 We haven't done anything in this space before but it's essentially I have worked on a couple of projects in this space and I really enjoyed this and just wanted to share with you guys what I've done, what I've learnt and some of the cool things that's going on and why you should maybe start working looking into this. So before I start, can you see that? Yes. Okay. The rest of the slides are not done. This is non-technical. I'm not going to be showing you any code. In case you want to see how to program some of your stuff, you can get back to me and I will show you some code and some schematics and stuff but generally it's pretty boring and even if I put it up nobody can believe it so I'll ignore it. And also these are very subjective views about the field because this is what I think about different aspects and if I tell you like, oh this is a very interesting field and I think that's what I do. So what do I want to talk about? First of all introduction to medical technologies, what are they and why it's interesting and some of the projects I have worked on in this field and I also added a section on some of the some ideas you can use if you want to get started on this field. So let's get started. What are military technologies? How many of you have at least heard of medical technologies or have considered working on anything that is remotely medical? Thank you, Jay. As you wish. What does it mean? Okay, that's the proof. Yeah. Okay. Let's see more. There are two kind of fantasies. So probably the most simple definition of a medical technology is that it is a product that solves a health problem and improves the quality of life. So this is a very broad definition and essentially that's what it is. Anything that improves your quality of life can be considered a medical technology, but it is generally encompassing preventive medicine or diagnostics or something like that. So some of the examples of medical technologies are diagnostics, there's prosthetics because nowadays there are quite a few like very advanced prosthetics that have machine learning involved, which actually learn from the user, user's behavior and how they want to move the hands. They can control the hand as if it's their own. There's also lots of improvements in biotechnology, which is also considered medical technology. But once I really am interested in telemedicine, so the massive improvements in internet connectivity and the amount of technology you can have at home, it is possible to diagnose and treat many diseases or have at least the first consultation at home itself. And that's a huge market and there's a lot of improvements in this. The next one is monitoring. That is about monitoring your vital signs, any of the information that the body looks out at, you record it, you look at it over time and you can diagnose what kind of issues you may have. The difference between monitoring and diagnostic is monitoring can be preventive as well. So you can start monitoring right now. So anyone who has like a valuable tracker is actually monitoring your steps. So essentially if you think about it, it's like preventing hypertension and diabetes, non-tariff. Similarly, there are monitoring solutions for different kinds of ailments and illnesses. And the last one which is relatively new is analytics in the medical space. So analytics here means we collect data from hundreds of users from clinical studies and you have the Apple health kit which collects data from your phones as well. So it gets all of this data and then you have to come up with smart insights based on that. So that is also another field which is up and coming and it's pretty interesting. So why is Medtec exciting? First of all, it's a growing market. It's worth 350 billion right now and it will be around 30% more by 2020. It's growing at German District. You look very happy. Okay, so... You look very uncomfortable. There's a camera in here. Okay, so also there has been like huge improvements in technology. The sensors are much better. The processors, the hardware that is required is much cheaper and faster. And also you have cloud computing, big data processing and stuff like that that makes it really interesting to work in this field. So the biggest improvement right now in terms of working on any of these projects is that there's a lot of user awareness. Now people understand that it's important to monitor themselves and they understand that they need to be aware of themselves so that they don't fall sick. So it's pretty easy to convince people that this is something you might want to use. This isn't this way like 10 years or 20 years back. So that's a really good improvement. So let's look into... I just want to give an overview of the METEC market but I'm just using monitoring solutions as an example. So most of the medical technology market is divided between fitness and clinical science. So fitness science is where your personal Fitbits and Pebble Watch can collect data and there are many other systems out there. So one of them is like this one. So this is a Mio Alpha. It is one of the first wearable wrist based watches that can record your heart rate. So this was the first one to come up with this technology. So this is really interesting and it lies somewhere in the fitness side. And on the other hand you have things like this which is a glucose monitor. So we did some user testing, that's