Builder project · Built at Pulse Foundry AI NYC X ABI Frameworks
EezyClaim
Turns messy clinical wound notes into clear billing decisions
- ✧ Audience Award

Tip: click @Pulse AI in the post to tag us.
From idea to reality
What we built
## What We Built We built a wound-care billing triage pipeline that helps identify which patients are eligible for Medicare Part B wound care billing. The system ingests patient data, extracts wound-related information from clinical notes and assessments, and routes each patient into one of three outcomes: - Auto accept: all required information is present and the patient is clearly eligible - Flag for review: the data is incomplete or ambiguous - Reject: the information is insufficient or not appropriate for billing The pipeline is designed to be practical for a real-world billing workflow, especially for a non-technical user such as a medical biller. --- ## How It Works ### 1. Data ingestion The system pulls patient, diagnosis, coverage, note, and assessment data from the mock PCC API. Because the API is rate-limited, the pipeline includes retry logic and backoff handling so it can recover from temporary failures instead of breaking. ### 2. Wound extraction We extract clinically relevant wound details from both: - Structured assessments - Free-text progress notes This includes: - Wound type - Stage (when applicable) - Location - Measurements - Drainage amount The extraction logic is rule-based and deterministic for most cases, which makes it fast and explainable. ### 3. Diagnosis-to-note mapping We link notes to active wound diagnoses by comparing the note content with the diagnosis description and wound-related terms. This helps determine whether a note is describing an active wound and which diagnosis it most likely corresponds to. ### 4. Routing logic Each patient is evaluated against the business rules for wound care billing eligibility. If the required clinical fields are documented and the patient has active Medicare Part B coverage, the patient is routed for billing. If something is missing or unclear, the system flags the case for review. ### 5. Presentation layer The output is meant to be easy for a biller to read. Instead of showing raw clinical data only, the system presents a clear decision with supporting context so the user can quickly understand what action to take. --- ## Why It Matters This solution matters because it reduces the manual effort involved in identifying billable wound-care patients. Instead of a staff member manually reviewing large volumes of notes and assessments, the pipeline automates the first layer of triage and surfaces only the patients that need attention. It improves: - Efficiency: faster identification of eligible patients - Consistency: standardized decision logic - Scalability: the architecture can handle more patients and more data over time - Usability: a simpler experience for non-technical users In short, the project turns messy clinical data into a structured, actionable workflow that supports billing operations more efficiently and accurately.