image
Author: Patrizia Cavazzoni, M.D., Director, Center for Drug Evaluation and Research
Artificial intelligence (AI) and machine learning (ML) are no longer concepts of the future. They are now part of the way we live and work. The U.S. Food and Drug Administration uses the term AI to describe the branch of computer science, statistics, and engineering that uses algorithms and models to perform tasks and exhibit behaviors such as learning, decision-making, and prediction. . ML is a subset of AI that uses data and algorithms without being explicitly programmed to mimic how humans learn.
The increasing data volume and complexity of AI/ML, coupled with advances in cutting-edge computing power and methodologies, have the potential to transform the way stakeholders develop, manufacture, use, and evaluate therapeutics. Ultimately, AI/ML will help us deliver safe, effective, and high-quality treatments to our patients faster.
For example, AI/ML can be used to scan medical literature for relevant findings and predict which individuals will respond better to treatment and which are at higher risk of side effects. Conversational agents or chatbots based on “generative” AI could answer people’s questions about participating in clinical trials or reporting adverse events. A digital or computerized patient “twin” can be used to model a medical intervention and provide biofeedback before the patient undergoes the intervention.
Regulatory use is real. Over 100 drug and biologics applications submitted to the FDA in 2021 included his AI/ML component. These applications spanned a variety of therapeutic areas, with sponsors incorporating the technology at various stages of development.
Like any evolving science and technology field, AI/ML in drug development presents challenges such as ethical and security considerations such as inappropriate data sharing and cybersecurity risks. There are also concerns about using algorithms that are somewhat opaque or have internal operations that are not visible to users or other parties. This can amplify errors and existing biases in the data. To promote fairness when using AI/ML techniques, we are committed to preventing and redressing discrimination, including algorithmic discrimination, that occurs when automated systems favor one category of people over another. Aiming to be To address these concerns, the FDA has released a discussion paper, “Using Artificial Intelligence and Machine Learning in Drug and Biologics Development.”
AI and ML in Pharmaceutical and Biologics Development
This discussion paper is a collaborative effort of the FDA’s Center for Drug Evaluation and Research, the Center for Biologics Evaluation and Research, and the Center for Devices and Radiation Health, which includes the Digital Health Center of Excellence. This document discusses the use of AI/ML in drug and biologics development with stakeholders in the medical product development community, including pharmaceutical companies, ethicists, academia, patients and patient advocacy groups, global regulators and other authorities. It is intended to facilitate discussion of And the development of medical equipment used for these treatments.
This paper contains an overview of current and potential future applications of AI/ML in therapeutics development. We also discuss possible concerns and risks associated with these innovations and how to address them. For example, the paper discusses the importance of human involvement, but that depends on how the technology is used. The paper also emphasizes taking a risk-based approach to assess and manage his AI/ML to foster innovation and protect public health.
This paper characterizes certain risks such as data bias, data inaccuracy and completeness used to train ML algorithms. Additionally, the paper outlines the role of model performance monitoring to ensure model reliability, relevance, and long-term consistency.
There are also issues to consider, requiring engagement and cooperation among the biomedical community. As a follow-up to this paper, we are planning a workshop to discuss how the community can work together to realize the potential of AI/ML in product development, keeping in mind potential challenges. We look forward to hearing from experts on this important subject.
CDER Discussion Paper on Regulatory Advanced Manufacturing Assessment Framework
To further address the use of AI in pharmaceutical manufacturing, CDER published another discussion paper, Artificial Intelligence in Pharmaceutical Manufacturing, as part of the framework of the Regulatory Advanced Manufacturing Evaluation (FRAME) initiative. AI technology is important in pharmaceutical manufacturing because it can enhance process control, identify warning signals early, and prevent product loss. We are also planning his second workshop for stakeholders to discuss discussion paper questions on AI in pharmaceutical manufacturing.
Our work in AI/ML extends beyond these efforts. We consult with product developers, engage with patients, and advance regulatory science in this area. As public health regulators, we want to encourage the safe development of these technologies to make important treatments more quickly and reliably available to Americans. The FDA’s work also supports the government’s ongoing efforts to ensure that technology improves the lives of Americans while promoting a consistent and comprehensive approach to AI-related risks and opportunities.
