Bridging research with multi-scale biomarkers and neuropsychiatric symptoms based on theory- and data-driven approaches
Neuropsychiatric disorders are a major burden for many people. Many studies have been conducted to investigate relationships between various biomarkers, such as genomics and brain MRI, and neuropsychiatric disorders, and to explore pathogenic mechanisms of neuropsychiatric disorders. However, relationships between markers and symptoms are highly complex and are still unknown. It has also been suggested that treatment responsiveness varies even within the same disease, and that there are common pathogenic mechanisms shared by multiple neuropsychiatric disorders, suggesting the need for a new disease classification. Our goal is to explain the complex relationship between biomarkers and symptoms by combining two artificial intelligence technologies for various neuropsychiatric disorders, and to determine the complex pathogenic processes of these diseases, which have not been revealed by conventional analysis. One of the two artificial intelligence technologies is a theory-driven approach that prepares an artificial neural network analogous to information processing of the human brain to reveal changes in information processing that occur in neuropsychiatric diseases. The other artificial intelligence technology is a data-driven approach that uses machine learning on big data to reveal hidden associations among the genome, brain MRI, and neuropsychiatric symptoms that cannot be found using conventional methods. Combining these technologies, we have prepared an artificial neural network for each subject and simulate developmental process leading to the onset of neuropsychiatric disorders in that subject. This research is expected to clarify developmental processes that cause neuropsychiatric disorders and to understand neuropsychiatric disorders from the perspective of information processing in the brain. This will enable application to therapies that regulate neural circuits in the brain, and is expected to contribute to creation of new disease classifications and personalized medicine. The research period is planned to be 36 months.