Supplier Audit Pharmacovigilance HTTP Interface and External

Data Characteristics

Supplier audit data in pharmacovigilance primarily focuses on supplier quality management systems, production compliance, and adverse event handling capabilities. Data sources typically include audit reports, non-conformity lists, Corrective and Preventive Action (CAPA) records, supplier qualification documents (e.g., GMP certificates, production licenses), and adverse event reports. Data update frequencies vary. Audit reports may update annually or biennially, while CAPA and adverse event reports may occur in real-time. Document structures are diverse. Audit reports often appear as structured documents (PDF, Word), containing evaluation metrics, findings, and recommendations. Qualification documents are scanned copies or electronic certificates. Adverse event reports usually follow standard formats like ICH E2B, including fields for drugs, patients, event descriptions, severity, and outcomes. Fields may contain multilingual descriptions. Units involve time (e.g., hours, days), quantity (e.g., batches), and severity levels.

Constraints Imposed by These Characteristics on "HTTP Interface and External Systems"

The diversity of supplier audit data challenges the data parsing capabilities of HTTP interfaces. The unstructured or semi-structured nature of audit reports and qualification documents requires interfaces to handle various file types like PDF and DOCX, extracting key information from them. Standardized formats for adverse event reports (e.g., XML, JSON) necessitate precise field mapping capabilities in the interface to ensure accurate data import. Audit reports and qualification documents update infrequently but can be large, requiring high limits for file upload size and processing timeout. The real-time nature of CAPA and adverse event reports demands interfaces that support high concurrency and low-latency data submission. Data may contain sensitive information, imposing strict security requirements on HTTP interfaces, including authentication and data encryption. The presence of multilingual fields also requires interfaces to correctly handle character encoding during data transmission and storage.

Configuration Guidelines

Configuration Item	Recommended Value	Rationale
`UPLOAD_FILE_MAX_SIZE`	`500 MB`	Audit reports and qualification documents may contain numerous images and scanned copies, resulting in large file sizes.
`PARSE_FILE_TIMEOUT_SECONDS`	`600 seconds`	Processing large PDF or DOCX files, especially with OCR recognition, can be time-consuming.
`maxContext`	`2000 characters`	Descriptions of individual paragraphs or key findings in audit reports can be long, requiring more context for semantic understanding.
`similarityThreshold`	`0.75`	Ensures precision in recall results when comparing historical supplier issues or similar adverse events.
`chunkOverlap`	`100 characters`	Handles closely related paragraphs in audit reports, preventing information fragmentation.
`requestTimeout`	`120 seconds`	External systems may respond slowly due to processing complex audit data or performing security checks.

Common Pitfalls

Workflows do not trigger as expected or processing results are empty after an interface call. This happens when the data structure returned by the external system does not match the field mapping preset in the FastGPT interface, preventing critical information from being correctly extracted.
Uploading large audit report files returns an HTTP status code 413 Request Entity Too Large. This occurs when the UPLOAD_FILE_MAX_SIZE parameter is not configured correctly, causing the server to reject oversized files.
After submitting an adverse event report via API, relevant context is missing or incomplete when queried in the FastGPT platform. This may be because the context parameter was not passed correctly during the API call or maxContext was configured too small, failing to capture the event description completely.

Verification Steps

Upload a typical supplier audit report (e.g., a 50 MB PDF file). Verify successful parsing and extraction of key information such as audit conclusions and non-conformity numbers.
Simulate submitting a multilingual adverse event report through an external system. Query the report in FastGPT and confirm that multilingual content displays correctly without corruption.
Create a workflow with complex business logic. Trigger it via the HTTP interface. Verify the workflow execution logs to ensure each step processes audit data or adverse event information as expected.

Note: The values provided are common starting points. Measure them against specific samples.

Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-21.