IKBFU Scientists together with colleagues from the University of Plovdiv (Bulgaria) and Ural Federal University have developed a new method for classification of patients with major depressive disorder. This method allows doctors to detect the disease with 93% accuracy. Artificial intelligence analyzes MRI brain images. Experts trained it to identify differences in neural connections in healthy and sick people. The effectiveness of this approach has been successfully confirmed in the course of work with real patients. The experiments involved 49 healthy people and 35 patients with clinical form of depression.
|Alexander Hramov, Head of the IKBFU Neurotechnology and Artificial Intelligence Center:|
We can classify both the near connections between neighboring areas and the far connections, which determine the nature of the whole structure of connections in the brains of healthy and sick people. The proposed classifier achieves an accuracy of 93%
The developers plan to continue their research to find changes in the brain that occur in patients with bipolar disorder (BAD). This disorder resembles MDD in its symptoms, which often leads to misdiagnosis. This is a serious medical problem because treatment for depression and BAD is different. If scientists succeed, a neural network will be able to distinguish one from the other. Another challenge for the specialists is to determine exactly what signs the program uses to diagnose depression. The peculiarity of AI is that users do not fully know its logic. This will allow controlling the quality of the neural network.
+7 (4012) 59-55-95 #7600
Личный кабинет для