Signed in as:
filler@godaddy.com
Building an AI-Driven Medical Coding Platform
My Role: Director of Product Management (Client-side lead)
My Contribution: I oversaw and led the entire product lifecycle, from strategic discovery to final development handoff as the founder of Sinjan Solutions
Duration: 5 years (2019 to 2023)
Platform: Web App
Context & Problem
Medical coding is a highly manual process where coders translate clinical documentation into standard codes (ICD, CPT, etc.) for billing and compliance.
This process is:
- Time-consuming (average claim takes 10–15 minutes to code manually)
- Prone to human error (leading to claim denials and revenue loss)
- Difficult to scale with growing healthcare data
Goal:
Design an autonomous medical coding platform that uses AI to predict, auto-code, and flag exceptions helping healthcare organizations reduce manual effort and increase claim accuracy.
Discovery & Research
Activities:
- Conducted stakeholder interviews with coders and operations heads to understand workflow pain points.
- Observed coding processes across multiple teams to map user journeys and bottlenecks.
- Analyzed analytics (claim turnaround time, error rates, denial causes).
Challenges Identified
- Coders needed an intuitive interface for error-free, fast data entry.
- Clients sought clear differentiation between AI and manual efforts, along with real-time tracking of progress and outcomes.
- The system had to ensure compliance with medical coding standards.
Key Findings:
- Coders prioritized speed, accuracy, and error reduction.
- Clients valued transparency, real-time updates, and detailed reports.
- Both groups demanded a seamless, intuitive user interface that reduced complexity.
Solution
The platform combined AI-driven automation with manual coding capabilities, addressing the identified challenges:
Features for Coders:
- A user-friendly interface for data entry, enabling coders to input ICD and CPT codes effortlessly.
- Validation features like NCCI and LCD to reduce errors and ensure compliance.
- Rules engine to manually create ever changing rules.
- Provider portal to ask questions to the providers.
Features for Clients:
- Dashboards displaying real-time progress of AI-handled and manually completed tasks.
- Detailed reports breaking down automation coverage and manual interventions.
Execution
Tools and Techniques
- Design: Figma, Axure RP
- Management: Jira, Confluence
- Methodology: Agile with iterative design and development cycles
Process
- Conceptualization: Created wireframes and prototypes based on user feedback.
- User Testing: Conducted usability testing with coders and stakeholders to refine the design.
- Implementation: Worked closely with developers to translate designs into a functional product.
- Iteration: Incorporated feedback to improve features and workflows post-launch.
Coder Work Queue
This is a screenshot of the coder work queue where they can view the encounters assigned to them. The encounters are sorted in descending order of date of service so that the older ones get processed first.
Manager Dashboard Wireframe
This wireframe displays the total work done by the coders in a day and the key metrics. It shows how much time on an average a coder takes to work an encounter and with their existing velocity, in how many days or weeks can they complete their backlog.