rPPG Health - Contactless Physiological Telemetry Engine

AI-driven contactless physiological monitoring platform using remote photoplethysmography (rPPG) to extract blood volume pulse (BVP) waveforms directly from standard RGB webcams, with the next stage focused on CDSCO regulatory compliance.

Milestone:Next Stage: CDSCO RegulatoryREGULATORY

Hardware Requirement:720p/1080p Standard WebcamsZERO_COST_HW

HRV Metric Output:RMSSD, SDNN, LF/HF Power RatioCLINICAL_GRADE

// REAL-TIME PERFORMANCE METRICSREFRESH_RATE: 1000ms

VIDEO_SAMPLING_RATELIVE_TELEMETRY

0fps

Standard consumer webcam compatibility

SIGNAL_STABILITY_GAINLIVE_TELEMETRY

+0%

FaceMesh dynamic ROI auto-reanchoring

PHYSICAL_CONTACTLIVE_TELEMETRY

0sensors

100% optical contactless diagnostic

HEART_RATE_MAELIVE_TELEMETRY

0.0bpm

Benchmarked against FDA-approved pulse oximeters

// TOPOLOGICAL PIPELINE MAP

SYSTEM_PIPELINE_ARCHITECTURE_GRAPH[RPPG-HEALTH_TOPOLOGY]

NODES: ONLINEFLOW: DUPLEX_STREAM

01 // INGESTIONSTEP_1

→Uncompressed 30 FPS RGB Camera Stream
→MediaPipe FaceMesh 468-Point Mesh Tracker
→Dynamic Forehead & Malar ROI Auto-Cropper
→Ambient Illumination & Motion Normalizer

SENSOR / INFLOWSTANDBY

02 // INFERENCE & PROCESSINGSTEP_2

→Plane-Orthogonal-to-Skin (POS) Algorithm
→Chrominance-Based (CHROM) Pulse Decomposition
→4th Order Butterworth Bandpass Filter (0.75-2.5Hz)
→Fast Fourier Transform (FFT) Power Peak Locator
→Random Forest Blood Pressure / Vitals Proxy Regressors

ACCELERATED_COREACTIVE_NODE

03 // TRANSPORT & BUSSTEP_3

→FastAPI Streaming WebSocket Interface
→NumPy In-Memory High-Speed Circular Ring Buffer
→ZeroMQ IPC Data Bus for Multi-Process Isolation
→Encrypted SQLite Clinical Audit Log Ledger

LOW_LATENCYSTANDBY

04 // CLIENT & INTERFACESSTEP_4

→Pure Canvas High-Performance BVP Waveform Renderer
→Live HRV Spectral Power Tachogram (RMSSD, LF/HF)
→Real-Time Systolic Peak Marker & Quality Indicator
→Audit-Grade Telemetry Inspector (Prep for CDSCO Track)

TELEMETRY_OUTSTANDBY

DATA_BUS:G.711u / WebSocket / Async Event Loop

CLICK_NODE_TO_INSPECT_VECTOR_STATE

$ cat /etc/manifests/rppg-health.json

ENCODING: UTF-8

{
  "project_id": "rppg-health",
  "status": "CDSCO_IN_REVIEW",
  "timeframe": "2026 - Present",
  "pipeline_stages": {
    "ingestion": [
      "Uncompressed 30 FPS RGB Camera Stream",
      "MediaPipe FaceMesh 468-Point Mesh Tracker",
      "Dynamic Forehead & Malar ROI Auto-Cropper",
      "Ambient Illumination & Motion Normalizer"
    ],
    "inference": [
      "Plane-Orthogonal-to-Skin (POS) Algorithm",
      "Chrominance-Based (CHROM) Pulse Decomposition",
      "4th Order Butterworth Bandpass Filter (0.75-2.5Hz)",
      "Fast Fourier Transform (FFT) Power Peak Locator",
      "Random Forest Blood Pressure / Vitals Proxy Regressors"
    ],
    "transport": [
      "FastAPI Streaming WebSocket Interface",
      "NumPy In-Memory High-Speed Circular Ring Buffer",
      "ZeroMQ IPC Data Bus for Multi-Process Isolation",
      "Encrypted SQLite Clinical Audit Log Ledger"
    ],
    "client": [
      "Pure Canvas High-Performance BVP Waveform Renderer",
      "Live HRV Spectral Power Tachogram (RMSSD, LF/HF)",
      "Real-Time Systolic Peak Marker & Quality Indicator",
      "Audit-Grade Telemetry Inspector (Prep for CDSCO Track)"
    ]
  },
  "telemetry": {
    "VIDEO_SAMPLING_RATE": "30fps",
    "SIGNAL_STABILITY_GAIN": "+28%",
    "PHYSICAL_CONTACT": "0sensors",
    "HEART_RATE_MAE": "1.8bpm"
  }
}

JOURNALCTL_OUTPUT_LOGS :: RPPG-HEALTH

LEVEL: INFO / DIAGNOSTIC

01[2026-02-02 11:04:01.012] OPTICAL_FEED: Camera locked @ 1920x1080 30.00 FPS (Lighting: 420 Lux)

02[2026-02-02 11:04:01.120] FACEMESH: 468 facial landmarks detected (Confidence: 0.984)

03[2026-02-02 11:04:01.185] ROI_ISOLATION: Forehead (P70-P107) + Dual Malar cheeks isolated

04[2026-02-02 11:04:01.300] POS_ENGINE: Skin reflection orthogonal projections calculated

05[2026-02-02 11:04:01.350] BANDPASS_FILTER: Butterworth [0.75Hz - 2.5Hz] attenuated high-freq motion noise

06[2026-02-02 11:04:01.420] SPECTRAL_FFT: Dominant pulse peak locked @ 1.216Hz -> Pulse: 73.0 BPM

07[2026-02-02 11:04:01.460] HRV_ESTIMATE: RMSSD: 41.8ms | LF/HF Ratio: 1.12 (Parasympathetic nominal)

08[2026-02-02 11:04:01.500] AUDIT_TELEMETRY: Frame signed with SHA-256 in prep for CDSCO review [VALIDATED]

// SUBSYSTEM ENGINEERING SPECIFICATIONS

>1. Solving Motion Artifacts & Ambient Noise

▪Developed dynamic ROI tracking using MediaPipe FaceMesh to maintain accurate malar/forehead spatial anchors through ±25° head yaw/pitch.
▪Combined POS and CHROM algorithms with temporal smoothing to eliminate lighting flicker and low-frequency motion drift.
▪Benchmarked against gold-standard clinical pulse oximeters, achieving Mean Absolute Error under 2.0 BPM across diverse skin tones.

>2. Next Stage: CDSCO Regulatory Preparation

▪Structuring the data acquisition pipeline in alignment with upcoming CDSCO medical device regulatory standards.
▪Implementing deterministic latency, SHA-256 frame signing, and immutable audit logs required for the CDSCO regulatory phase.

← Return to System DashboardTERMINAL_NODE: 0xrppg-health [STABLE]