OXP Live Modernization¶

Source: 2/18 Meeting, OXP DME Automated Order Process Workflow (internal spec)
Priority: Tier 3 — strategic but separate from patient-facing product
Status: Planned
Reference: repo_docs/warmfusion/plans/OXP_DME Automated Order Process Workflow.pdf

Job To Be Done¶

OrthoXpress needs their legacy Cold Fusion order system (OXP Live) rewritten in modern technology with AI-driven automation across the full DME order lifecycle.

Pain¶

OXP Live is a Cold Fusion application — aging technology with shrinking developer pool
Ravi (outsourced IT, 20 years with OrthoXpress) maintains it but doesn't re-architect
Every data entry point in the order process is currently manual
No semantic search for insurance clauses buried in patient notes
No AI-assisted product selection, eligibility verification, or authorization

Value¶

The automated order process spec describes AI augmentation at every stage of the DME lifecycle. This transforms OXP Live from a status-tracking system into an intelligent operations platform.

Current: Manual Order Pipeline¶

flowchart LR
    A[Fax/Email<br/>arrives] --> B[Manual data<br/>entry into<br/>OXPLive]
    B --> C[Manual<br/>benefits<br/>lookup]
    C --> D[Manual<br/>auth checklist<br/>+ submission]
    D --> E[Manual<br/>payment<br/>collection]
    E --> F[Manual<br/>delivery<br/>scheduling]
    F --> G[Paper<br/>forms at<br/>delivery]

Future: AI-Augmented Pipeline¶

flowchart LR
    A[Fax/Email<br/>arrives] --> B[AI OCR +<br/>auto-populate<br/>OXPLive]
    B --> C[RPA bot<br/>scrapes payer<br/>portals]
    C --> D[AI builds<br/>checklist +<br/>submits auth]
    D --> E[AI generates<br/>statement +<br/>payment link]
    E --> F[Smart<br/>scheduling<br/>AI]
    F --> G[Digital forms<br/>+ patient<br/>education]

Order Lifecycle (6 Phases)¶

The OXP DME Automated Order Process Workflow defines 6 phases, each with current manual data entry points and AI automation opportunities.

1. INTAKE¶

Current: Manual entry of patient demographics, insurance, doctor info, prescriptions into OXPLive from fax/email/portal.

AI automation opportunities:

Field	Current	AI Opportunity
Patient Demographics	Manual entry	OCR + NLP auto-extract from facesheet
Insurance Information	Manual entry	Document AI auto-read card, validate format
Doctor Information	Manual entry	Entity extraction: parse physician NPI, contact
Injury/Condition	Manual entry	Medical NLP: extract ICD-10 codes, diagnosis
Product Information	Manual lookup	Product Matching AI: map Rx to catalog SKU
Sizing/Measurements	Manual entry	OCR + Validation: extract measurements, apply sizing rules

AI-Assisted Product Selection (4-step pipeline):

Intent Identification — AI reads Rx, medical records, email to identify "Base Product" category
Candidate Filtering — System presents subset of products matching clinical criteria
Gap Analysis — AI checks "Measurement Required" flag; if true, requests Brace Measurement Form
Final SKU Resolution — AI applies "Sizing Rules" to select final billable SKU

Product Database Enhancement required: - Sizing Rules (measurement → SKU mapping) - Measurement Required Flag - Fitting Required Flag (based on HCPCS code + payer) - Clinical Indicators (ICD-10 codes for AI matching) - HCPCS Code Mapping with add-on code eligibility rules - Educational Content Links (video tutorials, PDFs, care instructions)

Decision Point Automation: - Small Ticket/Cash Pay Routing — AI analyzes product code + insurance type, checks historical auth requirements, auto-routes to Inventory (bypass Auth) or Authorization - Confidence score: High (>95%) = auto-route, Low = human review

2. AUTHORIZATION¶

Authorization Knowledge Base — requires a Payer-Product Requirements Matrix: - Payer + Plan Type (e.g., Medicare, Blue Cross PPO, Workers Comp) - Product/HCPCS Code - Required Documentation (Rx, Medical Records, CMN, ABN, Pre-Auth Form, Measurement Form) - Clinical Criteria (ICD-10 codes, medical necessity language) - Pre-Auth Required Flag - Timely Filing Window (days allowed after service date) - Coverage Limitations (quantity limits, rental vs. purchase rules, replacement timelines)

Dynamic Pricing Database (learned from EOB data): - Allowable Amount per Payer + Plan Type + HCPCS Code - Cash Pay Rate with tiered discount programs - Add-On Code Pricing (e.g., L2006, L2020, L2030) - Payer-specific bundling rules - Confidence Scoring (10+ EOBs = high confidence, <3 = low) - EOB Learning Engine auto-updates rates when 3+ consistent EOBs differ by >5%

