Designing for MedTech

By Abhi Ghavalkar, Co-founder of Third Space Therapy

Note: This is an edited transcript from the MedTech Summit talk Abhi presented on July 8, 2025.

Just a quick background about myself. I was born and raised in Oman in the Middle East. I moved to the UK when I was 17 for school. Two weeks into school I decided I wanted to be a professional soccer player, so I dropped out of school and tried pursuing that professionally for a year, getting injured, retiring early as I like to say. And then getting my bachelor's in product design engineering. I went to school at Loughborough in the UK. My degree was a mix of mechanical and manufacturing engineering with bits of industrial design. I worked in aerospace for a year with GKN in Bristol, before graduating in 2019 with my bachelor's. And I did my master's in design at UC Berkeley, graduated in 2022. I'm currently the CEO and co-founder at Third Space. We provide mental health services for patients on Medicaid across the US.

My path to becoming an entrepreneur was very unfortunate, I like to say, but I'm glad we made it there. I graduated in 2019 with my bachelor's right in the middle of Brexit in the UK when no one wanted to sponsor an international student's visa to remain in the country. Being born in Oman, I didn't have birthright citizenship. An Indian passport by descent, but I had no family there because we all lived in Oman. So I wasn't able to go back to the country where I was born. I wasn't able to go back to the country where I held a passport from. So I ended up having to file a human rights petition to stay back in the UK almost as a refugee because I had nowhere else to go, and when no one else was willing to sponsor a job for me to stay back in the country, I said, screw it, I'll go make my own job. And I became a founder.

So that's all I've done ever since. I've only worked on healthcare ventures and the common thread across all my companies has been to increase access to care for people who need it most. I've worked on healthcare software services and also hardware. I'll walk through an example of the design process for a medical device from my company Prana.

The COVID Ventilator Crisis

So as you might remember, during COVID there was a huge ventilator shortage worldwide. Doctors were forced to turn patients away because they didn't have enough vents. And they had to make really difficult decisions about who gets access to a ventilator.

We'd found an academic paper that had proven that you could split a hospital ventilator four ways using simple parts you find lying around in a hospital. This was done by two MDs several years ago. But they had some limitations around being able to control individualized airflow, preventing cross-contamination and monitoring multiple patients on the same device.

So I thought, okay, using my design and engineering background, it's probably more of a design challenge to solve rather than a medical challenge because that's already been taken care of by the MDs prior.

So we went and designed a few CAD models. I mean, these are not out of Call of Duty. These are CAD models of the ventilator splitter that we thought would be applicable to that scenario where we could control individualized airflow using valves, filters and things in place to prevent cross-contamination. And we essentially sent these prototypes along with our design plans to universities across the UK. So we reached out to Oxford, Cambridge, Imperial, Kings, UCL, and several others as well. And nobody really wanted to help us out at the time, either because they were too busy with their own efforts or they didn't think it was going to work.

From Plumbing Parts to FDA Approval

But we didn't want to take that no as an answer. So my co-founder went to a plumbing shop and for less than a hundred bucks, we found some plumbing parts that acted as connectors. We used balloons to simulate lungs, we connected this device to a vacuum cleaner and reversed it so that it was blowing out air instead of sucking in air, to act as a ventilator. And with that, we were able to prove that you could control individualized airflow across each branch. And we sent a video of the balloon and plumbing part prototype to the FDA. They replied positively saying, "Hey, if you do this with medical grade parts in a clinical setting, we could potentially grant you emergency use authorization."

And I was like, America, I have to move to America. This is the place to be. We used some of that feedback we got from that initial iteration and created what we thought would be the next version using the medical grade parts that they'd want us to use. And along with the screenshot of the FDA email, we were given a grant by UC Berkeley to actually go and work on this idea for a little while.

Pivoting to COPD

So that grant was to go and develop a workable product that we could go and test with some clinical partners in the city. And we knew that the ventilator shortage due to COVID was being addressed by a lot of aerospace and automotive manufacturers. So we thought of also COVID proofing the idea.

