Background: The objective of Phase I/II dose-finding design is to determine the Optimal Biological Dose(OBD) that is acceptably safe and demonstrates sufficient anti-tumor activity by maximizing a pre-specified utility function that balances safety and efficacy. In oncology, treatment is typically delivered over multiple cycles, which complicates OBD identification, moreover, achieving higher anti-tumor activity requires toxicity that is under control. Method: To address this challenge, we propose and evaluate the improved utility with pre-specified weights and adaptive penalties methods to regulate toxicity according to doses. Results: This approach enables the dynamic optimal dose that could be recommended based on trade-off between safety and efficacy. Conclusions: The simulation studies demonstrate that the proposed model outperforms the simply utility method across several performance metrics.