Dynamic Security for Medical Record Sharing

Information routinely collected by health care organizations is used by researchers to analyze the causes of illness and evaluate the effectiveness of potential cures. Medical information sharing systems are built to encourage hospitals to contribute patient data for use in clinical studies. These organizations possess a wide variety of environments and risk assessments, and require sufficient assurances of patient privacy. This thesis introduces mechanisms to dynamically generate an applicable security policy for medical information sharing systems. We present implementation-independent mechanisms that are capable of interoperating with different security settings at different sites to produce security configurations with significantly different characteristics and vulnerabilities. We also present a rules-based agent to assist in the selection process. This approach gives maximum freedom to generate the appropriate system according to the tradeoffs between cost, patient privacy, and data accessibility. Thesis Supervisor: Peter Szolovits Title: Director, Clinical Decision Making Group