why I have all this stuff. I didn't just go and buy this yesterday. So this is on more of a clinical side. So what differentiates these two is majorly legal aspects. So it's not the features, it's more about the legal aspects. It's like if you claim that you can be used by doctors then you need to go to FDA clearance and many other things. So it's based on what claims you make. So if a wearable device says that, okay I claim that I can reduce diabetes by so much percentage or something like that then they'll also have to go through quite a few regulatory clearances. So that's what differentiates the two sides of most of the matter. But at the same time things have been changing recently. People have been trying to approach the market which is the midsection which hasn't been, there aren't any good products in that market. So one good example would be this one. It's called Biffy Mobile. So it's a wearable ICU on your list. So you just wear something, it's like a whole hand-arm based device which can measure everything that you need to measure in an ICU. So it is somewhat clinical but at the same time it is also usable by the patients at home. So they can actually monitor themselves at home using this system. Again this is not for everyone, this is a very specialized use case. Similarly on the other side of things, there are fitness devices that are becoming more and more clinical type. One example is this really interesting product called Scandoo Scout. So they came up with this like one and a half years back and they launched it on IndieGoGo and they have like they've made millions for this. So what this essentially does is they call it the first medical grade tripod. So it is able to measure all kinds of vital signs. You can measure your temperature, your heart rate, blood pressure and all of that stuff. But why I say that it's not just fitness and it's moving towards the clinical side. Moving towards the clinical side is that they are aiming to initially release the product as a fitness device and then slowly collect enough data from the users and then go for regulatory effort. So this is a new model many startups are using to get a new speed because generally the clinical equipment are managed by big companies and getting into the space is pretty hard. So that's why it's really exciting to work on any of these projects because there are so many options for you to get into this. You can go into the fitness side or clinical side based on you know what your goals are. So any like I just put this dot like anything on the right is being regulated. So the thing is a product needn't be just here. It can move over time. So that's what Scandu is doing. So they want to build the first clinical tripod that's from start right. They want to call it a tripod that can measure all your vital signs from home. But they want it to be clinically permissible data. That means you can use, the doctors can use this data to make diagnosis. If you have a Fitbit, doctors can look at it and say they can't make any diagnostic based on your fitness device model. I will get to this but even Fitbit has a lot of lawsuits because they don't have accuracy. Their accuracy is pretty poor. We'll get to that but essentially what it means is that there are quite a few opportunities in this model. So next I want to get into what makes a good MedTech product. So according to me these three main fields that you need to address if you want to be a good MedTech product. And first is security privacy. Second is safety and third is accuracy. This has nothing to do with user empowerment and all. This is just technical factually these things have to be present in case you want to be a good MedTech product. So few years back this is what you could see on Fitbit pages. All the activities you record are public by default and people did not know that. So this is an example where you are trusting a device with very personal data and your health data tends to be very close to you. If you have any ailments you don't want other people to know. So yeah this is an example of things to avoid as developers when you're building in this space. Another one is pretty similar but this is not technically MedTech. When I was searching for this I just bumped into this and I was like oh I have to share this. So there's this Durex made experiment called Fondelle. So this is an iOS controlled vibrating underwear that long distance couples can use to be close to each other kind of thing. It's already pretty funny but here it gets even funnier. They did not think too much about security. So this is what the creator had to say. So basically you could hack into the system and then... Yeah exactly. Once this happens there's no going back. You can't say oh if it's the problem please start using my product name. So yeah security and privacy is really important in this space. So one of the regulatory requirements for making sure it's secure and the data is secure and private is the HIPAA compliance which is Health Insurance Portability and Accountability which is a US based requirement but it's followed worldwide. So if you have this compliance even in Singapore it is one of the requirements to be FDA-free rather than all that. So are you saying it's a US legislation? It's a US legislation but you have to like FDA which is also used in Singapore right now but US is the one which gives you