³Ô¹ÏÍøÕ¾

Can AI personalise mental health treatments for university students?

Department of Health

Professor Helen Christensen from the University of NSW wants to help students reduce their psychological distress, especially at exam time.

This is important because psychological distress can lead to poorer outcomes for university students, including:

  • performing badly
  • dropping out
  • using more alcohol and cigarettes
  • increased suicide risk
  • more severe mental health conditions in the future.

We don’t know which mental health therapy is best for each student

‘There are a range of mental health therapies that could help students,’ Helen tells us. ‘The problem is we don’t know which therapy is best for whom, and which is going to work first.’

For this reason, ‘students often have to have a number of goes at different therapies before they find one that works for them.’

AI can help

AI’s ability to process a large amount of diverse information could help solve this problem. ‘AI can find connections among different levels and types of information. These connections may not emerge using traditional statistical and data models,’ Helen tells us.

This gives AI ‘potential to find characteristics of students that might predict their reaction to different types of therapies.’

To test this potential, Helen teamed up with computer scientist Professor Svetha Venkatesh from Deakin University to create the Vibe Up trial.

Vibe Up

Vibe Up uses a smartphone app to deliver 1 of 3 mental health therapies to university students. As a control, for some students the app just tests how they feel. The therapies are 2-week, self-directed programs designed to ease psychological distress using:

  • mindfulness

  • physical activity

  • sleep hygiene.

The aim of the trial is to find out which treatments are best for people based on their level of psychological distress. This is measured by the Depression Anxiety Stress Scales questionnaire. The researchers also want to find out which therapy:

  • reduces psychological distress the fastest
  • works best for individual people.

An adaptive trial

To answer this question, Helen and Svetha use AI to deliver the trial faster and more efficiently. Over a series of mini trials, an algorithm calculates results and uses this information to adapt the trial.

‘Our algorithm allocates more people to treatments that are working well. This lets it differentiate the best treatment faster,’ Svetha explains.

‘Importantly, the results show that physical activity and mindfulness can be effective for severe psychological distress,’ Helen says. ‘This is contrary to the views of many clinicians.’

Vibe Up is a new model for digital mental healthcare

‘The results from Vibe Up are as good as you might get with brief face-to-face treatment,’ Helen concludes.

‘I think that is because useful treatments are given at the time students need them.’

Using AI, digital treatments ‘deliver a therapy that is more likely to work for each person. As a result, students are more likely to stick to the treatment and have better outcomes.’

Svetha adds, ‘As a computer scientist, I feel inspired if we can apply the fundamental AI we developed to solve a problem. It’s been a massive amount of work and very satisfying to do.’

The Medical Research Future Fund awarded $4.995 million to Helen and Svetha’s research study ‘Optimising treatments in mental health using AI’.

/Media Release. View in full .