Towards Safer Medical Device Software Systems: Industry-Wide Learning from Failures and the Use of Safety-Cases to Support Process Compliance

Abstract

Software safety is checked today in regulatory audits, which verify software development process compliance to regulatory requirements. Ensuring safety is a critical task in complex life-supporting systems and despite many existing ways of assuring it, unanticipated failure will always be possible. Checking process compliance to required standards ensures the quality of the processes by which software is developed but does not necessarily indicate the quality of the resultant software. Since medical device domain is facing an increasing amount of device recalls due to software failures, our goal is to explore the underlying reasons for this and suggest two improvements within this paper. First, we will introduce complicated and complex systems to illustrate why there will always be unforeseeable and unanticipated situations that could cause the failure of the entire system. We will then describe how medical device software systems are reviewed for compliance and safety today, highlighting the shortcomings in the current methods adopted in the medical device domain and suggest the use of systems thinking. We then propose two improvements to medical device software development where process compliance is supported by safety cases and industry-wide learning from experience.