Leveraging large language models to compare perspectives on integrating QSP and AI/ML

Area of Science:

• Pharmacology
• Computational Biology
• Artificial Intelligence

Background:

• Two recent papers present conflicting views on integrating Quantitative Systems Pharmacology (QSP) and Artificial Intelligence/Machine Learning (AI/ML) in drug development.
• One paper positions QSP as central, using AI/ML for computational enhancement, while the other suggests AI/ML as an alternative mechanistic framework.

Purpose of the Study:

• To analyze and reconcile contrasting perspectives on QSP and AI/ML integration using Large Language Models (LLMs).
• To determine the optimal roles and synergy between QSP and AI/ML across the drug development pipeline.

Main Methods:

• Utilized Large Language Models (LLMs) to compare core arguments of two papers on QSP and AI/ML integration.
• Conducted repeated, neutral prompt tests to assess LLM's comparative analysis of QSP and AI/ML methodologies.
• Synthesized human expertise with AI-driven analysis to evaluate the findings.

Main Results:

• LLM analysis indicated that QSP provides mechanistic rigor and regulatory clarity, suitable for specific drug development stages.
• AI/ML excels in high-dimensional data analysis and exploratory modeling, also suited for distinct phases.
• A hybrid approach harmonizing data-driven AI/ML insights with QSP's mechanistic integrity is proposed as optimal.

Conclusions:

• LLMs can serve as valuable tools for synthesizing complex scientific information and offering less biased viewpoints.
• The synergy between QSP and AI/ML, potentially guided by LLM analysis, can drive innovation in therapeutic discovery and optimization.
• Further community discussion is encouraged to align QSP and AI/ML within model-informed drug development (MIDD).