Machine Learning Applied to Datasets of Human Activity Recognition: Data Analysis in Health Care

BACKGROUND: Older adults who have poor health, such as those in personal conditions motivate them to remain active and productive, both at home and in geriatric homes, they need a combination of advanced methods of visual monitoring, optimization, pattern recognition and learning, that provide safe and comfortable environments and that once serve as a tool to facilitate the work of family members and workers. It should be noted that this also seeks to recreate a technology that gives these adults autonomy in indoor environments., OBJECTIVE: Generate a prediction model of activities of daily living through classification techniques and selection of characteristics, to contribute to the development in this area of knowledge, especially in the field of health, to carry out an accurate monitoring of activities of the elderly or people with some type of disability. Technological developments allow predictive analysis of activities of daily life, contributing to the identification of patterns in advance, to take actions to improve the quality of life of the elderly., METHOD: The vanKasteren, CASAS Kyoto and CASAS Aruba datasets were used, which have certain variability in terms of occupation and the number of activities of daily life to be identified, to validate a predictive model capable of supporting their identification. activities in indoor environments., RESULTS: After implementing 12 classifiers, among which the following stand out: Classification Via Regression, OneR, Attribute Selected, J48, Random SubSpace, RandomForest, RandomCommittee, Bagging, Random Tree, JRip, LMT and REP Tree, are analyzed in the light of precision and recall quality metrics, those classifiers that show better results when identifying activities of daily life. For the specific case of this experimentation, the Classification Via Regression and OneR classifiers obtain the best results., CONCLUSION: The efficiency of the predictive model based on classification is concluded, showing the results of the two classifiers Classification Via Regression and OneR with quality metrics higher than 90% even when the datasets vary in occupation and number of activities. Copyright© Bentham Science Publishers; For any queries, please email at [email protected].