Device personalization for heterogeneous populations: leveraging physician expertise and national population data to identify medical device patient user groups

Interaction with patient-facing medical devices requires integrated performance of both physical and cognitive tasks that are highly dependent on many user characteristics. Quantifying product use variability early in the design process is useful for device designers seeking to personalize medical device features to maximize performance and improve health outcomes, but three barriers make this difficult. First, patient populations often differ from the general population and can be difficult to recruit and access for human performance evaluations. Second, users of patient-facing devices are highly heterogeneous, leading to highly varied and complicated use-cases. Finally, there are numerous unique tasks for each medical device that make it resource prohibitive to perform comprehensive device use evaluations, particularly in the early stages of design. To address these challenges, a method for modeling highly prevalent patient user sub-populations to be used in targeted device personalization is proposed. In this study, Internal Medicine domain-expert input is used to identify patient characteristics critical to the performance of generic physical and cognitive tasks required for commonly prescribed medical devices. Then, a novel approach to quantify those characteristics is proposed, utilizing variables from the US National Health and Nutrition Examination Survey (NHANES) dataset, demonstrating a means to characterize specific patient populations. The data are statistically clustered to identify meaningful, task-specific device user sub-populations. The approach is demonstrated on the diabetes population for tasks related to hand-held self-management glucometer device use. The cluster results are then discussed, including their practical application to design personalization.

Automated summary

Interaction with patient-facing medical devices involves complex physical and cognitive tasks that are highly dependent on user characteristics. Quantifying product use variability early in design is challenging due to patient population differences, user heterogeneity, and resource constraints, but modeling prevalent patient sub-populations using NHANES dataset can help personalize device features and improve health outcomes.