We’ve all gotten used to medical devices that make diagnostics and treatment more efficient, from simple glucometers to sophisticated X-ray machines. But what do we know about how all these wonderful devices were developed? I certainly didn’t know much.
I’ve had the privilege of talking to Dmitry Parshkov, a system architect, serial entrepreneur, investor and owner of Parshkov, Inc. who shared his views on system engineering approaches, medical devices development and the most recent trends in this rather conservative industry.
What does the hardware development process look like in your company?
Well, we can briefly go through the entire process, from determining the concept to its realization and market implementation.
We usually kick off projects with teams that have already gone through the preliminary stage of basic opportunity identification. That is, they have identified some problem that needs a solution and qualified it as a real one by doing some research and/or testing. Our task then is to find an optimal solution to the client’s problem and help them implement it.
The first stage of the development process is eliciting the project requirements. To do so, we start with identifying the stakeholders as well as their needs and constraints. For example, for a medical device development project, we have to consider the requirements of the customer, the end users and the certification agency.
After determining all the needs and constraints, we formalize them and turn them into a set of technical requirements and metrics. For example, if the customer says she wants us to design a fast red car, we translate it into formal requirements of color = rgb(255, 0, 0) and speed.max = 220 mph. On that basis, we can now draft a few solution concepts and thoroughly inspect them to identify the optimal one. A great tool for concept benchmarking is the House of Quality matrix that enables precise assessment of possible solutions. The proof of concept stage also includes various tests of the most viable variants.
Next comes the stage of system architecture where we design the entire system based on the chosen concept solution. This includes bringing the industrial design in compliance with the sustainability requirements—that is, we need to consider the full lifecycle of each detail up to its utilization. After that, we develop a prototype, test it, demonstrate it to the customer, and make amendments, if needed. Finally, we’re ready to develop a production sample along with full documentation that can subsequently be used for mass production.
The basic process outline is more or less the same both for medical and non-medical devices. However, when we’re talking about non-medical devices, the risks generally boil down to financial losses, while in the medical sphere, the stakes are much higher and so is our responsibility as developers.
How is medical device design different from other hardware development?
The key thing to know about the medical devices market is that it’s strictly regulated by the authorities (for example, in the US, the regulatory body is FDA.) Since we’re dealing with human health and life, the responsibility of a medical device developer is immense. Not only the device itself, but the entire process of development must comply with the regulations.
So one cannot simply decide to develop a drug or a device that would enable 3x cheaper medical treatment. In many cases, market players do not need such advancements, especially if this concerns some rare diseases. Each technology used in the development process must be well-known and time-tested—otherwise, you’ll need to spend lots of time and money to justify its usage and prove safety.
For example, if I need to choose a display or a button for my future medical device, I cannot just pick up the most recent cutting-edge solution available in the market. We have to rely on predicate devices (PDs), i.e. tested solutions already approved by the regulators. Each innovation you use will make the whole process more complicated. So, we can say that an engineer at Apple and a developer of a medical device virtually have two different mindsets.
Another common problem in designing medical devices is managing multiple stakeholders’ requirements and constraints. Apart from the direct customer, we have to deal with a regulatory body (FDA), end users (patients) and insurance companies. Bringing together many stakeholders who speak different languages, profession-wise, and finding the middle ground is a challenging task.
We cannot make a medical doctor get into the intricacies of designing a complex system, nor can we make an engineer to be fully competent in business processes. Everyone should be doing their own job—but we need to make sure we all have a common basic understanding of the project. This is one of the primary tasks of a system architect who should be able to hold multiple project variables simultaneously and join parts of the system in a way that enables the desired outcome.
Could you elaborate a bit more on the certification process of medical devices?
From a formal standpoint, we have a number of standards to comply with, such as ISO 15288, a general standard codifying systems and software engineering processes, as well as specialized medical device standards: ISO 13485 (Europe) and CFR 21 (the US).
In the US, medical devices are classified into three categories by the potential damage they can inflict. For example, medical gloves fall into FDA Class I as more or less safe, surgical needles are Class II since they can be dangerous, and some complex invasive accessories will be qualified as Class III. So, when you’re developing a medical device, you need to keep in mind these classes and choose solutions that have already been tested and proved safe.
For instance, if you’re designing a surgical robot hand, it can be explained in terms of predicate devices: a manipulator, a scalpel, etc., which makes it a combination of well-known solutions. But if you’re developing something brand new and unique, like the da Vinci surgical system, this will be qualified as a de novo device, and you’ll most likely have a hard time explaining to the regulator why it will be useful and won’t do any harm.
That’s why designing a complex medical device or a system might take years—apart from the tricky process of development which has to be thoroughly documented in the design index file, it has to undergo multiple laboratory and clinical tests before it can be certified for mass usage.
On average, the development of an FDA-approved device costs $28M, depending on the class. Quite often, the design itself turns out cheaper than the certification process. For example, if you’re required to test your device on 50 cadavers (which is the usual requirement), the expenses might amount to $5M.
So, designing a medical device is extremely responsible, time-consuming and expensive. The sooner you start to work with an FDA agent, the more undesirable outcomes you can avoid.
What are the current trends in the medical devices market?
One could assume that the medical device trends are in line with the general technological trends such as AI, blockchain, etc. However, in reality, it is different. Medical device engineers are not so well-familiar with cutting-edge technology as their counterparts in other industries—it’s a whole different market. Even such giants as Siemens have to be conservative when it comes to designing medical devices.
Depending on the country, all innovations in the medical field are regulated to a certain degree. In the US, one of the major players in the market is insurance companies. They are very influential when it comes to implementing new medical technology.
Still, I cannot say there’re no innovations at all—they just get implemented at a slower pace. For instance, our team has developed a portable EMScanner that can be installed in an ambulance car to diagnose strokes. Just a couple of years ago you couldn’t certify a device that uses machine learning (ML) algorithms—it was simply impossible since we do not understand what’s going on inside the black box.
But today, the situation is slowly changing. We already have the first predicates using ML, and there seems to be an emerging trend. However, one of the major tasks ML technology is “allowed” to solve in medicine is prioritizing rather than decision-making which is up to humans.
For example, if a radiologist has to study 100 X-rays, they would normally do this in the order they were delivered. However, in some cases, an urgent diagnosis is needed, and even one hour can be critical to saving someone’s life. An ML system can handle the primary processing of X-ray photos and detect the ones that require urgent action. Such solutions are already being successfully used.
Another prominent trend is, of course, telemedicine, which enables not only differential diagnostics but also operating patients at a distance. Telemedicine makes it possible to provide qualified medical services to mobility-impaired patients or patients in remote areas. Especially during the pandemics, this technology has shown its huge potential.
To sum up, even though the medical technology market is evolving at a slower pace as compared to other industries, innovations inevitably find their way. And there’s more to come!