Abstract:
The following thesis presents the research to promote generalizability and explain
ability across multiple drug-disease contexts to integrate transfer learning into a re
inforcement learning framework. To address and overcome the limitations of tradi
tional drug repurposing approaches, task adaptive attention mechanisms and transfer
learning framework techniques are integrated. The proposed methodologies increase
the efficiency and efficacy of drug repurposing attempts by making the models more
adaptable to various conditions while focusing on the enhancing transparency, adapt
ability, and model performance. The evaluations that demonstrate interpretabil
ity, accuracy and efficacy compared to baseline reinforcement models. The results
achieved from this model indicate that it not only expands the application of RL in
the field of medical AI, but also makes these AI systems more transparent and trust
worthy.
