000041307 001__ 41307 000041307 005__ 20240409124034.0 000041307 0247_ $$2doi$$a10.6083/bpxhc41307 000041307 037__ $$aIR 000041307 041__ $$aeng 000041307 245__ $$aHow to: multiple animal pose-estimation of socially interacting prairie voles 000041307 260__ $$bOregon Health and Science University 000041307 269__ $$a2023-07-20 000041307 336__ $$aAbstract 000041307 520__ $$aEarly life sleep disruption (ELSD) from P14-P21 in the socially monogamous prairie vole impacts adult social behavior (Jones et al, 2019). The use of automated tools to study animal behavior is called computational neuroethology (Datta et al, 2019). One primary tool is “deep learning” which requires a sequence of neural networks to learn information from the image and make predictions on other frames (Mathis and Mathis, 2020). In the context of studying social behavior among multiple animals, the neural networks predict pose by finding individual parts of each animal. Recent work from our group extended earlier findings with pose-estimation of two cohabitating prairie voles separated by a mesh-divider (Bueno-Soares Jr, 2023). One primary limitation of the study is that the animals had to be physically separated for the program to accurately identify pose. Here, we demonstrate the functionality and use-cases of a new, more precise analysis pipeline, SLEAP (Social LEAP Estimates Animal Pose), that enables identification of pose from 2+ animals in the same space (Pereira et al, 2022). This particular program was chosen based on its advanced performance in identifying multiple animals across time. We report the required steps within the SLEAP GUI (Graphical User Interface) from training the algorithm with manual labels, evaluating model performance and correct identification errors. Moreover, we transform the data output from SLEAP to analyze two behavioral paradigms (social investigation and partner preference testing). Application of SLEAP in a diversity of animal behavioral paradigms will improve research productivity and reproducibility. 000041307 540__ $$fCC BY 000041307 542__ $$fIn copyright - joint owners 000041307 650__ $$aComputational Biology$$031511 000041307 650__ $$aNeurosciences$$022870 000041307 650__ $$aMachine Learning$$011449 000041307 650__ $$aSleep$$026066 000041307 650__ $$aDeep Learning$$012734 000041307 6531_ $$ahuddling 000041307 6531_ $$aELSD 000041307 6531_ $$asocial investigation 000041307 6531_ $$aanimal behavior paradigms 000041307 6531_ $$apartner preference testing 000041307 6531_ $$acomputational ethology 000041307 6531_ $$aprairie vole 000041307 691__ $$aSchool of Medicine$$041369 000041307 692__ $$aDepartment of Behavioral Neuroscience$$041394 000041307 7001_ $$aMilman, Noah$$uOregon Health and Science University$$041354 000041307 7001_ $$aRuffins, Matthias$$uOregon Health and Science University$$010958$$041354 000041307 7001_ $$aTinsley, Carolyn$$uOregon Health and Science University$$010958$$041354 000041307 7001_ $$aWickham, Peyton 000041307 711__ $$aResearch Week$$uOregon Health and Science University$$d2023 000041307 7201_ $$aLim, Miranda$$uOregon Health and Science University$$7Personal 000041307 7201_ $$aSoares, Lezio B. Jr.$$uUniversity of Michigan Medical School$$eCollaborator$$7Personal 000041307 8564_ $$95f7c2664-7298-4088-b6c6-8295a46b617a$$s63257$$uhttps://digitalcollections.ohsu.edu/record/41307/files/ResearchWeek.2023.Milman.Noah.pdf 000041307 980__ $$aResearch Week 000041307 981__ $$aPublished$$b2023-07-20