Falls are considered the main cause of fear and loss of independence among the elderly population and are also a major cause of morbidity, disability and health care utilization. In the majority of fall events external support is imperative in order to avoid major consequences. Therefore, the ability to automatically detect these fall events could help reducing the response time and significantly improve the prognosis of fall victims. This paper presents a unobtrusive smartphone based fall detection system that uses a combination of information derived from machine learning classification applied in a state machine algorithm. The data from the smartphone built-in accelerometer is continuously screened when the phone is in the user's belt or pocket. Upon the detection of a fall event, the user location is tracked and SMS and email notifications are sent to a set of contacts. The accuracy of the fall detection algorithm here proposed is near 97.5% for both the pocket and belt usage. In conclusion, the proposed solution can reliably detect fall events without disturbing the users with excessive false alarms, presenting also the advantage of not changing the user's routines, since no additional external sensors are required

Monitoring physical activity and energy expenditure is important for maintaining adequate activity levels with an impact in health and well-being. This paper presents a smartphone based method for classification of inactive postures and physical activities including the calculation of energy expenditure. The implemented solution considers two different positions for the smartphone, the user's pocket or belt. The signal from the accelerometer embedded in the smartphone is used to classify the activities resorting to a decision tree classifier. The average accuracy of the classification task for all activities is 99.5% for the pocket usage and 99.4% when the phone is used in the belt. Using the output of the activity classifier we also compute an estimation of the energy expended by the user. The proposed solution is a trustworthy smartphone based activity monitor, classifying the activities of daily living throughout the entire day and allowing to assess the associated energy expenditure without causing any change in user's routines

Strategies for fall risk assessment are currently not multifactorial neither implemented as a regular assessment of health status in clinics or hospital. The reason could be related with a lack of an easy to implement, complete and objective test to assess elderly's fall risk level. More recently, inertial wearable sensors have been used in combination with standard tests to evaluate the performance of the person during each phase of the test in an objective way. This paper proposes a methodology for collecting and analyzing the Timed-Up and Go (TUG) test instrumented with wearable inertial sensors. An automatic algorithm to segment the TUG test into three components was implemented prior to feature extraction. Overall, features from the walking and first turning phases of the tests could provide meaningful information to differentiate groups of high and low fall risk

The past years have witnessed a boost in fall detection-related research works, disclosing an extensive number of methodologies built upon similar principles but addressing particular use-cases. These use-cases frequently motivate algorithm fine-tuning, making the modelling stage a time and effort consuming process. This work contributes towards understanding the impact of several of the most frequent requirements for wearable-based fall detection solutions in their performance (usage positions, learning model, rate). We introduce a new machine learning pipeline, trained with a proprietary dataset, with a customisable modelling stage which enabled the assessment of performance over each combination of custom parameters. Finally, we benchmark a model deployed by our framework using the UMAFall dataset, achieving state-of-the-art results with an F1-score of 84.6% for the classification of the entire dataset, which included an unseen usage position (ankle), considering a sampling rate of 10 Hz and a Random Forest classifier.