[TECH LETTER] Medical Device Software incorporating Artificial Intelligence: Generating sufficient evidence under the MDR

GET THE TECH LETTER

 

Artificial Intelligence (AI) and Machine Learning (ML) technologies have the potential to transform medicine by aiding in the detection, diagnosis, and management of diseases. As digitalization of healthcare generates massive amounts of data, medical device manufacturers are increasingly incorporating AI technologies to automate the analysis of such data targeting to create innovative products and improve patient care. This turn towards AI-enabled medical device software (MDSW) is also evidenced by the plethora of studies evaluating the feasibility of artificial intelligence systems across a wide range of health-related indications.

 

While the interest in medical applications of AI is strong, inconsistent and incomplete collection of evidence remains one of the barriers to the assessment of the safety and performance of AI-MDSW by regulatory bodies.

According to the provisions of Article 61(1) of the MDR EU 2017/745, it is the responsibility of the manufacturer to specify and justify the level of Clinical Evidence necessary to demonstrate conformity of their medical device to the relevant General Safety and Performance Requirements (GSPRs); this level of clinical evidence should be appropriate in view of the device characteristics and intended purpose.

 

Determining the appropriate level of evidence might be challenging, especially in the case of AI-enabled MDSW which significantly differs from established medical device software in terms of technical and clinical aspects. At the same time, there is no explicit regulatory guidance for conformity assessment of AI technologies, delineating appropriate and practical evidence generation approaches.

 

Accordingly, this Technical Letter aims to provide an overview of the considerations for evaluating evidence regarding AI-MDSW.

 

GET THE TECH LETTER