Fall events are one of the main causes of injuries among the elderly. The purpose of this study has been to identify a computational framework for the real-time and automatic detection of the fall risk, allowing the fast adoption of properly intervention strategies, to reduce injuries and traumas due to the fall. A wearable, wireless and minimally invasive surface Electromyography (EMG)-based system has been used to measure four lower-limb muscles activities. Eleven young healthy subjects have simulated several fall events (through a movable platform) and normal Activities of Daily Living (ADLs) and their patterns have been analysed. Highly discriminative features extracted within the EMG signals for the pre impact fall evaluation have been explored and a threshold-based approach has been adopted, assuring the real-time functioning. The threshold level for each feature has been set to distinguish an instability condition from normal activities. The proposed system seems able to recognize all falls with an average lead-time of 840ms before the impact, in simulated and controlled fall conditions.

This paper presents a multi-feature approach for detection of key postures by using a MESA SR4000 time-offlight 3D sensor managed by a low-power embedded PC. Acquired data were pre-processed by using a well-established framework including self-calibration, segmentation and tracking functionalities. To accommodate different application scenarios, hierarchical coarse-to-fine features were extracted by exploiting two different descriptors: topological and volumetric. The topological descriptor encoded intrinsic topology of body postures in a skeleton-like representation based on geodesic distance. Instead, the volumetric descriptor used a cylindrical voxelization to describe postures in a histogram-based representation. Both synthetic and real datasets were used to evaluate performance. The complementary discrimination capabilities exhibited by the two descriptors allowed to achieve good results in four different application scenarios with a classification rate greater than 96.4%. © 2012 IEEE.

The aging population represents an emerging challenge for healthcare since elderly people frequently suffer from chronic diseases requiring continuous medical care and monitoring. Sensor networks are possible enabling technologies for ambient assisted living solutions helping elderly people to be independent and to feel more secure. This paper presents a multi-sensor system for the detection of people falls in home environment. Two kinds of sensors are used: a wearable wireless accelerometer with onboard fall detection algorithms and a time-of-flight camera. A coordinator node receives data from the two sub-sensory systems with their associated level of confidence and, on the basis of a data fusion logic, it operates the validation and correlation among the two sub-systems delivered data in order to rise overall system performance with respect to each single sensor sub-system. Achieved results show the effectiveness of the suggested multisensor approach for improving fall detection service in ambient assisted living contexts.

Purpose: Although no cure is available, cognitive rehabilitation (CR) for patients with mild¹ Alzheimer's disease (AD) appears to be an attractive treatment. In the last two years many technical solutions, such as serious games, have been studied for cognitive assistance^2,3. However, most of these initiatives (including commercial products such as Nintendo's Brain Age and Big Brain Academy) provide only memory challenges or random puzzles that are to be played a few minutes per day with the purpose of "improving brain performance". The objective of this work is the design and development of an Information and Communication Technology (ICT) platform that integrates advanced Natural User Interface (NUI) technologies for multi-domain cognitive rehabilitation (temporal and spatial orientation, visual and topographical memory, verbal memory and fluency, visual and hearing attention, etc.). Method: The platform architecture (Figure 1) is made up of a) a set-top-box connected to a TV monitor with internet connection, b) a commercial low-cost RGB-D camera (Microsoft Kinect), and c) an (optional) e-shirt (WWS Smartex) for monitoring vital signs. According to the specific rehabilitation program provided by the physician, the set-top-box automatically downloads customized sequences of exercises from a remote server, taking into account the patient's rehabilitation history and related factors. In order to make the system reliable, flexible, customized, and compliant with the international evaluation scales (MMSE: Mini Mental Scale Evaluation), few input parameters (amount of allowed errors and execution time, movement sensitivity) are set, and these parameters are based on the residual abilities of the patient. The system allows both autonomous execution of the required exercises and data reporting and storing of the daily performance for every exercise. In order to obtain more information during rehabilitation activities, the main vital signs (heart rate, breathing rate, electrocardiogram, etc.) are monitored if the subject wears the (optional) e-shirt. A Bluetooth radio link is used for transmitting clinical parameters to the set-top-box. Results & Discussion: The platform allows CR for AD patients without direct physician involvement in the rehabilitation session. (Performance metrics and clinical parameters are sent to the physician with a multimodal paradigm for clinical evaluations.) For proper interaction with the system, measurements from the patient are acquired by the RGB-D sensor at a distance range between 80 cm and 400 cm, allowing Natural User Interaction through 100% hands detection rate (Figure 2). The system allows an audiovisual link with the medical center, so that the physician can interact with the AD patient during CR, increasing the compliance and efficacy of CR and ensuring that the type and intensity of treatment are appropriate.

