In today’s world, where inequalities live left and right, how can we leverage AI to help us solve food disparities?
In this episode of Bite the Orange, Dr. Ilias Tagkopoulos, Founder and Chairman of PIPA LLC, talks about his work bridging the worlds of AI and food. He speaks about how AI can be a powerful tool to target food insecurity, negative environmental impacts from non-optimal agriculture methods, and production chain gaps in the food industry that impact access. He explains how PIPA uses data analytics to support R&D processes, clinical trials, and even personalized nutrition, a service they offer their clients to improve their final products and health outcomes overall. This use of data and AI has been more widely adopted throughout the last couple of years, and it’s brought about discussions around data privacy that Dr. Ilias believes are necessary but nothing to be terrified about.
Listen to the latest advancements that connect AI, life sciences, nutrition, and health!
FULL EPISODE
BTO_Ilias Tagkopoulos: Audio automatically transcribed by Sonix
BTO_Ilias Tagkopoulos: this mp3 audio file was automatically transcribed by Sonix with the best speech-to-text algorithms. This transcript may contain errors.
Emmanuel Fombu:
Welcome to Bite the Orange. Through our conversations, we create a roadmap for the future of health with the most impactful leaders in the space. This is your host, Dr. Manny Fombu. Let's make the future of healthcare a reality together.
Emmanuel Fombu:
Good morning, good afternoon, good evening, ladies and gentlemen. Welcome to another episode of Bite the Orange, and today we have a very special guest that has an incredible background and very passionate person, a great expert in what he's talking about. So today, I'd like to introduce Dr. Ilias Tagkopoulos. Welcome Dr. Tagkopoulos to the show.
Ilias Tagkopoulos:
Thanks, Manny. Very nice to be here.
Emmanuel Fombu:
I'm very happy I pulled that last name. So thank you. Now, that was ... that, is it okay if I go by Ilias?
Ilias Tagkopoulos:
Yes, please do.
Emmanuel Fombu:
All right, perfect, thanks. Tell us about yourself.
Ilias Tagkopoulos:
Yeah, I am a professor at UC Davis, but I'm of computer science and the Genome Center where we have an experiment in computational lab focusing on microbes, mostly. What we do here is more of AI and machine learning ... informatics trying to predict microbial evolution, antibiotic resistance, different traits coming from systems biology. And my second hat would be being a director of the USDA, the USDA NSF AI is for next-generation food system. That's a collaboration between Berkeley, Davis, Cornell, UIUC, ANR, and USDA itself, where we are focusing on how to make the food system better by using AI. And this has been, we have been given 20 million as an award from USDA, and we have more than 60 faculty participate. And the third hat that I have is being an entrepreneur by a few companies that we have co-founded, and one of them being PIPA, which is a company of AI and life sciences, focusing on nutrition and food.
Emmanuel Fombu:
Which is quite fascinating, and you have, you all wear met multiple hats, right, where you have a core piece. You actually trained neuroscience, or, correct?
Ilias Tagkopoulos:
Electrical engineering, computer science, correct.
Emmanuel Fombu:
I'd like to pick on the idea of food, right, and what you do on the food aspect of it. So tell me, what is the problem that you solve?
Ilias Tagkopoulos:
I would always put a little bit broadly at 40,000-foot level ... and the companies that we have co-founded. They are the two aspects of food. One is people and the other is the environment, and in both cases, we have our challenges. We are looking at say, people, there are a lot of people that are food insecure, including the United States, Europe, and other places, not only third-world countries. The numbers are staggering. You can see that one-third of the population of Earth, they are food insecure in one way or another. At the same time, the other one-third of nation has diabetes, has different comorbidities related to food, and there is some overlap also, which is also strange, like why people who have food insecurity have also issues with securing their food, having good food, having ... diabetes, ... we can say why. And this is the inbalance, the whole spectrum is not just a specific age bracket, but even, you know, young adolescent kids have these issues. So that is one of the main drivers of the, we need to do something and we need to do it now. The vitamins is another one, so the use of pesticides they are using are not optimal agricultural methods, molecular breeding, food processing, and bleeding to waste creates a big problem for the environment. That's the second thing that in parallel I would like to focus on and make better, I'll go forward.
Emmanuel Fombu:
... That was something that I think is pretty critical. If you look at the socioeconomic determinants of health and I think you have these deserts around the country and I think this will highlight, especially the pandemic period, where you saw a lot of people that had comorbid conditions like diabetes, that you ... people lose their lives in a spirit. So are you trying to solve this with AI so that you enhance in determining what was the best meal for a person or what do you do?
Ilias Tagkopoulos:
Excellent question, and the answer is all of the above. So you need to look at AI not as an enabler for micro-projects like small predictable scale in there, although that has value too. But in order to really seize the day and the potential of it to create a new, a new reality and a better reality that is more equitable and it is more efficient and it has a great benefit for human health, you need to see AI being the connective tissue across the whole spectrum that brings together the supply chain, the production, the processing, and the consumer, the consumption of food, and closing the loop throughout. And now this is nice in terms of strategy and theory, but how do you actually do it with people in the field? So what we have done is collaborating with the people that the inner network that they are talking and doing there down in the field and they are actually picking strawberries or apples or their food processing facility or their food bank that is distributing food. Just to give you an idea of one of the projects that we connect with the Yolo food bank is to look what is, if we are doing a more personalized instead of giving food out, which is, they're doing a terrific job of doing and this is the standard for all food banks, is how can AI describe this in a good way, and how can we actually give more specialized, more personalized food baskets for different types of individuals? Maybe you have kidney failure or maybe you are having diabetes and you need a different nutritional profile than the general nutrition profile, and this is an AI project we are running. Another one will be how to make the supply chain more resilient, how to have provenance, and how to actually do that in an efficient way, an automated way through AI.
