Virtual Control Groups · The Animal Use Reduction Strategy
Every behavioral study runs a control group — animals whose entire job is to behave normally. Virtual Control Groups synthesize that baseline from standardized behavioral data, so your animals go where they matter: the experimental arms. We're building it now, assay by assay, and we're recruiting early validation partners.
The Regulatory Moment
The agencies that govern preclinical research are moving the same way: fewer animals, and more weight on New Approach Methodologies. For many labs the question is shifting from whether to reduce to how, without giving up statistical power. The full regulatory landscape →
The FDA Modernization Act 2.0 (2022) removed the statutory requirement for animal testing before human trials, and FDA's 2025 roadmap encourages New Approach Methodology data. Once validated, a virtual control approach is designed to be a reduction strategy you can document.
In April 2025 the FDA announced a roadmap to reduce animal testing in preclinical safety studies and encourages New Approach Methodology data — and its ISTAND program formally evaluates novel drug-development tools, including AI-based tools.
Replacement, Reduction, Refinement has guided ethics review for decades. Virtual Control Groups are designed to turn the middle R from an aspiration into a line item.
What Is a Virtual Control Group?
Control animals in well-established assays are the most predictable subjects in science — that predictability is the entire reason they're valid controls. Which makes them exactly what statistical models learn best.
A Virtual Control Group is a validated, assay-specific baseline synthesized from standardized recordings of normally behaving animals — matched to your assay, your apparatus, and your conditions, and calibrated against your own lab's data before it ever replaces a live cohort.
The statistician's first objection — batch effects, drift, non-contemporaneous controls — is the right one to raise. It's exactly why virtual baselines are calibrated against your lab's own contemporaneous data and validated side-by-side against live cohorts before ever replacing one.
The goal: keep the statistical power, and keep the animals out of the control arm.
Any Assay You Already Run
PolarEye™, our high-definition smart camera, automates capture and scoring for the behavioral assays you already run — with zero protocol change. Every standardized session strengthens the baseline for that assay. This is how virtual controls get built: one real experiment at a time.
Running something else? If it has a published protocol, we're interested. Ask about your assay →
How the Platform Enables It
Hand-scored video from unstandardized rigs can't train a control arm. A validated virtual baseline demands measurement you can trust — identical capture physics, identical schema, environment on the record. That's the entire design brief of our stack.
Contrast generated by physics, not post-processing. Illumination and imaging identical across sessions, sites, and time of day — with environmental conditions* recorded alongside every session, so context is part of the dataset. Continuous tracking keeps each animal's identity attached to its data for the whole session.
Every device emits one standardized schema — whether the video came from PolarEye™ or the camera you already own. Every instrument, one dataset, by design.
Assay-specific models learn what normal looks like, are calibrated against your lab's own historical controls, and are validated side-by-side with live cohorts before replacing one.
The metadata is what makes a virtual baseline defensible: light, sound, ultrasonic noise, and climate are logged with every session*, so baselines are matched on context — not just on strain and assay — and unexpected variability has a place to be explained. See how we capture it →
*Environmental measures (ambient light, background sound, temperature, pressure, and humidity) require the optional Data Logger.
The Payoff
A virtual baseline is designed to draw on many standardized sessions instead of one small cohort, so control-arm statistics rest on more data and fewer live animals.
Get There First
Virtual Control Groups are built on standardized data — and every session you capture is a head start. Talk to us about your assays, your reduction plan, and where virtual controls fit in your program.