2026-10-12T01:33:39Z https://www.scidb.cn/oai

10.57760/sciencedb.35116 2026-05-15T14:05:44Z

2026-05-15 A Continuous Multimodal Wearable Dataset of Synchronized Cardiac, Kinematic and Heat-Production Signals from Growing Pigs in Open-Circuit Respirometry Chambers (Porcine-CaMo Autumn-2025 Cohort, 3 Chambers / 6 Pigs / 10 Periods / 106 Pig-Days) doi:10.57760/sciencedb.35116 en This dataset (Porcine-CaMo Autumn-2025 Cohort, paper-aligned release) provides continuous multimodal wearable recordings collected from 6 growing pigs (Sus scrofa domesticus) during September–October 2025 in open-circuit respirometry chambers. The experimental design comprised 10 sequential experimental periods over 40 calendar days, spanning chambers A1, B1 and B2, yielding 106 cumulative chamber pig-days (A1: 35 + B1: 41 + B2: 30) that mirror the scope of the companion research paper. All animals wore the same third-generation back-mounted device, simultaneously logging 512 Hz single-lead ECG (BMD101 chipset), 10 Hz three-axis gyroscope and 10 Hz three-axis accelerometer; the latter is also reduced to Vectorial Dynamic Body Acceleration (VeDBA). In parallel, the open-circuit respirometry chambers in B1 and B2 (A1's respirometry hardware was unavailable for this campaign) ran on a 5-min indoor/outdoor alternating sampling schedule, from which 5-min-resolution heat-production (HP) time series were derived via the Brouwer equation, computed by OpenCalori-Swine (https://github.com/zengzhengcheng/OpenCalori-Swine<br>Zenodo DOI: 10.5281/zenodo.20051163). ECG streams were processed with ECG TransUNet (a hybrid CNN-Transformer U-Net; https://github.com/zengzhengcheng/ECG-TransUNet<br>Zenodo DOI: 10.5281/zenodo.20051167) for automatic R-peak detection followed by adaptive statistical correction, producing one per-chamber R-peak label file (e.g. B1heartlabel, covering the full beat sequence for that chamber). Data are released in two parallel quality tiers: Corrected (the full set, one heartlabel per chamber) — R-peak series with electrode-detachment intervals and large-scale noise segments removed and small-scale (< 10 s) noise mean-imputed; and Perfect (a subset) — recordings whose entire file is uniformly clean and close to 100% accurate, copied separately into a perfect/ folder together with the matching .hdf, .huancun and heartlabel files for visualization replay. File lineage follows raw .txt → annotated .hdf → visualized .huancun → per-chamber heartlabel.csv (R-peak labels), accompanied by a merged 5-min per-pig analytical table aligning {animal metadata + HP + motion features + HRV features}, a standalone pig-metadata file, plus an ML-ready feature table derived from this cohort and a matching reference training-script set. The HP prediction model is trained per-cohort separately: this DOI ships the autumn-cohort training features and reference scripts but does not ship pre-trained model weights — downstream users retrain locally following the README and example Notebook (the companion January cohort, DOI 10.57760/sciencedb.35878, follows the same convention). The HRV-processing and IMU-processing pipelines are publicly open-sourced at https://github.com/zengzhengcheng/SwineSync-OpenSource<br>Zenodo DOI: 10.5281/zenodo.20051135; the heart-rate annotation GUI (SwineSync Studio annotator) is retained as an internal product tool distributed only as a Windows executable, while its upstream data-processing code is already open-sourced in the same repository. The release supports precision livestock farming, metabolism–behaviour coupling modelling, welfare phenotyping, and benchmarking of wearable animal-physiology algorithms. precision livestock farming; wearable physiological monitoring; continuous ECG; R-peak detection; heart rate variability; kinematics; VeDBA; open-circuit respirometry chamber; indirect calorimetry; heat production; multimodal data fusion; deep-learning ECG analysis; human-in-the-loop annotation; Brouwer equation; growing pigs Zeng Zhengcheng Tian Siqi Luo Xiangshi Zhang Shuai EMBARGO https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank