New in HealthShare Health Connect 2026.1

InterSystems FHIR Server can now validate finer-grained OAuth scopes that include search parameters, in conformance with the SMART App Launch STU2Opens in a new tab implementation guide. This validation provides access control with constraints on hundred of parameters, such as Observation codes, Patient birthdates, or Medication status. For example, to scope a request for only the current patient’s lab results you may use the following:

patient/Observation.rs?category=http://terminology.hl7.org/CodeSystem/observation-category|laboratory

To scope a request chaining all Observations of Patients born in 1990, you may use the following:

system/Observation.rs?patient.birthdate=1990

Additionally, the FHIR Server now validates finer-grained scopes for all resources and parameters available in FHIR STU3, R4, and R5, as well as the following FHIR operations and interactions:

• Operations: Patient $everything, Encounter $everything, List $find and $update-functional, Observation $lastn

• Interactions: Search, Read, Create, Conditional Create, Update, Conditional Update, Delete, Patch, Conditional Patch

The FHIR Server validates scope syntax according to the SMART App Launch STU2 order of operationsOpens in a new tab.

Requests made to the FHIR Server can now have their results filtered based on the client’s authorized scopes. For more information about configuring filtered searches, see FHIR Server Authorization Settings.

For example, a client may have the following scope, which permits access only to the current patient’s laboratory observations:

patient/Observation.rs?category=http://terminology.hl7.org/CodeSystem/observation-category|laboratory

If the client sends a request for the patient’s entire compartment using the scope GET <base_url>/Patient/id/$everything, only Observations in the compartment with matching laboratory codes are returned.

In this release, FHIR package and endpoint management has been enhanced in several ways:

• FHIR Server creation is now significantly faster.

• Users can now remove packages from an existing FHIR Server.

• Package updates on a FHIR Server now use stronger parameter conflict resolution:

  • Search parameters in the base STU3, R4, and R5 specifications cannot be overridden by identically named parameters.

  • Search parameters with the same name that originate from different packages are assigned unambiguous names.

  • When conflicts occur between different versions of the same FHIR package, the later version takes priority for search indexing and updates to the CapabilityStatement.

• CapabilityStatements now automatically populate definition elements in conformance with the FHIR specificationOpens in a new tab.

A flexible extension to X12 SNIP-level validation is now available for configuration, allowing users to invoke custom, field-level business rules directly within the standard validation workflow. This enhancement enables organizations to enforce additional data quality and compliance requirements without introducing separate post-processing steps. By executing custom validation during SNIP processing, users can maintain performance while meeting trading partner–specific and organizational validation standards.

This release introduces the Mirth Connect Conversion Tool, which accelerates migration from Mirth Connect to Health Connect. The tool supports faster, more structured migration engagements by providing automated assessment and partial conversion of existing integrations. It helps users understand migration scope upfront, reduce manual effort, and shorten time to value during integration engine replacement.

The CDA Validation Service is now available as an add-on for Health Connect Cloud (versions 2026.1+). It provides cloud-based validation of CDA 2.0 and C-CDA documents to improve data quality and interoperability in data exchange. This service helps healthcare organizations proactively identify structural and content-level data gaps before documents are exchanged, reported, or consumed by downstream systems. By improving document quality and alignment with USCDI-based CDA standards, organizations can reduce claim denials, minimize operational rework, strengthen compliance efforts, and increase trust in shared clinical data. Fully integrated with Health Connect Cloud, the service expands data integrity capabilities while supporting broader governance and quality initiatives.

This release introduces IntegratedML Custom Models, which extend IntegratedML by allowing users to define, train, and run custom Python machine learning models directly from SQL. Custom models support advanced use cases not covered by IntegratedML’s automated modeling, such as bespoke feature engineering, domain-specific algorithms, or externally developed model logic, while preserving in-database execution and SQL-based workflows.

Custom models are managed using the same IntegratedML SQL interface (CREATE MODEL, TRAIN MODEL, and PREDICT()), with model logic implemented in Python and executed within the InterSystems IRIS runtime environment. This approach enables seamless integration of custom ML models into production applications without exporting data or building external scoring pipelines.

This release introduces the ACORN-1Opens in a new tab algorithm to accelerate queries that use the HNSW index when filters are applied. The algorithm pushes evaluation of filter predicates into the HNSW graph traversal logic, allowing the engine to skip nodes that do not satisfy the filter criteria and return relevant results more quickly. This optimization is completely transparent to users and is enabled by the deep integration of vector search within the InterSystems IRIS SQL engine. Depending on the specific query and dataset, this enhancement can improve performance by one or two orders of magnitude with minimal impact on recall or precision.

InterSystems has enhanced the storage of date and time data in cubes, addressing several issues encountered when querying cube data through the SQL projection of the fact and dimension tables (including from third-party tools) and improving query performance. You can now also define date dimensions based on source class fields that use the efficient Posix data type.

