INTELLIGENT BIOMEDICAL DATA ANALYSIS AND PROCESSING Evaluation of artificial intelligence techniques for the classification of different activities of daily living and falls Ivanoe De Falco 1 • Giuseppe De Pietro 1 • Giovanna Sannino 1 Received: 25 September 2018 / Accepted: 20 December 2018 Ó Springer-Verlag London Ltd., part of Springer Nature 2019 Abstract Automatic detection of falls is extremely important, especially in the remote monitoring of elderly people, and will become more and more critical in the future, given the constant increase in the number of older adults. Within this framework, this paper deals with the task of evaluating several artificial intelligence techniques to automatically distinguish between different activities of daily living (ADLs) and different types of falls. To do this, UniMiB SHAR, a publicly available data set containing instances of nine different ADLs and of eight kinds of falls, is considered. We take into account five different classes of classification algorithms, namely tree-based, discriminant-based, support vector machines, K-nearest neighbors, and ensemble mechanisms, and we consider several representatives for each of these classes. These are all the classes contained in the Classification Learner app contained in MATLAB, which serves as the computational basis for our experiments. As a result, we apply 22 different classification algorithms coming from artificial intelligence under a fivefold cross-validation learning strategy, with the aim to individuate which the most suitable is for this data set. The numerical results show that the algorithm with the highest classification accuracy is the ensemble based on subspace as the ensemble method and on KNN as learner type. This shows an accuracy equal to 86.0%. Its results are better than those in the other papers in the literature that face this specific 17-class problem. Keywords Artificial intelligence  Classification  Fall detection  Activities of daily living 1 Introduction The report of the World Health Organization (WHO) [48] confirms that 28–35% of elderly subjects, with ages of 65 or more, fall each year. People over 70 years old are subject to a major probability of falls (32–42%). Additionally, people residing in communities fall less than those spending their lives in nursing homes. These latter are subject to falls with a percentage of 30–50%, and it has been noticed that in 40% of the cases there are recurrent falls. Also, the incidence of falls varies from country to country. Falls and the injuries related to them are a widespread and serious problem, especially for older people. They are potentially life-threatening events and may be simply the first signs of a single problem. Falls lead to hospitalization and increase cost and burden on society and may even lead to death. In fact, as reported in [34], falls and injuries related to them are the reasons why 40% of people who fall die. This percentage varies among countries. In the USA [5], for subjects with ages of 65 or more, the fall fatality rate is equal to 36.8 per 100,000 individuals, with higher incidence among men (46.2) than among women (31.1). In Canada [29], this rate is 9.4 per 10,000 for the same age group. Instead, in Finland, it is equal to 55.4 for male subjects and to 43.1 for female ones per 100,000 individ- uals [14]. Falls impact economically on the society, the family, and the community. Over the world, the costs related to falls are continuously growing. In general, due also to the & Ivanoe De Falco ivanoe.defalco@icar.cnr.it Giuseppe De Pietro giuseppe.depietro@icar.cnr.it Giovanna Sannino giovanna.sannino@icar.cnr.it 1 ICAR, National Research Council of Italy, Via P. Castellino, 111 80131 Naples, Italy 123 Neural Computing and Applications https://doi.org/10.1007/s00521-018-03973-1