The document discusses building a predictive model to forecast reimbursement decisions by major health technology assessment (HTA) agencies. It explains that a good model could improve predictions, disentangle multiple influences on decisions, and provide guidance to increase the likelihood of positive coverage. However, care must be taken to avoid "data mining" and ensure any model is based on meaningful factors and can predict outcomes for new cases. The success of a model will be judged by how well it predicts decisions and if it performs better than alternative approaches.