Medical ethicists from Oldenburg are investigating ethical aspects of the use of artificial intelligence in psychiatric practice. The focus is on the diseases Alzheimer's and depression.
Today, computers based on machine learning are already able to register the smallest changes in brain scans that can indicate the onset of Alzheimer's dementia - long before even trained specialists notice these changes. Artificial intelligence (AI) could also be used to treat other mental illnesses in the future. Researchers at the University of Oldenburg are now investigating the scientific, ethical and social significance of this development for neuropsychiatric research and practice in a new research project.
The Federal Ministry of Education and Research is funding the TESIComP project ("Theoretical, ethical and social implications of AI for neuropsychiatric research and practice") with more than 700,000 euros over three years. Medical ethicist Prof. Dr Mark Schweda from the Department of Health Services Research is leading the joint project, in which the University of Oldenburg is collaborating with the Rostock site of the German Centre for Neurodegenerative Diseases (DZNE) and Heidelberg University Hospital.
The research team is investigating how the understanding of psychiatry and neuropsychiatric diseases changes through the use of artificial intelligence in relation to Alzheimer's dementia on the one hand and depression on the other. The two diseases differ in one important respect: there are already recognised biomarkers for Alzheimer's dementia, i.e. objectively measurable parameters that indicate the presence of this disease. In particular, these include very specific changes in the brain that are visible on brain scans and can be used to train artificial intelligence. Defining measurable biomarkers in relation to depression, on the other hand, is much more difficult. The illness is usually diagnosed on the basis of descriptions from those affected. If they suffer from certain symptoms, they are considered to be depressed.
In interviews, the research team will ask scientists, sick people, hospital staff and developers of artificial intelligence what ethical problems they see and what expectations and fears they have of the new technology. For example, the researchers want to find out how traditional terms and categorisations of mental illness could change in the future through the use of machine learning methods. They are also investigating how patients and doctors would deal with the possibility of a computer influencing medical diagnosis or treatment decisions. Among other things, the consortium is using a prototype developed by the DZNE for a self-explanatory AI system for recognising Alzheimer's from MRI brain scans.
A panel of experts will then use the results of these interviews and discussions to develop ethical recommendations for the use of artificial intelligence in psychiatric practice.
