Techathon 6.0

Business Problem 

Current Challenges:

• Provider directories contain 40-80% inaccurate contact information causing member frustration and access issues
• Manual verification processes requiring staff to call hundreds of providers monthly for basic updates
• Multiple data entry points creating inconsistencies between online directories, mobile apps, and printed materials
• Regulatory requirements demanding frequent provider data updates with limited automation capabilities
• Time-consuming credential verification processes delaying provider network additions by weeks or months
• Member complaints about outdated provider information leading to unsuccessful appointment attempts

Desired Outcomes

• Automate provider data validation through intelligent web scraping and API calls
• Reduce manual verification time through AI assistance
• Achieve target provider contact information accuracy through continuous automated validation
• Create unified provider data management reducing inconsistencies across member-facing platforms
• Demonstrate reduction in provider directory maintenance costs through intelligent automation

Goal

Develop a simplified Agentic AI system that automates basic provider data validation using publicly available sources, demonstrates intelligent data quality improvement, and showcases the potential for full-scale provider data management automation with synthetic and public data sources.

• Provider data includes
• Demographics: Name, contact information
• Professional Details: Specialties, licenses, certifications
• Network Affiliations: Insurance networks, affiliations with other providers or groups
• Services Offered: Clinical focus, appointment availability
• Location and Facilities: Addresses, medical imaging facilities

Key Deliverable (Demo)

Demonstration Scenario: Automated validation and updating of 200 provider profiles using publicly available data sources.

• Input: Sample provider dataset with names, addresses, phone numbers, specialties, and basic credential information. Must include scanned pdf (unstructured data) formats.
• Process: AI agent automatically validates contact information via web scraping, checks credentials against public databases, identifies inconsistencies, and flags providers needing manual review
• Output: User interface to show updated provider profiles with confidence scores, actions status reports, prioritized list of providers requiring human attention, and generate communication email.
• Timeline: Complete validation cycle in under 30 minutes versus traditional manual work

Agentic AI Roles (Suggested / Illustrative)

Data Validation Agent:

• Performs automated web scraping of provider practice websites to verify current contact information and services
• Cross-references provider information against public databases including NPI registry and state licensing boards
• Conducts intelligent phone number and address validation using publicly available verification services
• Generates confidence scores for each data element based on source reliability and cross-validation results

Information Enrichment Agent:

• Searches public sources for additional provider information including education, board certifications, and specialties
• Analyzes provider websites and online profiles for updated practice information and service offerings
• Identifies potential network gaps by analyzing geographic distribution and specialty coverage
• Creates standardized provider profiles with enriched data from multiple public sources

Quality Assurance Agent:

• Compares provider information across multiple sources to identify discrepancies and inconsistencies
• Flags providers with suspicious or potentially fraudulent information for manual review
• Tracks data quality metrics and generates reports on validation success rates and common error patterns
• Prioritizes providers for manual verification based on member impact and data confidence levels

Directory Management Agent:

• Generates updated provider directory entries in multiple formats (web, mobile app, PDF)
• Creates automated alerts for providers requiring immediate attention or manual verification
• Produces summary reports showing validation results, data quality improvements, and recommended actions
• Manages workflow queues for human reviewers with prioritized tasks and supporting documentation

Data and System Assumptions

Publicly Available Data Sources:

• NPI Registry (CMS): Free API access for provider basic information, credentials, and practice locations
• State Medical Board Websites: Public license verification and disciplinary action information (can be scraped)
• Hospital/Health System Websites: Provider directory pages with current practice information and contact details
• Google My Business/Maps API: Practice location verification, phone numbers, and patient review data
• Medicare Provider Utilization Database: Public claims data showing provider specialties and practice patterns

Synthetic Data Generation:

• Provider Profile Generator: Create realistic provider datasets with names, addresses, specialties, and credential information
• Validation Scenario Creator: Generate common data quality issues like outdated phone numbers, moved practices, and credential changes
• Member Impact Simulator: Create synthetic member complaint data related to provider directory accuracy
• Network Coverage Generator: Generate geographic and specialty distribution data for network adequacy analysis

Evaluation Criteria

Submission Format

• Demo: Live prototype or 3–4 minute video.
• Documentation (brief deck or PDF):
  • System architecture diagram.
  • Agent roles and decision logic (auto-resolve vs. handoff).
  • Data schema & API assumptions (source PDF,KB article JSON).
  • Compliance guardrails (PII, content moderation).

