Advanced Sybil Attack Mitigation in VANETs via Proof of Work and Location-Based Techniques

Abstract

Pharmaceutical medication development is a time-consuming and challenging process. Unexpected negative drug reactions that arise during the process may cause the entire drug development pipeline to be stopped or restarted. As a result, it is essential to predict the adverse effects of the medicine in advance of its design. In our Deep Side approach, we use context-related (gene expression) information along with the chemical structure to forecast ADRs, taking into account variables such as dosage, time interval, and cell line. By using GEX and CS as integrated features, the proposed MMNN model outperforms models that solely use chemical structure (CS) fingerprints in terms of accuracy. The claimed accuracy is remarkable considering that our objective is to forecast the adverse effects that are independent of the situation. Finally, the SMILES Conv model outperforms all other approaches by applying convolution to the SMILES representation of drug chemical structure.