For more information about this change, refer to the Upgrade Impact ChecklistOpens in a new tab, as cubes must be recompiled after upgrading.

Table partitioning enables users to manage large tables more efficiently by splitting data across multiple databases based on a logical scheme. For example, you can partition a table by date ranges in a DateTime field, allowing older data to be moved to a database mounted on a lower-cost storage tier while keeping frequently accessed current data on premium storage. The data structure for partitioned tables also provides operational and performance benefits as tables grow very large (for example, beyond one billion rows).

As this is a broad new product capability, it will be delivered in phases. This release introduces core partitioning and bucketing capabilities, allowing you to create new partitioned tables, convert existing tables to partitioned tables, and move partitions to specific databases. The query optimizer takes advantage of these partitions, when appropriate, to improve query performance. Future releases will add functionality such as support for sharded tables and tables with columnar storage. See the documentation for more details.

This release introduces a new format for SQL query plans that uses plain text and indentation to represent nested modules, making plans easier to read. User-facing APIs, such as the EXPLAIN command and the $SYSTEM.SQL.Explain() API, return the new format by default. Lower-level and legacy APIs continue to use the previous pseudo-XML format to preserve compatibility with existing customer utilities. All API methods provide a parameter or qualifier flag that allows you to select either supported format.

The following demonstrates the new plan format:

   SELECT Name, Age, COUNT(*) FROM Sample.Person WHERE Name [ 'Michael' HAVING Age = MAX(Age)      
 
Warning:
    Complex condition [ on Name from table Sample.Person will not filter out any rows from index map Sample.Person.NameIDX.
 
    Cost: 1035181
 
    Module-FIRST:
    Call Module-C, which populates temp-file B.
      Module-C:
      Call Module-B, which populates bitmap temp-file A.
        Module-B:
        Read index map Sample.Person.NameIDX, looping on %SQLUPPER(Name) and ID.
        For each row:
            Test the [ condition on Name.
            Add ID bit to bitmap temp-file A.
            Accumulate the count(rows).
      Read bitmap temp-file A, looping on ID.
      For each row:
          Read master map Sample.Person.IDKEY, using the given idkey value.
          Add a row to temp-file B, subscripted by a counter,
              with node data of Age and Name.
          Accumulate the max(Age).
      Module-D:
      Read temp-file B, looping on a counter.
      For each row:
          Test the AND condition on Age.
          Output the row.

This release includes several low-level optimizations for processing large kill commands, reducing the number of physical reads and improving response times for these operations. These enhancements are particularly beneficial in ECP environments.

In addition, the write daemon no longer waits on operations that update IRISTEMP. This change helps prevent delays during write daemon checkpointing when IRISTEMP is heavily modified and resolves issues in which a kill of a large process-private global could appear to hang the system.

• When a specific host is selected, the Production Toolbar now displays actions specific to that host. A new setting also allows you to enable or disable connection lines to all hosts, or to show only the primary host connections when a host is selected.

• You can now change a host class using an action in the new toolbar.

• When different classes are used for hosts with the same name, the host name in the inspector panel (on the right) appears as a dropdown, allowing you to toggle between hosts with the same name. In the Production Configuration view, these hosts are grouped under the same name, with the number of same-name hosts displayed to the left of the host name.

• Global filtering in the left-side Collections panel now allows bulk filtering across hosts, productions, DTLs, and more.

• The File Path Directory selector is now enabled.

• You can now copy a host name as text using the copy icon next to the host name.

• Improved UI performance for larger productions, such as those with more than 2,000 hosts.

• User interface is localized to the supported languages.

• You can now collapse or expand production items (such as Rule Sets, Data Transformations, or Business Processes) in the Collections panel on the left.

• Introduced a new user experience for Message Viewer and Message Search.

• A history of recent searches is now available for easy access and reuse.

• When a historical search is selected to filter Message Viewer results, the Search History icon is highlighted to indicate that a filtered view is active.

• Expedite onboarding and project handovers by activating the new opt-in DTL Explainer feature. When a writable DTL is displayed in the DTL Editor, a Generate New button appears next to the Description. This button enables you to generate an AI Description for the displayed DTL using an LLM (OpenAI GPT-5.2). You can review the contents of the AI Description in a draft view. Afterward, you can insert or replace the description with the AI Description. Text within the description field can be edited as appropriate. This feature requires configuration with your OpenAI license to activate. Compatibility with other LLM Providers will be added in the future.

• New visual cues in the DTL Editor indicate whether a segment or field is required, optional, or deprecated.

• Hovering over a property in the DTL Editor now displays additional information about that property.

• Users can opt in to use the new BPL user experience.

• A new tree view visualizes the current context and allows navigation up and down the tree.

• The new BPL Editor user interface is now embedded in VS Code.

• The Rule Editor activity timeout has been increased to 30 minutes.

• Improved readability when editing target hosts in the Send action editor.