Artificial Intelligence/Machine Learning (AI/ML)-Based.

(SaMD) Action Plan – Summary

FDA New Perspectives

Predetermined Change SaMD Pre-Specifications (SPS): "what" aspects the manufacturer intends
Control Plan to change through learning.
Algorithm Change Protocol (ACP): "how" the algorithm will learn and
change while remaining safe and effective.

SPS/ACP should be Predetermined and disclosed in Premarket

notification, Changes Made on SPS/ACP should not fall under any of the
below Criteria that will require a new premarket submission:
(as per Link )
• A change that introduces a new risk or modifies an existing
risk that could result in significant harm;
• A change to risk controls to prevent significant harm; and
• A change that significantly affects clinical functionality or
performance specifications of the device.

Good Machine Learning FDA Will be releasing harmonization of Good Machine Learning Practice
Practice (GMLP) development in future.
Some standards FDA currently approve and Participates in development:
• Institute of Electrical and Electronics Engineers (IEEE) P2801
Artificial Intelligence Medical Device Working Group
• the International Organization for Standardization/ Joint
Technical Committee 1/ Subcommittee 42 (ISO/ IEC JTC 1/SC 42)
• (AAMI)/ British Standards Institution (BSI) Initiative on AI in
medical technology.
• Xavier AI World Consortium Collaborative Community and the
Pathology Innovation Collaborative Community
• International Medical Device Regulators Forum (IMDRF)

Patient-Centered FDA promotes Manufacturers to be more transparent of these devices to

Approach (Focused on users, and to patients more broadly, about the functioning of AI/ML-
Labelling) based devices to ensure that users understand the benefits, risks, and
limitations of these devices.

Also, to clearly describe the data that were used to train the algorithm,
the relevance of its inputs, the logic it employs the role intended to be
served by its output, and the evidence of the device’s performance in the
Product Label (when possible – Not Mandatory, FDA will release SAMD
labeling guidelines in future)

Regulatory Science In Summary FDA encourages methods that would identify and eliminate
Methods Related to bias and improve robustness and resilience of these algorithms to
Algorithm Bias & withstand changing clinical inputs and conditions.
Robustness
Also Expects a racially and ethnically diverse intended patient population
to train the algorithms to eliminate bias.
 Real-World Performance FDA will support the piloting of real-world performance monitoring by
(RWP) working with stakeholders on a voluntary basis (for AI/ML-based SaMD)

Source : https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-
and-machine-learning-software-medical-device