Skip to content

GRAFT

GRAFT (Generalized Runtime Analytics & Functional Twin) is an integrated platform for scientists and engineers working with bioreactor systems. It connects live sensor data from your bioreactor with trained machine learning models (LSTM, XGBoost, Linear Regression) to deliver predictive insights, run simulations, and evaluate model performance — all from a NiceGUI-based interface.

What You Can Do

Section titled “What You Can Do”

Real-Time Monitoring

Section titled “Real-Time Monitoring”

Stream and visualize live bioreactor parameters, continuously updated as your process runs.

ML Predictions

Section titled “ML Predictions”

Run models such as LSTM, XGBoost, and Linear Regression to forecast optical density or growth rate.

Simulations & Evaluation

Section titled “Simulations & Evaluation”

Run what-if scenarios with parameter overrides and evaluate model performance against ground-truth reference datasets.

Data Visualization

Section titled “Data Visualization”

Explore interactive charts, compare experiments, and analyze metrics across historical run data.

Section titled “Quick Links”

Project Layout

Section titled “Project Layout”
graft/
├── app/
│   ├── main.py
│   ├── backend/         # Data ingestion, WebSocket, DB, ML utilities
│   ├── configs/         # YAML configs + shared constants
│   ├── models/          # LSTM, XGBoost, LinearRegression, Gompertz
│   ├── services/        # Business logic for each dashboard tab
│   └── ui/              # NiceGUI dashboard UI and components
├── tests/               # Unit and integration tests
└── docs/                # This documentation