Edge AI and its Benefits

Interested in connecting with Adel? Reach out to him on Linkedin!

About NeuTigers: NeuTigers is a leading-edge company that is revolutionizing the application of state-of-the-art Artificial Intellectual Property from Princeton University. It specializes in solving healthcare and IoT problems using a unique energy/latency-efficient AI technology platform that can port intelligent services to diverse hardware platforms from sensors to mobile devices and cloud servers.

Key Questions and Topics from this Episode:

(01:02) Intro to Adel

(02:52) Intro to NeuTigers

(11:03) Can you share any use cases?

(14:16) Especially given your involvement in the medical industry, what’s your approach to security?

(19:35) What is Edge AI? What’re the benefits?

(23:20) In what industries do you see edge AI making the biggest impact?

(25:01) Is there any news you can share at NeuTigers?

---

Transcript:

– [Man voiceover] You are listening to the IoT For All Media Network.

– [Ryan] Hello everyone and welcome to another episode of the IoT For All Podcast on the IoT For All Media Network. I’m your host Ryan Chacon, one of the co-creators of IoT For All. Now, before we jump into this episode, please don’t forget to subscribe on your favorite podcast platform or join our newsletter at iotforall.com/newsletter to catch all the newest episodes ass soon as they come out. Are you tired of overspending on data plans? Do you need more consistent coverage? Are you over negotiating complicated contracts? Well, Simon IoT gets it. That’s why they offer customized transparent data services across the globe. Flexible contracts, taxes and fees included in one simple price and user-friendly data management. Your data is in your control. Their LTE SIM cards are scalable to your needs regardless of the industry you’re in or the devices you need connected. Learn more at simoniot.com/IFA So without further ado, please enjoy this episode of the IoT For All Podcasts. Welcome Adel to the IoT For All Show. How are things going on your end?

– [Adel] It’s going great, thank you so much.

– [Ryan] Yeah, it’s great to have you. let’s start off by having you do a quick introduction to our audience. Talk about your background experience, and anything you think would be relevant for our audience to get a better sense of who are listening to.

– [Adel] Sure, yeah. The, Basically my background was all in Life Science, so I spent my entire life developing drugs and medicines when I was, I used to, was actually in the pharmaceutical sector. But I will always have been like at the interface between the science and the technology, and always convinced that the two are complimentary to each other and they always followed these my PhD, which I did actually in France. Dedicated Special for the Oncology and then went to the pharmaceutical sector where basically I build a lot of these informatic research mixing the science and the technology together. And yeah, come to this country like in early 2000 to support some of the merger happening in this area and spend again 10 more years here and before really just taking off, leaving the corporate side, going actually to the Silicon valley where I stared there more than six weeks, six years, something . Spending my time doing a lot of med tech entrepreneurial activities, get my first time a CEO position there. And always again, I think at this area, how we can enable technology to do a better job in term of health outcome for people. So that’s basically my life had just come back here. You’re in the east coast back in 2018, where come back a little bit to school in Princeton, New Jersey. And there were what actually met my co-founder Professor Jar, which we started actually NeuTigers. This is basically a short story about 20 years so of my life.

– [Ryan] Fantastic. So tell me a little bit more about a NeuTigers. I know, obviously it’s out of Princeton, which is where the Tiger’s name comes from or the Tiger’s part of your name comes from. But tell us a little bit more about kind of the story behind the founding of the company, what you all do and kind of the opportunity you saw in the market to kind of, to start this started this project.

– [Adel] Yeah, I think the, I mean, I think when you’re looking to the history of what NeuTigers is doing right now, which is basically IoT based AI powered companies for many different use cases and core business is really related to healthcare. But one of the key challenge there was always, you know, how it can actually do more inference and trying to identify some of their, I mean onset of the symptoms before something big just happen. And I had been through a lot of personal stories and for traveling my life, when it comes to healthcare, I lost my sister from a breast cancer.

– [Ryan] I’m sorry.

