Bayesian Risk Identification Model (BRIM): A Predictive Model to Reduce Use Error Risk in Medical Device Interface Design

摘要

Increasing pressure from both regulatory agencies and the consumer market has expanded theneed for medical use error reduction. BRIM integrates user performance with risk managementto quantifiably predict human factors issues and illuminate design mitigation strategies duringdevelopment of medical devices. Upfront analytical modeling permits a significant reduction inrequired expertise and application of empirical methodologies. BRIM asserts that a common setof performance influencing conditions (PICs) determine how a user will interact with a medicaldevice and that a unique set of resulting human response failures (HRFs) manifest differentlydepending on the specific product interface design. Probability of HRF occurrence can bederived via a Bayesian Belief Network representation of PICs. By understanding the root causesof why a combination of interface, environment, or contextual influences lead to human error, wecan predict how a product will perform with respect to human interaction. And, by testingBRIM’s targeted set of design characteristics across human performance metrics, we can specifythis use error likelihood per product interface.