- Overview
- Medical Record Summarization
- About Medical Record Summarization
- A template's lifecycle
- Create a template
- Edit a template
- Deploy a template
- Send documents for processing
- Review a summary
- Medical Record Summarization (MRS) output data
- Manage deployments
- Customize application appearance
- Experimental features
- Template sections reference
- Clinical review reference
- System components
Healthcare Solutions user guide
After a summarization run completes and a reviewer approves or rejects a summary, Medical Record Summarization writes the results to a Data Fabric entity called MRS2_Output. Each record in this entity corresponds to one processed medical record and contains the generated summary, reviewer decisions, citations, and supporting metadata.
This page documents the MRS2_Output entity, used by current deployments. Deployments on an earlier version of Medical Record Summarization continue to write to the legacy MRS_Output entity, which is being deprecated. New integrations should target MRS2_Output.
The MRS2_Output entity
The MRS2_Output entity is automatically provisioned in your Data Fabric tenant when Medical Record Summarization is deployed. Each record is created when a summarization run completes for a document, and updated as the record moves through the clinical review workflow.
Output files
These fields contain the deliverables produced by the summarization process.
| Field | Type | Description |
|---|---|---|
SummaryFile | File | The AI-generated clinical summary produced by DeepRAG. This is the primary output of the summarization process. |
CitationsFile | File | A file containing citation mappings that link each finding in the summary to its source page in the original medical record document. |
OverrideFile | File | The reviewer-edited version of the summary. Populated only when a reviewer used the Override action on one or more sections before approving. If no override occurred, this field is empty. |
PacketFile | File | A bundled archive containing all output files for the summarization run. |
ExportFile | File | A JSON file containing the exported summary data. Populated when a reviewer exports the finalized summary. |
Status and workflow fields
These fields track where a record is in the summarization and review lifecycle.
| Field | Type | Description |
|---|---|---|
OutputStatus | Choice Set | The current processing or review status of the record. See OutputStatus values below. |
ReviewDurationMinutes | Number | The number of minutes elapsed between a reviewer opening the record and completing the review. |
CurrentAssigneeName | Text | Display name of the reviewer currently assigned to this record. |
CurrentAssigneeId | Text | User ID of the reviewer currently assigned to this record. |
StartedWorkingAt | Date-Time | When a reviewer opened the record and the review session began. |
ProcessedAt | Date-Time | When AI summarization processing completed for this record. |
CompletedAt | Date-Time | When the clinical review was completed (status transitioned to approved or rejected). |
CreatedDate | Date-Time | When this output record was first created. |
UpdatedDate | Date-Time | When this output record was last updated. |
OutputStatus values
| Value | Description |
|---|---|
pending | The document has been queued for processing but summarization has not started. |
processing | The summarization run is currently in progress. |
ready | Summarization is complete. The summary is available in the Clinical Review queue. |
completed | A reviewer has opened the record and the review is in progress. Despite its name, this is not a terminal state. |
approved | A reviewer has accepted the summary as accurate. Terminal state. |
rejected | A reviewer has determined the summary is inaccurate or unsuitable. Terminal state. |
faulted | The summarization run encountered an error. The record requires investigation or reprocessing. |
Source and identity fields
These fields identify the source document, the template used, and the run that produced the record.
| Field | Type | Description |
|---|---|---|
SourceFileName | Text | The filename of the source medical record document that was summarized. |
OutputId | Text | Unique identifier for this output record. |
TemplateId | Text | ID of the template used to generate this summary. |
TemplateName | Text | Name of the template used to generate this summary. |
TemplateVersionId | Text | ID of the specific template version active at the time of summarization. |
ManualBatchId | Text | Identifies the batch run that produced this record. Populated for manual upload runs; empty for scheduled deployment runs. |
PdfPageCount | Number | Page count of the source document. |
StorageBucketName | Text | Name of the Orchestrator storage bucket from which the source document was retrieved. |
IngestionFileId | Text | Links this output to the source ingestion file that was processed (MRS2IngestionFile.IngestionFileId). |
IngestionSourceId | Text | Identifies which ingestion source produced this output (MRS2IngestionSource.SourceId). |
ContextGroundingId | Text | Reference ID for the UiPath Context Grounding index entry for this output. Used to retrieve RAG-based citations. |
MRSPacketLocation | Text | Storage path of the packet file for this run. |
MaestroTraceId | Text | Trace ID used for LLMOps trace correlation. Set automatically by the backend after AI processing completes. |
Downstream automation patterns
After a reviewer completes their review, downstream processes typically need to retrieve the approved summary and deliver it to a system of record, handle rejected records by routing them for reprocessing, or archive the citation file for audit purposes.
The recommended approach is to configure a Data Fabric trigger on the MRS2_Output entity that fires when OutputStatus is updated to approved or rejected.
For general guidance on Data Fabric triggers, refer to About triggers in the Data Fabric documentation.