Phenomics

Integrating Heterogeneous Sensor Data for Dairy Cattle Research: The SLU Gigacow Infrastructure

---

Abstract

Precision livestock farming generates large volumes of sensor- and event-based data with considerable potential for genetics research and breeding, but their effective use is challenged by heterogeneity in farm environments, sensor technologies, and data structures, which can introduce systematic measurement bias. The objective of this study is to describe and evaluate the SLU Gigacow infrastructure as a framework for integrating heterogeneous on-farm data into standardized, research-ready phenotypes for dairy cattle research and genetic evaluation. SLU Gigacow currently includes genotypes and high-frequency production data from more than 10,000 dairy cows across 11 voluntarily recruited Swedish herds, with herd sizes ranging from 60 to 760 lactating cows and substantial variation in sensor types and management systems. An agile development methodology was applied to prioritize data sources and iteratively develop data integration pipelines. Data are interpreted through three complementary perspectives: location-based measurements (e.g. sensors installed in barns or milking systems), individual-based measurements directly linked to animals, and model-based transformations that convert raw observations into biologically meaningful traits. Animal attribution is achieved using RFID-based associations where possible, while probabilistic and temporal models are applied when direct linkage is not available. Examples include converting test-day milk records to daily yields using curve-fitting models and ongoing work to infer feed intake proxies from feeding time or presence in feeding areas. The infrastructure also enables the definition of multiple phenotypic definitions for the same underlying trait, allowing systematic comparison of how different measurement methods and definitions may influence phenotypes used in genetic analyses. We conclude that explicitly modeling measurement processes and defining trait-specific ontologies are essential for reducing bias, improving comparability across farms, and enabling robust downstream analyses. The SLU Gigacow infrastructure provides a scalable testbed for developing standardized phenotypes in preparation for international data-sharing initiatives such as ICAR Animal Data Exchange and ISO/TC 347, supporting future applications in genetics, management, and precision dairy research.

Keywords: 2026

How to Cite:

Ohlsson, I., Edvardsson Rasmussen, A., van Damme, R., Lundberg, A., de Koning, D. & Klingström, T., (2026) “Integrating Heterogeneous Sensor Data for Dairy Cattle Research: The SLU Gigacow Infrastructure”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2286986. doi: https://doi.org/10.31274/wcgalp.24188

Rights: 1