– [Adel] You know, something which has been build up for two years without noticing anything of something happened with my early uncle, which I actually also think we lost him, I think, from a heart attack. So really the, so then this comes through to them, to these questions, you know, how we can actually, I mean, monitor and just help you identify these early onset of all these major killer disease. So that was really the motivation, so about what they have been doing for the last 20 years. And again, I think when I met actually Professor Jha, I was actually coming back to the school, learning more about all this Edge computing. So this AI energy efficiencies, I met him actually during one of his courses. And he was dedicated to Edge computing, which is really a fascinating word in the sense that a lot of these solutions for the last five years were mostly cloud based. But a Professor Jha has been basically doing 30 years of this research and how we can actually make this complex kind of technology available or portable in more resource and the space, the constrained environment, like any Edge device, Edge devices to just basically a connected object, right? So, that’s where when it started and what NeuTigers is doing is basically packaging all this great innovation happening in the IoT medical devices, in addition to a lot of these innovations in deep learning. And then of course, I think focusing on some of the core problem, which never been solved before and healthcare, of course, I think was a big one and that’s the reason NeuTigers is focusing primary at that stage of developing or applying this commercial, commercially or this innovation from Princeton.

– [Ryan] That’s fantastic. I appreciate that kind of backstory. It’s been a, it seems like a very exciting kind of journey to where you all are now. I wanted to dive in a little bit more to kind of the offerings and talk about some of the things that you all are doing. So I know you have, or I guess, talk to us a little bit more about CovidDeep, kind of what that is and the challenges that you all came across when building that solution.

– [Adel] Yeah, I think that, again I think when I met Professor Jha, he was actually doing some proof of concept in the area of diabetes and mental health. And the really the importance is at that time, that’s basically when you have like a disease it’s actually printed in your body, right. And then you can capture some very informative information from your physiological signals. Whether it’s related to your heart electrical activities, your skin temperature, your oxygen saturations. Actually, all this information is basically digitally, I mean, related to any medical conditions. And that was actually like the initial premise, but in a it’s one thing of course to do, the other, it’s another thing just to prove it. And that’s what he started actually doing that in the area of diabetes. And how it works, we’re using some of these medical devices, whether it’s a smartwatch, which has a lot of these medical sensors, which has been by the way approved by, in general the regulations. So whether in Europe or here in the United State, when you have some of these connected watches now, which has any an EKG, I mean, this is like a clinically validated EKG replacing these market things you can actually have at the hospital. So the challenge then was how I can actually decipher all this different data coming from this physiological, which capture from these sensors and make sense of it and trying to differentiate somebody who can be healthy versus who can be for example, diabetics type one or diabetes type two. And that’s basically the premise of a lot of things. I think we’re doing here at NeuTigers and the way which comes actually to COVID-19. We pivoted actually after this proof of concept in diabetes and the move that have like early 2020, there was like a big professor actually in the pavier in Italy, who is the one of the major infectious disease and learn about some of the paper within the proof of concept. And he contacted us and back and saying, okay, can you do this actually for COVID-19. It was actually March 20 last year in the middle of the pandemic. You can imagine growing there a scare with my CTO, I think go in there, collect the data and the which works. I think what go in and connecting, giving actually the, a smartwatch to a different cohort. And the cohort are separated between people who are healthy versus the people who has this COVID-19, and as you know, there’s actually these three different kind of severity of the disease where they can be a symptomatic positive. That means you have the virus, but you don’t see any symptom. The severity of the people has, I mean, the symptoms where they can stay at home or they can be unfortunately need to be oxygenated or even incubated at the hospital. So we put this sensors device that was smart, smartwatch on those guys in general. Is 30 participant per cohort, just for the survey, we adopt the statistical power and you get all these data. And it’s pretty actually, that’s one of really the innovation there. The data collection is pretty quick. It’s like one month when you compare like a drug development where you need like years of collecting data before proving the, even the initial proof of concept of your product. Here, this kind of the pivotal clinical trial when you’re using this kind of technologies is between one to two months. So you collect this data and this is then where the secret sauce, I think where which we have a thing with Princeton University through some of these deep learning technology, we are able to generate models who can actually classify a person being like a negative or versus positive with these different severities of the disease. So you know, once you get this model, you can actually then because of this characteristic of the technology, we have licensed to Princeton University, we make in this very complex small networks, I think very light, still accurate, a very energy efficient. So you can actually embed them in any Edge device or any connected object, whether it’s, of course the smartwatch itself or it could be your course, your smartphone, it could be like even a ring, a connected ring. So you can actually then afterwards you can do the inference actually within the Edge device and you don’t need any connections to any cloud to privilege your security and your privacy.

