AI-Guided Digital Oncology — Nanoparticle Tumor Modeling
- mathematical modeling
- Bayesian
- MCMC
- Neural ODE
What it is
A hybrid modeling framework for nanoparticle transport in tumors that blends mechanistic math with machine learning:
- a 5-state ODE system (free, bound, binding-site, internalized nanoparticles, and tumor) with radial discretization and 16 biologically interpretable parameters,
- Bayesian parameter estimation via MCMC (128 walkers, 7000 steps, Gelman-Rubin convergence), and
- machine-learning validation with LSTM and Neural-ODE models.
It compares four treatment groups (Untreated, Saline, MNP, MNPFDG) and ranks their efficacy, with full reproducibility scripts. Companion code to your digital-oncology paper.
Stack
Python · SciPy ODEs · emcee (MCMC) · PyTorch · torchdiffeq · LSTM