Pharmacovigilance Profiling Technology for Patient Medical Records Based on Large Language Models
10.3870/j.issn.1004-0781.2025.04.009
- VernacularTitle:基于大语言模型的患者病历药物警戒画像技术研究
- Author:
Zhengshan WU
1
;
Shu ZHANG
;
Yimin LIN
;
Yi LEI
;
Qing WANG
;
Zhigang SUN
;
Lin ZHANG
Author Information
1. 福建省药品审评与监测评价中心,福州 350003
- Publication Type:Journal Article
- Keywords:
Pharmacovigilance;
Active monitoring;
Patient profiling;
Large language models;
Knowledge graph
- From:
Herald of Medicine
2025;44(4):554-560
- CountryChina
- Language:Chinese
-
Abstract:
Objective To enhance the efficiency and accuracy of post-marketing safety monitoring and evaluation of drugs in China by studying large language models-based patient medical record pharmacovigilance profiling techniques,providing scientific methods and technical support to ensure the safe use of drugs for patients.Methods This study constructs a pharmacovigilance profile that includes individual patient differences,medication details,and adverse reaction manifestations.It enhances a large language model with a knowledge graph in the field of pharmacovigilance and designs targeted prompts to guide the model to output pharmacovigilance profiles.Results Large language models demonstrate significant advantages in active monitoring,effectively processing and analyzing medical text data,and improving the monitoring and prediction capabilities of drug adverse reactions.Through the design of prompts,the model can more accurately depict patient pharmacovigilance profiles,providing decision support for medical professionals.Conclusions The study of large language model-based patient medical record pharmacovigilance profiling technology provides scientific evidence and technical support for the early detection and prevention of drug adverse reactions,helping to reduce medical costs,improve medical outcome prognoses,and opens new paths to ensure patient drug safety.