– [Ryan] That’s fantastic, that’s great. And are there any other use cases or kind of, you know applications of technology that you’d be willing to kind of share with our audience to talk a little bit more about what you all are doing?

– I mean, I think that’s one of the important things I think just want to, just to emphasize Ryan. First of all, I think that that’s really the kind of feedback and interaction we’re having. There’s a lot of education to do about this kind of solutions, healthcare solutions, which are more integrated technologies. It seems a little bit magic to some extent. But this is, a piece of NeuTigers who are really doing the hard work. I think working with a clinical expert basically generating this data, the proof of clinical studies and actually CovidDeep right now, it’s actually under review at the FDA. So, which is just telling you, it’s not a toy, it’s a reality today Using these very integrated AI and sensor-based medical devices, we can actually read the body and hopefully distinguish a particular medical condition from another one. But as you have said, right, I think that’s the, this is a technology which can be, I mean, transports to many verticals, not a one, for example, something which we heard even, or self-trust last year, we have been compacted by the smart, actually smart footware company. And to our big surprise, I think why actually are you connecting us? So, and the reason is if had like a smart insole we’ve embedded sensors there, which can capture basically the mobility of the person. So you have pressure sensors, of course you have mobility sensors, you have these GPS kind of sensors there.

– [Ryan] Hydrometers

– Et cetera, accelerometers. And actually, it’s, that’s what we heard actually. I mean, your feet is telling a lot about actually your health. It’s like 8,000 type of nerves, which are basically at the end of your body. And if something is happening, your feet knows it. So there’s plenty of application. One of the application for example, is a lot of, I think in health, the mobility disorders Parkinson’s, or these kind of things. The same things that exactly what taking this information from the sensors, right. And that we’re able, for example, to classify somebody, some being basically sitting without no activity or reading, or going upstairs, downstairs, or loaded, just going upstairs or downstairs, for example. So you can identify a lot of kind of mobility patterns, which can actually have different purposes. Health of course is one of them. But security, for example in plants I think to be sure that, you know, as a plant manager, you understand, you can assess actually the risk of your workforce, for example, particular environment. That’s the kind of very surprising applications when you have this kind of technology, which can be transport to many use cases. That’s basically another example of that, but there’s a lot of things to happy industrial IoT as well.

– [Ryan] Well, on the security side, how are you all kind of approaching the security elements with all of this medical data and all this information that you’re collecting? I know you have a platform that’s kind of, or at least a tool that kind of focuses on security. Can you talk a little bit more about kind of how you’re bringing that into everything that you have going on on the AI side and the, you know, the final application side?

