Deep Learning Enabled Digital Twin in Breast Cancer Treatment with Smart Cyber-Physical System for Adaptive Therapy Optimization and Tumor Evolution Forecasting (pp.316-331)
Y. Qiao1, X. Li1, L. Zhao1, C. Liu2, S.B. Sun 1,*
https://doi.org/10.30546/1683-6154.25.2.2026.3161General Surgery, Hongqi Hospital Affiliated to Mudanjiang Medical University, Mudanjiang 157000, China,
e-mail: 19545303079@163.com, starinnight@126.com, 342579584@163.com
2Department of Pharmacy, The Third Hospital of Heilongjiang, Heihe 164000, China, e-mail: lc8676263@163.com
∗Corresponding author e-mail: sun.shaobin@163.com
Abstract. This paper incorporates the Digital Twin (DT) technology, Deep Learning (DL), and Cyber-Physical Systems (CPS) to facilitate change in therapeutic approaches. To predict the evolution of the tumor and response to treatment based on continuously evolving clinical data, a hybrid mechanistic deep learning model is implemented to create a patient-specific DT. According to these predictions, a Deep Reinforcement Learning (DRL) agent is employed to determine optimal, individualized dosing regimens that will maximize tumor ablation and reduce systemic toxicity. The CPS applies such decisions in a closed-loop sensethink act methodology, which allows real-time adaptive control. In silico results demonstrate prolonged time-to-progression and reduced cumulative toxicity compared to conventional treatment strategies, highlighting the potential of the Digital Twin–Cyber-Physical System (DTCPS) systems in advancing personalized oncology.
Keywords: Digital Twin, Deep Learning, Breast Cancer, Cyber-Physical Systems, Adaptive Therapy, Personalized Medicine, Tumor Evolution Forecasting.
AMS Subject Classification: 92C50, 68T07, 93E20, 92C42.