Falls are very dangerous events among elderly people. Several automatic fall detectors have been developed to reduce the time of the medical intervention, but they cannot avoid the injures due to the fall. In this paper a study about the feasibility of a wearable system for the detection of the fall risk is presented. The aim is to detect the fall before to the impact on the floor, allowing the intervention of an impact reduction system. An electromyography-based system has been chosen because it can recognize the risk of fall faster than other more commonly used inertial sensors based system. The logical framework implements the feature extraction procedure, demonstrating a higher discriminative power of Co-contraction Indices and Integrated EMG for which at least of 70% of specificity and sensitivity are achieved in the classification process.

Continuous in-home monitoring of older adults living alone aims to improve their quality of life and independence, by detecting early signs of illness and functional decline or emergency conditions. To meet requirements for technology acceptance by seniors (unobtrusiveness, non-intrusiveness, and privacy-preservation), this study presents and discusses a new smart sensor system for the detection of abnormalities during daily activities, based on ultra-wideband radar providing rich, not privacy-sensitive, information useful for sensing both cardiorespiratory and body movements, regardless of ambient lighting conditions and physical obstructions (through-wall sensing). The radar sensing is a very promising technology, enabling the measurement of vital signs and body movements at a distance, and thus meeting both requirements of unobtrusiveness and accuracy. In particular, impulse-radio ultra-wideband radar has attracted considerable attention in recent years thanks to many properties that make it useful for assisted living purposes. The proposed sensing system, evaluated in meaningful assisted living scenarios by involving 30 participants, exhibited the ability to detect vital signs, to discriminate among dangerous situations and activities of daily living, and to accommodate individual physical characteristics and habits. The reported results show that vital signs can be detected also while carrying out daily activities or after a fall event (post-fall phase), with accuracy varying according to the level of movements, reaching up to 95% and 91% in detecting respiration and heart rates, respectively. Similarly, good results were achieved in fall detection by using the micro-motion signature and unsupervised learning, with sensitivity and specificity greater than 97% and 90%, respectively.

The paper presents a wearable system able to evaluate real time the risk of fall in elderly people, promoting the fast adoption of properly intervention strategies for reducing injuries (e.g. by activating an impact reduction system). A wireless and minimally invasive surface Electromyography-based system (EMG) has been used to measure four lower limb muscles activities. This work deals with the identification of highly discriminative features extracted from the EMG signals for the automatic detection of people instability. The framework prototype uses a threshold-based approach assuring real time functioning and permitting the detection of a typical imbalance condition about 200ms after the stimulus perturbation, in simulated and controlled fall conditions.

One distinctive feature of ambient assisted living-oriented systems is the ability to provide assistive services in smart environments as elderly people need in their daily life. Since Time-Of-Flight vision technologies are increasingly investigated as monitoring solution able to outperform traditional approaches, in this work a monitoring framework based on a Time-Of-Flight sensor network has been investigated with the aim to provide a wide-range solution suitable in several assisted living scenarios. Detector nodes are managed by a low-power embedded PC to process Time-Of-Flight streams and extract features related with person's activities. The feature level of detail is tuned in an application-driven manner in order to optimize both bandwidth and computational resources. The event detection capabilities were validated by using data collected in real-home environments.

Cognitive Rehabilitation (CR) is a relatively new approach to improve well-being for people with Alzheimer's disease (AD). At present only preliminary evidence regarding efficacy is available but it is enough to suggest that this kind of rehabilitation has the potential to bring about changes in behavior, enhance well-being and maintain involvement in daily life. The present work presents a digital platform integrating Natural User Interface (NUI) for motor and cognitive rehabilitation of patients with different disease condition. It is made up of an embedded PC connected to a TV monitor with internet connection, a low-cost 3D sensor (we use Microsoft Kinect in order to allow wide diffusion of the proposed solution), and an optional e-shirt with textile electrodes for clinical signs detection. The main contribution of this work is the design and implementation of an information and communication technologies (ICT) platform, through a customized Virtual Personal Trainer (VPT) allowing the patients to perform the rehabilitation practice at home. Moreover, the system provides an audio/visual link with the medical center, so the physician can interact with the patient during the rehabilitation practice, increasing the compliance and the efficacy and making sure that the type and intensity of treatment are appropriate. Customized algorithms for calibration, people segmentation, body skeletonization and hands tracking through the Kinect sensor have been implemented in order to infer knowledge about the reaction of the end-user to the Graphical User Interface (GUI) designedfor specific cognitive domains. For proper interaction, gestures of AD patients are acquired by the sensor in the nominal functioning range, allowing 100% hands detection rate, useful for an error free human-machine.

