Two-level Gaussian Process Regression – Response Metamodel for Epidemic Intervention Simulation

This repository contains the implementation of a two-level metamodeling framework developed for county-level opioid epidemic simulations in Pennsylvania. The framework approximates outputs of a large-scale agent-based model (FRED) across varying intervention dispensing rates of naloxone and buprenorphine. ├── pa_metamodel.ipynb # Main GPR metamodel notebook├── calibration/│ ├── calibration_pa.R # OUD Markov model and likelihood│ ├── imis_calibration.R # IMIS posterior estimation│ └── README.md # Calibration documentation├── resources/│ └── data_pa_int_XY.csv # FRED simulation outputs (X,Y = 1-5)└── README.md