RAMOS: A Model Validation and Sensitivity Analysis

Mayhew, L.D. & Taket, A. (1981). RAMOS: A Model Validation and Sensitivity Analysis. IIASA Working Paper. IIASA, Laxenburg, Austria: WP-81-100

Preview

Text WP-81-100.pdf Download (915kB) | Preview

Abstract

This paper focuses on the predictive capabilities of the model RAMOS (Resource Allocation Model Over Space). This model, which is being jointly developed by IIASA and the Operational Research Services of the UK Department of Health and Social Security (DHSS), is designed to predict the impact on hospitalization rates when population and resource availability are changing simultaneously in time and space. The performance of the model is critically examined using validation techniques and sensitivity analysis. The validation part is based on an experiment that tries to simulate the outputs of a regional health care system at a point earlier time. This "back-prediction" is then compared for accuracy with what actually occurred. It is shown that the model functions very well in achieving the purposes for which it was designed. Different model specifications are then tested in order to seek further improvements that remove some small but consistent biases in the outputs. Following this, a detailed sensitivity analysis is carried out on the main input variables and parameter, in order to check the internal consistency of the model when it is exposed to unrealistic extremes of change. The paper concludes by noting the mostly satisfactory performance of the model in both the validation tests and the sensitivity analysis but with some caveats and recommendations for further research.

Actions (login required)

View Item