Promising future for AI-driven addiction care

Alcohol

Karolina Bergström, freelance journalist
Published 7 Feb 2024

Artificial intelligence (AI) can function as a therapist, predict the outcomes of addiction treatments, and may soon be able to summarize medical records. The development of AI in addiction treatment is flourishing, but experts underscore the need to carefully weigh the benefits against the risks.

Chatbots can provide cooking advice and teach email composition, and AI-generated images can mimic anything from war scenes to the Pope wearing a designer down jacket. AI has become a significant part of many people’s lives in the past years. Now, a subtle AI revolution is also underway in addiction care.

A study at Yale University School of Medicine reveals that AI can efficiently identify patients at a higher risk of relapsing into alcohol abuse, determining whether additional support beyond conventional treatment is necessary. Apps like cognitive therapy based Vorvida and the Swedish app Previct offer AI-based treatment for alcohol and drug addiction, with simple digital interventions showing a 25 percent reduction in alcohol consumption, according to a Swedish study.

Picture of Joar Guterstam
Joar Guterstam

AI-guided selfhelp

Joar Guterstam, an addiction researcher at Karolinska Institutet and a senior physician at Beroendecentrum Stockholm, envisions AI-supported internet psychiatry as a facilitator of future addiction care.

“This development is so rapid that things that seemed far-fetched a year ago are possible today. For example, guided self-help based on artificial intelligence is already being tested within addiction care and psychiatry in general,” says Joar Guterstam.

Guterstam is cautiously optimistic about AI-based applications for addiction care. While he sees how these can conceivably work with conventional call processing and increase accessibility, he also acknowledges the risks in the rapid development.

– Instead of traditional scientific studies with development, publication, and evaluation, this often seems to be about “trial and error” and that the program learns step by step, which makes it more difficult to exercise traditional control. There, I see a challenge to weigh benefits against potential risks in the future, says Guterstam.

The limitations of chatbots

In the AI-driven addiction care of the future, artificial intelligence could serve as built-in decision and action support in electronic medical record systems, flagging ongoing treatments at risk of failure. This is at least what  Philip Lindner, a psychologist and associate professor at Karolinska Institutet, where he leads a research group on digital psychiatry, hopes.

Picture of Philip Lindner
Philip Lindner

– AI could also provide suggestions for messages that can be sent to patients; patients often describe how they miss contact between medical appointments. At the same time, it is easy for people to lose trust in new technology; it is enough for it to fail once, and the whole illusion of a working interaction is gone. So, if you are going to launch, for example, a chatbot, you have to be clear about its limitations, that it can help you with certain aspects, but not all, says Lindner.

AI-predicted outcome of treatment

In an ongoing project, Lindner has examined how artificial intelligence can predict the outcome of internet treatment for alcohol problems by analyzing written text responses from participants in a program.

– Previous research, including our own, has found it quite challenging to predict who will do well in treatment for alcohol problems. But in that case, you have only been able to use structured information such as age, gender, and input values. We believe that there is much greater predictive potential in what people freely express, even if advanced processing is required to exploit such data, says Philip Lindner.

Will free up resources

Picture of Katarina Krokeborg
Katarina Krokeborg

In December 2023, a conference called Teknologi og kunstig intelligens på rusfeltet, arranged by Norwegian policy network on alcohol and drugs Actis, was held in Oslo. It focused on the role of artificial intelligence in future addiction care.

Katarina Krokeborg, one of the conference’s speakers and department head of the addiction clinic at the hospital in Vestfold, envisions that various forms of treatments may occur on digital platforms based on self-help or combined with digital contact with the therapist.

But it is important to remember that human contact is a basic need, she emphasizes.

– Patients have very different needs, but if digital aids enable some to manage more independently, it will free up resources, which can be used for those who need more treatment, says Katarina Krokeborg.

 

 

The article is written by

Karolina Bergström, freelance journalist

 

This article is produced in collaboration between Alkohol & Narkotika and PopNAD. Here, you can read the text in Swedish.

 

 

FacebookXLinkedInEmailPrint