Discovering association rules from responded questionnaire for diagnosing geriatric depression

Due to the pressure from work load and daily life, there is an increase in geriatric depression population. However, some people may not notice or have no idea about the symptom of melancholia. More research input is needed to diagnose severity of melancholia at an early stage. To help users diagnose their physical fitness and mental health condition before outpatient service, two approaches are considered in this paper. One is from responded questionnaire. We apply data mining strategy to discover association rules from responded questionnaire, including geriatric depression, BAI, ASRM, and PSQI. The other is from user's recorded daily emotion. We devise user interfaces on smart phones for users to record their daily emotion. The proposed system can extract the association rules among negative emotion and help users understand their emotional variations. The discovered association rules can provide valuable information for psychiatrists to make more accurate diagnosis before outpatient service. To obtain informative analytical results, multitudes of simulations are performed on 2,500 data stored in our database. Simulation results under different combinations of score level, minimum support and minimum confidence are given for comparisons and to verify the feasibility and effectiveness of the proposed system.