A Smartphone-Based System for Detecting Falls Using Anomaly Detection Vincenzo Carletti (B ) , Antonio Greco, Alessia Saggese, and Mario Vento Department of Information Engineering, Electrical Engineering and Applied Mathematics, University of Salerno, Fisciano, Italy {vcarletti,agreco,asaggese,mvento}@unisa.it http://mivia.unisa.it Abstract. As reported by the World Health Organization, falls are a severe medical and financial issue; they represent the second leading cause of unintentional injury death, after road traffic injuries. There- fore, in recent years, the interest in realizing fall detection systems is considerably increased. Although the overall architecture of such sys- tems in terms of its basic components is consolidated, the definition of an effective method to detect falls is a challenging problem due to sev- eral difficulties arising when the system has to work in the real environ- ment. A very recent research trend is focused on the realization of fall detection systems running directly on a smartphone, so as to avoid the inconvenience of buying and carrying additional devices. In this paper we propose a novel smartphone-based fall detection system that consid- ers falls as anomalies with respect to a model of normal activities. Our method is compared with other very recent approaches in the state of the art and it is proved to be suitable to work on a smartphone placed in the trousers pocket. This result is confirmed both from the achieved accuracy and the required hardware resources. Keywords: Smartphone-base falls detection · Embedded pattern recognition · Anomaly detection · One-class classification 1 Introduction The World Health Organization (WHO) defines a fall as an event which results in a person coming to rest inadvertently on the ground or floor or other low level. Globally, falls are a major public health problem [1–3]. An estimated 424.000 fatal falls occur each year, making them the second leading cause of unintentional injury death, after road traffic injuries. Though not fatal, every year approxi- mately 37.3 million falls are severe enough to require medical attention. Such falls are responsible for over 17 million DALYs lost (the Disability-Adjusted Life Year is a measure of overall disease burden, expressed as the number of years lost due to ill-health). The largest morbidity occurs in people aged 65 years or older, young adults aged 15–29 years and children aged 15 years or younger. But, more c  Springer International Publishing AG 2017 S. Battiato et al. (Eds.): ICIAP 2017, Part II, LNCS 10485, pp. 490–499, 2017. https://doi.org/10.1007/978-3-319-68548-9 45