Objective <p>We present a novel dynamic risk-stratified breast cancer screening Markov model designed to maintain robust and epidemiologically consistent cancer prevalences, regardless of the number or composition of risk-stratification groups. This approach addresses a common limitation in existing models, where altering risk-group definitions or proportions can unintentionally distort overall incidence rates. By overcoming this constraint, the method lowers barriers to developing interactive, flexible and policy-relevant models that can be shared directly with decision makers. The paper serves as both a methodological contribution and a practical guide for implementation.</p> Methods <p>Our approach combines conditional transition probabilities with pre-stratified ‘at-risk’ states within a conventional Markov cohort framework. Lifetime risk is determined at the structural level, while the timing of onset is governed by age-specific conditional probabilities. These components are derived directly from Flemish cancer registry data, enabling close alignment with the target population and facilitating epidemiological validation.</p> Results <p>We validate the model against empirical breast cancer incidence data from Flanders, comparing predicted outcomes across age bands and cancer stages. The method consistently reproduces observed incidence patterns without additional calibration, maintaining stability when risk group definitions or sizes are varied.</p> Conclusions <p>Combining pre-stratified ‘at-risk’ states with conditional transition probabilities offers a simple yet powerful means of achieving epidemiological consistency in both risk-stratified and non-stratified cancer screening models. The method is data driven, transparent, and adaptable to different cancers or screening contexts, making it especially valuable for interactive models intended for use by policymakers.</p>