Integrating elicited patient preferences and clinical trial data in a quantitative model for benefit-risk assessment