Algorithms
Computational Software & Signal Processing
Turning raw physiological signal into defensible, low-latency inference.
Scroll
We build the computational layer that sits between noisy biosignal acquisition and a clinician-facing conclusion. The work spans digital signal processing, predictive physiological modelling, and real-time telemetry pipelines designed to hold their timing budget under realistic noise and motion artefact.
Capabilities
What this track actually involves
01Digital signal processing for ECG, PPG, EMG and impedance modalities
02Predictive physiological modelling and state-space estimation
03Low-latency telemetry and streaming inference pipelines
04Artifact rejection and signal-quality scoring
05Reproducible evaluation harnesses and benchmark datasets
Open Questions
What we are still trying to answer
?
How do we bound inference latency when the sensor, not the model, is the bottleneck?
?
Which physiological signals degrade gracefully under motion artefact, and which do not?
?
What does a defensible uncertainty estimate look like for a model that will inform care?
What Would Help
Where a collaborator changes the outcome
Signal-quality datasets from real-world acquisition
Embedded inference targets and power constraints
Clinicians willing to challenge our failure modes
Working on This Problem?
Tell us where you are stuck. Research collaborations, academic partnerships, and engineering pilots are all in scope.