On the Application of Artificial Intelligence/Machine Learning (AI/ML) in Late-Stage Clinical Development

Area of Science:

• Clinical Pharmacology
• Biostatistics
• Artificial Intelligence in Medicine

Background:

• Artificial intelligence and machine learning (AI/ML) are increasingly accessible for clinical development.
• Stakeholders require clarity on the realistic roles and standards for AI/ML in healthcare.
• Late-stage clinical research demands high standards of robustness, transparency, and traceability for AI/ML applications.

Purpose of the Study:

• To explore the application of AI/ML methods in late-stage clinical drug development.
• To summarize existing regulatory guidance and statistical work relevant to AI/ML in this context.
• To stimulate discussion on the general role of AI/ML analyses in drug development.

Main Methods:

• Review of current regulatory guidance and statistical literature on AI/ML in clinical development.
• Presentation of an industry case study applying ML methods to investigate baseline characteristics' influence on treatment effects.
• Utilization of standardized ML approaches with explainable AI (XAI) for intuitive graphical displays.

Main Results:

• Demonstration of ML methods' capability to analyze extensive baseline characteristics in late-stage clinical trials.
• Successful application of standardized, explainable AI methods for intuitive data visualization.
• Identification of potential for AI/ML to enhance understanding of treatment effects in drug development.

Conclusions:

• AI/ML methods can be applied robustly and transparently in late-stage clinical drug development.
• Standardized approaches and explainable AI are crucial for integrating AI/ML into clinical research.
• Further discussion is needed to define the optimal role and standards for AI/ML in drug development.