Process Automation: 1. Benefits Verification AI — RPA bots log into payer portals (Availity, Navinet), scrape benefits, auto-populate OXPLive, flag discrepancies 2. Standardized Checklist Generation — AI queries Authorization Knowledge Base, generates dynamic checklist with status indicators (Met, Missing, Needs Review) 3. AI Documentation Validator — OCR + NLP reads uploaded docs, extracts key data, matches against checklist, auto-marks items as "Met" 4. Gap Analysis — identifies missing docs, generates specific requests, auto-emails Territory Rep with deadlines 5. Clinical Criteria Validation — compares patient diagnosis (ICD-10) against payer's covered indications, flags potential denials, suggests alternatives 6. Pre-Authorization Submission — AI verifies checklist 100% complete, generates "Pre-Auth Package" with AI-written cover letter and medical necessity justification

2a. PRE-FULFILLMENT PATIENT PAYMENT COLLECTION¶

After authorization, before fulfillment: - AI Statement Builder — pulls patient responsibility calculation, generates itemized statement (product, HCPCS code, allowable, insurance payment, patient responsibility breakdown) - Cash vs. Insurance Comparison — AI calculates both options when patient has high deductible, recommends best option - Patient Responsibility Calculator — Insurance: (Allowable x Coinsurance %) + Copay + Unmet Deductible; Cash: Cash Pay Rate - Applicable Discounts - Multi-Channel Delivery — email, SMS, patient portal with embedded payment link - Payment Gateway Integration (Stripe, Square, PayPal) — credit/debit, ACH, Apple Pay, Google Pay - Payment Plan Option — for balances >$200, AI offers installment plan - Fulfillment Hold Logic — configurable rule to hold fulfillment until payment received; AI flags high-risk accounts for mandatory pre-payment

3. INVENTORY (FULFILLMENT)¶

Three delivery methods, each with AI automation:

A. Pending (Field Rep Delivery): - Pre-delivery payment verification - Smart scheduling AI (analyzes rep calendar, patient availability, product urgency) - Digital patient agreement with e-signature - Point-of-delivery mobile payment terminal

B. Ship from Stock: - Inventory management AI (real-time stock levels, predictive restocking) - Shipping automation (optimal carrier selection, auto-generated labels, carrier tracking API) - AI uploads delivery confirmation photo to OXPLive

C. Ship from Vendor: - AI generates PO and sends to vendor via EDI/email - Tracks vendor order status, monitors delivery timeline - Auto-escalates delays to Territory Rep

3a. POST-DELIVERY FOLLOW-UP & FITTING¶

Automated Post-Delivery Contact Workflow: - AI creates follow-up task for Territory Rep (24-48 hours post-delivery) - CRM-integrated call logging with guided conversation script - Fitting requirement detection (queries product database for "Fitting Required" flag by HCPCS code) - AI scheduler for fitting appointments

Fitting Documentation: - Digital patient agreement with e-signature - AI-generated Plan of Care (pulls ICD-10, product details, physician orders) - Fitter Acknowledgment Form (mobile-friendly guided digital form with photo upload) - Billing Code Compliance Check (cross-references HCPCS code with payer-specific fitting documentation requirements)

Automated Patient Education: - Product-Specific Content Library (video tutorials, PDF guides, care instructions, troubleshooting tips by product SKU/category) - AI selects relevant educational materials based on product delivered - Personalized email/SMS with product care instructions and how-to links

Semantic Search Use Case¶

The highest-value specific capability discussed at the 2/18 Meeting. OrthoXpress staff need to search patient notes and payer documentation for insurance coverage rules. Current keyword search misses relevant clauses because the same concept is described differently across payers.

Embedding-based search (same nomic-embed-text infrastructure used by the KB system) would find semantically similar clauses regardless of exact wording. This applies directly to the Authorization Knowledge Base and Payer-Product Requirements Matrix described above.

Relationship to HJ Platform¶

OXP Live modernization is adjacent to but separate from the patient-facing HJ product. Connection points:

Equipment assignments in OXP Live → automatically populate HJ care plans
Patient equipment status in HJ → visible in OXP Live order tracking
Post-delivery patient education → feeds into the Equipment Chat Agent knowledge base
Fitting documentation → triggers care plan activation in HJ

The OXP order process spec's "Automated Patient Education" phase (Section 3a.6) is where OXP Live and the HJ chat agent converge — the education content served post-delivery is the same content the chat agent uses for ongoing troubleshooting.