@article{fdi:010097493,
  title = {{E}nriching historical biologging datasets on seabirds using deep neural networks : a transformer-based approach to infer energy expenditure proxy from {GPS} and environmental data},
  author = {{M}uquet, {N}oemie and {B}runel, {A}drien and {B}arbraud, {C}. and {B}ugoni, {L}. and {D}elord, {K}. and {N}unes, {G}. {T}. and {Z}ucker, {J}ean-{D}aniel and {L}anco, {S}.},
  editor = {},
  language = {{ENG}},
  abstract = {{R}ecent advances in biologging have led to the widespread use of accelerometers, which generate high-resolution movement data essential for understanding animal behaviour. {D}erived from tri-axial accelerometry, {O}verall {D}ynamic {B}ody {A}cceleration ({ODBA}) serves as a proxy for energy expenditure that is less invasive and more cost-effective than alternative methods, contributing to a better understanding of individual survival and population dynamics. {H}owever, many long-term datasets lack accelerometry due to limitations in instrumentation or older technologies, thereby constraining their utility in multi-year ecological studies. {M}arine top predators, particularly seabirds, are valuable ecosystem sentinels due to their foraging strategies, energy constraints and sensitivity to environmental conditions. {U}nderstanding energy use during at-sea travel is essential for forecasting colony viability under climate change, thus supporting long-term ecological research and conservation efforts. {H}igh-resolution biologging has facilitated data-driven approaches to monitor seabird behaviour, but gaps in accelerometry data remain a challenge. {N}ew developments in deep neural networks ({DNN}s) offer opportunities to reconstruct missing biologging-derived metrics by capturing the complex spatiotemporal relationships between movement and environmental data. {O}ne way to model such dependencies is to use a transformer architecture, which processes all inputs jointly and produces representations that explicitly encode relationships between them. {I}n this study, we present {ODBAF}ormer, a transformer-based model designed to infer {ODBA} from {GPS} trajectories and environmental covariates. {W}e applied {ODBAF}ormer to seabird tracking data to address gaps in proxy data for energy expenditure. {W}e evaluated the model on unseen data and compared it with state-of-the-art machine-learning regression method {XGB}oost as a baseline. {W}e also explored its sensitivity to data heterogeneity and different environmental covariates combinations. {O}verall, {ODBAF}ormer accurately reconstructs {ODBA} time series and proxies for both energy budgets and spatial energy expenditure distribution, providing a scalable solution to enrich legacy datasets or for situations where the use of multisensor loggers is limited by cost or weight.},
  keywords = {accelerometry ; bioenergetics ; biologging ; deep learning ; marine ecosystem monitoring ; {ODBA} ; seabirds ; transformer models},
  booktitle = {},
  journal = {{M}ethods in {E}cology and {E}volution},
  volume = {[{E}arly access]},
  numero = {},
  pages = {[16 p.]},
  ISSN = {2041-210{X}},
  year = {2026},
  DOI = {10.1111/2041-210x.70336},
  URL = {https://www.documentation.ird.fr/hor/fdi:010097493},
}