Scientists at the Graduate School of Information Science and Technology at Osaka University used animal location tracking with artificial intelligence to automatically detect the walking behaviors of movement disorders that are shared between species. By automatically removing species-specific characteristics from gait data, the resulting data can be used to better understand neurological disorders that affect movement.
Machine learning algorithms, especially deep learning approaches that use multiple layers of artificial neurons, are well suited to distinguish between different data sources. For example, they can determine the species based on the characteristics of its tracks left in the snow. However, there are times when scientists are more concerned with what is the same, rather than what is different, in various data sets. This can be the case when trying to aggregate readings from different types of animals.
Today, a team of scientists led by Osaka University used machine learning to derive models from locomotion data created by worms, beetles, mice, and independent human subjects. species.
A central goal of comparative behavioral analysis is to identify human-like behavioral repertoires in animals. “
Takuya Maekawa, first author
This method can help scientists study human neurological conditions that cause motor dysfunctions, including those resulting from low dopamine levels. Animal movement data would generate much more information; however, the spatial and temporal scales of animal locomotion vary considerably from species to species. This means that the data cannot be directly compared to human behavior. To overcome this, the team designed a neural network with a gradient inversion layer that predicts a) whether or not the input locomotion data came from a sick animal and b) what species the data came from. entry. From there, the network was formed in such a way that it could not predict the species from which the input data was collected, resulting in the creation of a network unable to distinguish between species but capable of identify specific diseases. This allowed the network to extract the locomotion characteristics inherent in the disease.
Their experiments revealed features of interspecies locomotion shared by dopamine-deficient worms, mice, and humans. Despite their evolutionary differences, all of these organisms are unable to move while maintaining high speeds. In addition, the speed of these animals was found to be unstable during acceleration. Interestingly, these animals exhibit similar movement disorders in dopamine deficiency, even though they have different body scales and methods of locomotion. While previous studies had shown that dopamine deficiency was associated with movement disorders in all of these species, this research was the first to identify the shared locomotion characteristics caused by this deficiency.
“Our project shows that deep learning can be a powerful tool for extracting knowledge from data sets that appear too different to be compared by human researchers,” said author Takahiro Hara. The team anticipates that this work will be used to find other common features of the disorders that impact evolutionarily distant species.
Maekawa, T., et al. (2021) Analysis of interspecies behavior with attention and domain-based deep neural networks. Natural communications. doi.org/10.1038/s41467-021-25636-x.