One connected platform for COR Health, from device to AI model to web product.
Full-stack product engineering for the COR One: device integration, real-time sensor pipelines, cloud infrastructure, an AI model that reads ESR results, and the consumer web product, built as one system.
- Client
- COR Health, US
- Industry
- Connected health
- Scope
- Product engineering · IoT · AI · Cloud
- Engagement
- 2023 to now, ongoing
A device that could only launch with the whole stack behind it
The challenge
COR Health built the COR One, a home blood test that measures Erythrocyte Sedimentation Rate (ESR), a key marker for inflammation. Before it could reach customers it needed a complete technical platform: device integration, data pipelines, AI-assisted analysis, cloud infrastructure and a consumer web product.
None of it could be built in isolation. The device, the data, the model and the product had to work together, so COR needed one partner who could think and build across the entire stack.
Our approach
We took end-to-end ownership of the technical platform: firmware and device integration, sensor data pipelines, IoT cloud infrastructure, the ESR analysis model and the consumer product, engineered as one system rather than parts handed between teams.
The architecture was designed for reliability from day one, with data integrity and traceability built in rather than added later.
Every layer, engineered to talk to the next
Device signals flow into the data pipelines. The pipelines feed the AI-assisted ESR analysis model. The model's output surfaces in the consumer product. A complete system, not a patchwork.
- 01FirmwareCOR One firmware and device integration
- 02DataReal-time sensor pipelines
- 03CloudInfrastructure built for data integrity
- 04AI / MLESR analysis model
- 05ProductConsumer web product
The technical challenge across every layer
Device to cloud
The COR One had to send sensor data to the platform reliably: captured, validated and transmitted with the consistency a health product demands. Everything else was built on it.
Reading through the noise
Off-the-shelf image processing could not cope with the background noise in COR One readings. The model was trained and tuned on months of real device data until it read results reliably.
One owner for the system
COR did not need three specialists handing work between layers. They needed one team making architecture decisions that held up across firmware, cloud, AI and product at once.
From image processing to 99% accurate ESR readings
At launch, results were read with conventional image processing. The device housing is slightly transparent, so background noise leaked into the images and meant heavy manual checking for the first months.
Because the platform had captured structured readings from the start, those months became training data. The model now reads ESR results with 99% accuracy, a figure COR has confirmed, and every reading stays traceable.
A complete platform, and a partnership that continues
COR now has a complete connected platform, from the device in someone's home to the model that interprets their data and the product that delivers the result. Every reading and model output can be traced back to the device it came from, and the work continues after launch.
- End to endFull-stack deliveryFirmware and device integration through cloud to consumer product. One partner, one system.
- TraceableEvery readingReadings, model outputs and data handling structured so every result can be traced back to its source.
- OngoingPartnershipA continued engagement, not a handoff. We keep contributing to technical decisions as the product evolves.
We partnered with Code Pros to launch our new product, the COR One, which required full-stack engineering to bring it to life. […] Hardware is notoriously hard, and the COR One was no exception. Code Pros rose to the challenge.
Bob MesserschmidtFounder, COR Health
Read all our reviews
Building a connected product that spans hardware and software?
We have engineered connected health platforms end to end. Tell us where your product sits between device, cloud and consumer software. A 20-minute call is the first step.
Book a 20-min call