Enhanced Drug Toxicity Prediction via Reverse Transfer Learning and Graph-Based Visual Verification

DOI

10.2991/978-94-6239-707-1_29How to use a DOI?

Keywords

Graph Neural Networks; Transfer Learning; Tox21; OCR; Drug Verification; SIDER

Abstract

Predicting molecular toxicity and verifying drug identity in real-world scenarios is a fundamental challenge in ensuring pharmaceutical safety. This paper presents a combined framework addressing both issues through a two-stage pipeline process. First, we validate a reverse transfer learning module, demonstrating that a Graph Neural Network (GNN) pre-trained on high-level clinical phenotypes (SIDER 4.1 dataset) which captures rich molecular representations that are transferable to drug toxicity prediction (Tox21). Our base SIDER-transferred GINE model achieved a competitive ROC-AUC of 0.8125 on the Tox21 benchmark dataset. Second, to address the challenge of reading text from cylindrical medicine bottles prevalent in the current Indian market, we implemented an automated high-precision OCR pipeline. By employing a cascading deep learning strategy using EasyOCR combined with a heuristic smart-matching algorithm and an early-exit optimization, the visual system achieved an accuracy of 98.0% on a validation subset, demonstrating robustness against cylindrical distortion and specular reflection on medicine bottles.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Govind Bhatter
AU  - Animesh Shukla
AU  - Pratham Popatiya
AU  - Pratik Shah
AU  - Jignesh Patel
PY  - 2026
DA  - 2026/06/18
TI  - Enhanced Drug Toxicity Prediction via Reverse Transfer Learning and Graph-Based Visual Verification
BT  - Proceedings of the International Conference on Recent Advances in Intelligent and Sustainable Technologies (RAIST 2026)
PB  - Atlantis Press
SP  - 328
EP  - 338
SN  - 2589-4919
UR  - https://doi.org/10.2991/978-94-6239-707-1_29
DO  - 10.2991/978-94-6239-707-1_29
ID  - Bhatter2026
ER  -