Improving accuracy of patient synthetic data for testing medical cyber-physical systems

Abstract

Medical Cyber-Physical Systems (MCPS) integrate the cyber space and physical world elements for promoting support for health assurance activities. MCPS are life-critical systems, demanding a strong engineering effort to guarantee safety, what directly impacts on testing process. Testing MCPS using real patients is very expensive and complex, since their lives are involved. Thus, the use of patient synthetic data becomes a promising approach. In this paper we propose a model for improving accuracy of patient synthetic data for testing MCPS based on regression models. We use an existing Patient Baseline Model to generate vital signs of patients, but improving the statistical analysis. Using our approach we increased in about 73.9% the quality of the regression models and, consequently, their accuracies.