• 中国核心期刊数据库收录期刊
  • 中文科技期刊数据库收录期刊
  • 中国期刊全文数据库收录期刊
  • 中国学术期刊综合评价数据库统计源期刊等

快速检索引用检索图表检索高级检索

中国医药导刊 ›› 2024, Vol. 26 ›› Issue (11): 1080-1086.

• 监管科学 • 上一篇    下一篇

面向临床研究的统计软件系统开发

夏天雨1a,周怡1b,尤翀2*,周晓华1ab*
国家重点研发计划“面向药品现代化监管的智能化服务平台”项目团队1   

  1. 1.北京大学 公共卫生学院生物统计系a,北京国际数学研究中心b,北京 100871;
    2.复旦大学上海数学与交叉学科研究院,上海 200433;4.北京大学,北京 100871
  • 收稿日期:2024-09-30 修回日期:2024-11-21 出版日期:2024-11-28 发布日期:2024-11-28
  • 基金资助:

    国家重点研发计划(2021YFF0901400)

Development of A Statistical Software System for Clinical Research

  1. 1.Biostatisticsa BICMRb Peking University Beijing 100871, China
    2.SIMIS Fudan University Shanghai 200433, China
  • Received:2024-09-30 Revised:2024-11-21 Online:2024-11-28 Published:2024-11-28
  • Contact: Xiao-Hua Zhou E-mail:azhou@math.pku.edu.cn

摘要:

临床试验数据的统计分析是药品研发和审评过程中的核心环节,对促进医药产业高质量发展和提升药品现代化监管技术具有重要意义。然而,传统统计方法难以应对不完美随机化或数据不足等复杂情况。为此,本研究开发面向临床研究的统计分析系统,以提升药品监管审评和临床研究中药品有效性和安全性评估的科学性、准确性和灵活性。该系统引入国际前沿的因果推断方法、数据填补技术和机器学习算法,并结合智能化数据处理和图形展示,增强用户对复杂场景分析策略的理解与应用效率,确保研究成果的高质量。系统的创新性统计分析功能包括两个核心模块:针对临床试验随机化被破坏和真实世界数据环境,提供基于替代变量的主层方法和数据融合等因果推断策略,确保药物有效性评估结果的因果解释性和稳健性;在有限的临床数据基础上,提供可整合多源数据和安全监测信号的新型机器学习算法,构建药品不良反应信号挖掘和预测模型,增强对潜在不良反应的预警能力。本研究总结该系统功能与创新方法,以期为药品监管审评和临床研究中的药品有效性与安全性分析提供有力的技术支持。


关键词:  , 因果推断;药品有效性分析;药品不良反应监测;机器学习

Abstract:

Statistical analysis of clinical data plays a crucial role in drug development and drug evaluation being essential to the high-quality advancement of the pharmaceutical industry and the modernization of regulatory capabilities. However traditional statistical methods cannot cope with complex scenarios such as imperfect randomization or limited data which may arise during drug development. To address these challenges this study has developed a statistical analysis system tailored for clinical research aiming to enhance the scientific rigor accuracy and flexibility of clinical efficacy and safety evaluations under complex conditions in drug regulation and clinical trials. The system incorporates cutting-edge causal inference methodsdata imputation techniques and machine learning algorithms along with intelligent data processing and graphical visualization to improve users' understanding to complex analysis scenarios and improve application efficiency ensuring high-quality research outcomes. The system offers two core innovative features first in response to issues like imperfect randomization and real-world data environments in clinical trials it introduces novel causal inference strategies including principal stratification based on instrumental variables and data fusion to ensure causal interpretability and robustness in drug efficacy evaluations second it employs a new machine learning algorithm capable of integrating multi-source data and safety monitoring signals even with limited clinical data to develop models for adverse drug reaction signal detection and prediction thereby enhancing early warning capabilities for potential adverse reactions. This paper summarizes the system's functionalities and innovative methods providing strong technical support for efficacy and safety assessments in clinical research and trial regulation.


Key words:  , Causal inference; Drug efficacy analysis; Adverse drug reaction surveillance; Machine learning

中图分类号: