Intensive Care Unit Capital-Budgeting Workbench: An Open-Source Decision-Support Application for High-Acuity Investment Planning

Capital budgeting in intensive care units (ICUs) demands rapid, high-stakes investment decisions. We developed an open-source ICU Capital Budgeting Workbench that merges a deterministic finance engine with GPT-based natural-language processing (NLP) to streamline planning. The web application couples classic valuation metrics (e.g., net present value, internal rate of return, and payback) with an NLP module that converts multi-year free-text scenarios into structured projections. Users describe projects in everyday language; GPT parses these narratives into year-by-year cash flows and discount rates, after which the engine computes financial metrics and produces interactive sensitivity, scenario, and Monte Carlo analyses. In illustrative cases the workbench translated narrative ICU expansion proposals into five and ten-year cash-flow tables and NPV calculations within seconds, eliminating manual spreadsheet construction. Interactive dashboards let users test key assumptions, instantly revealing how occupancy, reimbursement, or inflation shifts influence returns. Compared with traditional ad-hoc spreadsheets, the tool demonstrated marked time savings and consistent analytic structure. This proof-of-concept shows how large language models can reduce transcription errors, standardize methodology, and embed uncertainty modeling in routine capital planning. Limitations include dependence on GPT’s parsing accuracy and the need for real-world validation with authentic hospital data. The ICU Capital Budgeting Workbench exemplifies practical AI integration for finance and operations leaders, offering transparent, reproducible, and scalable decision support for ICU equipment and facility investments. By replacing bespoke spreadsheets with a governed, open-source platform, it may improve efficiency and support better-informed, data-driven investment strategy.