Hybrid Machine Learning and Mathematical Modeling for Tumor Dynamics Prediction: Comparing SPIONs against mNP-FDG

TL;DR

This study combines machine learning and mathematical modeling to compare SPIONs and mNP-FDG in tumor control, highlighting their respective strengths in tumor suppression and eradication, and provides a user-friendly GUI for personalized treatment predictions.

Contribution

It introduces a hybrid approach integrating ML with continuum models to optimize tumor therapy strategies and offers an interactive tool for clinical decision-making.

Findings

mNP-FDG controls tumor growth faster initially

SPIONs are more effective for complete tumor eradication

The combined approach suggests joint therapy as optimal

Abstract

This is a Machine Learning guided study towards zone-specific ray therapy. Combining Machine Learning (Extreme Gradient Boosting) with continuum modeling (exponential and logistic growth), we find that while fluorodeoxyglucose-coated (mNP-FDG) can control cancerous tumor progression within 2 days compared to 18 days by Superparamagnetic Iron Oxide Nanoparticles (SPIONs), for complete termination of the tumor, SPIONS (20 days) are superior compared to mNP-FDG (more than 40 days). We also provide an interactive graphical user interface (GUI) developed with Tkinter/Python that allows users to input relevant data, such as treatment type and time, to receive real-time tumor volume predictions. Our ML-guided prediction indicates joint therapy as the optimum choice, with mNP-FDG ideal for taming the tumor spread, followed by SPIONs for complete eradication, facilitating personalized cancer treatment in clinical practice.