The paper presents an active vision system for the automatic detection of falls and the recognition of several postures for elderly homecare applications. A wall-mounted Time-Of-Flight camera provides accurate measurements of the acquired scene in all illumination conditions, allowing the reliable detection of critical events. Preliminarily, an off-line calibration procedure estimates the external camera parameters automatically without landmarks, calibration patterns or userintervention. The calibration procedure searches for different planes in the scene selecting the one that accomplishes the floor plane constraints. Subsequently, the moving regions are detected in real-time by applying a Bayesian segmentation to the whole 3D points cloud. The distance of the 3D human centroid from thefloor plane is evaluated by using the previously defined calibration parameters and the corresponding trend is used as feature in a thresholding-based clustering for fall detection. The fall detection shows high performances in terms of efficiency and reliability on a large real dataset in which almost one half of events are falls acquired in different conditions. The posture recognition is carried out by using both the 3D human centroid distance from the floor plane and the orientation of the body spine estimated by applying a topological approach to the range images. Experimental results on synthetic data validate the correctness of the proposed posture recognition approach.

The paper presents an active vision system for the detection of dangerous fall events and the recognition of four main human postures (lie, sit, stand, bend) in Ambient Assisted Living applications. The suggested vision system uses a Time-Of-Flight camera providing accurate 3D measurements of the scene in all illumination conditions. In order to accommodate different installation setups, the system recovers automatically the own 3D position and orientation in the space, according to a floor detection strategy, without human intervention and calibration tools (landmarks, patterns, etc.). The moving people are detected in the 3D points cloud by applying segmentation/tracking methods and metric filtering. The distance of the 3D human centroid from the floor plane is evaluated by using the previously estimated calibration parameters and the corresponding trend is used as feature in a thresholding-based clustering for fall detection. The system shows high performances in terms of efficiency and reliability on a large real dataset of falls acquired in different conditions. The posture recognition is carried out by using both the 3D human centroid distance from the floor plane and the orientation of the body torso estimated by applying a topological approach to the range images. Experimental results on synthetic data validate the soundness of the proposed posture recognition approach.

The paper presents a preliminary study on the feasibility of a pre-impact fall detector, able to evaluate the risk of fall real-time, allowing the fast adoption of properly strategies of intervention for reducing injuries (e.g. By activating an impact reduction system). A wearable, wireless and minimally invasive surface Electromyography-based system (EMG) has been used to measure four lower-limb muscles activities. This work deals with the identification of highly discriminative features extracted within the EMG signals for the automatic detection of people instability. The framework prototype uses a threshold-based approach assuring real-time functioning and allowing the detection of a typical imbalance condition about 200ms after the stimulus perturbation, in simulated and controlled fall conditions.

This paper presents a multi-sensor system for the detection of people falls in the home environment. Two kinds of devices are used: a MEMS wearable wireless accelerometer with onboard fall detection algorithms and a 3D Time-of-Flight camera. An embedded computing system receives the possible fall alarm data from the two sub-sensory systems and their associated level of confidence. The computing module hosts a data fusion software to operate the validation and correlation among the two subsystems delivered data in order to rise overall system efficiency performance with respect to each single sensor sub-system.

Wine aroma volatiles of two different typical Apulian wines made by autochthonous grape varieties (i.e. Negroamaro and Primitivo) were extracted by solid phase extraction (SPE) and analyzed using gas chromatography-mass spectrometry (GC-MS) in conjugation with an electronic nose (E-nose). Eighteen compounds were found over their own odour threshold and they were taken into account for further data analysis. Sensor data were analyzed by principal component analysis (PCA) to investigate the discrimination capability of the sensor array. The concentrations of volatile chemical compounds in wines determined by GC-MS have been correlated with electronic nose (E-nose) responses using partial least squares (PLSs) and quadratic response surface regression (RSR) analysis. By means of these regression models, relationships between E-nose responses and wine aroma compounds were established. Quite all of the 18 wine odorant concentration were predicted at a satisfactory extent; RSR technique gave better prediction results compared to PLS. © 2012 Elsevier B.V. All rights reserved.

The main goal of Ambient Assisted Living solutions is to provide assistive technologies and services in smart environments allowing elderly people to have high quality of life. Since 3D sensing technologies are increasingly investigated as monitoring solution able to outperform traditional approaches, in this work a noninvasive monitoring platform based on 3D sensors is presented providing a wide-range solution suitable in several assisted living scenarios. Detector nodes are managed by low-power embedded PCs in order to process 3D streams and extract postural features related to person's activities. The feature level of details is tuned in accordance with the current context in order to save bandwidth and computational resources. The platform architecture is conceived as a modular system suitable to be integrated into third-party middleware to provide monitoring functionalities in several scenarios. The event detection capabilities were validated by using both synthetic and real datasets collected in controlled and real-home environments. Results show the soundness of the presented solution to adapt to different application requirements, by correctly detecting events related to four relevant AAL services. © 2013 Alessandro Leone et al.

