Research, LifePulse

600K+

heartbeats detected in a single 24h recording

~90s

24-hour mouse ECG analysis time

10×

faster than human ECG, and far more complex

≥ Expert

model meeting/exceeding expert human analysis

Why mouse ECG?

Master the hardest signal. The easier ones become trivial.

Mouse hearts beat 10× faster than human hearts, 450 to 750 BPM. The ECG signals are dense, fast, noisy. If our AI can master this, human ECG interpretation becomes a solved problem.

Mouse models are the gold standard in cardiac research, used globally to study disease mechanisms, drug responses, and genetic conditions before clinical trials. Building the AI here means it's already validated in the environment that matters most for upstream science.

live signal comparison

Human ECG

~70 bpm

Mouse ECG

~612 bpm

Jerry · our research platform

A cloud-based SaaS purpose-built for 24-hour mouse ECG analysis.

Continuously updated, so your lab always has the latest detection capabilities. Used today, in production, by university researchers on real experimental data.

jerry.lifepulse.ai · 24-hour mouse ECG recording live

Jerry Research Platform, 24-hour Mouse ECG Analysis

Detection

Automated R-peak detection

Identifies 600,000+ heartbeats in a single 24-hour recording with sub-millisecond precision, processing in approximately 90 seconds.

Classification

Arrhythmia classification

Detects bradycardia, tachycardia, atrial fibrillation, pauses, PVCs, ventricular tachycardia. New detection models in active development.

HRV

Heart rate variability

Full HRV, SDNN, RMSSD, pNN50, Poincaré plots. The metrics cardiac researchers need to publish.

Model evolution

From signal processing to a proprietary AI model.

Our research has progressed through multiple generations of detection algorithms, ensemble voting systems, continuous wavelet transforms, culminating in a proprietary 1D Patch Transformer model.

In validation tests our AI model is meeting or exceeding the accuracy of expert human analysis. This breakthrough validates our core thesis: AI trained on the hardest cardiac signals can generalize to simpler ones.

Research to clinical

A deliberate path from preclinical validation to clinical deployment.

Preclinical validation

Training and validating AI models on 24-hour mouse ECG recordings with expert cardiologist oversight.

Cross-species generalization

Extending validated algorithms to human ECG data through clinical database integration.

Clinical feedback loop

Building physician review interfaces so expert corrections continuously improve the AI model.

Deployment at scale

Real-time AI cardiac diagnostics, starting with underserved rural facilities.

Research partnership

Validated alongside a leading academic medical center.

Our primary research collaboration is with the cardiac electrophysiology lab at a leading academic medical center. We process real experimental datasets, validate our algorithms against expert analysis, and iterate based on direct clinical feedback.

This partnership ensures our technology is shaped by the researchers who understand cardiac signals best.

Research highlights

What we're proving, in production.

≥ Expert R-peak

AI model meeting or exceeding expert human analysis in R-peak detection validation.

24h in 90 seconds

24-hour recordings processed in ~90 seconds, enabling same-day research insights.

Proprietary architecture

1D Patch Transformer trained on the most complex cardiac signals in electrophysiology.

Active drug-study analysis

Supporting real cardiac research at a leading university, today, in production.

Collaborate with us

Help us expand validation across species and clinical contexts.

Research

contact@lifepulse.ai

Research inquiries, academic, preclinical, and beyond.

Partnerships

partnerships@lifepulse.ai

Clinical sites, data partners, device makers.