Synheart Emotion Overview

​

What is Synheart Emotion?

• 😌 Baseline: Relaxed, peaceful emotional state
• 😰 Stress: Anxious, tense emotional state

​

Key Features

​

On-Device Processing

• All inference happens locally on your device
• No network calls required
• No raw biometric data leaves the device
• Privacy-first by design

​

Real-Time Performance

• < 1s inference latency per inference (ONNX models)
• ~0.8-7.24 MB model size per model

​

Research-Based

• Trained on WESAD dataset (wearable stress and affect detection)
• 78.4% accuracy on binary classification (Baseline vs Stress)
• 72.6% F1 score on WESAD validation set (LOSO CV)
• ExtraTrees classifiers with 14 HRV features
• ONNX format optimized for on-device inference
• Reproducible with published model card

​

Multi-Platform

​

Architecture

Wearable / Sensor
   └─(HR bpm, RR ms)──► Your App
                           │
                           ▼
                   Synheart Emotion SDK
            [Ring Buffer] → [Feature Extraction] → [Normalization]
                                     │
                                  [Model]
                                     │
                              EmotionResult

• Ring Buffer: Holds last 120s of HR/RR data (configurable, default: 120s window, 60s step)
• Feature Extractor: Computes 14 HRV features (time-domain, frequency-domain, non-linear)
• Scaler: Standardizes features using training μ/σ (built into ONNX model)
• Model: ExtraTrees (Extremely Randomized Trees) classifier
• Emitter: Throttles outputs (default: every 60s)

​

Quick Start Examples

from datetime import datetime
from synheart_emotion import EmotionEngine, EmotionConfig

# Initialize engine
config = EmotionConfig()
engine = EmotionEngine.from_pretrained(config)

# Push biosignal data
engine.push(
    hr=72.0,
    rr_intervals_ms=[850.0, 820.0, 830.0, 845.0, 825.0],
    timestamp=datetime.now()
)

# Get inference results
results = engine.consume_ready()
for result in results:
    print(f"Emotion: {result.emotion} ({result.confidence:.1%})")

import 'package:synheart_emotion/synheart_emotion.dart';

// Initialize the emotion engine (default: 120s window, 60s step)
final engine = EmotionEngine.fromPretrained(
  const EmotionConfig(),
);

// Push biometric data
engine.push(
  hr: 72.0,
  rrIntervalsMs: [823, 810, 798, 815, 820],
  timestamp: DateTime.now().toUtc(),
);

// Get emotion results (async for ONNX models)
final results = await engine.consumeReadyAsync();
for (final result in results) {
  print('Emotion: ${result.emotion} (${(result.confidence * 100).toStringAsFixed(1)}%)');
}

import ai.synheart.emotion.*

val config = EmotionConfig()
val engine = EmotionEngine.fromPretrained(config)

engine.push(
    hr = 72.0,
    rrIntervalsMs = listOf(850.0, 820.0, 830.0, 845.0, 825.0),
    timestamp = Date()
)

val results = engine.consumeReady()
results.forEach { result ->
    println("Emotion: ${result.emotion} (${result.confidence})")
}

import SynheartEmotion

let config = EmotionConfig()
let engine = try! EmotionEngine.fromPretrained(config: config)

engine.push(
    hr: 72.0,
    rrIntervalsMs: [850.0, 820.0, 830.0, 845.0, 825.0],
    timestamp: Date()
)

let results = engine.consumeReady()
results.forEach { result in
    print("Emotion: \(result.emotion) (\(result.confidence))")
}

​

Use Cases

​

Mental Health Apps

from synheart_emotion import EmotionEngine, Emotion

engine = EmotionEngine.from_pretrained()

# Stream biosignals from wearable
for biosignal in wearable_stream():
    engine.push(
        hr=biosignal.heart_rate,
        rr_intervals_ms=biosignal.rr_intervals,
        timestamp=biosignal.timestamp
    )

    results = engine.consume_ready()
    for result in results:
        if result.emotion == Emotion.STRESSED and result.confidence > 0.7:
            send_notification("Consider taking a break")

​

Wellness Coaching

// In your Flutter wellness app
final emotionEngine = EmotionEngine.fromPretrained();

// Integrate with Synheart Wear
synheartWear.streamHRV(windowSize: Duration(seconds: 60))
  .listen((metrics) {
    emotionEngine.push(
      hr: metrics.getMetric(MetricType.hr)!,
      rrIntervalsMs: metrics.rrIntervals,
      timestamp: DateTime.now(),
    );

    final results = await emotionEngine.consumeReadyAsync();
    if (results.isNotEmpty) {
      updateDashboard(results.first);
    }
  });

​

Research Applications

// Android research app
val emotionEngine = EmotionEngine.fromPretrained(
    EmotionConfig(
        windowDuration = 60_000L, // 60 seconds
        stepDuration = 5_000L      // 5 seconds
    )
)

// Log all results for analysis
val results = emotionEngine.consumeReady()
results.forEach { result ->
    database.insert(
        timestamp = result.timestamp,
        emotion = result.emotion,
        confidence = result.confidence
    )
}

​

Model Details

• Time-domain: RMSSD, Mean_RR, HRV_SDNN, pNN50
• Frequency-domain: HRV_HF, HRV_LF, HRV_HF_nu, HRV_LF_nu, HRV_LFHF, HRV_TP
• Non-linear: HRV_SD1SD2, HRV_Sampen, HRV_DFA_alpha1
• Heart Rate: HR (in BPM)

• ExtraTrees_120_60: 120-second window, 60-second step (default)
• ExtraTrees_60_5: 60-second window, 5-second step
• ExtraTrees_120_5: 120-second window, 5-second step

• Accuracy: 78.4% on WESAD validation set (LOSO CV)
• F1 Score: 72.6% on WESAD validation set (LOSO CV)
• Latency: < 1ms per inference (ONNX models)
• Model Size: ~0.8-7.24 MB per model

​

API Parity

​

Available SDKs

​

Privacy & Security

• On-Device Processing: All emotion inference happens locally
• No Data Retention: Raw biometric data is not retained after processing
• No Network Calls: No data is sent to external servers
• Privacy-First Design: No built-in storage - you control what gets persisted
• Not a Medical Device: This library is for wellness and research purposes only

​

Resources

• GitHub: synheart-ai/synheart-emotion
• Model Card: Model Details
• Technical Spec: RFC E1.1
• Issues: Report Bugs
• Discussions: Community Forum

​

Citation

@software{synheart_emotion,
  title = {Synheart Emotion: Multi-platform SDK for on-device emotion inference from biosignals},
  author = {Synheart AI Team},
  year = {2025},
  version = {0.1.0},
  url = {https://github.com/synheart-ai/synheart-emotion}
}