So moving away from acute respiratory distress syndrome, which is what we found as a symptom of COVID, we moved towards a condition called COPD or chronic obstructive pulmonary disease. It's caused by smoke inhalation, typically by inhaling smoke due to pollution or by cooking with solid fuel sources in low resource countries. It's the third leading cause of death in the world, and many deaths in the lower and middle income countries are caused by COPD. The other thing was trying to design for the COVID use case would mean that we would only use our product for an emergency use in the US. So we thought, where in the world would every day be emergency use? And the answer was, it was in Africa.

So through our network, we got connected to doctors and respiratory therapists in Ghana and Uganda. We worked with Dr. Alfred Aidoo at Komfo Anokye Teaching Hospital, and they were turning away patients every day because they didn't have enough ventilators to support COPD exacerbations.

And they basically said, "Let's save lives, my friends, every life matters." And what we learned at that time was that there were fewer than 2000 working ventilators across 41 countries across Africa compared to more than 170,000 in the US alone. And this is as of a New York Times article in 2020.

Research and Development

We got admitted into the NSF I-Corps program at Berkeley and we ended up speaking to several experts in pretty much most continents. It was a mix of clinicians, engineers, respiratory therapists and people in the supply chain just to figure out how we could go and address the shortage of vents in the African continent.

And we estimated that Africa as a whole would need anywhere between 700,000 to 1.4 million vents compared to the very small number they had at that time. And the problems were very clear. Hospitals had excess demand, but not enough capacity. They were overwhelmed with patients who needed access to vents, but they didn't have enough vents. Buying new ventilators was very expensive for hospitals in Africa, so they typically relied on donations from the western world. But these were ventilator devices that were at the end of their shelf life. And it costs a lot to admit a patient once they have the exacerbation, but when they turn them away and they come back with a critical emergency use, instead, that's even more expensive for the hospital to deal with.

So they wanted an intermediary solution to help on the preventative side. So with that, we went into the engineering phases of the product. We designed what our pneumatic circuit would look like. We knew we would have an inlet from the original CPAP device. We'll connect that to our splitter device and then the patient would be at the other end. I don't want to go too much into the engineering aspect, it's just showing you how the process gets refined over time. We essentially created the circuit, found parts that met the specifications for each part in the circuit, and tried to do that in a very capital efficient manner.

Clinical Testing and Partnerships

And it was at this time we formed partnerships with Stanford Children's Hospital and UCSF Health. UCSF actually invited us into the respiratory therapy department. We 3D printed some splitter valves and then tested it with simulator lungs and a CPAP device at their facility.

So it was encouraging to see how we'd moved away from that plumbing prototype to the UCSF clinic at this stage. We had a team in the UK supporting the work we were doing as well, and we were also testing the flow controls and pressure controls across the different limbs of the ventilator splitter device. And then we went even deeper into the details of individual components. So there were some sensors that we couldn't really change the specs for because we bought them off the shelf and then met all the clinical compliance standards.

But for the ones we designed, we said we could be even more efficient in terms of how we design it, both in terms of functionality and also cost. So we changed the way we designed our splitter from this kind of design to much more 3D printer friendly version of the design, with these wedges around the end, like these lips or chamfers that you could attach to existing ventilator parts very easily as well. And then you go further, I mean, I don't want to go too much into detail here with the biocompatibility and running simulations on COMSOL to look at what the flow and turbulence would look like through each of these valves and these pipes.

Digital Design and User Interface

We also looked at the digital design. So we knew that you couldn't monitor four patients using the existing CPAP setup. And we knew we had to monitor both pressure and flow and also have alarms for any errors or disconnections, if patients were connected to the wrong line or if they got disconnected accidentally. So, with that spec in mind, we had a little Arduino prototype here with a digital screen that we added. The next step was to design a Figma prototype of what the UI UX dashboard would look like.

And then we did the same on the industrial design side. We knew we had three high level requirements to meet. One was having a quick setup time because it was going to be used in mostly crisis situations every single day. It had to be compact, so it could be transported easily. And you had to have a large enough dashboard screen size so that you could read what the measurements were across all four patients.

So we designed the enclosure so it could also clip onto existing machines, and parts could be replaced easily for increasing life and service. And then we reconfigured our assembly of the pneumatic circuit so it fit flush within the enclosure itself.