– [Adel] Sure. First of all, I think cyber security I think is huge, right? I think what I want, just to make sure, I think with the technology we have licensed to this professor, everything we do at NeuTigers is actually focused in a IoT device, right. How it can apply only to like the right applications for monitoring and screening for medical conditions or how it can actually monitor somebody or how can actually monitor a particle supply chain, you predict the mechanisms for materials. For example, of course, I think that we are completely unprepared. I mean, the cyber security is already a huge problem by itself. But we are living in a very hyper connected world. They would be more than 80 billion and off the connected object, I think by 2025, we’re living in this hyper-connected world and from the cybersecurity now for the last five years, this is actually one of the primary target of the is the IoT device. Basically, I can hack your house just through your temperature monitor for example. Or I can hack my medical device, my heart rate and embedded heart rate, for example, device just directly there. So I think that this is really the key things. I think we completely unprepared, the industry won’t spend enough, I think to protect basically, or the cyber physical network system for all these connected objects. We were pushing these absolutely, I mean, unique as well. And in general, what you do, I think when you’re looking for liabilities, your back is just listening to the network, right. I think you don’t defy, but you particular liabilities, in one particular server to access some of the data. And then you start patching needs its own in this reactive mode, which is inefficient very costly, right. But the way actually this professor just actually just attack it, is really about, again combining the building blocks of all these cyber threats, you know. A cyber attack, it’s basically a piece of software. It’s a piece of code, which you can actually dissect like in different building blocks, right. And then you can, of course, so far there’s like, I think for the last five years, I think more than 40 and of IoT based attack and they are, when we generated this knowledge base of these building block, and we are able to use that and combine that with AI, learn from it with first stage is to reproduce actually the known attack. Just from this knowledge base of these building block, of these, all these different attack, which happened actually in the past. So, which was also, I think you’re already like a big, I mean achievement, but one of the things we were not even actually able to publish that because one of the things I think you can do then from there, okay. So from this knowledge base, you are about to reproduce, but what are the many possible. Because it’s a combination problem. You have all these different building blocks, some of them would come in from one attack, but then we can actually be combined to come actually from what we call zero the attack. That’s basically attack would never been, never happened yet. But you are absolutely feasible talking of course, to the expert, et cetera. So there were like 120 different new attacks, which were possible. And that’s what actually Shark and that’s the technology who’s doing this, it’s just basically building this knowlEdge base and then looking to identify a possible zero attack zero the attack in the future. And of course, I think one of the things we need to just very after is doing these defense, how we can actually design, I mean, these upfront defense, and then, so you can actually anticipate this attack in the future. And that’s what the, another tools that another version of Shark, which is more like the, attacking kind of piece that Gravitas is helping all these different aspect. So in the future now, what you’re going to have with this kind of tools in any cyber-physical system, you can actually not only install this silent kind of smart cyber agent. Its not only knows already what happened in the past. So, they can drop them right away, but more importantly, actually anticipate and have already the defense for something never happened before. That’s the uniqueness of what actually of Professor Jha has developed at Princeton. And which are using now again in many application, because of course healthcare, it’s a big, it’s a big one of course.

– [Ryan] Yeah, that makes a lot of sense. I appreciate that kind of information. That’s great to kind of understand a little bit more into what you’re doing on the security side. I did want to kind of expand on a few topics we’ve talked about here as we’re kind of finishing up. And the main topic I wanted to explore was a little bit more about Edge AI and talking to our audience briefly about what Edge AI actually is, and kind of the benefits of Edge AI versus traditional AI being done in the cloud and how that overall is impacting IoT.

– [Adel] Yeah, of course. I think when we’re talking about Edge AI, it’s basically any possible smart, connected object, right. It’s really how we can actually report or embed a lot of these intelligence services in connected object. And that’s what basically Edge is about. It’s not in the cloud, it’s at the Edge. So that that’s one, I think. And of course, I think it’s a big, big problem because most of these AI technologies or the, neural network model charging rated are not only, I think very big, I think that’s one, but also they are very energy consuming. So that’s the reason most of the time they need these cloud-based kind of infrastructure to do whatever, the training, generally this models, and also choose them to do the inference predictions on them. So this is where the innovation is coming through some of the kind of lab from Professor Jha Who’s actually taking all this fantastic capabilities now able just to put them back in these Edge kind of resource and energy constraint environment, with Edge AI. So what does he do, right. I think that’s the question. So start to ask why actually I’m doing this. I think if I can run that in the cloud, you know, but doesn’t care, I think where I think it’s can be around so long as I have the service I’m looking for. Actually, there’s three big things. I think that the first of all, the trends is really toward how it can actually get, you know, ensure my preserve, my privacy and my security. And as you know changing the cloud environment, there’s much more exposure, much more risk, and basically having put in these kind of capabilities in Edge device increase actually, and presume more actually our privacy security. Prophecy for example. So we have, again, a proof of concept, which we, I think we are in the way to productize is predicting depression, right. Using the same platform, basically connected devices who has all these smart sensors you’d capture the information about your physiological signals. And then you do this spoof of clinical. You get basically able to generate the models, can differentiate somebody from being baseline healthy versus somebody who can be in the rise to be a depressive, right. But imagine you have that in your iPhone, right. And that’s something which can be hacked, but I think all the inference you’re doing some days you’re okay, sometimes, unfortunately, you’re in this depression mode. You don’t want that to be hacked. Like that’s really one of the premise of the Edge AI, the other importance is preserving your privacy and the people are under the control of these data. We used to, unfortunately, to have, or bruising kind of history, data or habit buying habit just as stored by many different companies. I mean, a lot of people are not ready for their healthcare information. So we want to be that to be secure. They want to be private and that take control of all these standards of the data. And that’s really one of the added value of having all this new capabilities embedded in Edge devices.

