An Early Lab Solving
Hard Engineering Problems
Computational Biomedical Systems is a bioengineering research and venture lab. We design biomedical devices, embedded hardware, and computational software — and we are early enough that most of this is still being figured out.
Close the Gap Between a Signal and a Decision
Most biomedical measurement fails at the boundary between the physical world and the model that interprets it. The sensor is noisy, the modality is mismatched, the latency budget is fiction, and the person on the other end has no way to tell how confident the system actually is.
That boundary is where we work. We build the acquisition hardware, the signal processing, the embedded firmware, and the inference layer as one system — because optimising any one of them in isolation tends to move the problem somewhere less visible.
We are a venture lab, not a service company. We do not sell equipment, maintain devices, or run a help desk. We take on problems where the engineering is genuinely unresolved, and we are honest about which of those we have solved.
How We Hold Ourselves to Account
Evidence Over Assertion
A claim we cannot reproduce is not a claim we make. Bench data, versioned code, and a written record of what failed — in that order.
Design for Verification
Safety analysis and test strategy are part of the first sketch, not a phase bolted on once someone asks who signed off.
Hardware and Software, Together
We do not hand an algorithm to a hardware team and hope. Acquisition, firmware, and inference are designed as one system.
Collaboration by Default
We are early. Most of what we get right will come from partners who know the clinical or materials problem better than we do.
Standards We Design Against
We are an early-stage lab and do not yet hold the certifications a commercial medical device company would. These are the frameworks we design within from day one anyway.
Risk Management
We design against ISO 14971 risk-management principles and document hazard analysis from the first prototype, even where we are not yet claiming conformity.
Reproducibility
Every result is tied to a commit, a fixture, and a measurement setup someone else can rebuild. If it cannot be rebuilt, it is not a result.
Regulatory Awareness
We keep current on FDA and EU device pathways and involve regulatory counsel early — before architecture decisions foreclose the route to market.
Honest Status
Our research pages state what is exploratory, what is bench-proven, and what has never left the lab. We would rather under-promise than over-claim.
Bring Us Something Hard
We are most useful when the problem is genuinely under-determined and someone needs an engineering team willing to go build the thing.