Using Twitter to Predict the Stress Levels of College Students

A Q&A with a UCIPT Postdoctoral Researcher

How can your research on Twitter and human stress be applied to universities?

The stress level of college students is a huge concern for universities because stress can have significant negative impact on their health and academic performance. We recently conducted a research study and found that the content extracted from a person’s weekly tweets can be used to predict that person’s weekly stress level. This finding will help create technologies that monitor and predict levels of stress in college students in real time, and will be able to help universities allocate the appropriate resources to help students manage their stress levels and prevent future health complications.

What methods have you used in your research?

In our study, we followed 200 UCLA college students on Twitter for one full semester. We also asked them to complete weekly surveys that assessed their stress level. At the end of the study, we extracted and analyzed the sentiment and the emotions (anger, fear, love, and joy) expressed in the tweets from each student using various machine learning techniques. Finally, we conducted analyses that looked at the relationship between the students’ stress level and the sentiment and emotions expressed in their tweets.

How did you first get interested in conducting social media research on student stress?

This research has the potential to make a huge impact in people’s lives. Imagine that we can create a technology that can warn people about potential health risks based on their social media data and then create personalized health interventions that help them prevent illness. This technology can also be used for public health agencies to better target their health programs based on predicted needs. As a result, there is potential to significantly reduce healthcare costs.

What exciting trends are you finding in this area of research?

I think there are three main trends currently in this area of research. First, the methods learned from research studies like ours can be applied to other areas of healthcare, such as predicting depression or physical activity level. Second, social media data can be combined with other data sources (e.g., medical records) to improve the accuracy of predicting health risks. Finally, there is a need to create technology and tools geared toward researchers and clinicians that make analyzing social media data faster and easier. This will greatly improve the research and development process in this area.

Who are other researchers in this area that most inspire you?

I am extremely fortunate to work with leading researchers in the fields of psychology and computer science (Dr. Sean Young, Dr. Wei Wang). I draw inspiration not only from researchers in my field, but from people outside of my area as well such as Elon Musk and Steve Jobs. Ultimately, these people inspire me by consistently reminding me to challenge current methods of practice and not be afraid to think outside of the box to find innovative solutions to complex problems.