John Snow Labs 62d ago

Healthcare data is not analysis-ready — and most organisations rebuild the same pipelines from scratch. This session explains how the Tuva Project turns raw claims and EHR data into standardised, analytics-ready datasets using an open-source data model. Aaron Neiderhiser (CEO, Tuva Project) breaks down what it actually takes to build a scalable healthcare data platform — from fragmented data sources to consistent, reusable analytics logic. Most healthcare data lacks the concepts needed to answer real questions. There is no “readmission rate” or “cost of diabetes” in raw data — these must be defined, standardised, and implemented in code. Tuva solves this by creating a shared, open-source layer of healthcare concepts — covering encounters, conditions, episodes of care, and cost models — enabling organisations to analyse data consistently across systems. 📌 Applied Healthcare AI Summit 2026 — real-world AI and data systems in healthcare, from pilots to production. ⏱️ Key moments: 00:00 Why healthcare data is not analytics-ready 02:35 The core problem: missing concepts in raw data 08:23 From raw claims & EHR to analytics insights 09:50 Building a common data model across systems 14:47 Tuva architecture: concepts, pipelines, and dbt 18:04 From data transformation to real analytics use cases #HealthcareAnalytics #HealthcareData #OpenSourceAI #EHR #DataPlatform #DBT #ClinicalData #ClaimsData #DataEngineering #HealthcareAI #RealWorldEvidence