Plain Language Synopsis: This research applies model based engineering techniques to develop novel verification and hazard analysis methods, which help manufacturers establish the quality and safety in medical device user interface designs. Artifacts produced by these methods provide evidence for regulators to quickly and objectively assess the safety of devices' interaction with users.
Abstract: Designs of medical device Human Computer Interfaces (HCI) need to be robust and appropriately reactive to user actions. There is evidence that the HCI design of some devices on the market can cause use errors and erroneously process user input, which may subsequently lead to serious patient harm. Model based engineering (MBE) technology can be used to model HCI design decisions with mathematical precision. This technology can facilitate the development of HCI models that clearly define the device's interaction behavior with users; offering a formal (mathematical) basis to reason about and verify the safety of the design. Automatic tool support is available to facilitate such reasoning and verification activities. Tool artifacts provide manufacturers and regulators an objective and scientific basis for assessing the safety of medical device user interfaces. The authors have successfully applied MBE techniques to the analysis of medical device user interfaces in two studies. In the first study, automatic model extraction was applied to the user interface software of a marketed infusion pump to produce a model that resembles the pump's use interaction behavior. Automated formal proving on the model uncovered several design flaws in the pump's user interface that could lead to severe consequences including the pump ignoring key presses that might cause patient overdose. In the second study, the authors captured the user interface software design common in medical devices with a generic user interface model. Based on this generic model, a hazard analysis technique was proposed that integrates human cognition process models and general interaction design principles to guide more comprehensive and systematic identification of design flaws in user interfaces. Preliminary experiments showed that this hazard analysis technique can identify 3 times more software-related hazards in user interface designs, compared to standard hazard analysis techniques.