Unlocking Actionable Intelligence: Explainable Hybrid Graph Neural Networks for Dynamic Pharma Omnichannel Recommendations

The paper presents a novel hybrid graph-based framework combining Temporal Graph Neural Networks (TGNNs) and Heterogeneous Graph Transformers (HGTs) to model the complex, dynamic relationships in the pharmaceutical ecosystem. It enables explainable, causally attributed recommendations for commercial decision-making, including forecasting prescription potential, segmenting HCPs and patients, and identifying influence pathways. The approach aims to improve omnichannel engagement and drive patient-centric outcomes