When
Where
Sagar Samtani, Assistant Professor of Operations & Decision Technologies, Weimer Faculty Fellow,
Director of Kelley’s Data Science and Artificial Intelligence Lab, Kelly School of Business, Indiana University.
Title: From Subjective to Objective Measurements of Mental Health: An Artificial Intelligence (AI)-enabled Analytics Perspective
Abstract: Mental disorders affecting emotions and behaviors have become a significant concern among young adults and college-aged populations. For instance, young adults (18-25) are at high risk for depression, with 18.6% reporting experiencing major depressive episodes, of which nearly half did not receive treatment. In higher education, mental disorders can cause students to leave their educational journey and universities to incur substantial financial losses. Conventional mental health assessments such as PHQ-9 or GAD for identifying people displaying signs of depression or anxiety, respectively, require individuals to recognize key behavioral patterns, are based on self-reports, subjective, prone to misreporting, and may miss moment-to-moment behaviors. Analyzing sensor signals from smart devices (e.g., smartphones) could be the key element in uncovering an individual’s mental health conditions objectively, holistically, and cost-effectively. Artificial Intelligence (AI)-enabled analytics methods have recently shown significant promise in synthesizing sensor signal data from smart devices for various health applications.
In this talk, I will summarize a multi-year and inter-disciplinary AI-enabled mental health analytics initiative at Indiana University (IU) between the Kelley School of Business’ Data Science and AI Lab (DSAIL), IU’s Irsay Institute, and IU’s School of Public Health. I will specifically focus on DSAIL’s development of a Multi-View Agreement Self-Attentive Model (MV-ASAM) for identifying students exhibiting depressive behaviors. Based on deep multi-view learning and attention mechanisms, MV-ASAM captures dependencies between different views of smartphone sensors (e.g., screen time, accelerometers, etc.), aggregates sensor signals into higher-level human behaviors (e.g., physical, social, sleep), and identifies the sensors most critical for pinpointing depressive behaviors. We evaluated our proposed MV-ASAM against benchmark machine learning, DL, and multi-view learning models on the StudentLife and Tesserae datasets. Our proposed ASAM achieves an overall F1-score of 0.92 and an overall AUC of 0.94 and consistently outperforms the benchmark methods. Results from our experiments indicate that regular sleep patterns and properly reducing the amount of time for mobile phone usage at specific times of an academic semester are essential for college students to avoid depressive behaviors. I will conclude the talk by summarizing our team’s roadmap for developing an iOS mobile app (in collaboration with Apple) that includes the proposed MV-ASAM and other related methods, conducting a large-scale pilot study with the app at IU (2024-2025), and subsequently operationalizing mobile application for the IU community. Time will be reserved to discuss potential IU-UArizona collaborations on mobile application development and operationalization.
Bio: Dr. Sagar Samtani is an Assistant Professor and Arthur M. Weimer Fellow in the Department of Operations and Decision Technologies at the Kelley School of Business at Indiana University (IU). He is the founding Director of IU’s Data Science and Artificial Intelligence Lab (DSAIL). Samtani graduated with his Ph.D. from the Artificial Intelligence (AI) Lab in the University of Arizona’s Department of Management Information Systems (MIS). From 2014 – 2017, Samtani served as a National Science Foundation (NSF) CyberCorps Scholarship-for-Service (SFS) Fellow at the University of Arizona. Dr. Samtani’s research focuses on AI-enabled analytics for cybersecurity (open-source software security, Cyber Threat Intelligence (CTI), advanced cyberinfrastructure security, AI risk management, Dark Web analytics) and mental health applications. Samtani has published over 75 articles in leading Information Systems (IS) venues such as MIS Quarterly, ISR, and JMIS, cybersecurity venues such as IEEE TDSC, ACM TOPS, IEEE S&P, and Computers and Security, machine learning venues such as ACM KDD, IEEE TKDE, IEEE ICDM, IEEE Intelligent Systems, and health outlets such as IEEE ICDH and CHITA. His research has received over $5M in funding from NSF and private sources. Dr. Samtani has co-founded workshops on AI for Cybersecurity topics at ACM KDD and IEEE ICDM. He is deeply engaged with industry, serving on multiple advisory councils and regularly presenting at industry venues at RSA, BlackHat, DEFCON, Microsoft, JPMorgan Chase, IT Nation, and others. Dr. Samtani has won several awards for his research, including the ACM SIGMIS Doctoral Dissertation Award in 2019, Runner-Up for the INFORMS Nunamaker-Chen Dissertation Award in 2018, and multiple Best Paper awards. He has also won several teaching awards for his courses on AI for cybersecurity, CTI, and business analytics, including the IU Trustees Teaching Award. In 2022, Dr. Samtani was inducted into the NSF/CISA CyberCorps SFS Hall of Fame for his outstanding contributions to cybersecurity, was listed as a Top 50 Best Undergraduates Business Professor by Poets and Quants, and won the AIS Early Career Award for his early career contributions to the IS field. He has also won the IEEE Big Data Security Junior Researcher Award and the Outstanding Junior Faculty Award at IU (IU’s most prestigious award for junior faculty). He has received over 100 media citations from the Associated Press, WIRED, Forbes, Miami Herald, Fox, Science Magazine, AAAS, Dark Reading, the Indiana Chamber of Commerce, and other outlets. He is a member of INFORMS, AIS, ACM, IEEE, and INNS.