Quantifying Eligibility Pattern Shifts: a Data-Driven Paradigm for Early Risk Detection in Clinical Trials

Traditional Risk-Based Monitoring (RBM) strategies emphasise key risk indicators and site-level performance metrics but seldom address the heterogeneity of patient eligibility profiles. We present a data-driven framework that captures temporal and inter-site shifts in baseline inclusion characteristics. Central to this framework are two new metrics-Borderline Inclusion Index and Eligibility Distribution Divergence-that quantify departures from expected enrolment patterns. A Bayesian composite score synthesises these indicators to prioritise oversight actions. Through simulation experiments and a worked case study, we show that monitoring eligibility pattern shifts offers an early warning signal of operational or scientific risk and strengthens overall trial integrity. We operationalize the framework through an interactive Shiny web application that computes indicator-specific posteriors, generates composite site risk scores, and provides visual decision-support for centralized RBM implementation.