Tips for Participants

• Guardrails First: Enforce content moderation/PII redaction and grounded (SOP/KB) before enabling sensitive actions (rejection).
• Fast Wins: Target top 5-8 high-volume intents/properties for AHT reduction.
• Edge-Case Rigor: Demonstrate behavior for ambiguous/handwritten content, fuzzy match, missing data, and source failure scenarios.
• Modular Orchestration: Keep Worker Agents loosely coupled so new programs (clients) and tasks can be added safely.

Target KPIs (for pilot)

• Validation Accuracy: 80%+ success rate in identifying outdated provider contact information
• Processing Speed: Complete validation of 100 providers in under 5 minutes versus hours of manual effort
• Information extraction: Achieve 85%+ accuracy during information extraction from unstructured documents/scanned PDFs with 95% right confidence score   
• Processing Throughput: Handle 500+ provider validations per hour through automated pipeline

Example Flows to Implement (Pick 2–3)

Flow 1: Automated Provider Contact Information Validation

Trigger: Daily batch processing of 200 provider profiles for contact information accuracy

Process Steps:

• Data Validation Agent extracts provider practice information from synthetic dataset including names, addresses, and phone numbers
• Agent performs web scraping of provider practice websites and Google My Business listings to verify current contact information
• Cross-validation against NPI registry API to confirm provider identification and basic practice details
• Quality Assurance Agent compares information across sources and generates confidence scores for each data element
• Directory Management Agent creates validation report showing confirmed updates, discrepancies, and providers needing manual review
• Automated prioritization of providers for human verification based on member impact and data confidence levels

Flow 2: New Provider Credential Verification and Onboarding

Trigger: 25 new providers applying for network inclusion with basic credential documentation

Process Steps:

• Information Enrichment Agent extracts provider information from application forms and searches NPI registry for verification
• Automated lookup of provider licenses through state medical board websites and public credential databases
• Data Validation Agent performs background research on provider education, board certifications, and practice history
• Quality Assurance Agent cross-references information across multiple sources and identifies any red flags or inconsistencies
• Automated generation of provider profiles with enriched information and confidence ratings for credentialing decisions
• Directory Management Agent creates summary reports for credentialing committee with recommendations and supporting documentation
  

Flow 3: Provider Directory Quality Assessment and Improvement

Trigger: Weekly quality assessment of entire provider directory database (500 providers) for accuracy and completeness

Process Steps:

• Quality Assurance Agent analyzes all provider profiles to identify missing information, outdated data, and potential inconsistencies
• Data Validation Agent performs selective verification of providers identified as high-risk for accuracy issues
• Information Enrichment Agent attempts to fill data gaps through public source research and web scraping
• Automated generation of data quality metrics showing improvement trends and areas needing attention
• Directory Management Agent creates prioritized action lists for staff including specific providers and data elements requiring manual update
• Production of executive dashboard showing overall directory quality scores and improvement recommendations

Minimal Tech Stack (Reference/Illustrative)

• Programming Language: Python with libraries including BeautifulSoup, Requests, Pandas, and Scikit-learn for web scraping and data processing
• AI/ML Framework: OpenAI API (free tier) or Hugging Face Transformers for natural language processing and data matching
• Data extraction: VLM based data extraction for maximum performance
• Database: SQLite or PostgreSQL for local provider data storage and validation tracking
• Web Framework: Flask or FastAPI for creating simple dashboard and API endpoints
• NPI Registry API: Free CMS API for provider verification and basic information lookup
• Web Scraping Tools: Selenium or BeautifulSoup for automated provider website data extraction
• Google Maps API: Free tier for location verification and contact information validation
• State License APIs: Public APIs where available or web scraping for license verification
• Task Scheduling: Python Cron jobs or Celery for automated validation cycles
• Data Processing: Pandas and NumPy for data manipulation and quality analysis
• Confidence Scoring: Custom algorithms using data source reliability and cross-validation results
• Report Generation: Python libraries like ReportLab or Matplotlib for creating validation reports and dashboards