Sequential and stochastic-cost extension of single-stage budgeted prescriptive analytics

Extend single-stage budgeted prescriptive-analytics frameworks that allocate continuous treatment doses under deterministic total treatment costs to sequential intervention strategies with multi-stage horizons and random cumulative costs.

Background

The paper situates its contribution relative to prior predict-then-optimize work on allocating continuous treatment doses under a budget. That prior framework separates prediction from optimization but operates at a single decision point and assumes deterministic treatment costs. The authors identify sequential decisions and random cumulative costs as unresolved extensions relevant to maintenance schedules, treatment trajectories, and other multi-stage interventions.