Digital and AI transformation architecture
Healthcare data solutions in Microsoft Fabric bring structured, unstructured, clinical, imaging, and medical device data into the Fabric data lake. The system brings data into the data lake with open data standards by using Fast Healthcare Interoperability Resources (FHIR), Digital Imaging and Communications in Medicine (DICOM), and Observational Medical Outcomes Partnership (OMOP) data models, providing customers with one common architecture. This architecture also helps organizations connect to low-code/no-code tools from Microsoft.
Microsoft Fabric represents an end-to-end, integrated analytics platform that consolidates a comprehensive suite of data and analytics tools that are essential for organizations to harness the full potential of their data. It also establishes the groundwork for AI. Microsoft Fabric provides an all-in-one solution for data integration, data engineering, real-time analytics, data science, and business intelligence needs without compromising the privacy and security of your data. For more information, see Microsoft Fabric.
The following diagram illustrates the process of how data is ingested, stored, analyzed, and visualized in Fabric and the data platform for healthcare. The system ingests data through connectors, such as Microsoft Azure Data Factory and Microsoft Azure Health Data Services (FHIR, DICOM, MedTech services) that are integrated with OneLake in Fabric. Then, the system transforms and normalizes the data in OneLake into more structured and standard data model information.
The following table outlines important terminology in the transformation architecture that you should be familiar with.
| Component name | Description |
|---|---|
| Interoperability for health | Interoperability for health includes specific systems with the reuseable Microsoft and partner components. These systems are Azure Data Factory, Eventstream, Microsoft adaptors, connectors, and mapping and partner interoperability solutions. These components integrate unstructured data from various sources into a data lake. |
| Data Lake | Data is stored in raw form and transformed to industry common models by using Microsoft, partner, and custom transformation pipelines. |
| Data Governance | Microsoft Purview provides data governance across the entire data estate and helps govern, protect, and manage organization data. |
| Health Data Services | Azure Health Data Services, such as FHIR, DICOM, and MedTech services, are built on the data lake to maintain structured information. |
| Health AI Services | Transformed data is infused with AI services to enrich the data and provide a layer of intelligence interaction with consumers. Some examples of AI services are text analytics for health, OpenAI, and terminology services. |
| Microsoft Power Platform | Applications can access and update the information by using a low-code/no-code platform. |
| Microsoft Fabric | You can manage your data in one place with a suite of analytic experiences that work together, including: Data Factory, Synapse Data Engineering, Synapse Data Warehouse, Synapse Data Science, Synapse Real-Time Analytics, Microsoft Power BI, and Data Activator. |
| Healthcare data solutions in Microsoft Fabric | Allows you to ingest, transform, and manage your data in OneLake. The capabilities of this solution include: FHIR data ingestion, DICOM data ingestion, Healthcare data foundations, unstructured clinical notes enrichment, OMOP analytics, and Microsoft Dynamics 365 Customer Insights - Data preparation. |
| Microsoft Fabric + Azure Databricks | Azure Databricks can directly access OneLake through the same Azure Data Lake Storage APIs and SDKs. You can virtualize the customer's existing Data Lake Storage that's built through Databricks into OneLake by using shortcuts. You can have the system auto discover Delta Lake data that directly lands in OneLake, or the data that you virtualize through shortcuts, as tables. |
Healthcare data solutions in Microsoft Fabric have a flexible architecture. You can deploy only parts of the overall architecture that meet your requirements. As you turn on different capabilities of this solution, the following types of components become available in your workspace:
Lakehouses - Multiple lakehouses are available when you deploy the healthcare data foundations capability. The system uses these lakehouses to transform the raw, unstructured healthcare data into data that's structured for analysis.
Notebooks - Each capability that you turn on adds one or more notebooks to your workspace. The prebuilt notebooks contain the transformation logic that you run to transform your data. You can run these notebooks individually; however, you can best orchestrate them by using the data pipelines. These pipelines are available with the capabilities that you deploy. We recommend that you don't modify the prebuilt notebooks; however, you can create your own notebooks to contain your own logic and queries that you use in the workspace.
Data pipelines - Some capabilities that you deploy provide prebuilt data pipelines that run the logic in the prebuilt notebooks to transform your data. The prebuilt pipeline templates showcase how the system can construct the transformation activities into an end-to-end flow. You must not modify the pipeline templates because they're ready to run and process the sample data. To fully automate your processing, you can create your own pipelines as needed to run the prebuilt notebooks alongside any of your own notebooks.