Technology

Physics-informed AI, grounded in real measurement.

Most medical AI is statistical pattern-matching on legacy data. Algocyte is built on the physics of the measurement itself, then constrained by causal biological models, so longitudinal interpretation stays aligned with clinical reality.

Algocyte Proxima device

At a glance

6
Layers in the stack
2 populations
Independent validation (NHANES & UK Biobank)
1
Personal baseline per patient

The stack

Six layers, one signal.

Hardware-level measurement, physics-informed networks and causal biological models, engineered to work as one pipeline.

01

Multi-sensor measurement

Each cartridge is read by multiple sensors. Raw physical signals, not pre-processed indices, feed the model.

02

Physics-informed network

Neural networks are constrained by the physics of the sensor and the biology of the analyte, less data, better generalisation.

03

Causal biological model

A causal layer keeps interpretation consistent with what is biologically plausible across time and across patients.

04

Personal baseline

Each user's own time series defines their reference range. Drift is meaningful even within population norms.

05

Population context

Anonymised population datasets provide comparator distributions and signature priors.

06

Clinical workflow

Alerts support existing escalation paths, with EHR integration in development. The clinician stays in control.

Designed for the patient's home

Decentralised by design

Built for a system where the patient is at home.

The device, the model and the workflow all assume the patient is at home, not in a clinic, not in a lab.

  • Connects to portable and wearable devices, with EHR integration in development
  • Combines subjective & objective data: symptoms, lifestyle, immunology, history
  • Actionable insights across the care continuum
  • Structured clinical context with provenance
Explore the ecosystem

Privacy & governance.

UK GDPR

All personal data is processed under UK GDPR. Identifiable data is encrypted in transit and at rest, and never sold.

Anonymised research

Research datasets are anonymised before any model is trained. Re-identification is engineered to be infeasible.

Auditability

Every interpretation is traceable to the measurement that produced it and the model version that delivered it.

Bring regular immune monitoring to your patients.

Book a 20-minute briefing with our clinical team to see Algocyte in action.