AI-Native EMR Onboarding Platform NADIA-ANE

Acryl officially launched NADIA-ANE, an AI-Native EMR (ANE) onboarding platform, on May 12, 2026.
The platform compresses the previously costly and time-consuming process of deploying medical AI—long hindered by hospital-specific database schema differences—into a short, productized execution.

Powered by Acryl's proprietary medical-specialized foundation model ALLM.H combined with a semantic mapping engine, NADIA-ANE simultaneously proves world No.1 (SOTA) performance and an unmatched 0.43% safety floor on the international benchmark EHRSQL 2024, establishing itself as a next-generation EMR construction solution.

1. Drastic Reduction in SI Development Cost and Time

• Automatically converts hospitals' existing, divergent DB schemas—Excel, DDL, OMOP CDM, and more—into a single standardized format, nadia-ane-schema-v1.
• Value: Transforms the most significant historical barrier to AI adoption—massive custom SI (System Integration) development costs and timelines—into an automated, productized pipeline, dramatically shortening initial AI EMR build time.

2. World No.1 SQL Generation Accuracy and Industry-Leading Safety

• Achieved State-of-the-Art (SOTA) on the international medical benchmark EHRSQL 2024
  — SQL generation accuracy of 89.08%, surpassing the previous record set by the LG AI Research·KAIST joint team.
• Value: Constrains the hallucination rate—the tendency to fabricate answers—to just 0.43%, 10–30× lower than the typical medical LLM range of 5–15%. This eliminates misdiagnosis risk and delivers the uncompromising reliability essential to clinical environments.

3. Meets In-Hospital Security Requirements with Self-Operating MLOps

• On-premise deployment by default, designed for air-gapped (network-segregated) hospital environments. Delivered through a unified web interface integrated with Acryl's AI operations platform, Jonathan.
• Value: Fundamentally eliminates the risk of leaking sensitive patient data, and through intuitive visualization tools, empowers the hospital's own IT organization to independently operate and manage its AI EMR environment.

4. Native Understanding of Clinical Language

• Combines Acryl's medical-specialized model ALLM.H with the semantic mapping engine NADIA-SL-Engine, deploying role-specialized multi-agents (question understanding · schema linkage · SQL generation).
• Value: Precisely connects the complex natural-language queries, homonyms, and specialized medical vocabulary clinicians use every day to actual database attributes—without ambiguity—delivering accurate insights instantly.

5. Global Validation References and Proven Market Scalability

• Confirmed a world record on MIT's ICU database MIMIC-IV, and currently onboarding a major Korean university hospital's Clinical Data Warehouse (CDW) (linked with Dr. Answer 3.0 and K-ARPA).
• Value: Built on technology validated at top-tier Korean tertiary hospitals, NADIA-ANE is immediately applicable to the global and U.S. small/mid-sized hospital markets, where budget and network-segregation constraints make Big Tech cloud solutions difficult to deploy.