Return on investment (ROI) analysis in the face of uncertainty

Calculation of return on investment should be as easy as knowing the cost and the benefit and doing the math. Unfortunately in medical informatics rarely do we know the benefit with absolute certainty. Often the benefit is a combination of "fuzzy" ill defined concepts. In order to support the ROI analysis for medical information systems we developed a probabilistic time varying model that allows each of the fuzzy concepts of savings to be expressed as a likelihood distribution rather than a fixed value. The model and an interactive simulator were created using the graphical programming language G (National Instruments, Dallas,TX). This tools allows rapid prototyping and drag and drop ease of customizing the model to the particular clinical setting. The resulting ROI is expressed in terms of likelihoods, potential risk and confidence intervals providing a unique view of the financial analysis not typically available. In this paper an example is illustrated for perioperative information systems.