Título: Wearable Sensor-Based Sleep Quality Recommendation Based on Association Rules
Autores: Aleksander Romanha Santos, Emely Pujolli da Silva, Rosana Veroneze, Fernando J. Von Zuben
Resumo: Frequent pattern mining and the subsequent proposition of association rules are used here to promote sleep quality from wearable data sensors. The idea is to automatically provide users with relevant and personalized information regarding possible contextual factors that impact their sleep, including physiological monitoring data. This objective was achieved by revealing existing associations between contextual factors and sleep characteristics to the user. Based on data provided by users of Samsumg Galaxy Watch4 over 21 days, the FPGrowth algorithm was used to mine frequent itemsets, followed by the generation of association rules. Due to the potentially large number of generated rules, a summarization is performed by the FPMax variant. FPMax mines the maximal frequent itemsets, thus generating a condensed version of the results capable of preserving as much information as possible. The obtained result is a set of association rules capable of disclosing the users sleep patterns, thus indicating what to improve in his/her routine to get better sleep.
Palavras-chave: Wearable Sensoring; Recommender System; Frequent Pattern Mining; Association Rules.
Páginas: 8
Código DOI: 10.21528/CBIC2023-139
Artigo em pdf: CBIC_2023_paper139.pdf
Arquivo BibTeX: CBIC_2023_139.bib