A novel ultra-wideband radar sensor system for simultaneous detection of falls and vital signs is presented. The suggested system is able to deal with real-life conditions, such as lack of real-fall data for training, body movements, several people present, and privacy issues. Micro-Doppler features, extracted from time-frequency spectrograms, are used to classify human actions as related to normal or abnormal activities (falls). A deep learning framework is used to extract and classify such features, also taking into account the specific way the older adult performs activity-of-daily-living actions. Preliminary validation results are very encouraging, showing the effectiveness to achieve good detection performance in assisted living scenarios.

In recent years several world-wide ambient assisted living (AAL) programs have been activated in order to improve the quality of life of older people, and to strengthen the industrial base through the use of information and communication technologies. An important issue is extending the time that older people can live in their home environment, by increasing their autonomy and helping them to carry out activities of daily livings (ADLs). Research in the automatic detection of falls has received a lot of attention, with the object of enhancing safety, emergency response and independence of the elderly, at the same time comparing the social and economic costs related to fall accidents. In this work, an algorithmic framework to detect falls by using a 3D time-of-flight vision technology is presented. The proposed system presented complementary working requirements with respect to traditional worn and non-worn fall-detection devices. The vision system used a state-of-the-art 3D range camera for elderly movement measurement and detection of critical events, such as falls. The depth images provided by the active sensor allowed reliable segmentation and tracking of elderly movements, by using well-established imaging methods. Moreover, the range camera provided 3D metric information in all illumination conditions (even night vision), allowing the overcoming of some typical limitations of passive vision (shadows, camouflage, occlusions, brightness fluctuations, perspective ambiguity). A self-calibration algorithm guarantees different setup mountings of the range camera by non-technical users. A large dataset of simulated fall events and ADLs in real dwellings was collected and the proposed fall-detection system demonstrated high performance in terms of sensitivity and specificity. © 2011 IPEM.

Facial Expression Recognition is still one of the challenging fields in pattern recognition and machine learning science. Despite efforts made in developing various methods for this topic, existing approaches lack generalizability and almost all studies focus on more traditional hand-crafted features extraction to characterize facial expressions. Moreover, effective classifiers to model the spatial and temporary patterns embedded in facial expressions ignore the effects of facial attributes, such as age, on expression recognition even though research indicates that facial expression manifestation varies with ages. Although there are large amount of benchmark datasets available for the recognition of facial expressions, only few datasets contains faces of older adults. Consequently the current scientific literature has not exhausted this topic. Recently, deep learning methods have been attracting more and more researchers due to their great success in various computer vision tasks, mainly because they avoid a process of feature definition and extraction which is often very difficult due to the wide variability of the facial expressions. Based on the deep learning theory, a neural network for facial expression recognition in older adults is constructed by combining a Stacked Denoising Auto-Encoder method to pre-train the network and a supervised training that provides a fine-tuning adjustment of the network. For the supervised classification layer, the M-class softmax classifier was implemented, where M is the number of expressions to be recognized. The performance are evaluated on two benchmark datasets (FACES and Lifespan), that are the only ones that contain facial expressions of the elderly. The achieved results show the superiority of the proposed deep learning approach compared to the conventional non-deep learning based facial expression recognition methods used in this context.

Falls are one of the main causes of disability and death among the elderly. Several inertial-based wearable devices for automatic fall and pre-fall detection have been realized. They use the threshold-based approach above all and their mean lead-time before the impact is about 200-500 ms. The main purpose of the work was to develop a framework for fall risk assessment considering the lower limb surface electromyography. The user's muscle behavior was chosen because it may allow a faster recognition of an imbalance event than the user's kinematic evaluation. Moreover, a machine learning scheme was adopted to overcome the drawbacks of well-known threshold approaches, in which the algorithm parameters have to be set according to the users' specific physical characteristics. Ten kinds of time-domain features, commonly used in the analysis of the lower-limb muscle activity, were investigated and the Markov Random Field based Fisher-Markov selector was used to reduce the signal processing complexity. The supervised classification phase was obtained through a low computational cost and a high classification accuracy Linear Discriminant Analysis. The developed system showed high performance in terms of sensitivity and specificity (about 90%) in controlled conditions, with a mean lead-time before the impact of about 775 ms.

