Real-Time Predictive Analytics for Oncolytic Virus Replication and Tumor Lysis Using Machine Learning

Oncolytic virotherapy uses genetically engineered viruses to selectively destroy tumor cells while sparing healthy tissue. To address the complex virus-tumor-immune interactions, this study proposes a real-time predictive framework using advanced machine learning and deep learning to optimize treatment outcomes. It analyzes clinical trial data and records from 1,200 patients in Northern India, mainly Punjab, incorporating genomic, proteomic, and tumor microenvironmental features. Kaggle's Brain Tumor MRI Dataset is also used to improve classification of Glioma, Meningioma, Pituitary, and No Tumor cases. Key goals include quantifying tumor lysis efficiency as a function of virus burst size and cell susceptibility, enhancing predictive accuracy of tumor regression patterns with multi-scale data integration, and simulating tumor lysis dynamics in response to oncolytic virus activity. Models like RF (98.3%), SVM (98.5%), ANN (99%), KNN (99.6%), GBM (99.7%), and CNN (100%) were used. Techniques like RFE, Permutation Importance, and Grad-CAM enhanced accuracy and interpretability.