Safety-driven software design and engineering in medical cyber-physical-human systems

Abstract

This dissertation addresses the questions of how to ensure safety in medical Cyber-Physical-Human System (CPHS) development from various perspectives including (1) System Design for Networked Medical CPHS, (2) Safety-driven Requirement and Software Traceability, and (3) System and Human Behavior Modeling.

For (1) System Design for Networked Medical CPHS, we present two works including failsafe system design for networked medical systems and a communication middleware that provides protected communication. First, we present a system design framework to handle medical device interoperation hazards for networked systems. The hazards include temporal safety hazards, device interoperation hazards, and potential network failures when performing medical tasks. Second, we present a communication middleware that utilizes existing medical knowledge to provide safe communication among system components and prevent accidental or malicious modifications of the exchanged information. Moreover, the communication middleware unifies the communication mechanisms among Statechart models, software, and medical devices.

For (2) Safety-driven Requirement and Software Traceability, we present two works. The first work is a safety-driven requirement traceability framework to trace safety requirements, system design, and safety analysis. The safety-driven requirement traceability framework provides mechanisms and algorithms that ensure the safety analysis is always up-to-date when a requirement or a system design model changes. As such, the safety analysis can provide the proof of system safety while the system is evolving. In the second work, we design a requirement traceability framework for distributed systems based on information fow. The requirement traceability framework traces the exchanged information across software artifacts written in different forms such as Statechart or object-oriented programming languages. The framework also manages changes in the exchanged information and provides the impact analysis algorithms to detect impacted system components and requirements in the systems.

For (3) System and Human Behavior Modeling, we developed a set of common preventable medical errors severed to guide the design, model, and verification of a medical CPHS.