Use of deep learning to predict chronic wasting disease status based on animal movement

Authors: BLAHA, RAYMOND - Mississippi State University; Silva, Christopher; CUNNINGHAM, STEPHANIE - Mississippi State University; DEVIVO, MELIA - Department Of Fish And Wildlife; EDMUNDS, DAVID - Colorado State University; BOUDREAU, MELANIE - Mississippi State University

Submitted to: Movement Ecology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/27/2026
Publication Date: 6/10/2026
Citation: Blaha, R.L., Silva, C.J., Cunningham, S.A., DeVivo, M.T., Edmunds, D.R., Boudreau, M.R. 2026. Use of deep learning to predict chronic wasting disease status based on animal movement. Movement Ecology. 14. Article 57. https://doi.org/10.1186/s40462-026-00668-4.
DOI: https://doi.org/10.1186/s40462-026-00668-4

Interpretive Summary: Chronic Wasting Disease (CWD) is a cervid (deer, elk, moose, etc.) disease impacting the population and management of North American deer. CWD-infected cervids appear normal during most of the disease course, which makes them difficult to detect. Artificial intelligence (AI)-based approaches can be used to identify changes in behavior patterns associated with CWD-infected deer. We analyzed the global positioning system (GPS) data from a herd of radio collared mule deer. These deer were tested for CWD, so their CWD status was known. We trained an AI model using the movement data of CWD-negative mule deer (velocity, direction, and sinuosity). After training the unsupervised autoencoder (AE), the model was able to identify CWD-positive deer with 85% accuracy and 89% precision. When a conditional autoencoder (cAE) was trained, the accuracy dropped to 54-77% accuracy and the precision to 54%. Deer in the early stages of CWD reduced their use of space, while deer in the later stages of the disease moved more slowly and erratically. Thus, AEs can be used to accurately identify CWD-positive deer and to determine the approximate stage of the disease by analyzing their anomalous movements. There is potential for use of these models in real time CWD surveillance of wild populations.

Technical Abstract: Chronic Wasting Disease (CWD) is an invariably fatal prion disease of cervids that impacts cervid populations and wildlife management across North America. Since infected cervids often remain asymptomatic for months, diagnostic testing often requires broad sampling efforts to track and manage the disease. Movement based anomaly detection from Global Positioning System (GPS) collaring data offers a potential tool for tracking CWD. Here we evaluated whether deep learning behavioral anomaly detection models such as autoencoders (AE) and conditional autoencoders (cAE) could effectively identify anomalous movement changes in free-ranging mule deer (Odocoileus hemionus) that may have been associated with CWD infection. Unsupervised AEs achieved =84% accuracy and 89% precision irrespective of whether the model was trained using only CWD- or a 40/60 split of CWD+ / CWD- individuals. Important movement features in distinguishing between CWD+ and CWD- animals included metrics related to velocity, direction, and sinuosity. In contrast, supervised cAEs only achieved 54-77% accuracy and =53% precision across models trained with incidence rates of CWD+ individuals ranging from 10 - 40%; the models also had inconsistent results reducing their generality. Finally, using the best fitting model (AE trained using only CWD- animals), we found that early-stage animals exhibited redacted space use, whereas late-stage individuals presented more pronounced declines in velocity and altered directional patterns. These findings indicate that AEs can accurately identify CWD related behavioral anomalies using movement data, that animals show variation in what movement metrics matter depending on the stage of the disease, and that there is potential for use of these models in real time CWD surveillance of wild populations.