Tachyhealth is focusing on Artificial Intelligence solutions for the healthcare payers and providers. It’s pioneering digital health solutions, having real cases trials, then scaling them up when successful.

Role I played:

Evidence-Based Coding Logic
Challenge: Ensure codes comply with country-specific billing rules (e.g., CMS in the US, OPCS in the UK).

Solution:

• Rule-Based Post-Processing:
  • Validates AI predictions against official coding guidelines (e.g., ICD-10-CM Official Guidelines).
  • Flags conflicting codes (e.g., “Z79.4 (Insulin use) requires E11.65 (Type 2 Diabetes)”).
• Dynamic Code Updates:
  • Integrates with CMS/NHS APIs to auto-update coding rules annually.

System Integration & Deployment

A. Cloud-Based AI Inference

Objective: Provide real-time coding without requiring hospital IT changes.

Technical Stack:

• API-First Design: RESTful endpoints for EMR integration (FHIR/HL7 compatible).
• Scalable Backend:
  • Kubernetes for auto-scaling (handles 100K+ requests/day).
  • Redis for low-latency caching of frequent codes.
• Security:
  • HIPAA/GDPR Compliance: AES-256 encryption, zero data retention policy.
  • SOC 2 Type II Certified infrastructure.

Validation & Compliance

A. Accuracy Testing

• Benchmarked against human coders:
  • 98% Precision on common codes (e.g., hypertension, diabetes).
  • 92% Recall on complex cases (e.g., post-op complications).
• Continuous Monitoring:
  • Detects code drift (e.g., new COVID-19 variants) and retrains models.

B. Regulatory Approvals

• FDA Cleared (SaMD Class II) for automated coding.
• CE Marked for EU compliance.

Business Impact

80% faster coding (30 sec vs. 5 min manually).
15% increase in revenue capture (reduces under-coding).
Adopted by 200+ clinics across US/EU.

• Seamlessly integrate with any hospital’s workflow
• All while maintaining strict compliance with global healthcare regulations.

Core Technical Implementation

A. The AI Engine

• Developed a hybrid NLP system combining:
  • Fine-tuned BioClinicalBERT for code prediction
  • Rule-based validation against official coding guidelines
  • Continuous learning from user feedback

B. Cloud Infrastructure

• Built on AWS HIPAA-compliant architecture
• Implemented auto-scaling to handle peak loads
• Designed zero-data-persistence pipelines for security