In this paper an algorithmic framework for posture analysis using a single view 3D TOF camera is presented. The 3D human posture parameters are recovered automatically from range data without the usage of body markers. A topological approach is investigated in order to define descriptors suitable to estimate location of body parts and orientation of body articulations. Two Morse function are exploited, the first one provides an Euclidean distance mapping helpful to deal with body self-occlusions. The second Morse function is based on geodesic distance and provides an extended Discrete Reeb Graph description of the main body parts that are head, torso, arms and legs. Geodesic distance function exhibits the property of invariance under isometric transformations that typically occur when the human body changes its posture. The geodesic map of the body is obtained with a two steps procedure. Firstly, a Delaunay meshing is carried out starting from the depth map provided by the 3D TOF camera; secondly, geodesic distances are computed applying Dijkstra algorithm to previously computed mesh. Moreover, a re-meshing method is proposed in order to deal with self-occlusion problem which occurs in the depth data when a human body is partially occluded by other body segments. Experimental results on both synthetic and real data validate the effectiveness of the proposed approach to classifying four main postures: standing, lying, sitting and bending. © 2011 IEEE.

This paper presents a heterogeneous sensor platform for the detection of anomalies in circadian rhythm. Three detectors with different sensing principles are considered: a 3D time-of-flight camera, a MEMS wearable wireless accelerometer and a Ultra-wideband radar. Starting from human postural information obtained by each detector, a simulator of activities and related postures has been designed and implemented within this work. The use of a simulator is motivated by the lack of datasets containing long-term data for the analyzed context. The simulator is able to generate posture sequences calibrated on real experiments performed by each detector involved in the platform. Finally, a reasoner layer infers knowledge by using a suitable activity recognition module. Moreover, with an unsupervised clustering technique, the reasoner is able to detect specific circadian anomalies, thereby providing a tool for clinical evaluations. Experimental evaluation shows the effectiveness of the implemented solution, especially analyzing the performances related to the detection of anomalies varying sensing technology.

A non-invasive system for human posture recognition suitable to be used in several in-home scenarios is proposed and validation results presented. 3D point cloud sequences were acquired by using a time-of-flight sensor in a privacy preserving modality and near real-time processed with a low power embedded PC. To satisfy different application requirements in terms of discrimination capabilities, covered distance range and processing speed, a twofold discrimination approach was investigated in which features were hierarchical arranged from coarse to fine exploiting both topological and volumetric spatial representations. The topological representation encoded the intrinsic topology of the body's shape in a skeleton-based structure, guarantying invariance to scale, rotations and postural changes, and achieving a high level of detail with a moderate computational cost. In the volumetric representation, on the other hand, postures were described in terms of 3D cylindrical histograms working within a wider range of distances in a faster way and also guarantying good invariance properties. The discrimination capabilities of the approach were evaluated in four different real-home scenarios especially related with ambient assisted living and homecare fields, namely dangerous event detection, anomalous behavior detection, activities recognition, natural human-ambient interaction, and also in terms of invariance to viewpoint changes, representation capabilities and classification performance, achieving promising results. The two approaches exhibited complementary characteristics showing high reliability with classification rates greater than 97% in four application scenarios for which the posture recognition is a fundamental function. © 2012 Elsevier Ltd. All rights reserved.

A non-invasive technique for posture classification suitable to be used in several in-home scenarios is proposed and preliminary validation results are presented. 3D point cloud sequences were acquired using a single time-of-flight sensor working in a privacy preserving modality and they were processed with a low power embedded PC. In order to satisfy different application requirements (e.g. covered distance range, processing speed and discrimination capabilities), a twofold discrimination approach was investigated in which features were hierarchically arranged from coarse to fine by exploiting both topological and volumetric representations. The topological representation encoded the intrinsic topology of the body's shape using a skeleton-based structure, thus guaranteeing invariance to scale, rotations and postural changes and achieving a high level of detail with a moderate computational cost. On the other hand, using the volumetric representation features were described in terms of 3D cylindrical histograms working within a wider range of distances in a faster way and also guaranteeing good invariance properties. The discrimination capabilities were evaluated in four different real-home scenarios related with the fields of ambient assisted living and homecare, namely "dangerous event detection", "anomalous behaviour detection", "activities recognition" and "natural human-ambient interaction". For each mentioned scenario, the discrimination capabilities were evaluated in terms of invariance to viewpoint changes, representation capabilities and classification performance, achieving promising results. The two feature representation approaches exhibited complementary characteristics showing high reliability with classification rates greater than 97%. © 2013 Elsevier B.V.

