Data Characteristics for this Category
Regulatory and SOP documents related to respiratory system diseases primarily originate from internal medical institution management regulations, technical guidelines published by national health commissions, clinical pathways, and drug instructions from drug administration agencies. Data update frequency is relatively stable. National-level guidelines typically update every few years, while hospital internal SOPs are revised annually or quarterly based on clinical practice and the latest research. Document structures are predominantly unstructured text, often containing numerous charts, flowcharts, and specialized terminology. Fields and units are highly specialized, for example, dosage units (mg/kg, IU), time units (h, min), and physiological indicator units (mmHg, L/min). The same concept may also have multiple forms of expression.
Constraints Imposed by these Characteristics on "Tool Calling and Plugins"
The specialized nature of respiratory system regulatory documents requires tool calling to accurately identify and extract specific medical entities, such as drug names, disease diagnoses, treatment plans, examination indicators, and their normal ranges. Unstructured flowcharts and tables are common in documents, meaning traditional text parsing tools may be insufficient to fully understand their logical relationships. This necessitates more advanced structured information extraction capabilities. The update frequency is moderate, but revisions often involve critical parameter or process changes. Plugins must support version control and incremental updates to ensure the timeliness of recalled information. Additionally, regulations from different medical institutions may have subtle differences, requiring plugins to possess a degree of contextual understanding and multi-source information fusion capabilities to avoid misdiagnosis or misleading information. The diversity of specialized units and fields requires tools to correctly handle unit inconsistencies during numerical comparisons or conversions.
Configuration Guidelines
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
maxContext | 4096 tokens | Respiratory SOPs often contain complex logic, requiring a larger context to understand the full text. |
Chunk size (Segment Length) | 800–1200 characters | Balances semantic completeness with segment retrieval efficiency. |
Recall count (Retrieval Count) | Top 8 entries | Ensures coverage of multiple sections or regulations of relevant policies. |
Similarity threshold (Similarity Threshold) | 0.75 | Identifies highly relevant policy clauses and filters out irrelevant information. |
Rerank result count (Reranked Return Count) | Top 5 entries | Optimizes the quality and conciseness of the final presented results. |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Handles the parsing requirements for large PDF or scanned documents. |
Three Common Mistakes
- Tool calling returns "Enterprise has no patent information record." This occurs because the tool's query parameters do not match the field names in the target data source, preventing the query conditions from being passed correctly.
- Plugin execution fails, with logs showing "MongoServerError: The dollar ($) p." This may be due to incompatibility between the database version required by the plugin and the database version of the FastGPT runtime environment, leading to unrecognized operators.
- The Markdown to file plugin prompts "File upload failed." This happens when the file size exceeds the
UPLOAD_FILE_MAX_SIZEconfiguration limit, or file path permissions are insufficient.
How to Confirm Correct Configuration
- For typical respiratory diseases (e.g., asthma, COPD), simulate clinical scenarios by asking questions and verify that the treatment plans and drug dosages returned by tool calling are consistent with the actual SOP content.
- Upload regulatory documents containing charts and flowcharts. Test if the document parsing plugin can correctly extract key information and check if the extracted results include text descriptions from the charts.
- Simulate a policy update scenario by revising some key parameters. Then, ask questions again to confirm that tool calling retrieves the latest revised policy content.
- Invoke tools involving numerical calculations or unit conversions. Input parameters with different units and verify that the numerical values and units of the output results are correct.
Note: The values provided are common starting points. Measure them against your own samples for optimal performance.
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.