UL’s Professor Pepijn van de Ven will discuss his research into using simple AI models to inform mental health interventions.
Lately, there has been a lot of talk in the technology world about the use of AI in healthcare.
Last month, prominent generative AI companies OpenAI and Anthropic launched dedicated healthcare-focused services for their respective chatbots.
Both ChatGPT Health and Claude for Healthcare’s features were developed to help users with tasks like understanding test results and preparing appointments, but some are eyeing AI’s potential in more focused areas of the healthcare field.
One such researcher is Professor Pepin van de Ven, Professor in the Department of Electronic and Computer Engineering at the University of Limerick (UL).
Mr Van de Ven, who has a background in electronics engineering and a PhD in artificial intelligence, is currently course leader on Ireland’s national master’s degree in AI, run by UL in close collaboration with ICT Skillnet, and founding director of UL’s D2iCE research center, which researches the development and deployment of AI with a focus on the ethical, sustainable and trustworthy use of AI in society.
Van de Ven’s research currently focuses on the use of AI in mental health interventions.
“I’ve been very fortunate to have had the opportunity to work with pioneers of what we call Internet interventions, interventions delivered over the web,” he told SiliconRepublic.com.
“For the past 15 years, I have contributed to research programs focused on the use of smart technologies in the delivery of mental health interventions, with partners in Europe, Australia, the Americas and, of course, Ireland.”
He explains that the contribution he and his team have made to these projects revolves around using artificial intelligence to improve the delivery of the aforementioned interventions.
“For example, we find that AI can do time-consuming patient screenings that clinicians would otherwise have to do, freeing up clinician time for patient contact,” he says. “Screening interviews like this tend to use a series of questionnaires, which can be a real burden for patients. We’re doing a lot of work in analyzing questionnaires that are typically used in mental health during screening to see if they can be shortened.”
“When considering the use of AI, we need to think very carefully about its use to prevent unintended consequences.”
Advantages and precautions
Van de Ven believes his research is important because it has the potential to help an area of medicine that has long lacked proper attention.
“Unfortunately, there is still significant stigma around mental health, and services tend to be under-resourced. The thoughtful use of AI has the potential to lower the barrier to access these services, and also make the delivery of these services more efficient.”
“As our population ages, the need for health services, including of course mental health services, will only increase. I think the simple fact is that the only way we can guarantee quality services for everyone is through the use of AI.”
He says one of the misconceptions people have about his work is the belief that “AI is equivalent to generative technologies like ChatGPT.”
“Given the impressive advances in generative AI, this misconception has led to a lot of hesitation about using AI,” he says. “The model we are using is very simple compared to ChatGPT.”
He explained that using simple AI models in such sensitive areas reduces the risk of harm to patients, adding that he cautioned against using generative AI and large language models to replace human staff in services such as counseling.
“We have to be very careful,” he says. “I support the judicious use of AI to support the role of healthcare providers and enable them to spend as much time with patients as possible.
“We’ve all heard stories of people using generative models like ChatGPT to discuss mental health issues and actually confide in these AI models. And unfortunately, this has had devastating consequences in some cases.”
For example, in December, OpenAI was sued over claims that ChatGPT encouraged a mentally ill man to kill his mother and himself.
“Currently, we cannot guarantee how generative models will respond to prompts, so further research and careful testing is needed before such usage becomes mainstream.
“Like most other technologies, any AI model has the potential to cause harm, but the simple models we develop are useful for very narrow tasks, and often do so in a way that clinicians can understand,” he says. “As a result, their ability to cause harm is limited and well understood.”
persona
One of the projects in which Van de Ven and his team are the only non-Danish partners involved is the Persona project, which Van de Ven says aims to adapt the fully online mental health services already in use in the Danish healthcare system to a “so-called stepped care model.”
He explains that the model provides support to patients across three different steps or levels.
At the lowest level, patient engagement is self-directed, while the second level incorporates a blended approach where patients can access self-directed treatment while also having therapists available for online sessions.
The last step or level is the “traditional approach,” where patients see a therapist for each session, albeit in an online format, he says.
“The hope is that this stepped care approach will result in more efficient use of health care resources, giving us the opportunity to treat more people with available resources,” he says. “Our role in this project is to create an AI model that can predict the type of intervention a patient will need based on an evaluation of the information people provide when receiving a service.
“In the future, we hope that our model will be able to inform which stage of a stepped care model a patient should receive.”
Regarding Persona’s current progress, Van de Ven says that the Danish project partners have created a new intervention suitable for delivery at these three different levels, as well as an entirely new mobile platform to support the delivery of the intervention.
“After two years of effort, trials have recently begun and are progressing well. We hope to obtain a lot of interesting data in the near future to further improve the performance of our AI models.”
Speaking of the future, what are Van de Ven’s hopes for the long-term impact of his work?
“We hope to do the right thing by improving the services provided to people with mental illness and their loved ones,” he says. “Internet intervention and AI will play an important role in this process, but AI is truly a double-edged sword.
“When considering the use of AI, we need to think very carefully about its use to prevent unintended consequences.”
Don’t miss out on the knowledge you need to succeed. Please sign up for daily briefsSilicon Republic’s digest of science and technology news you need to know.
