What this category's data looks like
Daily data for this category comes from the national medical security information platform, hospital HIS and medical insurance interfaces, medical insurance catalog and coding database, and cross-provincial medical treatment settlement platform. The update frequency is daily. Typical documents include medical insurance settlement lists, expense details, medical record homepages, prescriptions, denied payment lists, and settlement reconciliation statements. Core fields cover diagnosis and treatment item codes, pharmaceutical and consumable codes, expense amounts, insured person information, and medical institution information. All fields follow unified national medical insurance standards, with no custom special units.
What constraints do these characteristics impose on the "knowledge base retrieval and recall" link
The daily updated data requires the retrieval system to support incremental synchronization, to avoid resource consumption caused by full pulls. Documents contain a large number of standardized coding fields, so retrieval matching must balance exact code matching and semantic matching to ensure accurate correspondence between policy basis and expense details. A single audit may involve associated queries of multiple policy clauses, so the retrieval scope must be limited to the compliant medical insurance policy knowledge base to prevent interference from irrelevant content. Medical insurance policy clauses are expressed precisely, so recall results must be sorted by relevance to avoid non-core clauses affecting audit efficiency.
How to set the configurations
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Value | Basis for This Setting |
|---|---|---|
similarity | 0.72–0.88 | Medical insurance policy clauses are expressed precisely. This threshold balances exact matching and recall scope, avoiding missed or false judgments |
limit | 3–6 | The policy basis corresponding to a single medical insurance cost settlement audit usually does not exceed 6 documents. Limiting the quantity improves audit efficiency |
usingReRank | Enable | Recalled policy clauses must be sorted by relevance to ensure core policies are displayed first |
datasetSearchUsingExtensionQuery | Disable | Medical insurance policy terminology is precise, so query scope expansion should be avoided to prevent irrelevant content from being introduced |
collectionIds | Specify the ID of the corresponding medical insurance policy knowledge base | Limit the retrieval scope to only compliant official policy documents to ensure the authority of audit basis |
Three common errors
- Phenomenon: A
quote type errorerror occurs when referencing configuration variables, and the interface prompts that the field format is abnormal. Cause: Variables are not encapsulated according to the field format of medical insurance policy documents. For example, correct reference markers are not added when directly referencing coding fields. - Phenomenon: Recall results for the same audit question are inconsistent between the old and new knowledge bases. The new library can recall normally, while the old library cannot. Cause: Core fields such as diagnosis and treatment item codes are not retained when documents in the old knowledge base are chunked, or the chunk length exceeds the matching threshold range.
- Phenomenon: The audit result returned by the conversation does not display the knowledge base reference source. Cause: The reference display switch for retrieval results is not enabled, or the associated reference output function is not enabled in the configuration.
How to confirm the configuration is complete
- Submit a single medical insurance cost settlement audit test request, and check whether the returned results include matching policy basis documents.
- View the retrieval log to confirm that the number of recall results conforms to the set
limitparameter range. - Verify the variable reference configuration to confirm that there are no format errors and that medical-related fields can be parsed normally.
- Enable the reference display configuration to confirm that the returned results include corresponding knowledge base source information.
The values provided here are common starting points and should be measured against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-28.