Abstract

Analytical and artificial intelligence (AI) methods are becoming more widely utilized in healthcare fraud detection. However, most existing methods assume clean and legitimate data streams, and do not consider the existence of smart adversaries. These adversaries may attempt to influence data which in turn may impact analytical and AI method outcomes. This paper presents a decision theoretic approach for AI methods in adversarial environments. Proposed adversarial risk analysis based framework allows incomplete information and adversarial perturbations on the data inputs. We solve the adversary’s poisoning decision problem where he manipulates batch data inputted into the methods, and discuss potential defender strategies to improve the security of existing AI frameworks.