heartKIT

On-Device Heart Intelligence

Train, test, and deploy real-time cardiac models on Ambiq’s ultra-low-power SoCs. Perfect for fitness wearables, wellness devices, and remote patient monitoring—covering denoise, ECG/PPG segmentation, rhythm, and beat classification.

heartKIT Highlights

01

Real-Time Edge AI

Run cardiac models locally on Ambiq’s ultra-low-power SoCs for instant, reliable insights on wearables and RPM devices. Low latency, battery-friendly inference—no cloud required.

02

Day One Ready

Kickstart projects with pre-trained models, datasets, and task-level demos. Clone, run, and showcase results in minutes—with configuration recipes you can adapt to your use case.

03

Extensible by Design

Tune tasks, models, datasets, and training via simple YAML. Add your own data or define new tasks with HeartKIT’s extensible factories to build custom workflows with minimal code.

04

Optimized to Deploy

Ship efficient inference with optimized architectures and deployment routines. Export compact models for HeliaAOT, HeliaRT, or TFLM and validate with provided metrics—perfect for battery-powered devices.

Tasks & Capabilities

Denoise

  • Clean ECG/PPG signals with real-time denoising/dereverberation to boost downstream accuracy—ideal for noisy, everyday wear.

Segmentation

  • Locate beats and intervals from ECG/PPG for reliable event boundaries and feature extraction.

Rhythm Classification

  • Classify rhythms such as AFIB and AFL on-device for immediate, private insights.

Beat Classification

  • Label beats (NORM, PAC, PVC, NOISE) for fine-grained analytics and alerts.

Video Library

Design Resources

Additional Documentation

  • Set favorites
  • Get update notifications
  • Use search filters
Preparing to download