С��Ƭ��Ƶ

Before leaving the house, you most likely check to ensure you have your ID, your shoes and most importantly your smartphone. In the past decade, American smartphone usage has grown more than 50% according to a . Smartphones have become as commonplace as a wallet or car keys and С��Ƭ��Ƶ researchers are taking advantage of this new commodity by using cell phone data to study individuals’ behavioral patterns during the COVID-19 pandemic and link cell phone use behaviors to mental health. 

Ruoming Jin, Ph.D., partnered with Deric Kenne, Ph.D., in an exploratory research effort to develop a computer learning framework that collects mobile sensor data and tracks participating smartphone users’ movements while keeping personal information private. 

Image

“We have an interest in understanding college students’ behavior and how they behaved during the pandemic as a representation of the overall population,” said Jin.

The pilot-study is funded by a $150,000 grant from the National Science Foundation as well as funding from the University Research Council.

Jin, a professor in the Department of Computer Sciences in С��Ƭ��Ƶ’s College of Arts and Sciences, explained that study participants will download an app allowing sensor-based metadata to be pulled and analyzed in the first stage, and in a second stage, the participants will help test the app  which can predict their behavior and mental wellness through  federated learning machine process, a process emphasized in privacy protection.

Image

“In the last few years there’s been a lot of interest in building a federated learning framework,” Jin said, “which essentially allows every person’s personalized data to be used in the learning framework without sharing all data to the cloud.”

Jin explained that by using a federated learning framework, mobile data can be collected and interpreted without including personalized information. Study participants’ personal details will be protected while the metadata, things like location, screen time and sensor data, will contribute to the overall machine learning process.

“We cannot see the content of what you really do, only the profile,” Jin said. 

Sensor data will be used to give researchers a sense of physical behaviors, whether an individual is sitting, standing or riding a bicycle. In terms of pandemic responses, it can be used to see how often the person is at home or how much time is spent on their phones. 

The app will also prompt participants to fill out short surveys and complete self evaluations to gauge anxiety and mental health effects. 

“The app will periodically ask questions about what you are doing, and send surveys to learn the person’s mental state,” Jin said. “Those data points will help us to potentially link the person’s behaviors to their mental health.” 

Image

Jin explained that beyond the COVID-19 framework, the app could be developed as a potential mental health resource for students that would be specified to that individuals’ physical behaviors and mental health responses. 

Kenne, an associate professor in the College of Public Health, said the mental health component has the potential to act as an early intervention resource for students.

“If we’ve got students walking around with cell phones and we can detect certain levels of depression or anxiety, we can give the student feedback that there might be issues of depression creeping up,” Kenne said. “Depression and anxiety is different for everyone, it can ebb and flow and goes in waves. If this works it’s an opportunity to pick up on those things very early and be able to intervene if necessary.”

Kenne explained that intervention from the app could look like a message sent from the app or possibly a peer-led care team that could reach out to students to prevent a mental health issue from becoming more severe. 

“This pilot study will help us work out kinks with the app; maybe students don’t respond to messaging through the app, so we can tweak things going forward,” Kenne said. “I see years and years of research evolving from this initial study.”

Kenne said the popularity of smartphone use among several generations opens a large demographic range for future studies. 

“There is such broad applicability with this technology. We are starting with the student population because it’s convenient for us and it’s important, but we potentially could be reaching populations from 10-years-old all the way to senior citizens,” Kenne said. “Everybody could be part of this at some point.”

This study is a collaboration with New Jersey Institute of Technology. Jin explains there will be students from both campuses contributing, and the study will involve three months of tracking sensor data.

For more information on С��Ƭ��Ƶ’s Department of Computer Science visit: /cs

For more information on the Center for Public Policy & Health in С��Ƭ��Ƶ's College of Public Health visit: /mhsu