Tips for safe and effective medical devices

In 2021, the FDA, in collaboration with Health Canada and the UK Medicines and Health Products Regulatory Agency, published 10 guiding principles to encourage the development of “good machine learning practices”. The intention is to promote “safe, effective and high-quality medical devices” that use AI and ML.

This is welcome advice for healthcare systems now relying on AI/ML medical devices, as well as digital health startups and medtech companies looking to develop their own devices. For them, and others interested in entering this field, a closer look at the guidelines provided by the FDA could help start or further enrich the journey.

The 10 FDA guidelines cover a variety of issues, but many of them revolve around the model itself, requirements related to cybersecurity and risk reduction, and the need to involve multiple people and disciplines. in the development and maintenance of medical devices that rely on AI. and ML.

DISCOVER: 5 Steps to Secure Medical Internet of Things Devices.

Get started with the model for AI-supported medical devices

AI/ML medical devices are based on a model – a program or algorithm formed by exposure to large amounts of data. The model makes predictions based on the data. If correct, the model learns and is reinforced; if incorrect, it adjusts to increase its accuracy. This learning process takes time, requiring millions or billions of permutations and adjustments. The more the model is trained, the more successful it is in establishing correct diagnoses and reducing the number of incorrect diagnoses.

FDA guidelines help ensure that the algorithms selected for the model are those that best suit the characteristics of the data and that the parameters are changed to produce the expected results.

In addition, the FDA recommends that the data collected for the model be representative of the intended patient population, that the model reference clinically relevant data, and that the samples be large enough – and of sufficient quality – to allow experts to get an overview of the data. All this requires the active participation of various actors and experts who can ensure that the model is sufficiently robust and useful.

Multidisciplinary expertise: Because medical devices have a wide variety of users and targets, it is important for the development team to involve a variety of actors with relevant expertise. These experts can help develop a full understanding of how the device will fit into the clinical workflow and highlight potential issues early in the design process where modifications are less costly. It is equally important to fully understand all the risks associated with patients, in order to ensure that the smart medical devices being built are safe and effective. Without the expertise of a multidisciplinary team, developers risk missing or misunderstanding some of the desired benefits and potential risks.

Cyber ​​security: FDA guidelines highlight the importance of implementing robust cybersecurity practices. He advises paying attention to the “fundamentals,” including basic software engineering practices, robust data management practices, and careful attention to cybersecurity throughout the design and development process. Testing must demonstrate device performance under clinically relevant conditions, which requires statistically sound test plans in addition to ensuring data quality is integrated and tested. The model design should also support the active mitigation of known risks. Authenticity and integrity of data must be guaranteed, not only during the design process, but also during device deployment. Real-world monitoring can improve safety and performance and reduce bias and risk.

RELATED: Learn how AI is helping healthcare organizations reduce preventable patient harm.

Implementing AI in healthcare is a laudable journey

Implementing AI/ML in healthcare is not easy. It requires a significant investment and requires the collaboration of a large number of players. All members of the multidisciplinary team should gain an understanding of the model, the results, and their potential implications. Also, as the model learns, it will necessarily change, which may require additional training of healthcare workers and patients using the devices. But the result – better monitoring of patients and clinical outcomes – will be worth it.