the clearance. So here for example AWS is certified. So if you store data in AWS using the certified system it is fine. Heroku on its own is not certified but there are add-ons like it's called Actable. If you want to know I'll share the name later. But it's an add-on on top of Heroku that makes sure that all the services you use are encrypted and it is HIPAA compliant. So this is a good thing to keep in mind because if you are... Suppose you're building a valuable product and you're collecting any data you can be sued if you do not store it in one of these HIPAA compliant systems. And also be empathetic to the user. I mean put yourself in their shoes and make sure that you don't reveal things that they haven't explicitly asked you to reveal. Next one is safety. Don't have the users that make sense because many a times METEC products work with users who already have some kind of health ailments so make sure that you don't worsen the situation. So for that we have the FDA clearance. So FDA clearance is when any product or new technology that claims to be in the clinical side of things has to grow and get cleared. Again, there are loopholes to this. There's one very recent one which is called 510K which is prior art. That means if someone else has got a similar product through FDA clearance already then you do not have to go through all kinds of clinical trials and things like that. So here's an example of somewhere this background. It's called Striker orthopedics. This is a HIP replacement for the HIP joint like if you have practice and stuff. They did not go through clinical trials saying that hey, they are very similar to the ones that are already in the market but once they actually put it on patients they had huge amounts of allergies. It broke inside the body and caused more harm. So it's very important to take safety into consideration more so than other products. And lastly, accuracy. So this is the crucial difference between fitness and clinical devices. So for clinical devices each type of device are different thresholds and based on the thresholds there are also how permissible it is in medical report. So for example for this guy which is a home-based glucose monitor the error threshold is 20%. So it can be just 80% accurate and that's good enough. But that also means that doctors cannot declare you diabetic based on these things. So there are certain rules about which devices can be permissible in the clinic and in case anybody wants to build a product that is for that particular space and you want to target doctors then it's very important to research into what are the regulations by FDA for that particular segment of products. One of my favorite rule of thumb is take non-invasive. Don't put anything into the user's body and most likely you will have very less to worry about here. Safety, accuracy and... So now I just want to get into some of my projects and just give an idea about some example projects that can be done in the space and why I'm so crazy about this. Okay, so one of my first projects I worked on was something called Clearbridge Vigil Science. So this is where I actually got introduced to this topic and this is basically building a wearable ECG monitor and it's completely wireless. So if you remember I showed you a picture before which is this. This is called a halter monitor and this monitor has anything between 4 to 12 bits all around your chest and it measures your ECG for a prolonged period of time. But it's extremely inconvenient. You need to put on these weird sticky things with gel and people really hate this and they really go through this and if they realize that they have some health conditions. But suppose you want to make this more available to the wider market. We built this ecosystem of products called Cardio Leaves. So it's like a plaster kind of thing you just put it on your chest. You can use it on its own or you can use this. So you can use the shirt as well. So the shirt has embedded electrodes in it. So if you just wear it it's like a skin like shirt and then you just plug this device on top of it and it works. You can't see it but basically it reads the heart rate inside. So this is one example of a product that is starting off in the fitness side of things and they're slowly trying to get clearance and move into the clinical side. So this is another one of those examples. Another project which I've worked on which is actually this is for a hackathon which is a test strip scan. Can we switch to Singapore favorite? Yes. You know them? No. Okay so this is something I built for a hackathon which is basically a test strip scan. So test strips are these tiny extremely cheap you know tests that you can administer on yourself or you can have like suppose you have a blood donation camp and you want to quickly make sure that people don't have any known pathogen there are test strips for that. So you just put a drop of blood on it, wait for some time and then it will show lines something like pregnancy and stuff. All of these work on a similar concept that they have some kind of enzyme that binds to certain type of molecules in your blood or saliva or urine or some sort of volume fluid and it shows up as a marker. So here what I built was I hacked the CV drive, removed the insides, added some 3D printed parts and stuff and it basically has a photodiode which scans the tray as it moves