Jordan Leahy
Back to Work

SmarterDx

Redesigned clinical diagnostics to boost accuracy 60%

RCMLLMICD-10SepsisClinical AI
SmarterDx
01
WHAT IS PREBILL IN RCM

What is PreBill in RCM?

1

CLINICAL DOCUMENT INTEGRITY

Pre-bill Design Focused

Once a patient has been discharged and the encounter has been coded.
Common sense ML is applied to flag new revenue & quality improvements.
2

REVENUE OPPORTUNITY?

Clinical UI Storytelling

Once a patient has been discharged and the encounter has been coded.
If a new Pdx opportunity exists with higher revenue, then an internal CDI team needs to verify.
3

SEND TO CLIENT

New or resequenced primary diagnosis?

Select evidence and send to client
If client approves, revenue is attributed to efforts
02
The Team
1

My Role

Senior Product Designer

  • Owned information architecture, visual design, and usability testing.
  • Defined the "explainable AI" layer for every recommendation.
2

Stakeholders

  • Clinical SMEs (nurses, coders)
  • Data science (model development and feature importance)
  • Product (RCM strategy)
  • Engineering (EHR integration)
3

Scope

2.5 months

  • Deployed to internal CDI team
  • Clinical validation studies
4

Tools

FigmaFigJamProduct DiscoveryJira
03
The Challenge

The Validation Gap

1

The Context

Each patient encounter explodes into over 30,000 data points, creating a massive validation burden.
Despite their expertise, manual review inevitably leads to missed diagnoses and lost revenue.
2

Clinical Impact

Incomplete documentation resulting in revenue leakage.
AI models that flag diagnoses without explaining 'why'.
Slow review cycles that delay billing and increase denials.
3

Design Challenges

Consolidating all data into a single trusted view.
Showing the 'why' behind AI suggestions.
Identifying workflows that speed up decision making without reducing accuracy.
04
The Process

From Research to Validation

1

Clinical Research & Discovery

Facilitated design thinking workshops to investigate why valid clinical opportunities were being rejected.
By combining these collaborative sessions with real-time shadowing, I uncovered the specific evidence gaps causing clients to disagree with our findings.
2

ANALYZED MVP & IDENTIFIED MODULAR PATTERNS

Moving beyond 'One Size Fits All'

Audited the existing "single view" MVP and identified that a one-size-fits-all approach failed for complex conditions.
Define a modular UI framework where clinicians see tailored components based on the diagnosis (e.g., Sepsis, Metabolic Encephalopathy, CHF) rather than a generic chart.
3

Prototype & Clinical Testing

Iterative testing with CDI staff using high-fidelity interactive prototypes.
Identified which features had the highest impact on improving revenue, allowing us to double down on what worked.
4

Implementation & Validation

Deployment and real-world performance monitoring.
Ensured pixel-perfect handoff, integrated clinical workflows, and monitored agreement rates via HEX data dashboards.
05
The Solution

Dynamic Clinical UI

1

Dynamic Clinical Layouts

Moved beyond a single generic view to a dynamic system where the interface changes structure based on the specific clinical condition.
Designed tailored workflows for Sepsis, Metabolic Encephalopathy, CHF, and STI to match their unique validation criteria.
2

Evidence-Driven Priority

The UI architecture is determined by the specific clinical data points that impact the DRG assignment.
It visually prioritizes the evidence needed to prove or disprove a Primary Diagnosis, shifting the focus from passive review to active validation.
3

Data Dense Visualization

Developed a rich, interactive visualization engine that synthesizes and plots key clinical diagnostic markers over time, allowing for rapid trend identification.
4

Clinical Decision Support

Integrated evidence-based treatment recommendations directly into the physician's workflow to reduce cognitive load and standardize care pathways.
5

Team Collaboration Tools

Enabled seamless care coordination with real-time patient status sharing and structured handoff communication tools between shifting care teams.
06
The Impact

Clinical Confidence

1

60% Accuracy Boost

Redesigned clinical diagnostics to boost accuracy significantly through thoughtful design.
2

Leading Clinical Feedback

"Having a data dense visualization of real data that makes sense makes the world of difference. Being able to see all the sepsis-related data and tell that story of, this is the primary diagnosis, makes the world of difference." — Stephen, Lead Clinical Document Specialist
07
Reflection & Learnings

The Retrospective

1

What Went Well

Early clinical stakeholder involvement led to accurate requirements.
Iterative design process prevented major usability issues.
Real-world testing validated design decisions.
Cross-functional collaboration improved solution quality.
2

Challenges

EMR integration complexity required creative solutions.
Managing clinical validation timelines and requirements involved constant negotiation.
Addressing varying workflows across different hospital units was critical.
3

Next Steps

Expand to additional infection prediction models.
Develop pediatric sepsis detection capabilities.
Add predictive antibiotic resistance modeling.