As originally asked by Jerrold Shapiro.

In his 2011 book, “The Lean Startup” Eric Ries promotes the concept of rapidly iterating the product until the company fully understands the customer’s needs and fulfills them. His many examples are drawn from the software industry, where the code for a web page can be changed overnight and where the number of people who come to that website each day is so large that hypotheses as to which version of the site, or of a feature on the site, can be statistically tested. Does anyone know if this concept is applicable to medical devices, where changes take a much longer time to create, implement and get approved for human use?

Karan Verma
Head – Business Development, South Asia at GE Healthcare
For several reasons, I’ve preferred reading Steve Blank’s books ( Four Steps to Epiphany, Startup Owner’s Manual) as he highlight’s the philosophy of Lean Start up in a more deeper, meaningful way. My key learnings, which I actively use in my current role as Global Product Manager in healthcare business:
1. State your assumptions – especially related to key success factors – what are your beliefs on why things will go right for the NPI or feature you’re planning
2. Map them on a X (extent of confidence) & Y (impact of it is proven wrong) and ensure you test all assumptions that can adversely impact, even if your confidence is high
3. Keep iterating with customers during the design process – show, ask, see & learn
4. Be willing to try many things – and be willing to pivot

Kirk W. Schmidt
Ra Medical Systems Europe Manager
The wants and needs of customer’s are the core and the reason for manufacturing medical devices. As a legitimate by product of fulfilling them a company becomes successful. Therefor these issues should be clear to the best possible extent before going into sales. Yet, given the strong volatility of customer wants and needs, a soon to come second generation of the same medical device should be in the pipeline, the manufacturer having a finger on the customers pulse. Some manufactures claimed they could rapid prototype within weeks, not months. Fulfilling customer’s wants and needs is an evolution not only a goal. The “the concept of rapidly iterating the product until the company fully understands the customer’s needs and fulfills them” could be understood as rapid flexibility and adjustment to changing wants and needs? That can work…..

Nancy Knettell
Founder and Principal at SoftwareCyber510K
Even if there was not an overarching safety and regulatory process that guides software for medical devices, there is a big difference between small iterations of well understood software and poorly designed software that is hastily thrown over the wall in the guise of being nimble.

While the use of design documentation appears to be a lot of work to those looking into the process, re-writing code over and over again because it is poorly designed from the beginning is a total waste of expensive resources and can add more time than you can think.

It is so easy to ensure you are being nimble when you develop software products. Think in terms of solving problems that people will pay money for. Do your marketing research. Prototype if you have to. Then develop the product to solve those problems. But don’t overburden your first release with a million “I think it should haves” to sell it. Add the features later for an upgrade. It is a big myth that correct software process overburdens software. It is actually the other way around.

Ed D.
QA/CAPA Manager/Design Engineer
The thought is terrifying.

Jerrold Shapiro
President and CEO, Fem-Medical LLC
Virginia, Eric Ries told me how he advised a large medical device company to iterate the design of a large piece of equipment, such as an MRI machine. He advised them to make two versions of the outer case, take both to a trade show of MRI users, put each in a separate room and ask physicians to evaluate each separately. No scanning was being done, and no patients were involved, so this could be done without IRB or IDE clearance. In many cases, you can’t get device performance results without a real device in a person or animal. For some organs and systems, there is no exact animal model, so the work must be done in humans. I think certain design aspects, such as human factors, can be learned quickly whereas others, such as detection of a given disease with a certain degree of specificity and sensitivity, can not be done quickly. I agree with other commenters that much of the work should be done before the product is designed to find out what the users want, need and will pay for. But if the concept for a medical device or test is really new, such as CT or MRI imaging, and requires the physician interpreting the data to learn a whole new way of doing so (such as looking at body cross-sections instead of frontal, lateral or saggital views of the shadows organs cast on an X-ray plate) then the demand may not come from the market. In the 1960s the inventors of CT could not raise money from conventional sources, so had to turn to the Beatles’ company, EMI, for funding. Now CT is considered routine.

Karl Schulmeisters
Principal and Founder at ExStreamVR
As pointed out above, Agile Startup is about not executing on unknowns. Given the cost of clinical trials for either pharma or medical devices, its not very smart to be going to clinical trial without a darned good idea of how well the device will work. Even by the time you get to “First in Man” the Agile/Lean Starttup approach would have you know pretty darned well what you expect to see.

so where would the “rapid innovation” and “Fail fast” apply in a Lean Medical Startup?

