Assessment of Daily Routine Uniformity in a Smart Home Environment Using Hierarchical Clustering

Identification of Daily Living Recurrent Behavioral Patterns Using Genetic Algorithms for Elderly Care

“…Figure 8 shows the distribution of the average durations of each ADL for each person in the dataset, while Figure 9 illustrates the frequency of appearance of each ADL. 𝑛−1 (13) where: 𝑛 is the number of days from the dataset in which the activity ai is performed, 𝑑𝑢𝑟𝑎𝑡𝑖𝑜𝑛(𝑎 𝑖 𝑗 ) is the duration of the activity 𝑎 𝑖 in day 𝑗 and 𝑚𝑒𝑎𝑛 is the average value of the durations of the activity 𝑎 𝑖 in 𝑛 days. Based on the standard deviation the lower and upper bounds of the interval of the duration for the activity 𝑎 𝑖 are computed as follows:…”

“…Many authors are addressing the detection of frequent daily activity patterns and routines of a person using unsupervised or supervised machine learning-based solutions. Mohan et al [13] propose an unsupervised approach consisting of two steps for extracting the daily routine. In the first step, the activity data corresponding to each day are segmented into groups of locations with homogeneous distribution using the superpixels extracted via energy-driven sampling, while the second step groups the activity segments using the hierarchical graph-based region growth algorithm.…”

“…Only a few research is looking at the activity length and sequence of such activities to detect the person's recurrent behavior patterns that are part of their daily routine and relevant deviations from it. The use of IoT sensors can generate large volumes of data that require efficient algorithms capable of processing this data and associated search space to identify recurrent behavioral patterns [13]. The collected data could be incomplete and inaccurate, and in this case, algorithms capable of handling data quality issues are required [13].…”

“…The use of IoT sensors can generate large volumes of data that require efficient algorithms capable of processing this data and associated search space to identify recurrent behavioral patterns [13]. The collected data could be incomplete and inaccurate, and in this case, algorithms capable of handling data quality issues are required [13]. Metaheuristics algorithms could be a viable solution for solving such types of problems since they provide a near-optimal solution for problems with incomplete and inaccurate data or when the computing power is limited, and the execution time must be low [14].…”