🔍 DX Navigator
A smarter way to search for diagnoses inside the EHR, so providers can land on the most specific, billable code without scrolling through a sea of irrelevant results.Problem
ICD-10 codes are the universal diagnosis labels used in healthcare; over 70,000 of them, each painfully specific. Pick the wrong one and you risk inaccurate records, denied claims and lost revenue.
For providers coding happens in the middle of patient visits, where every extra click adds stress, breaks focus and eats into already limited time.
Every provider knows the ICD-10 game: either you remember the exact code (unlikely) or scroll through pages of “unspecified” results, hoping you pick something close enough. That “close enough” often means compliance headaches, more denials and extra time fixing notes after hours. Our search wasn’t making it any easier.
How might we assist users in finding correct Dx codes so they get the highest reimbursement possible without over-coding?
How might we help users to find the most specific relevant codes in one search even if they don't remember/know the code number or description
Goal & Success Signals
Many new customers were coming from other EHRs with far better diagnosis search, tools that made it easy to code at the highest specificity. When they switched to us, they struggled. That meant more denials, lower reimbursement and some very loud escalations.
Our goal was simple: fix this fast. Success meant:
Matching (or beating) the specificity of competitor tools
Cutting the time to find the right code
Reducing denials and protecting reimbursement
Turning an escalation risk into a product win
We measured it through satisfaction scores, direct customer feedback and a visible drop in coding-related complaints after launch.
Role & Scope
I owned Dx Code Navigator design end-to-end, from mapping the problem with real customers, to testing multiple interaction models, to partnering with engineering through launch.
I planned and facilitated research sessions, synthesized findings into clear design principles, prototyped and iterated quickly, and worked side-by-side with engineering to get the details right.
The feature shipped as part of Elation Health’s premium EHR offering and is now live in production, helping providers code faster, more accurately, and with fewer headaches.
Research
Started by mapping what we knew, what we needed to learn and the levers we could pull. This clarified our value proposition, success metrics and constraints.
Next, I mapped the ideal provider journey, from opening the note to selecting the final code, with zero dead ends. This workflow became the backbone for testing and defined what “friction-free” should feel like.
We started with a simple but telling task:
“Find the ICD-10 code for Type 2 diabetes mellitus with peripheral angiopathy with gangrene (E11.52)”
We ran 30–60 minute moderated sessions with six providers from customer practices, plus an internal clinical leader. Each participant tried the task using our existing search and three new concepts.
Why this code? It’s long, specific, and mirrors the complexity that trips providers up in real life. It also requires modifiers (peripheral angiopathy, gangrene) that generic searches often bury.
What we learned from baseline testing:
Satisfaction with our current search was middling (⭐️3.78/5).
Every provider completed the task, but most needed multiple tries.
“Unspecified” codes were a tempting shortcut when the right result didn’t appear quickly.
We also looked at competitor patterns. eClinicalWorks, for example, uses IMO Smart Search, guiding providers through modifiers and surfacing RAF context along the way. This confirmed that guided narrowing and modifier prompts are table stakes for anyone who’s used modern EHR coding tools.
Exploration
I tested three concepts side-by-side. All aimed to get providers to the most specific code faster, but each took a different path.
Concept 1: Radio Filter Cards
A familiar eClinicalWorks-inspired approach. Search results were paired with quick-select filter cards (e.g., age, sex, modifiers).
Bet: Familiar mental model would make it easy to adopt.
Reality: 100% task completion, but many struggled to know which filters to hit first. Satisfaction: ⭐️ 3.75/5.
Concept 2 – Dual-Pane With Chips
Categories on the left, options on the right, with “chips” as building blocks. Typeahead narrowed results in both panels.
Bet: The side-by-side view plus live narrowing would feel faster and more in control.
Reality: Highest satisfaction at ⭐️3.91/5. Fewer difficulties and smiles when typing instantly reduced options.
Concept 3 – Scrollable Categories + Options
All categories in a scrollable left pane, options on the right. Type to narrow still available.
Bet: Continuous scroll would improve discoverability and keep clicks low.
Reality: ⭐️3.83/5 satisfaction, and the fewest reported difficulties overall, though some still preferred the chip-building clarity of Concept 2.
Across All Three Concepts:
Guided narrowing works, but flexibility matters ; providers want to bail out of the “steps” at any point and still find what they need.
Type-to-narrow is a clear win. Watching categories and options shrink in real time feels powerful.
“Billable” as a filter was more distracting than helpful, it led some to chase reimbursement over clinical accuracy.
Final Design
The shipped Dx Code Navigator pulled the best pieces from Concepts 2 and 3.
What I kept:
Dual-pane layout with scrollable categories for quick scanning and easy jumping between sections.
Type-to-narrow that filters both categories and options in real time.
Modifier prompts surfaced early, so providers could add specificity without digging.
“It is 100% better than what we have right now in Elation. It is faster.”
What I changed:
Removed “Billable” as a primary filter to keep the focus on clinical accuracy.
Defaulted Age and Sex filters to reduce irrelevant results.
Added RAF awareness in the results list without biasing the selection process.
“This is so useful for practices that don’t have a coder… it would mean more revenue for these practices because they would be submitting more accurate codes.”
Why it works:
Providers can take a guided path, or skip straight to typing, without losing context.
The consistent nav pattern (borrowed from other parts of Elation Note) makes it feel instantly familiar.
Every interaction reinforces speed without sacrificing specificity.
“I really really really enjoyed this concept ”
Impact
Provider satisfaction jumped from ⭐3.78 with the old search to ⭐3.91 in testing with the new design.
Support escalations tied to coding dropped after rollout, especially from new customers migrating from other EHRs. Providers called out that it’s no longer a mental drain, coding now feels like a quick, almost mindless step in their workflow.
On the business side, this solved a near-dealbreaker for a top-tier customer. Practices that had threatened to churn over coding struggles now had a tool that matched and in some cases beat the specificity of their previous EHR.
The feature shipped as part of Elation Health’s premium EHR offering and is now live in production, helping providers code faster, more accurately, and with fewer headaches.
Reflections
Even though this project started as a fire drill to prevent churn and resolve escalations, I made sure the solution wasn’t just a one-off fix. I pushed to test with providers across practice sizes, specialties and payment models (fee-for-service and value-based care) to ensure the design scaled beyond one account.
That broad validation paid off, the patterns work across workflows and specialties, making the DX Navigator a foundation we can build on.
It reminded me that moving fast doesn’t have to mean cutting corners. With the right focus, you can run the research, ship quickly and still end up with something that holds up long after launch.