Product Evolution

So that led to this beautiful looking version of the ventilator splitter device. We were awarded the World Changing Ideas Award by Fast Company in 2021 for that particular version. And this is us testing out that version in my bedroom at Berkeley. And I was living with three ventilators who were my roommates that Stanford and UCSF had donated to the project. So we had simulator lungs down here, like these little balloons you see. We connected them to the CPAP and then all the parts that we sourced and designed ourselves.

And this is what the final pneumatic circuit looked like when we took it into testing. From there we had one more iteration to make it more friendly in terms of usability from a respiratory therapist perspective because we wanted to minimize the time it would take them to get trained to use a device like this. So we had an inlet on the side instead of behind the device. We had colored strips for quick identification. So you knew which patient was attached to which line. We added a few more details to a dashboard itself. Then this is what a photo realistic rendering looks like of the device attached to an existing ventilator.

So this was the evolution of the product. So it began with a simple splitter design connected to a CPAP and simulator lungs. And we didn't really have any more custom parts, so we used rubber connectors to keep things in place, followed by going into the more pneumatic circuit that was close to the product being in implementation. And then finally we attached a bunch of sensors to monitor airflow and pressure, that were all connected to this Arduino circuit that we'd programmed ourselves.

It got funny to the end when we ran out of the grant funding and we had not enough money to go and build an enclosure. So I just cut open an ASOS shoebox and shoved everything inside it so we could go into testing. It seemed to work. We managed to hit all the specification requirements we wanted for flow rate control and pressure control. And it was great to see how far we'd come from the plumbing prototype in the UK about a year and a half ago to where we were with the ASOS shoebox prototype.

Still scrappy, but it was several generations ahead. We received a bunch of awards and accolades for the work we'd done with that project.

The Design Process

But the general theme here is one that you see across most design projects. In general, what you see is you always start with the design challenge or a problem, and you go through this process of diverging and converging as you get more and more focused and refined in scope and solution over time. Design always starts off as a very messy process. You go back in circles quite a bit. You go up and down, you go left and right, go in and out. And as the process goes on, you end up having more refinement and clarity.

You always start with a problem. You want to know what you're solving and from there, you end up designing around that.

Understanding What the Customer Wants

So there's many different ways to make a hole in a block of wood. And what I was trying to get at is at the end of the day, to your customer, it doesn't really matter. They're there to get a job done.

As long as there is a hole in that block of wood, they'll be happy. It doesn't matter if you use a drill bit, a hammer and a chisel, a nail, or a woodpecker to get that hole.

What you want to think about from a customer's perspective is when the customer uses your product to get a job done, they feel like they're doing something awesome and they've become empowered by using something you've created.

So you basically want to put yourself in the shoes of your customer to figure out what they might want to get done, and then design the product accordingly. Rather than saying, I think this is what the customer wants, so this is what I'm going to give them. And that's basically what the whole design process is about.

So trying to distill that insight to the point where you have something that the customer will actually pay for and use consistently. And a great way to do that is this storyboarding process that Walt Disney pioneered. It's a really good way to get in the shoes of the customer and understand what happens at every touch point with your user journey.

Customer Journey Mapping

There's a tool called the Customer Journey map that I like to use a lot. Go through all the different actions that a user takes with your product or your service, what thoughts they go through, what emotions they feel, what the points of friction are and the opportunities are in response to those pain points that you find.

So if you take Uber as a case study, and you look at what life was like before Uber, there's someone who's at a bar with some friends. They leave the bar to go back home. They take their phone out, they call taxi company A, there's no taxis available. They call taxi company B. They say they have a 45 minute wait. They don't want to wait that long. They call taxi company C, there's a 10 minute wait. They confirm the location and they agree to book that taxi. It takes 15 minutes instead of 10. So they're a little frustrated. They get back home, the driver asks them for a cash payment, they realize they don't have cash. So they go to an ATM, they withdraw cash, they pay the driver, and they go home.

And when you look at what the customer's thinking at every step, it begins with, the weather could be miserable. Like, oh my God, it's cold, my fingers are freezing. Why is this guy not picking up? Why the hell are there no taxis available with this company? Is there a 45 minute wait? Okay, I go with this one. But why is it taking so long? Oh my God, this is so embarrassing. I don't have cash. He only wants cash as a form of payment. I just want to go home and go to bed.