Anomalies in the circadian rhythm may cause psychological or neurological disorders, mainly in elderly people. Early detection of anomalies in circadian rhythm could be useful for the prevention of such problems. This work describes a multi-sensor platform for anomalies detection in circadian rhythm. Three detectors with different sensing principles are considered: a Time-Of-Flight camera, a MEMS wearable wireless accelerometer and an Ultra-Wideband radar. The inputs of the platform are sequences of human postures, even simulated, extensively used both for analysis of Activities of Daily Living and human behavior understanding. A postures simulator, calibrated on real experiments performed by each detector involved in the platform, has been implemented in order to compensate the lack of wide datasets containing long-term data for the analyzed context. Finally, a reasoner layer infers knowledge by using a suitable activity recognition module; by means of an unsupervised clustering technique, the reasoner is able to detect specific circadian anomalies, providing a tool for clinical evaluations. Experimental evaluation shows the effectiveness of the implemented solution and the ability to detect circadian anomalies at varying sensing technology.

This paper reports the description of a multi-sensor platform able to automatically assess the level of physical activity and sedentary time of older adults. The platform has a hierarchical network topology, compound by N detector nodes managing several ambient sensor nodes and one detector node that manages a wearable sensor node. The system provides also one coordinator node that receives high-level reports from detector nodes. The idea of using heterogeneous sensors is motivated by the fact that in this way we expands the number of end-users, as they may accept only a type of sensor technology. The objective assessment was conducted through two main algorithmic steps: (1) recognition of well-defined set of human activities, detected by a 3D vision sensor (ambient node) and a smart garment (wearable sensor node), and (2) estimation of a physiological measure, that is (MET)-minutes. Results obtained in terms of activity recognition (and subsequent physical activity/sedentary time assessment) showed that the integrated version of the platform performs better than each single sensor technology with an overall accuracy obtained using simultaneously data provided from both sensory technologies that is about 5% higher of single sub-system, thus confirming the advantage in using a coordinator node. Finally, an added value of this work is the capability of the platform in providing a sensing invariant interface (i.e., abstracted from any specific sensing technology), since the use of the activities enables the integration of a wide set of devices, providing that they are able to reproduce the same set of features.

This paper presents an open platform for continuous monitoring of clinical signs through a smart and noninvasive wearable device. In order to accomplish a communication in proximity, the Near Field Communication wireless technology is used, providing a fast link between the device and the host, avoiding the pairing (as typically occurs for Bluetooth protocol) and limiting the power consumption. The Arduino ecosystem has been used for prototyping since it allows an easy and open integration of ad-hoc functionalities. The first prototype of the platform has been customized for human body temperature measurement, assuring a lifetime of the battery for at least 2 months. Moreover, the acquisition and related transmission of other kind of clinical signs could be easily implemented, making the platform cost-effective in mobile scenarios.

In this paper, a computational framework for occupancy detection and profiling based exclusively on depth data is presented. 3D depth sensors offer many advantages against traditional video cameras. Occupants' privacy can be assured more effectively because depth information is unsuitable to reveal the person's identity. Notable low-level computer vision tasks can be simplified, thus lightening the computational load. The presented framework is suitable for wall-mounting setups as well as for ceiling-mounting setups, and scales well with the number of people. To take full advantage of depth data and to accommodate specificities of crowded environments, several improvements to the standard computer vision pipeline are suggested. Firstly, the running Gaussian average background model is adapted to work with depth distances in crowded scenes. Secondly, the classical complete linkage agglomerative clustering is boosted by adding edge-based constraints specifically designed for people segmentation in depth data. Thirdly, to reliable discriminate people, specific depth-based features are defined to be used with a Real AdaBoost classifier. The preliminary results achieved by using two different depth sensors and synthetic data are very promising, outperforming existing approaches. Relevant applications for building energy management, such as occupancy profiling and construction of trajectories and density maps, have been also demonstrated.

Currently available technological solutions do not allow to reliably detect falls in the elderly, due to still-open issues on both sensing and processing sides. In fact, the most promising sensing approaches raise concerns for privacy issues (e.g., vision-based approaches) or low acceptability rate (e.g., wearable-based approaches); whereas on the processing side, commonly used methodologies are based on supervised techniques trained with both positive (falls) and negative (non-fall) samples, both simulated by healthy young subjects. As a result of such a training protocol, fall detectors inevitably exhibit lower performance when used in real-life conditions, in which monitored subjects are older adults. In order to address the problem of fall detection under real-life conditions, this study investigates privacy-preserving unobtrusive sensing technologies together with an unsupervised methodology trained exclusively on daily activity (non-fall) data from the monitored elderly subject. Preliminary results are very encouraging, showing the effectiveness to achieve a good detection performance and, most importantly, which is more reproducible in real-life scenarios.

