Quantifying human performance for heterogeneous user populations using a structured expert elicitation

Heterogeneous product user populations are common across many safety-critical domains. Catering to variable user needs is critical for designing safe and effective systems. Despite this, it is common for a 1-size-fits-all approach to be applied in design for these populations. Quantifying risk of use error throughout the design process can justify design decisions that maximize system performance and safety. Many regulatory agencies require consideration of user variability in design validation activities. However, there are practical challenges for integrating variable users into these activities. Adequately representing populations requires significant time and monetary commitments for subject recruitment. In addition, population access may be difficult in some cases. In this work, an alternative to traditional human factors design validation efforts is presented. Expert elicitation is proposed as a cost-effective means to quantify heterogenous user performance in the formative product design stages. The approach relies on the generation of generic physical and cognitive tasks that can be applied across use cases. The approach is demonstrated on the diabetes population, specially focusing on medical device use. The output of the demonstration are performance distributions for 27 task-user group pairs that can be integrated into design validation efforts to identify human error risks that require mitigation.

Automated summary

Heterogeneous product user populations are common in safety-critical domains, but designing for variable user needs is often approached with a one-size-fits-all mindset. Quantifying the risk of use error can help justify design decisions for maximizing system performance and safety. Introducing expert elicitation as an alternative to traditional human factors design validation efforts, this study focuses on the diabetes population to demonstrate how performance distributions for task-user group pairs can help identify human error risks in design validation efforts.