Well for one in the usability design. I know of one industrial designer who has patents on both an IV system and an endoscopic system that used “Agile Design” principles in working with the consulting MDs to figure out the Human Factors engineering to both optimize usability and reduce error chances in the usage. Those products are now in the marketplace.

Nor does Agile require necessarily a short delivery horizon. The main part of it is that it requires a process that works iteratively to improve the product and is not unwilling to revisit basic design choices even well into the process IF such a change is driven by a Use Requirement.

Virginia Guest
Regulatory Affairs Consultant
Rapid iteration in medical devices means doing a lot of animal studies. Am I correct about that?

Aroop Kumar Dutta
ExCel Matrix Biological Devices P Ltd.
Yes it does.

Lee Balaklaw MD, MBA
President, Schooley Mitchell of Louisa
This is a fascinating discussion thread. I think that in particular with medical devices fully understanding the customer’s needs and fulfilling them is very important. The question is how much time was spent figuring out what the customer’s needs really are? For medical devices, is the customer the end user? For example, the facility may be the customer actually doing the purchasing and may be more concerned about the price point of the medical device. Whereas, the end user could be a clinician that is trying to meet a clinical need. How broad is the customer base or end user base? Are some of the end users power users or early adopters of technology? How does that translate for the middle market end user, who may not want all of the bells and whistles? Rather than spending a lot on product development, a full understanding of the customer’s need first may save a great deal of development costs. When a medical device is truly new and innovative, has the concept been first validated with research to prove some significant clinical utility, to find out if it really meets or anticipates a customer’s needs? Many medical devices over the years were great ideas, innovative, and met some customer’s needs but were very narrowly targeted in the marketplace. In some cases such innovative ideas medical products met customer needs but did not have solid research data to prove that they improved outcomes. In these cases such products did not succeed commercially.

Biplab Pal
CTO & Co-Founder MachineSense (Prophecy), Data Analytic for IIoT, PhD in ECE
For medical devices, it can never been so “lean”. 80% of the cost goes to beta trial of establishing reliability & reproducibility of the same results on different human beings. Such testing and then converging to FDA approved norm is so expensive, products in the other domain do not need to conform to this level of standard.

Mark Warren
Experienced Manufacturing Professional “Discovering Solutions with People”
Toyota and Deming – he is more of a secondary influence. Toyota was forced to use SPC in 1951 by the US Army as part of a truck contract. SPC was introduced into Japanese industry as part of the CCS Management Training program in 1949 by Homer Sarasohn. When he was leaving, they requested a replacement. Walter Shewhart was the one they commended. He declined, but had a young man he would recommend instead… Deming. The influence of Deming came only after Nissan won the Deming Prize – 1961. By then, TPS was fairly mature.
As for Drucker, I would say he was a great influence, mostly for his “Management by Objectives”, which we know as Hoshin Kanri. On a management philosophy angle, Matsushita should be given more credit of influencing thinking than Drucker.
…back to Eric’s book, the testing of assumptions, validating with experiments, iterative learning process; this is all part of the Job Methods program (JM), one of the three TWI programs that have been woven into lean. We are just rediscovering our what our grandparents learned so many years ago.
I admire Eric’s zeal, and wonder what he might be able to achieve if he knew a little more about the experiments of those that came before him.

Bernhard Kappe
Medical Device Companion Software, Agile in FDA
Mark, thanks for the correction!

Burrell (Bo) Clawson
I research patents & design products to get a patented competitive position: Over 30 patents.
Mark, thanks for the correction. I should have said Toyota developed its processes with help from Deming & Drucker.

No matter what, organized structure to managing and organizing business is needed.

Mark Warren
Experienced Manufacturing Professional “Discovering Solutions with People”
“Bo is absolutely right that TPS was developed from Deming and Drucker.”
This I must disagree with… Taiichi Ohno used the TWI programs (Training Within Industry) to build the foundations of the Toyota Production System. As much as I like Deming and Drucker, this is historically inaccurate.
Ohno was trying to implement flow (these principles were documented about 1910) and the missing piece to make it work was TWI. It is the synthesis of these principles that make up what we know today as lean.

