For the analysis of diseases associated with movement disorders, experimental model animals have been used instead of humans who have difficulty in intervention experiments.However, it was difficult to directly compare and analyze the movement behavior data of animals of different species because their body lengths and movement methods differ greatly.

　The research group has developed an artificial intelligence technology (neural network) that extracts behavioral characteristics that can distinguish between normal and diseased (dopamine deficient) individuals, although it is impossible to distinguish species from the movement trajectory of animals.The characteristic of distinguishing the disease was considered to be the characteristic of the disease common to all species.

　This technique was applied to dopamine-deficient and normal dopamine-deficient and normal individuals of humans, mice, red flour beetles (rice pests), and nematodes, respectively, to discover migration characteristics seen across species.Dopamine-deficient humans, mice, and nematodes commonly had motor disorders such as inability to move at high speeds and unstable speeds during acceleration.In addition, dopamine-deficient mice, nematodes, and insects commonly had movement disorders such as the inability to slow down smoothly before bending.

　The results of this research will enable us to confirm the effects of the treatment method using animals, which are easier to experiment with than humans, for the development of treatment methods for diseases associated with movement disorders such as Parkinson's disease.He added that the mechanism by which dopamine deficiency affects exercise may be evolutionarily conserved from nematodes to humans.

Paper information:[Nature Communications] Cross-species Behavior Analysis with Attention-based Domain-adversarial Deep Neural Networks