AI TRIAL REVEALS WHICH MENTAL HEALTH APPS ACTUALLY WORK FOR STRESSED STUDENTS

Turns out your phone might be smarter about your stress than you think

In a world-first study, Australian researchers have utilised artificial intelligence (AI) to determine which mental health apps effectively assist university students struggling with psychological distress, and which don’t provide significant benefits.

The clinical trial, led by the Black Dog Institute in partnership with UNSW and Deakin University, tested four self-guided smartphone programs focused on mindfulness, physical activity, and sleep hygiene (plus an active control) in more than 1,200 students.

The findings, published in JAMA Network Open, mark a significant leap in how we study and deliver mental health care.

“This is the first time AI has been used to match digital mental health tools to students’ individual needs,” said Black Dog Institute and UNSW Professor Jill Newby.

“We were able to learn, in real time, which brief programs worked best for different levels of distress — so we can deliver faster, more effective and personalised support.”

Delivered entirely through the Vibe Up smartphone app, the study used a cutting-edge adaptive trial design powered by AI.

Instead of the old-school model, long, rigid, and expensive, this system learned as it went.

It tested several app-based programs simultaneously and quickly identified which ones worked best for different levels of distress.

The Results

The results? Mindfulness and physical activity emerged as the most effective approaches for students experiencing severe distress, while those with mild distress responded best to sleep hygiene and moderate physical activity.

For moderate distress, no single winner emerged, suggesting that further work is needed to tailor interventions to this middle group.

Traditional clinical trials often last for years and require large sample sizes.

This AI-driven approach significantly reduced the timeline and the number of participants needed; researchers estimate that it would have taken 25% more people to achieve the same results using the old method.

Professor Sunil Gupta from Deakin University’s Applied Artificial Intelligence Institute said the technology offers a glimpse into a more personalised, efficient future for mental health.

“Our algorithm learns as it goes, sending each person to the treatments that show the most promise for them.

"This helps researchers quickly figure out which combination of treatment will work best for that individual.”

Next up, the team plans to test whether this approach can personalise care for depression using treatments like cognitive behavioural therapy (CBT) and lifestyle-based interventions.

They’re also exploring whether smartphone-tracked data, things like activity levels and sleep patterns, could help predict how well someone will respond to specific treatments.

In short, mental health support might soon be as adaptive as your Spotify playlist, learning what works for you, faster than ever.