Budget Control in Hospital Construction with AI: A Lesson from Government Project

A practical playbook is outlined in this chapter to utilize auditable light AI to keep government hospital construction within budget. Three components are fed by a standard monthly data template: (1) threshold alerts (R/A/G for cost, time/schedule and contingency), (2) a small predictive model forecast of next month's overrun percentage using 5-7 fully transparent features, and (3) a single page dashboard plus a regular review rhythm. When applied to a 24-month hospital program, the approach was able to reduce average monthly overrun from 8.0% to 5.3%, move numerous packages from RED/AMBER to GREEN and stabilize contingency usage. The results show significant differences between what has been completed and what is still to be done, along with increasing changes in scope as the earliest warning signs of potential problems. This is an easily adopted, audited and robust to staff turnover on public project methodology.