FDA’s top scientists published a review article in the Journal of American Medical Association (JAMA) reviewing the various uses of Artificial Intelligence (AI) and the regulatory trends and expectations. FDA acknowledges several current and potential uses of AI in drug development and clinical research, making these processes more efficient and precise.
In drug development, AI can help with drug target identification by analyzing vast amounts of data to identify biological targets and predict their role in diseases. It can also aid in screening and designing compounds by predicting how effective a drug might be and its potential side effects. AI can even assist in finding new uses of approved drugs, called drug repurposing, by analyzing various data sources. Additionally, AI could be used for modeling pharmacokinetics and pharmacodynamics, helping to optimize drug doses for specific populations, like those with rare diseases or pregnant women. It can also streamline advanced pharmaceutical manufacturing by improving process design, monitoring, and control.
In clinical research, AI offers multiple benefits, starting with helping identifying or screening potential participants for recruitment into clinical trials. By mining data from health records and social media, AI can match individuals to clinical trials, ensuring diversity and fair representation. AI can also help in participant selection and stratification, predicting outcomes based on characteristics and grouping patients to reduce trial variability. Furthermore, AI can be used to improve adherence and retention during trials by using tools like smartphone reminders, smart pillboxes, and digital markers to track patient involvement. It can also reduce the burden on participants by using passive data collection methods.
AI tools have been used to enhance clinical trial data collection, management, and analysis by identifying new disease characteristics, evaluating data, and ensuring accurate and secure data storage. It can even simulate clinical outcomes using virtual cohorts and digital twins. In postmarket safety surveillance, AI can play a role in detecting adverse events through social media and medical literature, helping evaluate drug safety after they hit the market. It also assists in determining the relationship between drugs and adverse events and in prioritizing safety cases for review.
AI has the potential to transform both drug development and clinical research, making these processes more effective and tailored to the needs of specific populations. As AI continues to advance in medicine and health care, it’s important to understand how well it works in real-world settings. This requires ongoing efforts beyond the FDA, involving the entire health care system to keep up with fast-moving technology. Without proper attention, AI could fail like other technologies that haven’t lived up to their promise, or even cause harm if it prioritizes profit over patient outcomes. The FDA intends to have strong oversight to maintain public trust and ensuring AI benefits health care. The FDA also intends to interact with both the health care industry and academia to ensure AI remains safe and effective for patients.
