With emerging technologies which can scan body parts to detect certain cancers and many other applications, Artificial Intelligence (AI) is becoming increasingly valuable in the medical device and healthcare space.

We asked Senior Design Development Engineer, John, about some of the basics of AI, what’s holding AI back in the medical sector and how AI might impact the role of design engineers in the future:

What’s the difference between AI and connected devices?

People often confuse AI and connected devices (or the IoMT) however, it’s important to know the differences and associated implications for medical device development.

Connected Devices: The device or endpoint can monitor specific data on a patient and send that data to a retriever, such as the cloud, to be downloaded by a clinician who can then monitor the results and make decisions on a therapy or treatment.

AI / Machine Learning: Instead of sending the data to a clinician to process, it goes via an algorithm that can either process it fully (which the regulations don’t currently allow) or, partially so that it lightens the physicians load by making suggestions i.e. *This is potentially a cyst – review this patient*.

Why don’t medical device regulations currently allow full AI processing?

Regulations don’t currently allow full AI in medical devices because of the risk involved however, many argue that a AI is better or more reliable than a human resource in some ways because AI doesn’t suffer in the same ways that humans do i.e they don’t come to work a bit hungover, tired, or in a bad mood following an argument with their partner…

Although humans currently have higher cognitive and mental processing ability, computers are unbiased and by now, they aren’t far away from human ability. Actually, AI is set to outperform humans by 2040.

Currently, the law in the UK dictates that any data coming from a medical device has to be reviewed by a human. So, even though we could design a very simple device that incorporates AI to interpret a standard urine test for example, the regulations would not allow it.

What is holding AI back?

Regulations are slow whereas the tech and med tech industries are incredibly fast paced and ever developing new technology and systems therefore, regulators cannot catch up. So, although AI as a technology is ready to go for a number of medical applications, the regulations won’t allow it. Regulations are already years behind whereas technology continues to advance.

Interoperability is another issue we’re currently facing. We have all of these different systems, software’s, programmes and bits of code globally, but they are inoperable with each other. They have all been designed in isolation therefore, there is no fluid language which makes for a big technical hold back when it comes to AI. For example, a heart scanner may be gathering data however, a different system to the one its being used with may not be able to interpret that data without it being re-written.

Do you think that regulators will come to a compromise before they are forced to?

The main factor that will drive this is cost. Regulators receive a lot of pressure from the governments who fund them therefore, if there’s a significant cost benefit associated with the change, this will often influence them to speed things up.

Risk is also a big factor not only for regulators but also for users. People’s views on new technologies often take some time to adapt due to lacking trust for example, when someone goes for a specific procedure like a scan, they expect a certain protocol and for the data to be reviewed by a doctor. If a computer has reviewed it, they may lack trust and request a second opinion from a human regardless.

Do you think it’ll take a while for that shift in perspective toward AI from a patient point of view?

I think people will continue to welcome developments in AI very gradually.

If you reflect on 50 years ago, people would be very sceptical of wearing technology such as a smart watch to measure parameters such as heart rate or even driverless vehicles, whereas now, people are not only fine with it but, actively pursuing it.

Gradually, people will become more comfortable with AI. It’s already happening and is happening fast in other less-regulated sectors. The wider population are yet to see the true benefits of AI in the medical device and healthcare space.

Do you feel that AI could threaten your role as a design engineer in the future? Will it be a help or hindrance?

We can’t be complacent – we must accept that AI is going to be a part of our day to day in some form and embrace it. It already is to some degree; we heavily rely on CAD and FEA whereas in the past, you would require advanced mechanical engineers to quickly calculate elements such as stresses and strains.

I think if we’re complacent and we don’t get more comfortable with AI as well as continue to build expertise in the area, not just in medical but in all sectors, we will inevitably get left behind. These days, basic CAD packages can easily be downloaded by anyone.

I think we should be mindful of continuously upskilling ourselves in avenues such as risk assessment or creative thinking – activities that are more difficult for a computer to do (at least in the shorter term).

John Harverson 21 October 2021

Share

Get in Touch with John Harverson

Senior Design Development Engineer

Holding a Bachelor’s degree in Industrial and Product Design as well as certificates in Project Management and UX Design, John is an advocate of user-centred design and finding innovative solutions within regulated fields. With over a decade’s experience in the medical device industry from working as a Consultant helping companies develop innovative medical devices, to head of an Industrial Design company developing imaging equipment for early detection of breast cancer, John also manages our ISO 13485 & ISO 9001 QMS.

Get In Touch

Our latest insights

Latest Posts

Sustainable Medical Device Development: Legislation, Directives, Regulation & Standards

Read More...

How our Innovative Problem Solving Strategies can help you

Read More...

Share