In the US the percentage of people living with loneliness is rapidly increasing. This is particularly true when it comes to young people, about half of whom report living with loneliness.
Chronic loneliness is linked to a number of health conditions including depression and Type 2 diabetes, with the CDC reporting that it is more dangerous than obesity and as damaging as smoking 15 cigarettes a day. However, identifying people living with loneliness is not always simple.
However, new research published in JMIR found that data from Fitbit and smartphone devices could help identify college students living with loneliness. Researchers were able to mine this data and identify relevant behavioral characteristics.
“Results showed that fine-grained behavioral features extracted from mobile and wearable time series data can detect low and high levels of loneliness with high accuracy and that these features can distinguish the behavior of students with high levels of loneliness from those with low levels of loneliness,” researchers of the study wrote.
Researchers found that their machine learning algorithm was able to predict with 80.2% accuracy binary levels of loneliness, and changes in that loneliness level with 88.4% accuracy.
Fitbit and smartphone device data revealed that students with high levels of loneliness were spending less time outside of campus at night and on weekends, and were in fewer social places during week days.
“Passive sensing has the potential for detecting loneliness in college students and identifying the associated behavioral patterns,” researchers wrote. “These findings highlight intervention opportunities through mobile technology to reduce the impact of loneliness on individuals’ health and well-being.”
HOW IT WAS DONE
Researchers recruited 160 college students. At the beginning of the semester long study students were given a questionnaire on loneliness.
Scientists then extracted data from Fitbits and smartphones. The data mined included activity and mobility, communication and usage and sleep behavior data.
Researchers then used three analytical methods to conduct the study. The first was an analysis of the surveys provided by college students. Secondly, they used an algorithm to “extract behavior patterns associated with loneliness.”
“We applied Apriori, a well-known frequent itemset algorithm for discovering associations among items in transactional datasets, to extract patterns between the overall loneliness level and combined questions as well as combined behavioral patterns that were most associated with the level of loneliness,” authors wrote.
Lastly, researchers used a machine learning classification to infer the level of loneness and if it had changed over time.
Loneliness is a hot topic in the digital health space. Much of the focus has been on using technology to curb the impact of loneliness. For example, robotic pets have often been used to help seniors fight off loneliness.
This study married the topic of loneliness with another popular digital health conversation — using phone and Fitbit data to monitor mental health. For example, Dr. John Torous who heads up the digital psychiatry division of the department of psychiatry at Beth Israel Medical School, and his team are working on a tool that will allow patients to give their provider access to phone information, such as if they are answering text messages, GPS of their activities and phone call records.