PP042 Personalized therapy sequencing in triple-negative breast cancer using microenvironment-guided reinforcement learning and spatial transcriptomic feedback

Triple-negative breast cancer (TNBC) often develops resistance to systemic therapies due to dynamic remodeling of the tumor microenvironment (TME). Despite its clinical relevance, treatment selection models typically assume a static TME and fail to account for evolving spatial interactions between immune, stromal, and malignant components. This study aims to develop an adaptive AI-based framework using multi-agent reinforcement learning (MARL) to simulate therapy-guided TME progression and optimize treatment sequencing in TNBC by integrating histopathologic and spatial transcriptomic data.