Burrell (Bo) Clawson
I research patents & design products to get a patented competitive position: Over 30 patents.
We are in this forum because of a desire to do innovative work that leapfrogs the competition. So which comes first, innovation or startup? My view is the former.

I do not think I would use a particular “Lean Startup” book or principle to guide what is necessary to do to get the innovation conceived and proven.

That is just me of course. I can’t conceive of starting a company without already having been able to prove out the concept.

Paul M. Stein
Chief Scientist, Inventor, and Entrepreneur – Dedicated to the Treatment of Critical Unmet Medical Needs
Earl Bakken of Medtronic used to push his particular system of a slightly less precisely focused project management, “Ready-Fire-Aim”.

Bernhard Kappe
Medical Device Companion Software, Agile in FDA
Minimally viable product is a tricky concept in Lean Startup. The idea is to prototype features, get feedback, eliminate this that are not absolutely needed, and refine the prototype/get feedback until you get it right. That way, you eliminate unnecessary features and make sure you nail the ones you do have. You can add features as well and get feedback on those. This guards against one of the biggest issues with any new product – feature bloat. Launch is tougher. As mentioned above, early marketing experiments that work outside of a highly regulated space (think dropbox’s concept video used to test market demand) would be a no no if you need FDA clearance or approval.

One point to remember – as Steve Blank says, market type changes everything. If you are launching a product into an existing market, or launching version n+1 of a product, the threshold of what you release is much higher. Fast feedback loops prelaunch are still much better than waterfall, but you’re not dealing with the high uncertainty scenario lean startup is meant to address. Resegmented or new markets are a better match for Lean Startup. In that case, getting a minimal viable product in front of a small number of users/customers and iterating it until they love it (and then going for launch) is a viable strategy. If you can iterate and release earlier and more often than the competition that will invariably arise, you are much more likely to win.

Bo is absolutely right that TPS was developed from Deming and Drucker. In this case, the fast release cycle comes from another Deming inspired idea – Boyd’s OODA loop. Boyd was a Korean war fighter pilot who based his theory on his observation on F-86 Sabre vs. MIG-15 fighter battles.

OODA stands for Observe-Orient-Decide-Act. Boyd suggested that in any dynamic conflict, the opponents went through this continuous cycle as they selected courses of action. You’d look at what was going on, figure out what it meant, decide how to respond to it, and then actually respond. Then you’d observe the results of your action, beginning the cycle again.

The upshot of this is that if you can cycle through the OODA Loop faster than your opponent, with each cycle his reactions will be less and less appropriate to the situation at hand. Eventually he’ll be so out of phase with what he should be doing that it is easy to beat him.

Burrell (Bo) Clawson
I research patents & design products to get a patented competitive position: Over 30 patents.
Douglas Stinson accurately noted “Apple was never first with anything.”

And that is true with most if not all products I’ve worked on.

There have been a few times I have conceived a small product no one has ever done before. Those I remember as being unique in that I had to really carefully find out how customers would think of it and handle it and use the product in all manner of ways. There is no “program” or plan for how to develop an entirely new product the world has never seen before.

Lots of people have to give input when you run into a truly unique new product. There will be people who misuse, misconstrue and be totally confounded if you give them a unique new product and you have things to learn from all of them. It is user-centric design.

Scott Collins, PhD
CEO at StartupAdvisorNetwork.com
Bottom line, lean startup philosophies can be very helpful for certain parts of certain products in the medical industry and should be considered. I often break down a new product design into its component parts and apply principles that make sense for each. For example, lean startup techniques can work well in prototype development prior to a fully vetted, FDA approved product. This can hold true for software, hardware, and technique development. Once the market has accepted a given prototype, the formal development process continues. Hans Ludi mentioned earlier in this chain that 2/3 products fail due to a lack of understanding the market. Tag to this the often long development and regulatory approval processes in our field and you will understand that getting the design/prototype correct is of paramount importance. Pharma companies have putting millions into techniques that allow them to fail fast and thus reduce costly clinical trials necessary to get their products to market. Devices, including software, are being increasingly regulated by the FDA. Whenever reasonable, understand your market and product and take steps to assure your design meets this market before taking the time and money necessary to meet these regulatory challenges.