T/JNBDA 0008-2025 Active Group standards

T/JNBDA 0008-2025 Guide for Building a Medical Standardization Artificial Intelligence System

Publish Date: 2025-11-13 Implement Date: 2025-11-13 For services related to genuine standard inquiry, procurement, translation, and other related services in China, please Contact Us

Price: $ 20.00

Out of stock

Basic Information

Standard Code: T/JNBDA 0008-2025

Standard Type: Group standards

Standard Status: Active

is_force_gb: no

CCS Name: -

ICS Name: Integration, terminology, standardization, and literature (vocabulary)

Publish Date: 2025-11-13

Implement Date: 2025-11-13

Scope

This standard specifies the general principles, construction requirements, main functions, and data security for the construction of standardized medical artificial intelligence systems. This standard applies to the construction of standardized medical artificial intelligence systems. The standardized medical artificial intelligence systems described in this standard are used as auxiliary systems and are not intended to serve as the basis for final decision-making. By accessing multi-source heterogeneous medical data through multiple channels and preprocessing files using OCR technology, the data are subjected to deep vector representation based on medical domain knowledge graphs and segmentation rules. The generated vector data are stored in dedicated vector databases. At the same time, associated structured data and original documents are stored in relational/non-relational databases and file storage systems, respectively. Through recall testing, secondary annotation, and review of data samples, the content of the knowledge base is continuously iterated and optimized. The artificial intelligence system receives patient information and pending prescriptions through manual entry or system interfaces. It uses large-scale language models (LLMs) to perform semantic analysis and query optimization on the input data, and then conducts precise retrieval from the constructed knowledge base. Through the Rerank model, the retrieved knowledge fragments are reordered and prioritized based on the weight of the data. Subsequently, the high-value information and the original questions are handed over to professional medical LLM large models, which use their reasoning capabilities and prompt engineering techniques for comprehensive analysis to ultimately generate structured medication review reports. At the same time, the system also has the ability to automatically audit historical prescriptions. It pulls data from the hospital information system (HIS) for analysis through sched