it. And then it can be based on the bar. Normally there will be 2 bars, there will be a controlled bar and there will also be a regular one which shows what is your actual reading and then it will just figure out what approximately what is your glucose levelers or what is your white blood count and all that else. So this is something else that I play around with this. But the most interesting project for me which I worked on was this one called EasyMob which is my final project but I worked on it for quite a few years. It is a wearable continuous monitoring system which was aiming to bridge the gap between the clinical and fitness side. So it should be able to get high resolution data from regular use. So we tried to get multiple versions out. So this is the first version called EasyMob Cardio which has heart rate, blood oxygen, body surface temperature and various other things. So this was part of some course in the university. So I just want to use this as an example to showcase my learnings in this field. Before we get started maybe I can just show you a quick video of what it looked like. So this one is an early prototype which is basically just a heart rate sensor. And it has no casing or anything. There is no ordinary interface. So it is wirelessly connected to an Arduino board because at this point we are just using the Arduino's to the hardware part. And then it is using our own custom circuit to figure out heart rate sensor. There are quite a few schematics online if you want to get into this space as well. But we were also trying to measure the oxygen here and also actively monitor it. So this is what it looked like and one of the earlier prototypes. So now I just want to take you through some of the key learnings that I had working in this field. And one of the tools I used throughout this process was design thinking and a bit of beam methodology. But essentially designing or something we were introduced to very early in the NUS as part of some course and we found it really useful for applying it to this project. So what is design thinking? It is a powerful tool to help make the right product. It is basically keeping design in mind throughout the whole development cycle of the product. So it comprises of these five stages. One is prototype, which is learn more about the audience. You need to define the problem. You have to ideate about how do you solve that problem, prototype your solution and test it to the users. Now again, this is an iterative thing. So after you test it, you learn new things and then you again go back to prototyping or ideation based on what's required. And this was created in Stanford D-School. And it is pretty popular among people who are into product development. I'm sure many of you have already used this technique. So I was pretty lucky because I got to visit D-School as well. So let me go through the learnings. First learning which was eye opening for me at that time was always to the users first. And by this, it doesn't mean that go talk to your users if they want this product or not. That is testing. This is basically do not have a solution in mind and try to find the correct user for that solution. Try to go find the user you want to work with who probably won't solve and then build a solution around that problem. Because if you take the other way around, it's very hard to figure out, especially in this field because there's so many different elements out there. So we initially decided we wanted to go with people who have end stage renal disorder. These are people who are on dialysis machines all the time. And they have to be on this machine a couple of hours every other day. And this is really painful for them and this thing does a lot of life. They can't spend time in the family and stuff. So we really wanted to solve this problem. But before we went with that, we conducted a huge number of interviews. We spent like a considerable amount of time just interviewing people from different fields. We had to find out that we talked to people and then we got more ideas and we talked to more people just to figure out what was required. And again, it's not 100 interviews in one shot, it's over the whole course. So we did a little digging around and found that the main causes of ESRD are diabetes and hypertension. So the next lesson is pivot. So what happens especially for me was that I was very tied into this idea of helping ESRD patients that I did not want to change my music rule. Especially since I've already done so much work in that particular field and interviewed so many people. But the other thing that design thinking teaches us is we need to iterate quickly. And if things don't work, once we test with users, we should be able to just modify the product or completely change it. In our case, we have to completely change it because we realize that once you have already reached the end stage renewal disorder phase, there's no going back. Basically, it's a terminal condition. So we decided that it would be much more effective if we could come up with a way to prevent this in the first place. And the current industry practice is that they give these kind of devices to users at risk for people who have diabetes or hypertension, which are in the morning in the schools and they ask them to take everything to monitor themselves. So we decided we wanted to solve this problem and we phrased