Published Research and Conference Presentations

Peer-reviewed publications by ArcaScience's team on benefit-risk methodology, pharmacovigilance AI, and regulatory science.

Peer-Reviewed Publications

Signal Detection 2025

Domain-Specific AI Models vs General-Purpose LLMs for Pharmacovigilance Tasks

Comparative analysis demonstrating superior performance of purpose-trained domain models for adverse event extraction, MedDRA auto-coding, and causality assessment. Evaluation across 10,000+ FAERS case narratives shows 92% precision vs 67% for GPT-4.

Journal: Artificial Intelligence in Medicine Authors: Chen S, Torres M, Liu J, Patel P
View on PubMed
BRA Methodology 2024

Multi-Criteria Decision Analysis in Pharmaceutical Benefit-Risk Assessment

Systematic review and implementation guide for MCDA methods in benefit-risk assessment. Covers swing weighting, SMAA, and stochastic acceptability analysis with regulatory acceptance considerations for FDA and EMA submissions.

Journal: Value in Health Authors: Chen S, Kowalski A, Rodriguez E
View on PubMed
Signal Detection 2024

AI-Driven Disproportionality Analysis for Post-Marketing Signal Detection

Novel methods integrating traditional disproportionality metrics (PRR, ROR, MGPS, BCPNN) with deep learning for enhanced signal detection. Validation on 15 years of FAERS data demonstrates 3x improvement in early signal detection vs standard methods.

Journal: Journal of Pharmacoepidemiology Authors: Torres M, Patel P, Chen S
View on PubMed
AI in Pharma 2024

Natural Language Processing for Adverse Event Extraction from Regulatory Documents

Transformer-based NLP models for automated extraction of adverse event terms from unstructured clinical narratives, literature, and regulatory submissions. Domain-specific pre-training on pharmacovigilance corpora achieves 94% F1 score.

Journal: BMC Medical Informatics and Decision Making Authors: Liu J, Rodriguez E, Torres M
View on PubMed
BRA Methodology 2023

Quantitative Benefit-Risk Assessment Using BRAT Framework: A Platform Approach

Practical implementation of the Benefit-Risk Action Team (BRAT) framework within a software platform. Demonstrates automated value tree construction, effects table generation, and scenario modeling aligned with FDA and EMA guidance.

Journal: Drug Safety Authors: Chen S, Patel P, Kowalski A, Torres M
View on PubMed
Regulatory Science 2023

Automating PSUR/PBRER Generation: A Regulatory Science Perspective

Framework for automated generation of Periodic Safety Update Reports (PSUR) and Periodic Benefit-Risk Evaluation Reports (PBRER) compliant with ICH E2C(R2). Validation study across 20 submissions demonstrates regulatory acceptance with 60% time reduction.

Journal: Therapeutic Innovation & Regulatory Science Authors: Rodriguez E, Chen S, Liu J
View on PubMed

Conference Presentations

DIA Annual Meeting

"AI-Powered Benefit-Risk Assessment: From Research to Regulatory Practice"

San Diego, CA — June 2025

ICPE (Pharmacoepidemiology)

"Real-World Evidence Integration in Automated Signal Detection Systems"

Atlanta, GA — August 2025

ISoP Annual Meeting

"Quantitative Frameworks for Benefit-Risk Assessment in Pharmacovigilance"

Barcelona — October 2025

ISPOR Europe

"From Regulatory BRA to HTA Value Dossiers: Automated Evidence Synthesis"

Barcelona — November 2025

Discuss Our Methodology With Our Science Team

Talk to the scientists behind these publications about how ArcaScience's methodology can be applied to your regulatory challenges.

Talk to a Scientist Our Methodology