Remodeling Healthcare Referrals with Fivetran, Agentic AI, and Databricks Genie

Each well being system stands to profit immensely from AI: from enhancing affected person outcomes and lowering clinician burnout to optimizing the income cycle. Regardless of an inflow of expertise, hospitals and well being methods proceed to wrestle to comprehend this worth.

For healthcare supply organizations, the core downside is a extremely fragmented information and AI property. Affected person information within the digital well being report (EHR), medical imaging in PACS methods, streaming telemetry from bedside screens, and unstructured medical notes are all siloed. This fragmentation results in disjointed safety insurance policies, the lock-in of crucial operational and medical information, and AI sprawl with out correct enterprise semantics.

Moreover, constructing and productionizing HIPAA-compliant AI brokers is extremely tough, particularly when making an attempt to steadiness computational price versus medical high quality and security.

Maximizing AI Potential Begins with Your Knowledge

Databricks unifies and democratizes your whole information and AI, designed to deal with the complexity of healthcare environments safely and effectively. However to construct a single supply of fact, you first must unlock your information from legacy silos. Fivetran will help. 

Historically, extracting information from complicated EHRs meant engineers spending weeks constructing brittle, customized ETL scripts. Right now, Fivetran’s Epic Readability connector accelerates this course of by offering a totally managed, zero-code pipeline. It mechanically and securely extracts wealthy medical and operational information from Epic Readability and syncs it instantly into Databricks.  

By automating change information seize (CDC) and schema mapping, Fivetran eliminates guide extraction efforts, permitting well being methods to movement crucial information into Databricks in minutes moderately than months.

With Fivetran feeding the platform, the Databricks technique is constructed on two foundational steps:

• Develop a typical information basis: Unify your hospital’s information structure with the lakehouse structure for massive-scale information warehousing, Lakeflow for ingest and streaming (excellent for real-time HL7/FHIR feeds and IoT bedside screens), and Lakebase for serverless Postgres. This unified basis delivers the bottom Whole Value of Possession (TCO) and breaks the cycle of vendor lock-in typical of legacy healthcare IT.
• Unify your whole governance: Leverage Unity Catalog to supply unified, strict governance for all belongings. In healthcare, that is crucial for PHI and HIPAA compliance. Unity Catalog governs not simply conventional tables, but in addition predictive fashions, AI brokers, PDFs (like scanned referral paperwork), and uncooked recordsdata.

Optimizing Referral Administration: Begin with Your Enterprise Context

As soon as a ruled information basis is established, well being methods can ship their distinctive medical and operational context to AI. By making use of AI that really understands your particular hospital protocols, affected person populations, and payer contracts, Databricks permits transformative use circumstances for healthcare supply.

For hospital management, sufferers receiving well timed specialty care whereas retaining them inside the well being system’s community is a prime precedence. Optimizing referral administration to cease community leakage is an enormous problem. As much as 50% of referrals fall right into a “black gap” as a result of the info is fragmented throughout paper faxes, Epic Readability databases, and exterior payer methods. 

They merely lack visibility into whether or not a affected person really scheduled or attended their follow-up.

By tapping into the complete potential of Fivetran-ingested information inside Databricks, leaders can unlock immense enterprise worth. AI fashions can:

• Predict referral no-shows
• Establish root causes of out-of-network leakage, and
• Robotically suggest the subsequent finest in-network specialist based mostly on availability and proximity.  

As soon as information has been unified underneath a typical platform, your information workforce might want to mannequin that information and put together it for reporting and activation. Fivetran affords a strong toolset for remodeling the info, together with in-platform Quickstart information fashions for frequent SaaS platforms, integrations with dbt and Coalesce, and internet new capabilities like:

• AI Column Builder to counterpoint your information with further fields, powered by your AI supplier of selection like ChatGPT, Claude, and Gemini
• Viewers Builder to construct customized viewers segments

Then, when that information is prepared, you possibly can push it into the functions the place your workforce works, to assist them use information to drive choices of their day by day platforms. 

Knowledge activation ensures that information shouldn’t be siloed in dashboards, static experiences, or information warehouses. Healthcare directors and medical professionals must entry delicate, well timed information with a minimal of friction or delay. It isn’t sensible for medical doctors and nurses — or their workers — to study SQL and spend hours trawling by means of enterprise intelligence platforms when their consideration is urgently wanted elsewhere.

Fivetran Activations helps the end-to-end administration of referral lifecycles by integrating information from orders, scheduling, and accomplished visits. With that information, you possibly can mannequin affected person journeys throughout suppliers, flows, capability and leakage throughout supplier networks, and standardize referral metrics.

Crucial insights from these fashions can then be despatched to operational methods of every kind, whether or not to tell finish customers or set off automated actions.

Constructing Manufacturing AI Brokers and Democratizing Insights for Care Groups

With Fivetran and Databricks, we’re making it easy to empower well being methods to construct and deploy safe, clinical-grade information and AI apps.

By interacting together with your information in pure language with AI/BI Genie, hospital directors, nursing managers, and medical administrators can actually speak to their information. They’ll discover questions on affected person movement, readmission charges, or ED wait instances and obtain immediate insights through pure language, information tables, and visualizations.

Well being methods can deploy manufacturing AI brokers that obtain increased high quality at a decrease price. For instance, just like how AstraZeneca utilized Agent Bricks to parse by means of over 400,000 complicated medical trial paperwork, a hospital can deploy brokers to extract structured information factors from many years of historic unstructured medical information—with out requiring their IT groups to write down complicated code.

By accelerating information ingestion with Fivetran and unifying analytics and AI on Databricks, well being methods can transfer from fragmented IT to a very clever healthcare supply community.

Be part of us at HIMSS!

In case you’re headed to HIMSS26, try our session on the Databricks sales space (#5453), Wednesday, March 12:00 – 12:00 PM PT.  And if we don’t get an opportunity to see you in Vegas, be a part of us at our upcoming webinar, Wednesday April 15 at 9am PDT (hyperlink).