6th International Workshop on Mining Actionable Insights from Social Networks

Special Edition on

Healthcare Social Analytics


Talk: Human-Centered Machine Learning for Dangerous Mental Health Behaviors Online

Abstract. Research and industry both use machine learning (ML) to identify and intervene in physically dangerous health behaviors discussed on social media, such as advocating for self-injury or violence. There is an urgent need to innovate data-driven systems to handle the volume and risk of this content in social networks and its propagation to others in the community. However, traditional approaches to prediction have mixed success, in part because technical solutions oversimplify complex behavior and the unique interactions of communities with both individuals and platforms. This talk will focus on recent research on human-centered machine learning as a lens to make these predictions more ethical and compassionate, as well as technically rigorous. I’ll talk a bit about my work in ML for dangerous mental illness behaviors in online communities, like opioid abuse, suicidal ideation, and promoting eating disorders. Then, I’ll discuss some alarming gaps in data science pipelines for generating labels for training data. Our recent work has found challenges in construct validity that jeopardize the state-of-the-art – and I’ll discuss how we’re attempting to fix this. Together, these inform an agenda for human-centered machine learning that is scientifically and technically rich and more considerate of social contexts in data.

Bio. Dr. Stevie Chancellor is an Assistant Professor in the Department of Computer Science & Engineering at the University of Minnesota. Her research combines approaches from HCI and ML to build and critically evaluate human-centered systems, focusing on high-risk health behaviors in online communities. Her work has been featured in The Atlantic, Wired, Smithsonian Magazine, and Gizmodo. Dr. Chancellor recently received her doctorate in Human-Centered Computing from Georgia Tech and completed a postdoctoral appointment at Northwestern University.

Talk: Obtaining insights about non-medical use of prescription medications from social media via natural language processing

Abstract. Prescription medication (PM) misuse is a major public health crisis that has reached epidemic proportions in many countries including the United States. Commonly abused PMs include opioids, depressants and stimulants, and the consequences range from minor side effects such as nausea to serious adverse outcomes including addiction and death. Despite the enormity of the problem, there is a lack of surveillance mechanisms that would enable investigations on the factors contributing to PM abuse, the natural history of the individuals who develop substance use disorders, and the characteristics of the populations affected (eg., age and gender) by distinct classes of abuse-prone PMs. Social media is a potentially useful resource for conducting real-time automatic surveillance of nonmedical PM use, but utilizing social media effectively requires the development of advanced natural language processing and machine learning methods. In this talk, I will present our research, funded by the National Institute on Drug Abuse (NIDA) towards developing the computational infrastructure required for effectively utilize social media data for this task. Our infrastructure involves manual annotation, supervised classification, information extraction, and combining them with social media meta-data such as geolocation. I will present our various data characterization methods and their validations against traditional sources of data such as the CDC Wonder database and the National Survey on Drug Use and Health.

Bio. Dr. Sarker is an assistant professor at the Department of Biomedical Informatics, School of Medicine, Emory University. He also serves as program faculty at the Department of Computer Science, Emory University and Department of Biomedical Engineering, Emory University and Georgia Institute of Technology. His research interests lie at the intersection of natural language processing, applied machine learning and social media. His is currently leading a number of research projects in this space funded primarily by the National Institute on Drug Abuse of the National Institutes of Health, the Centers for Disease Control and Prevention and Emory University.

Talk: Twitter as a tool for understanding sexual assault and family violence

Abstract. Family violence remains a significant public health concern as a result of the COVID-10 pandemic. Statistics Canada reports that individuals, especially women and children, face a higher risk of family violence during COVID 19. The campaigns of "signal for help" and "safe word" were launched to facilitate resources of help for family violence survivors. In the context of telemedicine, useful, validated, and real-time online resources are pivotal to support individuals and families experiencing family violence as well as to strengthen the communication between survivors and non-profit agencies serving family violence victims. This talk will discuss the use of Twitter data in sexual assault and family violence research, limitations, and future directions in using social media data in this field.

Bio. Jia Xue is an Assistant Professor at the Factor-Inwentash Faculty of Social Work co-appointed with the Faculty of Information since 2018 after completing her research fellowship at JF Kennedy School of Government at Harvard University. She received her Ph.D. from the School of Social Policy & Practice and a M.A. in Statistics from Wharton School at the University of Pennsylvania after she completed her law degree from Tsinghua University Law School (China) in 2011. Jia is the founding director of the Artificial Intelligence for Justice lab. She is also affiliated with Schwartz Reisman Institute for Technology and Society. Her research focuses on applying computational and AI approaches to examine various facets of intimate partner violence, domestic abuse, and sexual assault, addressing gender biases in AI, and studying rape myth culture and school bullying in Chinese societies. Jia’s research has been published in scholarly journals such as Journal of Interpersonal Violence, Journal of Family Violence, Violence against Women, Child Abuse & Neglect, American Journal of Public Health, British Journal of Social Work, Journal of Medical Internet Research, and Children and Youth Services Review. Her research has been funded by the Connaught New Researcher Award, Richard B. Splane Applied Social Policy and Social Innovation Fund Faculty Research Grant, SSHRC Insight Development Grant, CIHR COVID-19 Rapid Research Fund, and Tsinghua University-U of T joint call. Her Google Scholar is here. She is currently on Connaught Committee, Research and Innovation and Strategic initiatives at the University of Toronto (2020-2023), serves as the Human-Centered Data Science liaison at the Faculty of Information, the committee for Tsinghua Alumni Academia Club (TAAC), and the faculty chair for U of T – Tsinghua Forum in 2020.