Development of computational model for description of magnetic drug targeting for cancer therapy: Modeling and validation

Abstract

Computation of blood flow containing ferrofluid would be useful for analysis of drug carrier motion for cancer therapy. A thorough understanding nanoparticles behavior is challenging and needs to be addressed by developing sophisticated theoretical methods. A hybrid modeling for analysis of blood motion containing ferrofluid was implemented via mechanistic modeling combined with artificial intelligence. The system of analysis also considered external magnetic force for control of nanoparticles motion in the blood vessel. This research focuses on the analysis of velocity field based on a dataset consisting of variables x(m), y(m), and U(m/s). The objective is to develop accurate predictive models using Gaussian Process Regression (GPR), Kernel ridge regression (KRR), and Polynomial Regression (PR). The Dragonfly Algorithm (DA) was employed for hyper-parameter optimizing. The results demonstrate the performance of these models in relation to R² score, RMSE, and MAE. The GPR model achieves the highest score of 0.99603 in terms of R², indicating excellent predictive accuracy. It also exhibits the lowest RMSE of 7.1443x10^-3 and MAE of 5.35436 x10^-3, suggesting minimal deviations between the expected and predicted velocity values. The PR model also has a significant performance with an R² test score of 0.99348, RMSE of 9.1376 x10^-3, and MAE of 7.22828 x10^-3. The aforementioned results underscore the effectiveness of these models in accurately forecasting velocity based on the provided input variables.