Predictive Apriori Algorithm in Youth Suicide Prevention by Screening Depressive Symptoms from Patient Health Questionnaire-9

• Main Authors: Yaowarat Sirisathitkul, Putthiporn Thanathamathee, Saifon Aekwarangkoon
• Format: Article
• Language: English
• Published: UIKTEN 2019-11-01
• Series: TEM Journal
• Subjects: depression; feature selection; predictive apriori algorithm; random forest; suicidal risk
• Online Access: http://www.temjournal.com/content/84/TEMJournalNovember2019_1449_1455.pdf
• ISSN: 2217-8309; 2217-8333

This study employed the Predictive A priori algorithm in identifying significant questions of Patient Health Questionnaire-9 (PHQ-9) for suicide tendency prediction by using PHQ-9 and suicidal screening form (8Q). The random forest was applied to calculate the classification accuracy of PHQ-9 and 3 feature selection algorithms were applied to determine the attribute importance. The Predictive Apriori algorithm was applied to find the meaningful association rules. The classification accuracy of PHQ-9 is 92.12% and item no. 1 and no. 9 of PHQ-9 are less important. The significant risk factors associated with suicidal ideation are Item no. 2, no. 4, and no. 5.