Aidbox for Deep 6 AI Architecture

Deep 6 AI uses AI to quickly match patients to clinical trials by analyzing medical records, helping researchers find the right participants faster and speeding up medical breakthroughs.

The Challenge:
Deep 6 AI’s clinical trial recruitment platform needed to:

  • Cut data loading times
  • Scale to handle 1B+ FHIR resources
  • Validate data against custom IGs in real-time

The Solution: Migrating to Aidbox FHIR server on AWS EKS

Tech Stack:

  • Aidbox FHIR Server (replacing legacy implementation)
  • AWS EKS + RDS PostgreSQL (auto-scaling for 2,400 resources/sec!)
  • Automated IG Validation (90% fewer errors)
  • Grafana/Prometheus (real-time monitoring)

Results:

  • 50% faster data loads (1.2B resources in 6 days)
  • 90% cleaner data via synchronous validation
  • Multi-tenant ready SaaS architecture

Key Takeaways for Devs:
1️⃣ Validation matters: Custom FHIR IGs caught errors during ingestion—not downstream.
2️⃣ K8s + FHIR = scale: 48 Aidbox instances handled 1TB+ of clinical data.
3️⃣ Bulk FHIR > REST: Their bulk API approach crushed traditional throughput limits.

👉 Full Case Study: Migrating to Aidbox: How Deep 6 AI enhanced performance of its AI pipeline for clinical trial recruitment