Now, when you look at how the customer feels, when they're going through those thoughts, the end goal of the customer is to arrive at home, get from A to B using transportation to go home. And that only happens after going through about 16 different steps before they finally get that job done. And throughout their entire journey, their experience is just miserable. And what you want to try and do is minimize the time in the red and maximize the time in the green in terms of how the customer feels at every touch point with your service or your product.

So the pain points from that journey were, number one, waiting outside for an unknown duration of time in an uncomfortable environment, calling multiple companies individually, not knowing about trust and safety, long wait times, lack of confidence, directions manually, sometimes having to carry cash, and then just uncomfortable interactions overall.

But with all the pain points that a customer experiences, there's also many opportunities that arise. Have the opportunity to book something from a mobile phone instead of having to talk to someone on the phone. You can do a centralized search instead of having to call several people individually. Instead of in the cab of a stranger, you can essentially have identification and authorization done at scale as well. Essentially what happens is journey that looks like this gets condensed.

Something that looks like this with much fewer steps: you take your phone out, you enter the destination, confirm the pickup location, payments completed within the app itself, get track your taxi and you have a very accurate estimated time of arrival as well. And you get home and you've taken out several steps in that journey by looking very closely at what the customer feels and thinks at every single step.

That's essentially what we want to be doing when we're designing any interaction with our users, whether it's working with patients, with providers, or other stakeholders in the med tech space as well. And at every step, what you're trying to do is uncover insights and validate assumptions. And one of the best ways of doing this is through hypothesis testing. And you want to try and be as specific and as quantifiable as you can with the hypothesis that you are testing.

Hypothesis Testing

So instead of saying, hey, I'm going to go and test whether I need water to swim. That's not a very good hypothesis. But if you go and try and say, CFOs at public hospitals need to make a 200% margin on overnight stays to pay their costs. That's something you can go and validate through conversations you have with CFOs at public hospitals with the data they share with you. That's how you want to design the conversations around understanding what that customer journey looks like.

The idea of going through this exercise was as you speak to more and more people, you get more pieces of the puzzle and the big picture becomes clearer essentially with time, right?

What you don't want to do is fall into the trap of feature-itis where you've got a hundred insights from a hundred different people and you want to become all things to everyone, which is not the way to go.

I prefer the paperclip approach, which I think is one of the best designs of all time. It's a simple piece of metal wire that was bent three times and has so much functionality, and that's how you want to be.

Crafting the Perfect Experience

And once you have an understanding of what the journey looks like, what the pain points are, what the opportunities are, the final step becomes crafting the ideal experience for that customer, or for your user.

There was a podcast with Reid Hoffman called Masters of Scale, and on the very first episode, he had Airbnb's CEO, Brian Chesky on it, and he talked about handcrafting the perfect experience for your users. So doing things that don't scale before you actually start thinking about scale.

So Brian essentially goes through this exercise, or he went through this exercise in the early days of Airbnb saying, what can we do to surprise our guests at Airbnb? So what would it take for Airbnb to design something for their customer that would make them tell literally every single person they've ever met about Airbnb? So how do you delight the user to the point where they can't stop talking about your product?

So it goes through what a check-in experience at Airbnb looks like, and for that check-in experience itself, what would a one star experience look like on Airbnb? So imagine you've booked an Airbnb, you show up at this Airbnb, knock on the door, no one answers. You wait 10 more minutes, you ring the doorbell, you knock on the door again. No one answers. And you realize the host has stood you up. He hasn't given you information on how to access the apartment if he's not in. And you basically give them a one star review saying, I'm never using Airbnb again. I want my money back.

A three star experience would be if you get to the Airbnb, knock on the door, there's a 15, 20 minute delay. But the host does let you in. At least you got in, but it wasn't the best check-in experience.

A five star check-in experience would be if knock on the door, the Airbnb host immediately lets you in and welcomes you into their apartment. It's like, yep, this is what Airbnb was designed for. It worked. No big deal. And that's as high as star ratings go on Airbnb. It's a five star experience at most.

