CSS Health Document AI
by CSS Versicherung · Primary insurer
Document understanding pipeline that reads incoming medical invoices and reports, extracts structured data, and validates them against policy coverage automatically.
2 free views left this month
Business context
- Purpose of use
Automate health invoice processing while protecting sensitive medical data.
- Value chain steps
- Claims managementPortfolio administration
- Insurance branches
- Supplementary health insuranceDaily sickness allowance insurance
- Application scope
- Internal only
- Business criticality
- Mission critical
- Country or market unit
- Switzerland
Value and impact
- Added value category
- Cost reductionEfficiency gainQuality improvement
- Added value estimation
Automates structured capture for the majority of incoming health invoices.
- Estimated time savings
Roughly 8 minutes saved per document
- Estimated weekly case volume
- 22,000
- Time to value (months)
- 10
- Competitive advantage gain
Domain-tuned extraction for Swiss medical tariff systems (TARMED, TARDOC).
- Quality gain
Fewer keying errors and faster reimbursement cycles.
Technical details
- AI category
- Computer visionNatural language processing
- AI model name
Claude 3.5 Sonnet
- Model version
2024-10
- AI platform provider
- AWS Bedrock
- Cloud provider
- Amazon Web Services
- Technology stack
- AWS BedrockTextractStep Functions
- Interfaces used
Claims intake queue
- Architectural insights
Event-driven extraction with confidence-based human review thresholds.
- Automation level
- 4/5
- Estimated IT difficulty
- 4/5
Compliance and governance
- EU AI Act risk category
- High risk
- EU AI Act role
- Deployer
- Relevant policies
- Swiss Federal Act on Data ProtectionGeneral Data Protection RegulationEU Artificial Intelligence Act
- Responsibilities
Health claims operations owns exception handling and audit sampling.
- Data used
Medical invoices, tariff catalogues, policy coverage rules.
Metadata
- Development type
- Scaled application
- Go-live date
- Nov 1, 2024
- License model
- Software as a service
- Original source
- https://www.css.ch/news
Community
Average rating
Comments
Log in to join the discussion.
No comments yet. Be the first to comment.
Related solutions
by Zurich Insurance Group
AI-driven claims triage and automated damage assessment for motor and property claims. The system classifies incoming claims, estimates severity, and routes complex cases to specialists while straight-through processing simple ones.
by Baloise
Anomaly detection platform that scores claims for fraud likelihood using network analysis and behavioural patterns, flagging suspicious cases for special investigation.
by Die Mobiliar
Computer vision model that estimates property and natural-hazard damage from customer-submitted photographs to accelerate settlement after weather events.