Project brief
A clinical-grade 12-lead ECG acquisition system for hospital and ambulatory cardiac monitoring, transmitting data wirelessly via BLE 5.0 to a tablet display application. A TensorFlow Lite model running on the device processor performs real-time arrhythmia detection including atrial fibrillation, ST-elevation, ventricular tachycardia, and PVC. HIPAA-compliant cloud storage maintains records for 10 years.
The supplied record describes a medical monitoring design. This explanatory account does not independently validate diagnostic performance, clinical suitability, or privacy compliance.
The engineering challenge
Synchronize multichannel ECG acquisition, artifact processing, and monitored data delivery.
Engineering approach
- Organize acquisition, filtering, and model inference into separate processing stages.
- Preserve record context while streaming to the tablet interface.
- Link stored waveforms, annotations, and alert states for review.
Features & capabilities
- 12-lead ECG acquisition at 1000 Hz per channel using ADS1298 ECG IC
- AI arrhythmia detection: AF, ST-elevation, VT, PVC (92% specificity/sensitivity)
- BLE 5.0 wireless (Coded PHY, 10m reliable range in hospital environment)
- Real-time cloud storage (HIPAA-compliant) with 10-year retention
- Android tablet application with rhythm strip scrolling and annotation
- Alert escalation: on-device alarm → nurse call integration → pager
- Dry electrode motion artifact suppression using adaptive filter
- IEC 60601-1 Third Edition electrical safety compliance framework
Software & engineering tools
nRF52840, ADS1298 ECG AFE, TensorFlow Lite, Android tablet app (Kotlin), AWS HealthLake backend