MedGemma-Micro MacBook M2 · LiteRT & iOS Core ML
CHIP ACCEL MacBook M2 · XNNPACK
MODEL SIZE 301.93 MB
BUDGET < 350 MB (48.1 MB Headroom)
SIGNATURE Dual-Signature

PPG Waveform Monitor

90s Continuous Pulse Stream @ 25 Hz (2250 Samples)
Normal Sinus Rhythm
T: 0.0s - 90.0s Gain: 1.0x (Calibrated) 60 FPS
Simulate Cardiac Condition:
⌚
Wear OS Telemetry Ingestion (Galaxy Watch 4/5/6) Raw BioActive ADC (650k counts) • ChannelClient Ingestion • Anti-Aliased Resampling
Buffer Active
90s Rolling Ring Buffer: 2,250 / 2,250 samples 100% Full
Stream Galaxy Watch 4 Scenario:
HEART RATE BPM
72.0
Resting Rhythm
rMSSD (HRV) ms
38.4
Parasympathetic Vagal Tone
SDNN ms
41.2
RR Regularity Index
SENSOR LATENCY ms
14.8
1D-Conformer ANE / NPU Pass

Multi-Task Arrhythmia Classifier

1D-Conformer Biosignal Encoder (medgemma_micro_cardio_350m.tflite)

Cardiology Clinical Assistant

medgemma_micro_cardio_350m.tflite (11-Layer Transformer · M2 LiteRT)
TFLite 768-D Semantic Engine Active
Clinical Presets:
🩺
MedGemma-Micro 350M LiteRT M2

Hello! I am MedGemma-Micro running directly on your MacBook M2 via TensorFlow Lite (LiteRT with XNNPACK CPU acceleration).

You are actively testing the unified multi-signature model medgemma_micro_cardio_350m.tflite (301.93 MB, strict sub-350MB budget). You can:

  • Classify Arrhythmias: Test 90s PPG pulse streams across Normal Sinus, AFib, Bradycardia, Tachycardia, and PVC.
  • Wear OS Smartwatch Bench: Ingest streaming Galaxy Watch 4/5/6 PPG data, test 100 Hz decimation, and off-wrist lead-off rejection.
  • Cardiology Q&A: Inquire about medications, DASH diet, exercise guidelines, sleep apnea, or emergency triage (matched across 1,552 indexed clinical guidelines with 768-D semantic embeddings).
  • Run Automated M2 Benchmark: Click the "🚀 Run M2 Benchmark" button in the top navigation bar to execute the full 4-stage validation suite.
Ready
⚠️ Medical Disclaimer: For educational purposes only, not a prescription or treatment plan. Do not start, stop, or change any medication without your doctor’s approval.