For farms conducting research trials, data integrity is everything. The Feed Intake Data Platform is a comprehensive, end-to-end solution designed to automate the collection, processing, and visualization of high-frequency feed intake data. Custom-built for the RIC2Discover feeding system, this platform leverages RIC2Discover's native API to extract granular data, ensuring researchers spend less time wrangling CSV files and more time analyzing verified results.

The RIC2Discover system produces a staggering volume of raw data—every individual visit, every gram of intake, and every second of bin occupancy is recorded across dozens of stations simultaneously. For a research farm, this isn't just a storage challenge; it's a processing bottleneck. Raw data is often riddled with noise, requiring complex cleaning to filter out mechanical drifts, and sophisticated aggregation to turn second-level events into meaningful daily intake metrics and behavioural patterns. Without a system that monitors these streams in real time, small hardware drifts or subtle animal behavioural shifts can quickly snowball into significant data inaccuracies.

We architected a unified solution consisting of two distinct parts that run together on a central data hub. This hub acts as the "brain" of the research facility. Because it interfaces with the RIC2Discover hardware via a secure online API, the platform offers total deployment flexibility: it can be hosted on a physical server at the farm, in the cloud, or on institutional off-site hardware.

The first part of the solution is a resilient ETL engine. Its job is to ensure that no data point is ever lost and that every record is research-ready before it reaches the database.

• API Syncing: The pipeline "pulls" raw data from the RIC2Discover cloud API at configurable intervals, centralizing data into a single, high-performance database.

• Data Sanitization: As data flows in, the pipeline automatically cleans and aggregates it. It converts millisecond-level "events" (like a bin door opening) into validated "intake records," filtering out noise and mechanical errors.

• Stateful Resilience: If the farm's internet or the server restarts, the pipeline uses intelligent state management to remember exactly where it left off, backfilling any missing data automatically.

The second part is a custom-built R Shiny application. This is the visual interface where researchers and farm staff interact with the data processed by the pipeline.

• Animal Welfare Alerts: Algorithms analyze daily intake against rolling averages to instantly flag animals with sudden drops in consumption—often the first sign of illness.

• Equipment Health Monitoring: The dashboard tracks the calibration history of every feed bin. It automatically flags bins due for maintenance or those reporting anomalous data (e.g., negative weights or drift), protecting trial validity.

• Interactive Visuals:

  • Cumulative Intake Curves: Monitor feeding behaviour throughout the day in real-time.

  • Bin Weight Diagnostics: Visualize sensor data to diagnose mechanical issues remotely.

• Data Quality Reports: Automated filters isolate "feed stealers" (animals eating from unassigned bins) and other behavioural outliers that could skew trial results.

• Flexible Deployment: Pulls data via API, allowing for installation on-premise, in the cloud, or on private off-site servers without loss of functionality.

• Audit Trails: Every data point includes upload and update timestamps, providing a clear chain of custody for research audits.

• Multi-Architecture Support: Fully containerized with Docker, supporting both standard servers (AMD64) and power-efficient edge devices (ARM64).

• Secure Access: Integrated with ShinyProxy for secure, user-based authentication and resource management.

The primary goal of this platform is to move the burden of data management away from the research team. By automating the high-frequency ingestion and cleaning of RIC2Discover data, the facility gains a level of operational oversight that manual exports cannot provide. Researchers can verify trial conditions in real-time, maintenance can be scheduled based on actual sensor performance, and anomalies are caught hours after they occur rather than weeks later during post-trial analysis. Ultimately, this infrastructure ensures that the final dataset is as reliable as the equipment used to collect it.

Focus on the science, not on IT. We build the infrastructure so you can focus on the breakthroughs.

Let's start your private implementation, contact us to request a demo.