A team from the University of Michigan has developed a latest software tool to assist researchers across the life sciences more efficiently analyze animal behaviors.

The open-source software, LabGym, capitalizes on artificial intelligence to discover, categorize and count defined behaviors across various animal model systems.

Scientists have to measure animal behaviors for a wide range of reasons, from understanding all of the ways a selected drug may affect an organism to mapping how circuits within the brain communicate to provide a selected behavior.

Researchers within the lab of U-M faculty member Bing Ye, for instance, analyze movements and behaviors in Drosophila melanogaster-;or fruit flies-;as a model to check the event and functions of the nervous system. Because fruit flies and humans share many genes, these studies of fruit flies often offer insights into human health and disease.

“Behavior is a function of the brain. So analyzing animal behavior provides essential details about how the brain works and the way it changes in response to disease,” said Yujia Hu, a neuroscientist in Ye’s lab on the U-M Life Sciences Institute and lead creator of a Feb. 24 Cell Reports Methods study describing the brand new software.

But identifying and counting animal behaviors manually is time-consuming and highly subjective to the researcher who’s analyzing the behavior. And while a couple of software programs exist to routinely quantify animal behaviors, they present challenges.

Lots of these behavior evaluation programs are based on pre-set definitions of a behavior. If a Drosophila larva rolls 360 degrees, for instance, some programs will count a roll. But why is not 270 degrees also a roll? Many programs don’t necessarily have the pliability to count that, without the user knowing the best way to recode this system.”

Bing Ye, Professor, Cell and Developmental Biology, University of Michigan

Considering more like a scientist

To beat these challenges, Hu and his colleagues decided to design a latest program that more closely replicates the human cognition process-;that “thinks” more like a scientist would-;and is more user-friendly for biologists who may not have expertise in coding. Using LabGym, researchers can input examples of the behavior they need to research and teach the software what it should count. This system then uses deep learning to enhance its ability to acknowledge and quantify the behavior.

One latest development in LabGym that helps it apply this more flexible cognition is the usage of each video data and a so-called “pattern image” to enhance this system’s reliability. Scientists use videos of animals to research their behavior, but videos involve time series data that will be difficult for AI programs to research.

To assist this system discover behaviors more easily, Hu created a still image that shows the pattern of the animal’s movement by merging outlines of the animal’s position at different timepoints. The team found that combining the video data with the pattern images increased this system’s accuracy in recognizing behavior types.

LabGym can also be designed to overlook irrelevant background information and consider each the animal’s overall movement and the changes in position over space and time, much as a human researcher would. This system can even track multiple animals concurrently.

Species flexibility improves utility

One other key feature of LabGym is its species flexibility, Ye said. While it was designed using Drosophila, it will not be restricted to anyone species.

“That is actually rare,” he said. “It’s written for biologists, in order that they can adapt it to the species and the behavior they need to check without having any programming skills or high-powered computing.”

After hearing a presentation concerning the program’s early development, U-M pharmacologist Carrie Ferrario offered to assist Ye and his team test and refine this system within the rodent model system she works with.

Ferrario, an associate professor of pharmacology and adjunct associate professor of psychology, studies the neural mechanisms that contribute to addiction and obesity, using rats as a model system. To finish the obligatory statement of drug-induced behaviors within the animals, she and her lab members have needed to rely largely on hand-scoring, which is subjective and very time-consuming.

“I have been trying to unravel this problem since graduate school, and the technology just wasn’t there, when it comes to artificial intelligence, deep learning and computation,” Ferrario said. “This program solved an existing problem for me, however it also has really broad utility. I see the potential for it to be useful in almost limitless conditions to research animal behavior.”

The team next plans to further refine this system to enhance its performance under much more complex conditions, comparable to observing animals in nature.

Source:

Journal reference:

Hu, Y., et al. (2023) LabGym: Quantification of user-defined animal behaviors using learning-based holistic assessment. Cell Reports Methods. doi.org/10.1016/j.crmeth.2023.100415.