is 0.15^° (0.16^°) for the inverted pendulum, and 0.39^° (0.59^° ) for the shank, 0.82^° (1.06^° ) for the thigh, 0.87^° (1.09^° ) for the HAT (head-arm-trunk) in the three-link model. The mean value of RMSE without calibration is 1.02^° for the inverted pendulum, and 11.01^° (shank), 11.39^° (thigh) and 12.21^° (HAT) in the three-link model. These results suggest that, after the calibration procedure, one SAA per segment is enough to estimate 2D joint angles accurately in a kinematic chain of any number of links.

{\rm N}

In human movement modeling, the problem of multi-link kinematics estimation by means of inertial measurement units has been investigated by several authors through efficient sensor fusion algorithms. In this perspective a single inertial measurement unit per link is required. This set-up is not cost-effective compared with a solution in which a single-axis accelerometer per link is used. In this paper, a novel fast technique is presented for the estimation of the sway angle in a multi-link chain by using a single-axis accelerometer per segment and by setting the boundary conditions through an ad hoc algorithm. The technique, based on the windowing of the accelerometer output, was firstly tested on a mechanical arm equipped with a single-axis accelerometer and a reference encoder. The technique is then tested on a subject performing a squat task for the knee flexion-extension angle evaluation by using two single-axis accelerometers placed on the thigh and shank segments, respectively. A stereo-photogrammetric system was used for validation. RMSEs (mean ± std) are 0.40 ± 0.02° (mean peak-to-peak range of 147.2 ± 4.9°) for the mechanical inverted pendulum and 1.01 ± 0.11° (mean peak-to-peak range of 59.29 ± 2.02°) for the knee flexion-extension angle. Results obtained in terms of RMSE were successfully compared with an Extended Kalman Filter applied to an inertial measurement unit. These results suggest the usability of the proposed algorithm in several fields, from automatic control to biomechanics, and open new opportunities to increase the accuracy of the existing tools for orientation evaluation.

-Link Chain

Falls are by far the leading cause of fractures and accidents in the home environment. The current Cochrane reviews and other systematic reviews report on more than 200 intervention studies about fall prevention. A recent meta-analysis has summarized the most important risk factors of accidental falls. However, falls and fall-related injuries remain a major challenge. One novel approach to recognize, analyze, and work better toward preventing falls could be the differentiation of the fall event into separate phases. This might aid in reconsidering ways to design preventive efforts and diagnostic approaches. From a conceptual point of view, falls can be separated into a pre-fall phase, a falling phase, an impact phase, a resting phase, and a recovery phase. Patient and external observers are often unable to give detailed comments concerning these phases. With new technological developments, it is now at least partly possible to examine the phases of falls separately and to generate new hypotheses.The article describes the practicality and the limitations of this approach using body-fixed sensor technology. The features of the different phases are outlined with selected real-world fall signals.ZusammenfassungStürze sind die mit Abstand häufigsten Ursachen von Frakturen und häuslichen Verletzungen im Alter. In den Cochrane Reviews und anderen systematischen Analysen wurden mehr als 200 randomisierte Interventionsstudien zur Sturzprävention erfasst. Eine neue Metaanalyse liegt für die Risikofaktoren von Stürzen vor. Dennoch bleiben Stürze und sturzbedingte Verletzungen eine große Herausforderung. Ein neuer Ansatz zur Erkennung, Analyse und Prävention von Stürzen ist es, Stürze in Abschnitte aufzuteilen. Dies könnte bei der Erstellung diagnostischer und präventiver Ansätze helfen. Phänomenologisch ist offenkundig, dass es eine Vorphase, Fallphase, Aufprallphase, Ruhephase und mögliche Erholungsphase gibt. Patienten und Fremdbeobachter sind allerdings nicht in der Lage, hierzu exakte Angaben zu machen. Durch technologische Neuentwicklungen ist es nunmehr möglich, diese Abschnitte zumindest teilweise zu beurteilen und daraus erste Hypothesen abzuleiten.Der Artikel beschreibt dabei die Praktikabilität und Beschränkungen der Verwendung von am Körper getragenen Sensoren. Die Sturzphasen werden anhand von Fallbeispielen verdeutlicht.

Quasi-Real Time Estimation of Angular Kinematics Using Single-Axis Accelerometers

Falls are by far the leading cause of fractures and accidents in the home environment. The current Cochrane reviews and other systematic reviews report on more than 200 intervention studies about fall prevention. A recent meta-analysis has summarized the most important risk factors of accidental falls. However, falls and fall-related injuries remain a major challenge. One novel approach to recognize, analyze, and work better toward preventing falls could be the differentiation of the fall event into separate phases. This might aid in reconsidering ways to design preventive efforts and diagnostic approaches. From a conceptual point of view, falls can be separated into a pre-fall phase, a falling phase, an impact phase, a resting phase, and a recovery phase. Patient and external observers are often unable to give detailed comments concerning these phases. With new technological developments, it is now at least partly possible to examine the phases of falls separately and to generate new hypotheses.The article describes the practicality and the limitations of this approach using body-fixed sensor technology. The features of the different phases are outlined with selected real-world fall signals.ZusammenfassungStürze sind die mit Abstand häufigsten Ursachen von Frakturen und häuslichen Verletzungen im Alter. In den Cochrane Reviews und anderen systematischen Analysen wurden mehr als 200 randomisierte Interventionsstudien zur Sturzprävention erfasst. Eine neue Metaanalyse liegt für die Risikofaktoren von Stürzen vor. Dennoch bleiben Stürze und sturzbedingte Verletzungen eine große Herausforderung. Ein neuer Ansatz zur Erkennung, Analyse und Prävention von Stürzen ist es, Stürze in Abschnitte aufzuteilen. Dies könnte bei der Erstellung diagnostischer und präventiver Ansätze helfen. Phänomenologisch ist offenkundig, dass es eine Vorphase, Fallphase, Aufprallphase, Ruhephase und mögliche Erholungsphase gibt. Patienten und Fremdbeobachter sind allerdings nicht in der Lage, hierzu exakte Angaben zu machen. Durch technologische Neuentwicklungen ist es nunmehr möglich, diese Abschnitte zumindest teilweise zu beurteilen und daraus erste Hypothesen abzuleiten.Der Artikel beschreibt dabei die Praktikabilität und Beschränkungen der Verwendung von am Körper getragenen Sensoren. Die Sturzphasen werden anhand von Fallbeispielen verdeutlicht.

Vorschlag für ein mehrphasensturzmodell auf der basis von sturzdokumentationen mit am körper getragenen sensor

Re: A new device for in situ static and dynamic calibration of force platforms by Hsieh et al. [Gait and Posture 33 (2011) 701–705] No re-calibration matrix is provided. Methods proposed in [7– 10] allow estimating, in different ways, global, local, linear, and non-linear re-calibration matrices. This is very useful in practice allowing data re-calibration by pre-multiplication matrix even