共用方式為

Set up healthcare data explorer (preview) in Microsoft Fabric

  • 發行項

[This article is prerelease documentation and is subject to change.]

Before you can use healthcare data explorer (preview) for data analysis, you need to set up the Microsoft Fabric tenant for the service.

Prerequisites

  • Contact the healthcare data explorer team at healthdata@microsoft.com to get an allowlisted tenant ID.

  • Make sure you're assigned to the Admin security role.

  • Get your data into the Fabric Lakehouse. The data format and organization requirements are as follows:

    • The data can have both structured (tabular) and unstructured parts. While the structured part is mandatory, the unstructured part isn't.

    • The schema for the structured (tabular) part of your data must align with Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) version 5.4, and it must be in the delta-parquet format. Healthcare data explorer (preview) requires data to be in the OMOP format. If your data isn't currently in this format, you can use the Healthcare data solutions experience to transform your data into the OMOP format. For more information, see Healthcare data solutions in Microsoft Fabric (preview).

    • For tabular data, healthcare data explorer (preview) currently considers only six of the more than 30 tables defined in OMOP CDM 5.4. The supported tables are person, observation, procedure_occurrence, visit_occurrence, condition_occurrence, and drug_exposure.

    • The unstructured part of the data must be at the root of the unstructured Files section of the lakehouse. For example:

      Patients Files for patients
      Patient-id-1 - .dcm (DICOM) files
      Patient-id-2 - .json/.txt clinical notes
      ... where patient IDs match the person_id column in the OMOP CDM's person table.

Enable healthcare data explorer (preview) for your organization

  1. On the Fabric navigation bar, select the Settings gear icon.

  2. Under Governance and insights, select Admin portal.

    A screenshot showing the gear icon and admin portal access from the Microsoft Fabric home page.

  3. On the Admin portal, select Tenant settings.

  4. Expand the Users can create Healthcare Data Explorer items (preview) item.

  5. Review and accept the Preview Terms, and slide the toggle to Enabled.

  6. In the Apply to section, you can either keep the default selection of The entire organization, or you can provide this permission selectively to specific security groups.

  7. Select Apply.

    A screenshot showing how to enable the capability and select a security group.

Enable Copilot and Azure OpenAI Service (preview)

  1. On the tenant settings page, scroll down to enable Copilot and Azure OpenAI Service (preview). Enable Users can use a preview of Copilot and other features powered by Azure OpenAI.

  2. Enable Data sent to Azure OpenAI can be processed outside your tenant’s geographic region, compliance boundary, or national cloud instance.

  3. Under Apply to, select The entire organization.

  4. Then, select Apply.

    A screenshot showing how to configure Azure OpenAI settings.

Create a workspace

A workspace acts as the foundational environment for hosting lakehouse datasets and healthcare data explorer (preview) instances. You can use the data in the lakehouse to look at large amounts of patient information to spot patterns, decide on the best treatments, and ensure delivery of effective care.

  1. On the Fabric home page, select Workspaces on the left navigation pane.

  2. Select + New workspace.

  3. Complete the fields in the Create a workspace form.

    Under Advanced, select Premium capacity or Fabric capacity. You need Premium capacity or Fabric capacity (P1 or F64 or higher capacity SKU) to run Azure OpenAI capabilities in healthcare data explorer (preview).

  4. Select Apply.

    A screenshot showing the settings required for a workspace.

Create a healthcare data explorer (preview) item

  1. On the Industry Solutions Fabric home page, select the Healthcare Data Explorer (Preview) tile.

    A screenshot displaying the healthcare data explorer (preview) tile.

  2. Name the healthcare data explorer (preview) item.

  3. Select Create.

Connect a lakehouse

To complete the healthcare data explorer (preview) set-up process, you need to connect the underlying data that you want to explore and analyze. The dataset must be in a lakehouse in the same workspace as the healthcare data explorer (preview) item. The data within the lakehouse must conform to the OMOP data standard. For more information, see Prerequisites.

  1. On the new healthcare data explorer (preview) item home page, select Connect a lakehouse.

  2. Select the lakehouse that you wish to connect, and then select Connect.

    A screenshot showing how to connect to a lakehouse.

Next steps

Build patient cohorts with generative AI in healthcare data explorer (preview)