Danish researchers develop algorithms for mental health

Associate Professor Melanie Ganz-Benjaminsen from the Department of Computer Science at the University of Copenhagen has just received a grant of DKK 20 million from the Lundbeck Foundation's Collaborative Projects program. The grant will support the development of the TRUSTMIND research project, which investigates the use of artificial intelligence (AI) in mental health – with a particular focus on depression.

The TRUSTMIND project builds on research from the Neurobiology Research Unit at Rigshospitalet, with which Melanie Ganz-Benjaminsen is also affiliated. Rigshospitalet will therefore provide the necessary data for the project.

European healthcare systems are under pressure, especially in the field of psychiatry. Long waiting times for diagnosis and treatment are a challenge, and in Denmark, patients wait an average of 92 weeks for a specialist consultation. With the TRUSTMIND project, the researchers aim to investigate how AI can be used to improve the prioritization of healthcare resources – but with a critical eye on the technology's limitations, explains Melanie Ganz-Benjaminsen.

- We are not just developing AI algorithms to solve the problems, but to understand their weaknesses and challenges in a realistic context, explains Melanie Ganz-Benjaminsen.

Artificially intelligent decisions?

The purpose of the project is not to create ready-to-use clinical tools, but rather to simulate a development process for AI algorithms in mental health. With access to extensive datasets from Danish registers and existing research projects, the project will identify the errors and limitations that can arise when complex algorithms are applied to real-world data.

According to Melanie Ganz-Benjaminsen, the goal is to create a test environment that can reveal the challenges that can be expected if artificial intelligence is used for decision-making in the healthcare sector.

- We draw on experiences from other sectors that already use algorithms for decision-making, such as recruitment and the legal system. Our project is particularly relevant in psychiatry, where predictive algorithms are already being developed in Denmark to reduce readmissions, says Melanie Ganz-Benjaminsen.