A study in Ireland has used artificial intelligence to help predict how people with depression respond to online therapy, which could help clinicians select the most beneficial therapy for patients early on. 

Depression affects millions of people worldwide and, although there are effective treatments, half of the patients do not respond to treatment and only one-third achieve remission. Improving treatment selection earlier can greatly reduce disease burden.  

Researchers from Trinity College Dublin (TCD), an EARA member, have developed a machine learning model, a type of artificial intelligence specialised in learningpatterns from data, that could help clinicians predict if patients responded well to clinician-guided internet-based cognitive behavioural therapy (digital CBT) – a self-paced and easy-to-use anxiety and depression management tool, designed to help people understand how their thoughts and behaviours impact their mood. 

A total of 883 patients receiving medication or digital CBT for four weeks self-reported data such as social and health factors (like stressful events, social support, exercise, diet), depression symptoms (low mood, lack of interest), among others. Using these data, the researchers designed an AI model that could predict 19% of the variance in patients’ improvement, specifically the ones undergoing CBT, since the model was unable to predict patients’ responses to medication. 

“While 19% may seem modest, given the scale of the global depression treatment gap, even small improvements in our ability to allocate treatments more effectively could have a substantial impact on health and wellbeing, quality of life, and the economic burden of disease,” said Claire Gillan, from TCD and leader of the study published in JAMA Network Open.