Towards a safety case for runtime risk and uncertainty management in safety-critical systems

Many safety-critical systems have a human-in-the-loop for some part of their operation, and rely on the higher cognitive abilities of the human operator for fault diagnosis and risk-management decision-making. Although these operators are often experts on the processes being controlled, they still sometimes misjudge situations or make poor decisions. There is thus potential for Safety Decision Support Systems (SDSS) to help operators, building on past successes with Clinical Decision Support Systems in the health care industry. Such SDSS could help operators more accurately assess the system's state along with any associated risk and uncertainty. However, such a system supporting a safety critical operation inevitably attracts its own safety assurance obligations. This paper will outline those challenges and suggest an initial safety case architecture for SDSS.