- Overview
- Medical Record Summarization
- About Medical Record Summarization
- Roles and permissions
- A template's lifecycle
- Create a template
- Edit a template
- About policies
- Create a policy
- Edit a policy
- Refining decision model rules
- Deploy a template
- Send documents for processing
- Review a summary
- Medical Record Summarization (MRS) output data
- Manage deployments
- Manage review workflows
- Customize application appearance
- Experimental features
- Template sections reference
- Policies reference
- Clinical review reference
- System components
Create a policy in Medical Record Summarization by uploading a policy document and assigning it a decision profile that defines the available outcomes.
Create a policy by uploading a policy document and assigning it a decision profile. When you create the policy, the AI reads the document and builds a decision model from it. The run takes a few minutes, after which the generated rule tree is available for review.
Prerequisites
- You are a member of the MRSAdmin or MRSProcessOwner local group in Automation Cloud. See Roles and permissions.
- Your policy document is a single PDF file.
Name the policy and upload the document
-
Select Policies in the left navigation sidebar.
-
Select Add new policy.
-
In Policy name, enter a name for the policy.
-
Under Policy document, drop your PDF into the upload area, or select the area to browse for it.
Only PDF files are accepted, and a policy takes a single document. Once a file is attached, select the upload area again to replace it.
-
Select Continue.
Both a name and a document are required before Continue becomes available.
Choose a decision profile
The next step asks which decision profile the policy uses. A decision profile pins the question the AI answers about your document and the outcomes it may choose from, and it is reusable across policies.
-
Review the existing profiles. Each one shows its decision question, whether it is a Direct decision or a Scored decision, how many outcomes it defines, and how many policies already use it.
-
Expand a profile to see its full definition.
-
Do one of the following:
- To reuse an existing profile, select it and then select Create policy.
- To define a new profile, select Add new decision profile and then select Continue. The wizard advances to the profile definition step, described in Define a new decision profile below.
Define a new decision profile
-
In Decision profile name, enter a name.
-
In Decision question, enter the question the AI should answer as it reads the policy document.
This question shapes how every rule is interpreted. Keep it to a single yes/no or which-one question.
-
Under How should decisions be made?, select the model type:
Option Description Direct decision The AI picks one of your outcomes for each case. Scored decision Conditions contribute points, and the total lands in a score band that names the outcome. -
Define the outcomes. These are the only answers the AI may give, so cover every case the policy can produce.
- For a Direct decision profile, enter an Outcome label for each outcome and select its polarity - Positive, Negative, or Neutral. Polarity affects only how the outcome is styled; it does not affect which outcome is chosen.
- For a Scored decision profile, enter an Outcome label, a Min, and a Max for each score band. Bands are matched in order from top to bottom and bounds are inclusive. Leave the last Max blank for no upper limit.
Select Add outcome to add a row, or the remove icon to delete one.
-
Optionally, enter Notes.
Notes are hints for the AI about this document type, read alongside the policy document. Use them to add detail about the decision profile, its outcomes, and where this decision fits within the larger use case. Accurate, concise notes help the AI extract the rules that serve the decision.
-
Select Create policy.
Result
Creating the policy starts a run in which the AI reads the document and builds the decision model from it. This takes a few minutes. The policy appears on the Policies page with a Pending decision model status, which becomes Active once generation completes, or Failed if it does not.
Next steps
- Edit a policy - Review and refine the generated rule tree before using the policy in production.
- Edit a template - Add the policy to a template as a Policy section.