Best practices in testing of AI based SaMD

Today’s world of medical devices uses a lot of digital technology such as IoT, AI. These technologies help them become smarter due to the ability to improvise and enhance based on the specific data. However, along with the benefits, technology has introduced new risks as well. The AI models derive algorithm parameters by learning from training data. The accuracy and generality of the algorithm are governed by training data set factors, such as patients’ age, gender, ethnicity, etc. Any bias in the data will be propagated to the algorithm, impacting the outcome under real-world clinical situations. Therefore, good practices must be implemented and adopted when developing such medical solutions in order to ensure patient safety.

Key Takeaways :

  1. Introduction to AI based SaMDs
  2. Appreciating the importance of software development and QMS best practices to get SaMD certified
  3. Understanding the challenges in the SaMD world with the advent of AI
  4. Best practices and tools used in testing AI based SaMD
  5. Normal pitfalls in testing AI based SaMD.
Anupama Ananthasairam Anupama Ananthasairam