Software Performance Engineering of a Web service-based Clinical Decision Support infrastructure

This paper has two contributions: a) it proposes a web services-based infrastructure to support Clinical Decision Support Systems (CDSSs) for processing multi-domain medical data from the obstetrical, perinatal and neonatal care domains, and b) applies Software Performance Engineering (SPE) to the proposed infrastructure. This extends a XML-based framework for medical data interoperability and integration of CDSSs into the Neonatal Intensive Care Unit, developed previously by the authors. The framework integrates CDSSs, such as Artificial Neural Networks (ANNs), Case-Based Reasoning (CBR) tools, and alert detection systems. The goal is to reduce medical errors, to support the physician's decision-making process and to improve ultimately patient care. We applied SPE from the early design stages in order to ensure that the system will meet its performance requirements, and to identify possible solutions for relieving the performance limitations of this prototype system. The performance evaluation is based on a layered queuing network model of the proposed web services-based infrastructure.