喻园管理论坛2024年第48期(总第980期)

演讲主题: What Symptoms and How Often? An Interpretable AI Approach for Depression Detection in Social Media

主 讲 人: 颜志军，北京理工大学管理学院教授

主 持 人: 杨彦武，信息管理与数据科学系主任、教授

活动时间: 2024年5月10日（周五）16:00-18:00

活动地点: 管院大楼406室

主讲人简介：

颜志军，北京理工大学管理学院教授、博士生导师。兼任中国信息经济学会常务理事、信息系统协会中国分会（CNAIS）常务理事、中国优选法统筹法与经济数学研究会理事、管理科学与工程学会理事，Internet Research、Electronic Commerce Research and Application、《北京理工大学学报》（社会科学版）编委。研究领域包括互联网医疗、健康管理、健康大数据分析。先后主持国家自然科学基金重点国际合作项目、科技部重点研发计划子课题、北京市自然科学基金等重要科研课题40 余项。在MIS Quarterly, Production and Operations Management, Journal of Management Information System 、《管理科学学报》、《系统工程理论与实践》、《中国管理科学》等国内外重要学术期刊和会议发表论文100 余篇，曾获多项省部级科技奖、中国信息经济学会优秀成果奖。

活动简介：

Depression is a common mental disorder involving a depressed mood or loss of pleasure for long periods of time, which induces grave financial and societal ramifications. Social media-based depression detection is an effective method for early intervention to mitigate those consequences. Such a high-stake decision inherently necessitates interpretability. Although a few studies explain this decision based on the importance of input features, these explanations misalign with the depression diagnosis criteria that are based on depressive symptoms. To fill this gap, we develop a novel Focused Temporal Prototype Network (FTPNet) to detect depression and interpret depressive symptoms as well as how often they occur. Extensive empirical analyses using large-scale datasets show that FTPNet outperforms comprehensive benchmark methods with an F1-score of 0.864. Our result also reveals new symptoms that are unnoted in the survey approach, such as sharing admiration for a different life. We further conduct a user study to demonstrate improved interpretability over the benchmark. This study contributes to Information System (IS) literature with a novel interpretable depression detection method in social media. In practice, our proposed method can be implemented in social media platforms to provide personalized online resources for detected depressed users.