A Smartphone-Based Personalized Activity Recommender System for Patients with Depression

Abstract

According to the World Health Organization (WHO) there is currently rapid growth in the proportion of the population suffering from depression Hence depression might not be disregarded making awareness of negative emotions a helpful treatment In recent years many studies have used smartphones to identify and track emotion states in both healthy participants and patients with mental disorders However most previous approaches did not provide additional feedback about appropriate activity that helping users to improve emotion to the user As a result this study was devoted to building a smartphone-based context-aware relaxation activity recommender system that increased users’ awareness of their emotions and helped them to alleviate negative emotions This study developed the context-aware activity recommender method and implemented it on smartphones In order to consider the variety and applicability of activities in our system our proposed method created recommendation lists by referring to users’ environmental situation and activity histories and preferences of similar users Finally we implemented a personalized smartphone-based activity recommender system Using application usage patterns our system instantaneously made an appropriate recommendation upon identifying users’ negative emotions and then helped users to alleviate these emotions This experiment was divided into two parts In the first we collected healthy participants’ data to evaluate the accuracy and availability of the system Here the results achieved a mean reciprocal rank (MRR) of 0 38 In the second part we collaborated with clinical psychiatrists to gather data on patients with depression and discussed the usability and validity of system In this case we obtained a MRR of 0 28 Furthermore we found that the improvement rates of depression stress and anxiety were 15 28% 5 63% and 10 25% respectively In this study we successfully developed a smartphone-based context-aware personalized activity recommender system Through data collection over a period of 14 days and a preliminary clinical experiment we verified the availability and validity of our system Finally we expect that this recommender system can increase users’ awareness of their emotions assist with emotion management and help clinical psychiatrists with tracking users’ mental states

Date of Award	2015 Aug 18
Original language	English
Supervisor	Jung-Hsien Chiang (Supervisor)