An Algorithm to Determine Which Psychological Therapy Will Help Someone’s Mental Health

Can we create an algorithm that predicts which psychological therapy would work for someone?

The project

At any given time, 15% of the UK population experience depression or anxiety – conditions that can have a profoundly negative impact on people’s lives, and at their most severe, are life-threatening.

Research has shown that psychological therapies such as cognitive behaviour therapy, exposure therapy and psychotherapy are effective treatments for depression and anxiety. And in England, people can receive these through the NHS programme: Improving Access to Psychological Therapies (IAPT).

Most individuals using IAPT currently start out receiving low intensity treatment, and then those who don’t respond well are provided with higher intensity interventions.

But right now, 40% of people in IAPT don’t see an improvement – partly because we don’t have the tools to be able to predict which treatment will work best for a particular person.

This is the challenge that Dr Rob DeRubeis and Zachary Cohen are taking on.

The process

Rob and his team at University of Pennsylvania will run a ‘data tournament’ to create the algorithm – working with teams of mental health experts from around the world who specialise in personalising care.

The teams will use information from over 6,000 people referred to IAPT in Northern England to predict whether someone will benefit most from a low or high intensity treatment.

A sample of the data will be kept back in order to assess how effective the algorithms are at helping people get the right treatments.

The potential

The most powerful algorithm could be rolled out across IAPT services UK-wide, enabling clinicians to accurately recommend which treatments will work for which people – so more people can receive the help they need and get better faster.