How Post Market Surveillance Interacts with Statistical and Computational Models of Drift in AI Models
This article may be relevant to your organisation if you are working with evolving ML/AI and have insights that could help shape regulations ensuring adaptive models are both safe and explainable
Medical devices using machine learning or adaptive AI algorithms, like all other medical devices, must be approved for market, satisfying medical device regulators, with appropriate evidence, that they are safe and fit for intended purpose. Adaptive AI algorithms can often be seen as “black box AI” because they change as they learn, making it difficult to explain how they are making their decisions. This presents a challenge for regulators all over the world, not only of medical devices, but in many other AI tools that carry risk. To protect patient safety, it will be necessary to understand if the logic of an algorithm has changed significantly since it was approved. If it has, both manufacturers and regulators need to know whether it remains safe and fit for purpose and what to do about it. This project will research methods for determining if there has been a significant change in the way that an adaptive AI algorithm medical device is working and how the change should be reported to regulators.
Aim - to investigate the different forms of drift when assessing post-market surveillance of AI models.
Objectives:
Identify the different potential forms of drift when performing post-market surveillance of AI decision models (e.g. data drift, performance drift, bias-drift, population drift)
Apply a full empirical analysis to explore appropriate metrics to identify different forms of drift under different scenarios.
Work with regulators / innovators to identify how this analysis can best be used to inform research / development and regulation - in particular Predetermined Change Control Plans (PCCPs) and how it will impact the PMS SI.
If you are interested in this topic, RADIANT-CERSI would like to hear from you.
Please answer the questions below and leave your contact details to continue the conversation.