– [Ryan] Gotcha. That makes a lot of sense. And where do you kind of see Edge AI playing kind of a leading role in certain industries or are there certain industries where you see Edge AI playing a bigger role? And kind of, where do you see, I guess, the largest growth potential for Edge AI to kind of be involved in different use cases or particular use cases that may come to mind?

– [Adel] Yeah, I mean, it’s cybersecurity is again like a common ground. so there’s going to be a cybersecurity for IoT overall. I think it doesn’t matter whether it’s in healthcare or it’s a connected object, right. But when it comes to IoT AI, Edge AI, I think there’s really huge opportunities in the area of industrial IoT. We’re starting actually talking to the Hourspace, and I mean, producing like a helicopter, it’s a pretty complex. But already a piece of it like a blade of the helicopter. It’s in I mean, huge more information along the the different materials, which need to be used, et cetera, to put, to ensure the quality basket of these blades at the end, right. And one of the challenge is really how I can now do a lot of these supply chains, getting connected with a lot of smart sensors, and I mean collect information about the vibrations, the imagery, for example, to detect any defect. So this is where I think that this industry is completely behind when it comes to these AI efficiencies. I think along the value chain. We believe this is really a huge vertical, which we already have been tracking with some major player in the area.

– [Ryan] Yeah, I think that those are great points. Edge AI is something we’ve been talking kind of about here and there, but not into the detail I think it deserves with the role that it’s definitely starting to grow and play in IoT solutions. So this has been fantastic. I do want to ask, is there anything kind of new and exciting coming out of NeuTigers anytime soon or anything our audience should pay attention to? And at the same time, what’s the best way for people to follow up, to ask questions or learn more?

– [Adel] Yeah, sure. I think the thank you for asking that Ryan, appreciate that. I think a NeuTigers actually where raising our series a money right now.

– [Nice] Nice

– [Abel] I think that’s a great moment, I think, to be involved and especially for the people interested about really the new normal and healthcare is going to be more about virtual care. More and more delivered through telemedicine for example are heavily or you see better there. And we are one of the pioneer in this area. The second of course, I think we have, or a NeuTigers website, you know www.neutigers.com. I think you can learn a lot of things about what we do. And the next for us really innovation is really about really addressing more medical on that need. We’re working for example, in new category of the product, instead of just doing screening, which is like CovidDeep right away, whether you’re positive or negative, we’re going actually, we’re developing new category product, using these sensor days technology to monitor the exacerbations or found medical conditions. Like one of them that one of the next product it’s for sickle cell anemia, which is a horrible disease. And we’re working with the community of these disease to into basically their quality of life using the standard technology.

– [Ryan] That’s fantastic. Well, I wish you all the best of luck with kind of all those projects, as well as a series of raising round. Sounds like a lot of exciting things going on and I’m sure that money when you’re able to complete the round, we’ll go to some good use. So thanks again for taking the time to connect today and kind of sharing your insights on what you have going on, as well as all the Edge AI, you know, in-depth knowledge that you were able to kind of share with our audience there. So I really appreciate it. And thanks again for being on the show.

– [Adel] Yeah, thank you, Ryan. And just for past one thing final thought. I like actually your model about IoT For All, because I think one of the really also motivation we have, how we can actually democratize to the access to these very complex, I think technologies. And that’s the reason also for us NeuTigers is the intelligence for all I like this and in the model. And thank you so much.

– [Ryan] Thank you.

– [Adel] I’m certainly looking forward, I think perhaps the next day I’m coming back in the show as well.

– [Ryan] Absolutely yeah, absolutely. I think a lot of what we’re doing aligns with what you’re trying to accomplish at at least a philosophical level. So this has been great and I definitely would love to have you back at some point in the future.

– [Adel] Yeah, thank you so much and have a good day.

– [Ryan] Everyone, thanks again for joining us this week on the IoT For All Podcast. I hope you enjoyed this episode and if you did, please leave us a rating or review and be sure to subscribe to our podcast on whichever platform you’re listening to us on. Also, if you have a guest you’d like to see on the show, please drop us a note at ryan@iotforall.com and we’ll do everything we can to get them as a featured guest. Other than that, thanks again for listening and we’ll see you next time.