it as it's painful, inconvenient and costly. So these are the things we wanted to solve. And with that, we came up with this solution, which is we wanted all the stuff we already talked about, but we also wanted to aim towards building normal basis, new course monitoring and things like that. So this would be the ideal product to solve this situation. At the same time, it's important to make sure that there is a good market space for this. And this is what we did a quick analysis and figured out that there's this middle market between fitness and medical devices, which is currently very sparse. There aren't many devices you can use yourself, which are clinically accessible data. And we wanted to try to aim for that space. So next comes the most important, most fun part, which is prototype and testing. So you'll see different levels of ability here. The first one is like user story mapping. We just sit around and figure out what kind of things do users want to do. And then we decided to work on higher mockups of the thing and go test. So this is really useful for testing software. Suppose you want to test hardware as well. There's nothing, you couldn't have done too much. You could have given up. But we were lucky to have access to a 3D printer. So we printed some 3D models of what we thought it would look like. And we decided, hey guys, wear this for a while and tell me what you think about it. So we have different designs and stuff. You want to get this possible. Just 3D printed bands to see what they feel about the look and feel. So initially we thought of a funky watch like stuff. So then we talked to the users and they told us that hey, we want to wear our watches. Also, we don't want something to replace our watch. And we don't want to be branded as sick. So they want something that's empowering and not something that marks the mass difference. So that's some of the cool learnings we had. And after that, we obviously went around building the real product. So this is like different versions of the product. This is the final one, which is a 3D printed box. The other one is a cardboard box. So you see, sophistication is increasing with time. But this is a fully functional prototype which did what we set up. Though we never actually got to the manufacturing part of it. And the last and final one for me would be that it is really important to iterate quickly. When you're in a harder product, it becomes, it is not as easy as software for you to quickly change based on the user requirement. But at the same time, we tried our best to keep changing. So we had something that we called the golden unit, which was the last functional prototype. But we kept building new versions of the product every couple of weeks. So different times it was just one schematic and we kept trying to change. And that really helped us try out a lot of different versions of the product. And another one that really helped me with iterating was that NUS gave us lots of, not traveling money, but us to go to different markets, talk to investors, go for different computations and all this stuff, which really helped us talk to manufacturers and other stakeholders in this market. And that gave us a lot of clarity. This is my quick overview of what I did and why this should be interesting. So how many of you want to try something in this video? Any ideas for what you want to do? So anyway, these are just random topics which I have been putting off for very long time. That way I should get to this sometime and just putting it out there. So if anybody wants to pick up something and work on it, it would be really awesome. One is analytics with activity data. So right now almost everyone here has a clever lot of fitness tracker. I have one here, I have two here right now. So it gives you a lot of data. It tells you when you sleep, when you wake up, it tells you how many steps you took, how active you were in the last minute. But what if you could get more information? For example, are you a morning person or a person who likes to work at night? What is the optimal sleep amount for most active day? And in many of these apps you say like how do I feel today? So you can say like how does the number of steps and sleep affect my happiness level? So there are so many other things that are not tackled by many of the systems. And all these data are available to you. So if you want to build something on this, it will be really useful as well. And this is a completely software sandwich. If you want to play around with a little bit more hardware, this would be where first everybody here is at. Like many people have pebbles and pebbles is hardware hacking friendly. So if you have seen there is a four pin in the bottom of the table. So this pin is a magnetic adapter that connects to their custom serial port kind of thing. So they have their own protocol and stuff but if you are able to talk in that protocol, you can connect to this and you can have smart straps. So this one you see here is a smart strap that enables SMS sending facility. You can also have smart straps that measure your heart rate. You can basically think about using this as a screen and processor and power source for whatever valuable project you want to work on. At the same time, I should warn you, you might end up burning your weapon, which might happen when you are playing with hardware. Another one which