Sundew Algorithms implements bio-inspired adaptive gating for stream processing with statistical validation, layered precision uplift, and hardware readiness. Tested across healthcare, IoT, ECG, financial, and network security domains.
Comprehensive benchmarking across 6 datasets with 10+ presets, featuring bootstrap confidence intervals and statistical validation for production readiness.
| Dataset | Preset | Recall | Energy Savings | Precision (95% CI) |
|---|---|---|---|---|
| Heart Disease | custom_health_hd82 | 0.196 | 82.0% | 0.755 (0.680-0.828) |
| Breast Cancer | custom_breast_probe | 0.118 | 77.2% | 0.385 (0.294-0.475) |
| IoT Sensors | auto_tuned | 0.500 | 88.2% | 0.670 (0.574-0.758) |
| Network Security | aggressive | 0.233 | 89.2% | 0.461 (0.355-0.562) |
| Financial | aggressive | 0.164 | 90.1% | 0.219 (0.144-0.304) |
| MIT-BIH ECG | auto_tuned | 0.218 | 88.8% | 0.340 (0.328-0.352) |
* 95% confidence intervals from 1000 bootstrap samples
Comprehensive visual documentation of the Sundew Algorithms system design and processing pipeline.
Comprehensive testing across healthcare, IoT, financial, and security domains with statistical rigor.
Cardiovascular risk assessment with clinical features.
Tumor characteristics with enriched features.
Cardiac arrhythmia detection from ECG signals.
Multi-sensor monitoring with anomaly detection.
Intrusion detection and anomaly patterns.
Market data with volatility and anomalies.
# Install from PyPI
pip install sundew-algorithms
# Or with uv (recommended)
uv pip install sundew-algorithmsimport sundew
# Load preset configuration
config = sundew.get_preset('custom_health_hd82')
# Create and run algorithm
algorithm = sundew.SundewAlgorithm(config)
result = algorithm.process(event_data)# Execute full evidence suite (all datasets, presets, analyses)
uv run python tools/run_full_evidence.py
# Generate specific benchmarks
uv run python benchmarks/run_dataset_suite.py --presets tuned_v2 auto_tuned aggressive
# Create architecture diagrams
uv run python create_architecture_diagram.py