AI/ML Challenges in Precision Medicine: FDA Paper

Share

Researchers from FDA, PharmaLex, and Quinten Health say there remain challenges to address before artificial intelligence/machine learning (AI/ML) has an increased impact on precision medicine. Writing in Therapeutic Innovation & Regulatory Science, the researchers summarize the topics discussed at a 2022 panel of FDA and industry experts about the challenges of successfully applying AI/ML to precision medicine.

“Bias is an important issue,” they say, “such as selection bias where data are not representative of the target population…. It is important to adopt strategies during early planning to mitigate these biases.”

The paper says that when selecting data to train and validate AI/ML tools, it is important to employ methods and systems that ensure sufficient data quality. In addition, it says, the demonstration of external validity is desired for improving the generalizability of performance of AI/ML algorithms when applied to patients and settings different from those on which the development or training was performed.

Read more