I find pretty hard is fall detection. Surprisingly, because it is really difficult to figure out if someone has fallen down. This is very important for products that are targeted towards the Anthony because for them if they fall down, it's a serious problem. They might want to make an SOS call kind of thing. There are quite a few products out there in the market but none of what you can tell, none of them are fool to. So what they do is they are just accelerometers. If they detect a jerk, they will suddenly start beeping. If you don't press the silence button within a minute or so, then it will cause something on those lines, which is not really that sophisticated. And also it doesn't integrate with any of the valuable systems right now because I'm sure that when we grow old, we'll want to be wearing these kind of devices or something even better. So something that integrates with these devices is something that will be very useful. And also in the same space, building stuff for elderly is pretty important right now because as you know, most countries have an aging population and the kind of solutions out there for this market aren't great. Like most people are trying to build things for people who are fitness fanatics and things like that but not so much for elderly. And that is one space in case you guys are interested, you can. This is some random sketches I had. You don't need to know too much about this, just an example. So that's basically it. If you have any other ideas, just send it to me. I can add it to this link and share it with you. So that's it. Thank you. Yes. After this specializing specific need, I want to measure accelerometer. I want to get accelerometer data out of a wearable wrist strap. And that is it. This is for you, my friend. So this is a hacker-friendly wearable watch called Kronos made by Texas Instruments so you can actually take the insides out and go grab it however you want. Awesome. Thank you very much. You can borrow it for a while. Yeah, so this is pretty... The second big part of my problem is that the customer in my case or the person who is supposed to wear this is a small child. A child will be between the age of between three to ten years old. That means a really small wrist. Nothing like that exists. Even the smallest one you can get for girls with small wrists does not fit a child under the age of ten years old. They will fall, they will slip over their hands. The children don't have established wrists. So the hand doesn't branch out like this, right? So, you know, to add the specific problem that I'm trying to take is something called hand flapping which will naturally make the flippant van fall off. So every time I get the van, the hand flaps the child makes the van fall. One question. Why do you want to measure that? Okay, back story. The Orbson Spectrum, a child now, is ten years old. So the Orbson Spectrum, and they have a number of problems. One of the problems is insufficient deep bone or muscle stimulation, also known as proprioceptive stimulation of the muscles. So we self-stimulate, also called stimming. Stimming behaviors have multiple things. They will make funny noises, funny sounds, they will shake their heads, some of them will butt their heads against the walls. These are all self-stimulating behaviors to get neurological feedback that their body is not supplying to their muscles. One of these symptoms for stimming is hand flapping. There are lots of interventions both dietary as well as occupational therapy, therapeutic behaviors to alleviate the symptoms of stimming, to settle down the central nervous system. So basically what I'm trying to do is correlate the specific intervention because there are many interventions, not all of them work, correlate the specific intervention with a positive response in the stimming behavior. That's it. I need to make sure that the interventions I'm doing is actually having a positive effect on the stimming behavior. Sure. So essentially what you need is an accelerometer, or that is a small enough accelerometer that you can actually... Do you attach to my charts? Yeah, exactly. So this one essentially is a good starting point. If it looks like a regular watch, it's not too big, but yeah, for a kid maybe. But at the same time, the chip that it uses is this guy and the whole chip is just this big and can be even made smaller because this has absorbent pins on it. But it can be made much smaller and yeah, I mean if you are actually looking for something like this, then you can build something like a plaster or something, you just strap it on to the kid's wrist. That could be something. So maybe yeah, if you're interested, you can borrow it for a while and I'll borrow that. Yeah. Is that a mic? Yes, it is. It's called... It's an MSP-430. It's a Texas instrument arm-based mic. We use this in our final version of EasyMonitor. I just have some extra lines. Okay, so I don't want to get into the technical aspects of it, but essentially it's all based on light absorption. So different molecules in your blood absorb lights in a different way. That's how you have spectrographs, right? Some light signatures are absorbed. Based on how much is absorbed, you can know the concentration of it. So that's how these sensors work actually. So these sensors, the green light is absorbed by the blood, the red blood cells. So every time there's a