Oncospark

Reducing Revenue Loss from Denied Claims
Summary
When Medicare denies a claim, it's usually because the documentation doesn't clearly justify why the procedure should be covered. With only 1 in 10 denials ever appealed, surgical clinics lose thousands in recoverable revenue. My team helped reimagine the appeal process to change that.
My Contribution
As the sole product designer on this contract project, I worked alongside a business strategist and 2 engineers to design an AI-assisted Prior Authorization Dashboard, from surfacing denied claims to auto-generating appeal letters, validated directly with clinic staff.
Timeline
From August 2025 to December 2025
Team
1 Product Designer
1 Business Strategist
2 Engineers
Tools
Figma Agent
Claude Code
Vercel
VSCode
Skills
User Research
Prototyping
Usability Testing
Full Stack
At a Glance

An AI-assisted prior authorization tool built to help surgical clinics recover revenue lost to denied claims.

Key Highlights

7/8 clinical professionals

Said they’d find the tool valuable at their clinic after testing the working end-to-end appeal prototype.

Working AI prototype

Built the end-to-end prior authorization workflow using Figma Make, Vercel, and Claude Code
Before: No centralized view of denied claims
Prior authorization requires clinical teams to prove that a treatment is medically necessary before insurers will cover it. When prior authorization requests are denied, clinical teams must determine why, gather supporting documentation, and appeal, often under tight deadlines.
After: Centralizing an AI-Enabled Prior Authorization Workflow
A centralized dashboard brings every case, denial reason, and deadline into one place, helping healthcare professionals quickly identify what needs attention and move directly from a denied claim into an AI-assisted appeal.
At a Glance

An AI-assisted prior authorization tool built to help surgical clinics recover revenue lost to denied claims.

Key Highlights

7/8 clinical professionals

Said they’d find the tool valuable at their clinic after testing the working end-to-end appeal prototype.

Working AI prototype

Built the end-to-end prior authorization workflow using Figma Make, Vercel, and Claude Code
Before: No centralized view of denied claims
Prior authorization requires clinical teams to prove that a treatment is medically necessary before insurers will cover it. When prior authorization requests are denied, clinical teams must determine why, gather supporting documentation, and appeal, often under tight deadlines.
After: Centralizing an AI-Enabled Prior Authorization Workflow
A centralized dashboard brings every case, denial reason, and deadline into one place, helping healthcare professionals quickly identify what needs attention and move directly from a denied claim into an AI-assisted appeal.
The Problem

Millions Lost to Preventable Denials

A Manual, multi-Step appeal process
Without centralized denial tracking, healthcare professionals spend 13+ hours each week identifying denied claims, decoding denial reasons, gathering documentation, and drafting appeals.
By the Numbers

4.1 million denials

Prior authorization requests denied by Medicare Advantage insurers in 2024.

13+ hours/week

Spent by physicians and staff completing prior authorization requirements.

Only 11.5% appealed

Yet 80.7% of appealed Medicare Advantage denials were overturned
User Research

We mapped the existing appeal workflow to understand where staff were losing the most time.

Our research showed that the biggest bottleneck came after a claim was denied. Staff had to manually interpret denial codes, gather supporting documentation, and build appeals from scratch, often without clear guidance on what would make an appeal successful.
Where the Appeal Process Breaks Down:
Documentation often doesn't clearly justify why a procedure should be covered

Staff manually decode CARC codes to figure out why a claim was denied

Building an appeal from scratch takes 40+ minutes for staff with no guidance

Only 1 in 9 denied requests are appealed despite 80.7% of appealed denials being overturned
Deep Dive

We mapped the current claim denial workflow to find where staff lose the most time.

Secondary market research showed tools like CoverMyMeds help large hospitals manage billing, but don't give smaller surgical clinics an integrated way to track denials, understand why claims are rejected, and file appeals in one place. No existing solution uses AI to interpret denial patterns or generate appeal responses.
Solution

Automated Claim Denial Workflow

An AI-powered prior authorization platform that turns denied claims into actionable appeals, helping clinical staff understand why a claim was denied, build an evidence-backed response, and manage the appeal from denial through submission.
Core Functionalities

Cited AI appeal generation

Pulls relevant clinical documentation to draft an evidence-backed appeal letter

End-to-end appeal tracking

Guides staff from denial through submission while predicting approval likelihood.
Key Design Decisions

The system in practice: from denial to submitted appeal.

Across the workflow, we grounded each design decision in three principles: prioritize what matters, automate repetitive work, and preserve human oversight where judgment matters.

Before: Ordered table

Every field is manual, from writing the appeal rationale to uploading required documents
No way to know if the appeal is complete or likely to succeed 
Staff has to log into payer portal to understand reason for denial

After: Prioritized Dashboard

AI autogenerates the appeal letter with inline citations and retrieves available documents from the EHR
Predicted approval outcome and completeness checks give staff confidence 
AI auto-generates the appeal letter with inline citations

Before: Blank appeal form

No way to see which claims are financially at risk or time-sensitive at a glance
Staff had to scroll through every row to spot denied or urgent cases
No visibility into appeal deadlines or how much revenue was on the line

After: Guided submission flow

Surfaces revenue at risk, denial counts, and expiring appeals as headline metrics
Countdown-based due dates make time-sensitive appeals impossible to miss
One-click filters like "Denied" and "Within 7 Days" replace manual scanning with triage

Before: Static checklist

Staff had to write the appeal rationale and gather documents from scratch, with no structure
No signal on whether an appeal was strong enough to succeed before submitting it
Every supporting document had to be tracked down manually from the EHR

After: AI-drafted appeal letter

A predicted approval outcome gives staff confidence before submitting
Required documents are auto-retrieved from the EHR and pre-checked off the list
A guided 3-step flow (Review Documents → Generate Appeal → Submit) replaces a blank form
Final Prototype

We mapped the current claim denial workflow to find where staff lose the most time.

Secondary market research showed tools like CoverMyMeds help large hospitals manage billing, but don't give smaller surgical clinics an integrated way to track denials, understand why claims are rejected, and file appeals in one place. No existing solution uses AI to interpret denial patterns or generate appeal responses.
The Updated Appeal Process
Staff begin by reviewing the denied claim and required documentation, with available records automatically pulled from the EHR. The system then drafts an appeal letter with inline citations for staff to review and edit before submission. Finally, staff can submit the appeal and track its status, with predicted approval and completeness checks surfaced throughout the process.
Prioritizing which case denials need attention first
Instead of scanning claims one by one, staff can see revenue at risk, approaching deadlines, and denial reasons in one place. Expanding a claim surfaces the context and required documents needed to decide what to appeal and begin the process directly from the dashboard.
Outcome

From a manual, hours-long process to an AI-assisted workflow adopted by clinical operations leadership.

The Prior Authorization Dashboard was approved for development by OnCoSpark's clinical operations leadership. In usability testing, 7 of 8 participants said they'd find the tool incredibly valuable if deployed, and the project was recognized with 2nd place at the 2025 Fung Engineering Health Innovation Competition.
Reflection

Denied claims aren't just paperwork, they're real revenue clinics depend on.

This project taught me how much of a healthcare workflow's complexity lives in the small, unglamorous details, a missing CARC code lookup, an unclear deadline, a document nobody flagged as required. Designing for staff working under time pressure meant every screen had to answer "what do I do next" without making them think. Translating that complexity into a system real clinics could trust and adopt required as much research and stakeholder alignment as it did craft.

Explore more of my work

Explore more of my work

Back

Overview

At a Glance

The Problem

User Insights

Deep Dive

Impact & Outcome

Reflection