Search documentation
Find a documentation page by the words on it.
Documentation
Manage Claims Overview
The Overview sub-tab of Manage Claims is a working view of everything Serelora holds for one patient, in three parts: the documents, every claim in one table, and a graph that shows how they connect. Use it when you want to audit a chart rather than read it, for example to see which document an entry came from or whether a file produced anything at all.
Step 1: Open the Overview
In an open chart, open the Actions menu and select Manage Claims. The Overview sub-tab is the first of three. Under it are three views: Documents, Claims and Graph. It is available to roles that can manage claims: Admin, Physician, Physician Assistant and Medical Assistant by default. See Manage Claims.
Step 2: Check what each document produced
The Documents view lists the patient's documents with Filename, Category, Author, Type, Status, Document date, Chunks, Mentions and Claims. On a narrow pane, Chunks, Mentions and Type drop away first, then Author.
The last three columns are counts. Chunks is how many passages the document was split into. Mentions is how many facts Serelora picked out of them. Claims is how many chart entries those became. A document that shows a status of uploaded but has zero mentions and zero claims produced nothing, and the reason is on its row in the Documents tab. See Why a Document Was Not Read Into the Chart.
The refresh button reloads the list and its counts. Here the list is read-only: you cannot retry a document or open it from this view. Use the Documents tab for that. See Documents.
Step 3: Filter all of the patient's claims
The Claims view is one table of every entry on the chart. Above it are three filters:
- Search statement matches the entry's wording.
- Category limits the table to one kind of entry.
- Epistemic status limits it to Asserted, Superseded, Contested or Retired, or All statuses. These are the claim statuses in the Glossary.
The columns are Statement, Category, Confidence, Status, Evidence and Date. Evidence is how many source passages support the entry. Confidence is the extraction model's estimate, shown as a bar and a percentage. Click a heading to sort. Select a row, or View at its end, and the entry's claim detail opens beside the chart. See Claim Detail and Where a Fact Came From.
Unlike the chart tabs, this table is not limited to entries that currently stand. By default it lists entries of every status, including superseded and retired ones. Choose a status to narrow it. If nothing matches, the table says No claims match search or No claims match filters.
Step 4: Explore the Graph
The Graph view draws the patient's claims as a network. Scroll to zoom, drag to pan, and click a node to see its details in the panel on the right. The panel's guide names the five kinds of node: a Document, a Chunk of text from it, a Mention picked out of the chunk, an Instance, which is one dated statement of a claim, and an Identity, the claim itself. Following the links from an identity back to a document is the same trail as the evidence in the claim detail.
Type in Search by claim description… to find an entry. The graph loads claims in groups. The line under the box says, for example, how many of the claims are showing, and Load more brings in the rest. In the details panel, View full claim details or View claim identity details opens the entry, and the document's name opens the document at the highlighted passage.
What this view is for
It is a read-only way to look at provenance. Nothing you do here changes the chart. To merge duplicates, use Duplicate claims. To change codes across the chart, use Manage codes. See Generating Codes for Claims.