The chapter presents an automated monitoring system for the detection of dangerous events of elderly people (such as falls) in AAL applications. In order to provide a self-contained technology solution not requiring neither the environment rearrangement, nor the presence of specialized staff, nor a priori information about elderly characteristics/habitude, the focus is placed on the classification of human postures and the detection of related adverse events. The people is detected through a non-wearable device (a TOF camera), overcoming the limitations of the wearable approaches (accelerometers, gyroscopes, etc.) for human monitoring (the devices are prone to be incorrectly worn or forgotten). The system shows high performances in terms of efficiency and reliability on a large real dataset of falls acquired in different conditions. The posture recognition is carried out by using a topological approach on the 3D points cloud. Experimental results validate the soundness of the posture recognition scheme. © 2011 Springer Science+Business Media B.V.

This work describes a heterogeneous sensor platform for elderly people useful in Ambient Assisted Living context for sleep disorder evaluation. The platform integrates hardware (ambient and wearable sensors), as well as software components (data simulation tool, reasoning). Three sensors with different sensing principles are considered: a Time-Of-Flight camera, a MEMS wearable wireless accelerometer and an Ultra-Wideband radar. The inputs of the platform are the postural information, even simulated, in common to all involved sensors (i.e., Standing , Bending, Sitting, Lying down). Since they are extensively used both for analysis of Activities of Daily Living and human behaviour understanding. A posture simulator, calibrated on real experiments performed by each sensor involved in the platform, has been implemented in order to compensate the lack of wide datasets containing long-term data. Moreover, the platform integrates a reasoning layer for automatic sleep disorder evaluation by using an unsupervised learning technique. The effectiveness of the platform was demonstrated by preliminary results, exhibiting high accuracy in sleep disorder evaluation using the three aforementioned sensors.

Falling is one of the main causes of trauma, disability, and death among older people. Inertial sensors-based devices are able to detect falls in controlled environments. Often this kind of solution presents poor performances in real conditions. The aim of this work is the development of a computationally low-cost algorithm for feature extraction and the implementation of a machine-learning scheme for people fall detection, by using a triaxial MEMS wearable wireless accelerometer. The proposed approach allows to generalize the detection of fall events in several practical conditions. It appears invariant to the age, weight, height of people, and to the relative positioning area (even in the upper part of the waist), overcoming the drawbacks of well-known threshold-based approaches in which several parameters need to be manually estimated according to the specific features of the end user. In order to limit the workload, the specific study on posture analysis has been avoided, and a polynomial kernel function is used while maintaining high performances in terms of specificity and sensitivity. The supervised clustering step is achieved by implementing an one-class support vector machine classifier in a stand-alone PC. © 2013 Gabriele Rescio et al.

Falls are the leading cause of disability and death among the elderly. Over the years, several inertial-based wearable devices for automatic fall and pre-fall detection have been devised. Under controlled condition, these systems show a high performance for unbalance detection (up to 100% of specificity and sensitivity), however the mean lead time before the impact is about 200-400 ms. Although this period of time is enough to active an impact reduction system (i.e wearable airbag) to minimize injury, it is necessary to increase it so as to improve the system efficiency and reliability. A user's muscle behavior analysis could be more strategic than that of a kinematic evaluation one, permitting a rapid recognition of an imbalance event. This also holds true for several research studies on muscles response during a state of imbalance, whereas a limit number of them deal with the development of wearable electromyography (EMG)-based systems for human imbalance detection, suitable for predicting a lack of balance in real time. With respects these limitations, the main purpose of this work has been the development of a low computational cost expert system for real time and automatic fall risk detection. The analysis of this is achieved through lower limb muscles behavior monitoring. A Machine Learning scheme has been chosen in order to overcome the well-known drawbacks of threshold approaches widely used in pre-fall systems, in which the algorithm parameters have to be set according to the users' specific physical characteristics. Ten kinds of time-domain features, commonly used in the analysis of the lower-limb muscle activity, have been investigated. With a view to reducing the processing complexity, the Markov Random Field (MRF) based Fisher-Markov feature selector was tested. It showed a high degree of accuracy in the EMG-based features selection for lack of balance detection. The Co-Contraction Index, Integrated EMG and Willison Amplitude features have been also considered. The supervised classification phase has been obtained through a low computational cost and a high classification accuracy level Linear Discriminant Analysis. The developed system shows high performance in terms of sensitivity and specificity (about 90%) in controlled conditions, with a mean lead-time before the impact of about 775 ms. Therefore, the feasibility of a quick and wearable surface EMG-based unbalance detection system, by using Machine Learning methodology, has been demonstrated. The system may recognize a fall event during the initial phase, increasing the decision making time and minimizing the incorrect and inappropriate activations of the protection system, in real life scenario.

