A Fortune 50 healthcare organization was developing an AI-enabled electronic medical record designed to improve providers’ situational awareness, communication, collaboration, and response.
Iterate simulation testing of an AI-enabled EMR
Challenge
We were tasked with guiding the evolution of an AI-enabled electronic medical record (EMR). In addition to recruiting and delivering a large sample set of voices and utterances for training the Large Language Model, we were also asked to provide recommendations for incremental improvements to system usability and design a roadmap for both the EMR and the AI-agent.
Approach
Our 4-year program of research involved usability testing, iterative design research, and simulated-use environments. We designed and operated both low-fidelity (Wizard of Oz) and high-fidelity (scenario-based trauma center simulation) test environments over the course of 10+ projects carried out in both the United States and United Kingdom.
Outcome
In addition to suggesting incremental improvements, we offered guidance on aligning notifications with users’ information needs and communication patterns. We collaborated on multiple professional conference presentations and conducted a summative evaluation, which was submitted to the FDA for 510(k) pre-market approval.