Published June 6, 2026 | Version v1

Graph Neural Network-Based Drug Candidate Shortlisting for Cancer Treatment

Description

This paper presents OncoScreen AI, a Graph Neural 
Network based system for predicting anticancer drug 
potency. Trained on real-world ChEMBL and NCI-60 
datasets, achieving 85% accuracy and 0.91 AUC-ROC, 
outperforming traditional ANN by 7%. Developed as 
final year project at Vidyavardhini's College of 
Engineering and Technology, Mumbai.

Files

IEEE Template 1234.pdf

Files (899.5 kB)

Name Size Download all
md5:e98f19cb3f243411b1ad94f26c85a806
899.5 kB Preview Download