Some images blurred for confidentiality reasons
|
Background
|
|
GE Healthcare is a global medical technology company with 55K employees, 4MM install base, $20B in revenue, serving mainly healthcare providers and institutions.
Since 2013, GEHC had been actively becoming a software engineering company, and was beginning to build AI & predictive analytics into some products. In 2019 an AI platform team was created with data science, software engineering, product management, and UX, to enable product teams and partner institutions to create inferencing models and predictive analytics for imaging. I stood up an additional UX team and scaled it along with the rest of the organization, growing to three designers and a researcher. |
|
|
Overview & Situation
|
|
Situation
My UX team's charter was to create the GE Healthcare AI platform, and support strategic AI solutions, such as Edison Smart Scheduler for administrators and Xray Critical Care Suite for healthcare providers. Problem However, AI experiences across GE products (ultrasound, Xray, MRI, radiology software, etc) were inconsistent in terms of editorial, notifications, and visual display of AI findings, resulting in unpredictable experiences for providers, and a lack of an AI "brand" for GE. My Role As Director of UX, I identified the need, drove a strategy shift with exec leadership, led workshops, & created first patterns. Duration 2 months to v1 patterns. 9 months to v2 patterns. |
AI Care Pathway app
(blurred for confidentiality)
|
|
Opportunities & Obstacles
|
|
I started by driving a shared understanding with AI platform & product team leadership, that a unified UX for AI across products would create more value for GE and our users.
The platform execs initially said design was up to the implementing product teams: the platform was just a service. However, after meeting with with the leadership of the 23 product teams creating AI features, I had broad commitment to adopt any standards. I used this product team commitment to gain agreement from the platform team execs on the value of UX alignment, and agreement that the platform organization had the cross-portfolio visibility that could drive governance. |
Edison AI vision
AI Platform delivery roadmap
|
|
Actions: Discovery research
|
|
But in order to be effective in aligning teams and creating interaction standards, we needed to know how healthcare providers perceived the value of AI and how they wanted to engage with it, or were currently using.
My team led the first AI research at GE, with global healthcare providers in India, Europe, and the US, across multiple healthcare institutions. We learned that receptivity to AI and notification preferences varied by region, hospital policy, and workload of the radiologists. We distilled these and other findings into a simplified core workflow and a set of archetypes. Based on this research we also created a draft set of principles, as well as design & editorial guidelines. |
Imaging AI user simplified journey
|
|
Actions: Stakeholder Alignment & Scope
|
|
I led a 2-day workshop with worldwide Product, platform, and UX leaders, sharing our research, aligning on principles, and patterns that created consistent AI editorial, notifications, iconography, and visual representation of findings.
We built a unified roadmap of AI features across these product teams, prioritized next steps for design and research, then aligned on dates and expectations. We published the first version of the AI standards on eds.gehealthcare.com, then obtained broader feedback, and began executing against the agreed design and research plan. |
AI visualization guidance - v1 document
|
Draw Toast kickoff: www.drawtoast.com
|
Global workshop via Miro & video
|
Pattern ideas & needs from product groups
|
Aligned Product roadmaps & UX Research & Design next steps
|
|
Actions: Refine & Expand Guidance
|
|
Based on our shared roadmap, My UX team and I led multiple design and research activities: a company-wide AI icon challenge, AI visual design working group, and deeper research with radiologists.
Some of our challenges were showing AI results without obscuring the diagnostic image. Thinking globally to represent AI as an icon without using the letters A & I. Mapping the severity of AI findings to notification preferences and hospital policies. Based on these workshops, design ideation, and deeper research with radiologists, we broadened and refined our visual standards, notification guidance, and editorial approach. |
AI visual pattern in GE Universal Viewer
|
Evaluation of visualization approaches
|
Icon design challenge: representing AI without text
|
|
Outcomes: Partnership, Launch, & Awards
|
|
|
By the end of the project the platform product team partnered with UX on governance of the AI portfolio, making design implementation part of strategy.
All 23 teams building AI features agreed to adopt the standards, and our flagship products Universal Viewer and Xray Critical Care Suite launched with our interaction and visual AI standards. We also patented three UX patterns, published the patterns and guidance in the Edison Design System, and won a DMI Design Value Award for our AI design process in 2020. 
Edison Design System homepage
|
DMI Value Award submission
|