heartbeat, the vessels expand and then contract. So when they expand, the amount of blood that the light has to go through increases and that causes a dip. And that's how you see those lights. Similar concept can be used for... using glucose as well. So actually... I don't know. This is actually one of the prototypes we use for scanning... for testing for glucose. So this is a 3D printed casing kind of thing. This is a lens and a light source. There's a sensor attached to the other. So we use to put a vial of different kinds of glucose solutions into the system and see how they change. And based on that, we can... Basically there's a vial in here and similarly you can have different sensors attached here and you can see how much it absorbs through... This is for how much it lets it pass through and how much is... This one is for how much it's reflected back. So things like this, you can have different sensor systems for... They're never able to achieve extremely high precision for glucose because it's pretty hard. Because the same wavelength is absorbed by water as well. So you need to be able to cancel the effects of the heart pumping and stuff. So yeah, it's a little bit more scientific. There are some solutions out there, but at the same time none of them have been activated yet. What was the accuracy between a clinical version of a glucose monitor compared to a watch? Oh, this one has no clue. Do you mean the one I was building? Yeah. So the one I was building had an accuracy of around 80% but only in clinical environments. In the sense that everything was standardized and stuff. Then it used to... As soon as you take it out, then the accuracy is way too much. So it's not too accurate. So these ones tend to be more accurate based on the fact that everything is packaged and stuff. But again, these strips that come in here, they have an expiry rate. And so with time, the accuracy decreases. That way it's 80%. When it comes out, you use it immediately, it might be much higher. So what's the difference in accuracy between Fitbit and Lopper? Fitbit doesn't... Oh, you mean this kind of food? Yeah. So accuracy-wise, it will be extremely accurate. But at the same time, it's not consistent throughout all kinds of activities. So suppose you are running around, it will not be accurate. Because the thing is, when you move your hand, there will be a change in blood flow in your hand. And this will leave you as a blood flow. So the most accurate one is to check the electrical response by the heart. So that's where you wear the straps and stuff. So you see long-distance runners, they wear the strap. Because these are not too accurate. Especially if you do any lifting activities and stuff. The muscles contract here and then completely goes out. The liability issues, if you misreport it... Exactly. So that's what I meant by fitness to medical market, right? So if you give a disclaimer that this cannot be used for medical use, then it is fine for you to release it to the market. But at the same time, as I mentioned, the flip side is no doctors pay or patients will ever use your product. That's the balancing effect. You need to know which kind of market you want to come into. I think this is one of the industries I specifically excluded from my professional career. Because I had a CS course, there was a story back in the day, of like, hey, y'all want to become developers, blah, blah, blah. So there's two industries in particular you should pay attention to. I do some thought to before you enter in one of us, like medical advice, medical software, and the other was basically like defense. Because there's probably only two where you can write software that kills people. And then it was a classic case of Derek and Tom, right? He knows the radiation dosing machine or the concurrency bug. And people were getting like hundreds of times the radiation dosing. And it's like, you know, if I'm doing that, screw up someone's website. Yeah, that's true. So that's why, like when I mentioned the three main requirements, right, safety is really important. And you need to, like even though FDA clearance is really costly, time consuming, it's useful to have it in place for consumers. As the workers, we hate it, but still it's good for our people. Another question I had was like when you have data, how do you put multiple sources out there, like all the standards, how do you make that useful? Like for example, right, I would like, like my state versus what, you know, I would like to capture that data using different applications. Yeah. So right now all applications, like all the, on the fitness side of things, all applications have different standards. Everyone has their own custom standard, but many of them synchronize with each other. So that is what Google Fit is trying to do. It's trying to collect data from all people and make it into an API service which more services can use. But at the same time, if you look at the medical side of things, along with the HIPAA compliance, there are also standards on how the data should be stored. So that way any software that is able to work with one data source can work with another data source just by switching. So there are some standards, but not in the fitness market yet. As developers, you can create whatever standard you want right now and store data as long as you're using it. Anything else gets? Okay, thank you.