Exploration of Predictive Potential of AI-enabled Portable System in Anticancer Drug Delivery: A Comparative Study with Modified Gompertz like Biphasic Response Model.

Abstract

Mathematical models are conventionally used to understand the of tumor behaviors, but they generally lack in precisely correlating drug efficacy with tumor response. Artificial intelligence (AI) has forged a new avenue in cancer management, but requires complex and heavy computing resources. In this paper, we have presented an AI enabled single board computer (SBC) and proposed a modified Gompertz like biphasic response model (MGBRM) for the prediction of anti-tumor activity of docetaxel-palmitate and its solid lipid nano-particles on breast cancer. Linear regression algorithm using C +  + library utilizing in-vivo experimental data over the span of 20 days was employed. A MGBRM was validated for no treatment, treatment with DTX-PL and DTX-PL-SLN using in-vivo data and compared with the AI model. The actual tumor volumes versus the numerically calculated tumor volumes from the modified Gompertz model exhibited good correlation coefficient with r² value of 0.999 for no treatment, 0.986 for DTX-PL and 0.998 for DTX-PL-SLN. In addition to that, the presented AI enabled SBC system also demonstrated good correlation with tumor volumes obtained through in-vivo experiment over a time. The r² for actual tumor volumes versus AI predicted tumor volumes for the studies conditions were close to 1. Both models were compared for biphasic response that can be useful to understand the numerical system parameters and black-box (AI) prediction for the tumor specific treatment. However, the modified MGBRM model is a leveraging step in predicting the tumor volumes in animals receiving treatment that was not feasible with the conventional model.