Mayo Clinic

My team and I at AKQA worked with Mayo Clinic to create their Digital Triaging Tool: a supportive, stress-free introduction to Mayo’s world-class care.

During our discovery phase, we talked with call center reps, patients, and business stakeholders to identify opportunities via prototypes and collaborative work sessions. 

It arose that patients typically encountered many steps, across multiple channels (web, email, calls), each punctuated by days of delay. 50% of prospective patients were not scheduled at Mayo clinic due to legacy system issues and challenges with human-based scheduling; this number did not even account for the portion of people who, after several days of back-and-forth, were declined for care. 

It was clear we needed to rigorously audit, and edit what information was needed, when, and to create solid foundations for a path to AI-assisted triaging. It was equally clear that any simplification could not come at the sacrifice of connection, compassion, or support.

Compassionate Connection 

Above all, we knew our users would be stretched and stressed. Conversational design was at the heart of the triage tool. Calm, compassionate language, guiding interactions, and a sensitive approach to accessibility were all implemented with the goal of mimicking the supportive and trusting experience of speaking with one of Mayo’s world-class specialists.

Authority through design

Mayo’s industry authority is unmatched, which fosters trust and optimism in their patients. We carried this through the design via elevated, assertive use of color, typographic sensitivity, and line work.

Confidence through clarity  

We wanted patients coming into the triage tool feeling like an expert. Pairing the ‘one-thing-a-page’ model with an intuitive vertical navigation kept users focused on the task at hand. By minimizing clutter, stripping unnecessary questions, and providing timely, prescient information inspired confidence through a sense of progress. 

Best-in-class Technical solutions

We turned to Optical Character Recognition to speed the insurance upload process up. 

Pivotally, we created bespoke Natural Language Processing AI models to fast-track patient-provider fit. We trained models to path a patient’s written description through one of 200+ diagnostic D-trees that mapped their issues to ~20,000 symptoms. Depending on the confidence of the match, users would be triaged to follow-up questions or directly to scheduling with the correct department.

Outcome

Creating a compassionate virtual exchange of information become ever more essential upon the launch of the triage tool, a few months into the pandemic. During a time of increased vulnerability and insecurity, Mayo could accelerate scheduling the right patient, with the right provider, at the right time, all in a welcoming, guiding manner.

Credits

Paul Sieka, Amy Teets, Brian Simpson, Eric Ku