PharmacoPlatform: ICH E2B(R3) ICSR 网关、不相称性评分模型与真实世界安全信号挖掘

PharmacoPlatform Global Pharmacovigilance & Signal Detection Pipeline

Multichannel safety intake, MedDRA NLP normalization, and disproportionality scoring

Central System: Pharmacovigilance Core - E2B(R3) Engine & Disproportionality Analyzer

Clinical IoMT Nodes

• Multichannel Safety Intake Sink
• Clinical NLP / MedDRA Auto-Coder
• Individual Case Safety Report (ICSR) Core
• Disproportionality Signal Engine
• Regulatory Gateway Dispatcher

Architectural Layers & Regulatory Standards

• Multisource Adverse Event Ingestion (Optical Character Recognition (OCR) / Multi-Tenant Intake APIs): Ingesting adverse event reports from clinical trials, spontaneous patient hotlines, scientific literature, and social platforms.
• Medical Ontology Normalization (Clinical Transformer Models / MedDRA Dictionary v26+): Mapping unstructured patient symptom descriptions into standardized Medical Dictionary for Regulatory Activities (MedDRA) terms.
• Statistical Signal Detection & Disproportionality (Bayesian Confidence Propagation Neural Network (BCPNN)): Computing Proportional Reporting Ratios (PRR) and empirical Bayes geometric means to detect emerging safety signals.
• Automated Regulatory Electronic Submission (AS2 Secure Gateway / E2B(R3) XML Validator): Validating and transmitting expedited 7-day and 15-day safety reports to FDA FAERS and EudraVigilance.

Key Metrics: < 15 min (Automated ICSR Processing Time) | 99.4% (MedDRA Coding Accuracy) | 100% (ICH E2B(R3) XML Schema Compliance) | Real-Time (Adverse Signal Disproportionality Mining)