The current diagnosis of depression is performed by the doctor in charge subjectively judging clinical symptoms such as depressed mood and decreased motivation, and an objective diagnostic method has not yet been established.In addition, antidepressant drug treatment is also performed by trial and error, and since there are about 3% of patients who do not respond to treatment, the development of objective diagnostic methods and antidepressant drug treatment responsiveness prediction methods based on brain science data has been developed. It has been demanded.

　In this study, brain function image analysis data using MRI of subjects of the Department of Psychiatry, Hiroshima University (depression patients and healthy subjects in total 134), blood biomarker candidate substances such as brain-derived neuronutrient factors, and psychological tests The clinical evaluation index based on the interview results was integratedly patterned using the Bayesian multiple co-clustering method using machine learning, which is one of the artificial intelligences.The results showed that depression patients could be classified into three subtypes based on resting brain activity and early childhood trauma experience in the default mode network of the brain centered on the right angle.

　In addition, one of these groups was found to be less effective against SSRIs, the most frequently used antidepressant in the treatment of depression.This result suggests the possibility of predicting the therapeutic effect of SSRIs before administration of antidepressants by evaluating the patient's MRI brain function image data and childhood trauma experience at the first visit, based on brain science data. It is expected to contribute to the development of objective diagnosis and treatment of new depression.

Paper information:[Scientific Reports] Identification of depression subtypes and relevant brain regions using a data-driven approach