Fall events can cause trauma, disability and death among older people. Accelerometer-based devices are able to detect falls in controlled environments. The paper presents a computationally low-power approach for feature extraction and supervised clustering for people fall detection by using a 3-axial MEMS wearable accelerometer, managed by an stand-alone PC through ZigBee connection. The paper extends a previous work in which fall events were detected according to a threshold-based scheme. The proposed approach allows to generalize the detection of falls in several practical conditions, after a short period of calibration. The clustering scheme appears invariant to age, weight, height of people and relative positioning area (even in the upper part of the waist), overcoming the drawbacks of well-known threshold-based approaches in which several parameters need to be manually estimated, according to the specific features of the end-user. In order to limit the workload, the specific study on posture analysis has been avoided and a polynomial kernel function is used while maintaining high performance in terms of specificity and sensitivity. The supervised clustering step is achieved by implementing an One- Class Support Vector Machine classifier. © 2013 IEEE.

Falling down events can cause trauma, disability and death among older people. Accelerometer-based devices are able to detect falls in controlled environments. This kind of solution often presents poor performance in real conditions. The aim of this work is the development of a computationally low-cost algorithm for feature extraction and the implementation of a Machine Learning scheme for people fall detection, by using a tri-axial MEMS wearable wireless accelerometer. The proposed approach allows to generalize the detection of fall events in several practical conditions. It appears invariant to the age, weight, height of people and to the relative positioning area (even in the upper part of the waist), overcoming the drawbacks of well-known threshold-based approaches in which several parameters need to be manually estimated according to the specific features of the end-user. In order to limit the workload, the specific study on posture analysis has been avoided and a polynomial kernel function is used while maintaining high performance in terms of specificity and sensitivity. The supervised clustering step is achieved by implementing an One-Class Support Vector Machine classifier in a stand-alone PC. © 2013 IEEE.

The aim of this work is the development of a computationally low-cost scheme for feature extraction and the implementation of an One-class Support Vector Machine classifier for people fall detection, by using a tri-axial MEMS wearable wireless accelerometer, managed by a stand-alone PC through ZigBee connection. The proposed approach allows the generalization of the detection of fall events in several practical conditions after a short period of calibration. The approach appears invariant to age, weight, people's height and the relative positioning area (even in the upper part of the waist) This overcomes the drawbacks of well-known threshold-based approaches in which several parameters need to be manually estimated according to the specific features of the end-user. In order to limit the workload, the specific study on posture analysis has been avoided and a polynomial kernel function is used, while maintaining high performances in terms of specificity and sensitivity. © 2013 IEEE.

The project proposes a comprehensive preventive program of dementia in elderlies with minor cognitive disorders due to neurodegenerative diseases. The program combines physical and cognitive training, by means of an integrated technological system composed of a Virtual Environment, simulating daily activities, a smart garment measuring physiological parameters, a bicycle ergometer and tailoring the system to the specific patient's status. Preliminary results confirm the feasibility of this intervention and appear promising in order to contrast age-related cognitive neurodegeneration.

The main goal of Ambient Assisted Living solutions is to provide assistive technologies and services in smart environments allowing to elderly people to have high quality of life. Since Time-Of-Flight vision technologies are increasingly investigated as monitoring solution able to outperform traditional approaches, in this work a monitoring framework based on a Time-Of-Flight sensor network has been investigated with the aim to provide a wide-range solution suitable in several assisted living scenarios. Detector nodes are managed by a low-power embedded PC to process Time-Of-Flight streams and cxtract features related with person's activities. The feature level of detail is tuned in an application-driven manner in order to optimize both bandwidth and computational resources. The event detection capabilities were validated by using data collected in real-home environments. © 2013 The Authors. Published by Elsevier B.V.

The paper presents an active vision system for human posture recognition, which is an important function of any assisted living system, suitable to be employed in indoor environments. Both hardware and software architectures are defined in order to meet constraints typically imposed by AAL (Ambient Assisted Living) contexts such as compactness, low-power consumption, installation simplicity, privacy preserving and non-intrusiveness. Two different approaches for feature extraction (topological and volumetric) are discussed and the related discrimination capabilities evaluated by using a statistical learning methodology. Experimental results show the soundness of the presented active vision-based solution in order to classify four main human postures (standing, sitting, bent, lying) with an adequate detail level for the specific AAL application. © 2011 IEEE.