But then you go one step further and you think, what would a six star experience look like on Airbnb? You do the basic, table stakes are you get into the apartment. What happens afterwards? The host treats you personally. They have a welcome gift on the dining table for you. They've got you a bottle of wine or a bottle of champagne, and they have all your favorite amenities stocked up in the bathroom. And you're like, wow, this is way more personalized than a hotel. I feel so cared for and thought about. So I'm definitely going to use Airbnb again. It was much better than I expected. That's a six star experience.

Now what would that look like if it was a seven star experience? The host welcomes you. They've cooked you a dinner based on your favorite cuisine because they know that from your Airbnb profile, they've got you some local trinkets based on your interest as well. They know you like surfing. So they've rented a surfboard and kept that in the apartment for you saying, "Hey, we've got this for you. Don't worry about it." They've made reservations to a hard to get restaurant in the city for you. They've driven you around in their car, given you a tour of the city themselves and then taken you out to dinner. So this is way beyond what you would experience at normal Airbnb. And this could be classed almost as a seven star experience.

Or what would a 10 star experience look like? So Brian says in that podcast you arrive at the airport, a parade of people welcoming you on the tarmac. When you get to the Airbnb, there's a press conference outside and you're on a camera and you're treated as a celebrity. And then you get to check into the apartment. So, that's a 10 star check-in experience on Airbnb.

But then what would an 11 star experience look like? You land at the airport. Elon Musk receives you on the tarmac and he flies you into space.

Now, nine, 10, and 11 star experiences are not feasible. This is what Brian says, but if you go through this crazy exercise, there's a sweet spot where it is somewhere between they showed up and they opened the door, which is a four to five star experience. And the extreme, which is, I went to space, which is not going to happen. But if you design the extreme and you come backwards, you can find a sweet spot that's achievable, but is above the status quo.

So knowing someone's preferences and saying, "Hey, I have a surfboard in the house." It's not that crazy, and it seems pretty reasonable, but it's also above the baseline of what most products and services offer today. And this is the kind of stuff that creates really great experiences that people end up talking about.

So this is what Brian says. "In order to scale, you have to do things that don't scale." And I highly recommend reading that episode or listening to that episode.

The Messy Nature of Design

So what I'm trying to say overall through this talk is design is a very messy process. You diverge and converge several times, and you are essentially trying to design the right thing, but also designing things right as part of the second half of that process.

If you double click on this double diamond diagram, you see there's multiple steps within each of those, and it essentially revolves around the same things I just touched upon today in terms of speaking to the right stakeholders, getting enough data points, and then synthesizing that info and creating a very unique experience by handcrafting every touchpoint for the customer.

And what you're trying to do is make sure you can get feedback as early as possible instead of having to wait until you have the perfect product and then realize that's not something that people even want in the first place. And that's because as time progresses, you have less freedom in your designs as things get constrained, and the cost to make a change also go up. So you want to try and do that as early as possible, which is why that iterative process is very effective to keep in mind as well.

And you can do this with software. So this is something I did with a telemedicine company and EHR that I built. It's very easy to move a post-it. The cost is basically zero. Time is also very close to zero to move a post-it in terms of changing what the user experience looks like compared to building a Figma prototype, a high fidelity wire frame and making changes to that versus going and programming something and then making changes at the final stages.

So try and keep it pretty low touch with the plumbing prototypes like we showed for the medical device. And this is what the product ended up looking like at the very end, once we knew what people wanted to buy.

You can do this for services as well. So as easy as setting up a landing page, getting signups before actually going and building something. Then when you have wire frames, sharing those wire frames with people to get some more feedback, versus then actually having people design the service and create content for you or anything like that.

And we've seen that happen with hardware as well, where you can make a "looks like" and a "works like" prototype where you fake some interactions and you make nice renderings to say, this is what it would look like in concept, but this is how it works with the technology available today.

And it's the same principle with the medical device, with the plumbing parts to what the final product look like.

Key Takeaways

But top three things I want to say to end the talk here are, number one, trust the design process. It is messy. So, just trust that it'll get refined with time. If you put in the effort and go through the right motions. Be scrappy and try to move fast. Collect feedback as fast as you can, also don't be attached to your ideas because at the end of the day, you're trying to solve a problem for the customer rather than push your idea onto them if that's not necessarily something they'd want.

Abhi Ghavalkar is the CEO and co-founder of Third Space Therapy, which provides mental health services for patients on Medicaid across the US.