Human-centered risk management for medical devices - new methods and tools.

Studies regarding adverse events with technical devices in the medical context showed, that in most of the cases non-usable interfaces are the cause for use deficiencies and therefore a potential harm for the patient and third parties. This is partially due to the lack of suitable methods for interlinking usability engineering and human-centered risk management. Especially regarding the early identification of human-induced errors and the systematic control of these failures, medical device manufacturers and in particular the developers have to be supported in order to guarantee reliable design and error-tolerant human-machine interfaces (HMI). In this context, we developed the HiFEM methodology and a corresponding software tool (mAIXuse) for model-based human risk analysis. Based on a two-fold approach, HiFEM provides a task-type-sensitive modeling structure with integrated temporal relations in order to represent and analyze the use process in a detailed way. The approach can be used from early developmental stages up to the validation process. Results of a comparative study with the HiFEM method and a classical process-failure mode and effect analysis (FMEA) depict, that the new modeling and analysis technique clearly outperforms the FMEA. Besides, we implemented a new method for systematic human risk control (mAIXcontrol). Accessing information from the method's knowledge base enables the operator to detect the most suitable countermeasures for a respective risk. Forty-one approved generic countermeasure principles have been indexed as a resulting combination of root causes and failures in a matrix. The methodology has been tested in comparison to a conventional approach as well. Evaluation of the matrix and the reassessment of the risk priority numbers by a blind expert demonstrate a substantial benefit of the new mAIXcontrol method.