Credit: University of Waterloo

Analysis by researchers at the University of Waterloo using artificial intelligence (AI) supports the conventional wisdom that taking care of yourself makes you feel good.

Researchers built an AI to identify key words in more than 700,000 anonymous online journal entries written by over 67,000 users of a mobile mood-tracking application. They found strong associations between positive moods and getting enough sleep, eating well and exercising.

Even activities such as getting a haircut or having a manicure were linked to feeling calmer and happier. So, it turns out that mom was right—taking care of yourself really does make you feel better!

"What appears to make people happy are simple things they can do for themselves, such as resting well, eating well, exercising, meditating, yoga," said Lukasz Golab, an engineering professor at Waterloo and supervisor on the study.

The AI model looked at key words associated with particular moods and then analyzed surrounding words and phrases to try to understand the context. For example, if the word 'job' was frequently used by people feeling stressed, the analysis would identify nearby words such as 'busy' and 'overworked' to help explain why people found their jobs stressful.

Golab said the strong associations between positive moods and taking care of yourself provide important insights for both individuals and governments.

"You may not be able to control the world around you, but you can control your lifestyle and your own good habits," he said. "And in terms of public health, it is useful to confirm with real evidence to make sound, data-driven decisions."

Golab said the technology at the heart of the study could be used as a screening tool to flag possible mental health issues in online social media posts.

The lead researcher on the study, "Micro-journal mining to understand triggers," was former Waterloo graduate student Liuyan Chen. Her paper with Golab appears in the journal Computing.

More information: Liuyan Chen et al. Micro-journal mining to understand mood triggers, Computing (2020). DOI: 10.1007/s00607-019-00777-6