AUTHOR=Huber Phillipe , Ausk Brandon J. , Tukei K. Lionel , Bain Steven D. , Gross Ted S. , Srinivasan Sundar TITLE=A convolutional neural network to characterize mouse hindlimb foot strikes during voluntary wheel running JOURNAL=Frontiers in Bioengineering and Biotechnology VOLUME=Volume 11 - 2023 YEAR=2023 URL=https://www.frontiersin.org/journals/bioengineering-and-biotechnology/articles/10.3389/fbioe.2023.1206008 DOI=10.3389/fbioe.2023.1206008 ISSN=2296-4185 ABSTRACT=Voluntary wheel running (VWR) is widely used to study how exercise impacts a variety of physiologies and pathologies in rodents. The primary activity readout of voluntary wheel running is aggregated wheel turns over a given time interval (most often, days). Given the typical running frequency of mice (~ 4 Hz) and the intermittency of voluntary running, aggregate wheel turn counts therefore provide minimal insight into the heterogeneity of voluntary activity. To overcome this limitation, we developed a 6-layer convolutional neural network (CNN) to determine hindlimb foot strike frequency of mice exposed to voluntary wheel running. Aged female C57BL/6 mice (22 mo, n=6) were first exposed to wireless angled running wheels for 2 hr/d, 5 d/wk for 3 wk, with all VWR activity recorded at 30 frames/s. To validate the CNN, we manually classified foot strikes within 4800 1 s videos (800 randomly chosen for each mouse) and converted those values to frequency. Upon iterative optimization of model architecture and training on a subset of classified videos (4400), the CNN model achieved an overall training set accuracy of 94%. Once trained, the CNN was validated on the remaining 400 videos (accuracy: 81%). We then applied transfer learning to the CNN to predict the foot strike frequency of young adult female C57BL6 mice (4 mo, n=6) whose activity and gait differed from old mice during VWR (accuracy: 68%). In summary, we have developed a novel quantitative tool that non-invasively characterizes voluntary wheel running activity at much greater resolution than was previously accessible. This enhanced resolution holds potential to overcome a primary barrier to relating intermittent and heterogeneous VWR activity with induced physiological responses.