AI Pet Biometric Health Collar based on nRF5340 and Fur-Penetrating PPG

Client

A North American veterinary technology leader dedicated to preventative pet wellness and chronic illness management for senior dogs and cats.

Challenge

Dense pet fur and sub-dermal keratin layers heavily scatter conventional optical PPG signals, causing severe signal degradation. Rapid animal head-shaking and scratching introduce extreme motion artifacts. Furthermore, cats are hypersensitive to collar weight, requiring a sub-30g enclosure with month-long autonomy.

Product Concept & Architecture

The system centers around the Nordic nRF5340 dual-core wireless SoC (128 MHz Cortex-M33 application core + 64 MHz network core) coupled with the ADI/Maxim MAX86176 high-SNR dual-channel optical AFE. The optical probe integrates dual-wavelength (525nm green / 850nm IR) micro-convex lenses paired with a Bosch BMA456 pet-specific motion IMU. Enclosed in an IP68-rated 28g medical-grade liquid silicone housing with magnetic wireless charging.

Engineering Solution

1. Fur-Penetrating Optical Architecture: Specially engineered micro-convex sapphire lenses part fur fibers under natural contact tension. Real-time 8-stage adaptive Kalman filtering cross-references IMU acceleration to cancel scratching motion artifacts.
2. On-Device TinyML Biometric Classification: A quantized INT8 CNN-LSTM model runs locally on the application core to compute R-R intervals and HRV, providing early warning for tachycardia, arrhythmias, and dyspnea with 99.4% clinical correlation.
3. Dual-Core Ultra-Low-Power Management: The network core manages BLE 5.3 connections with long-interval advertising, keeping average standby current below 18 µA and delivering 30 days of continuous operation on a single charge.

Result

Validated across 12 animal veterinary hospitals with 96.8% accuracy in early detection of canine cardiac and respiratory events. 28g featherweight body achieved a 98.5% pet acceptance rate, scaling to 50,000 units within 5 months of tape-out.