Emmanuel Fombu:
That, which is quite fascinating. So you're going through a personalized nutrition approach based on data, right? You're personalizing the experience for each person. This goes beyond whether you're healthy or you have any disease condition, right? So, and I think that's quite fascinating to think about things that way. But something more important I want to touch on with you is you have ... PIPA corp. ... You have specific programs. You have the LEAP program, for example, that you ... called. And so I want you to explain the programs of what you do in PIPA and what the benefits are, I know life science companies, for example, are something that ... could be very valuable, too. So tell me about that.
Ilias Tagkopoulos:
So PIPA was created in 2015 with it, to fill a need, and the need in 2015, it was that there were not a lot of AI companies that could actually translate into practice the perfect storm that was happening in other fields. So better results happened, started happening in face recognition and the Googling and the Facebook of the world come from computer vision, and it spread through deep learning architectures into many other fields. Companies like CPG companies and non-traditional high-tech computer companies, they were behind and PIPA started filling that gap, and we have finished more than 60 projects with different clients and partners on anything that you can imagine from manufacturing and farming and other areas. But the one thing that we focused pretty fast was AI, life sciences, nutrition, and health. So what we have developed over through these assignments, projects, and we consolidated as a product is everyone, what they want to do is to have innovation ... They want to come up with the new product to hit the market. The better, the better snack that is not only taste good, but also it's healthy for you. They might be probiotic that would be good for both your gut and your brain. So we created that product, and so what we did was we created an AI engine that is able to ingest information from all the papers over there, medias of papers, absence of full papers. They even get in databases, information that's already curated or information that is from raw ... to profiles and other profiles like clinical trials, putting them all together and coming up with the answer to any kind of question that would be relevant to R&D, starting with, for example, what are the novel biomarkers that reduce inflammation, for example? So ... this is the product that we are having. This is what we are providing to our clients, doing it both as a software, as a service, and use cases. We ... by informaticists, they're the scientists.
Emmanuel Fombu:
We have someone in the life science industry listening to this podcast right now, right? So, you have to be like, like a typical case study, something that would be exciting to you, like ..., what would be the ideal, beautiful use case?
Ilias Tagkopoulos:
Yeah, so to me, all it matters is the final impact to all the people, to our society. So to me, the idea of these cases is something that we can actually create as a product and say we, of course our client or either in partnership or after this engagement, and it can be something that it will help people to live a better life. So this can be, for example, gut health, think .... This is, let's say, IBS, ulcerative colitis or Crohn's disease or anything like that, right? So can you go from data and predict what would be the product? Maybe it's not, or it can be like the supplement. Now, it can be an ingredient, for an ingredient company, it will make these people have less of the symptoms or even have more much milder version of the disease, if not the disease at all. So what we do in that case, it's a very important step, first step is, we're taking all this data and we analyze it. And what we have found is that not everyone is same. So we may have the same symptoms, but the underlying cause of the disease is very, very different. So the first thing we do is we are segmenting those populations and then we are focusing, laser focusing the solution to one of those segments by using, again, AI, and bioformatics, and chemoformatics. Then that final product, whatever that is, the bioactive chemicals, compounds, or the probiotic bacteria, then our client may actually take them to the clinical trials that, in collaboration with PIPA, and then bring the product to market. So that, that whole pipeline thing, the final product is what we are looking for. We're not looking for one-off engagements where they are very ephemeral.
Emmanuel Fombu:
So I think this long-term kind of relationship, because you get to understand these conditions over a long period of time, right, that you improve the overall well-being of a patient that way that you are working with. And so ideally, when would you like to get involved with life science companies? You'd like to get about before the ... study? What would be the perfect time to reach out to you?
Ilias Tagkopoulos:
Yes, that's an excellent question. So first of all, we are ready to collaborate in any way that will have that specific vision and impact in the end, right? So we have companies that are coming early on and say, okay, we know that this is one area we want to focus to bring help for example, like help us tell us how we should design the study, what will be computational, what will be the experimental part, what are the milestones that will go forward? We can be there, or there are other companies that have done a clinical study already, it's not as successful and they need us to go there and make the data even more focused and better, suitable for their market lands. So again, ideally, the earliest this is, with a more very, very clear concept and goal, the better it is for us because we can optimize it. But again, this is not an ideal world or we're working with any case as long as it has this final goal that aligns with our mission.
Emmanuel Fombu:
And also just to add to that, but if a study is already done, when you have a great analytics ... platform and insights, so there's still a lot of insight you can get from datasets, right, from studies that maybe had happened in the past. So you offer those services as well, correct?
Ilias Tagkopoulos:
Correct, and sometimes it is funny, we can even predict the success or failure of clinical trial by knowing what the inclusion-exclusion criteria, what they have done, because we can see from the computation part that this is focused enough or is not very constrained and will not get the right pathways involved by the genes and the right people involved in order to be successful.
Emmanuel Fombu:
So ... just to add to that point, I think you said something that's very important. I mean, I've worked in designing clinical studies for several years in my life science side of the industry, and I think a lot of times how we design a study, we ... spend our time thinking about how you execute a study, right? Now, you have the right patient population in that study, you have the right secondary endpoints, primary endpoints. And so you do the studies and sometimes you end up messing it up because the product didn't work, is how you design the study, or, correct?
Ilias Tagkopoulos:
Very true. Very, very important, yes.
Emmanuel Fombu:
So also, what do you think it's a, in an ideal world, if I was, as you look at in terms of improving the way clinical research is designed and some of the services that you could provide from your company perspective, right, what would be the ideal state to play? If I'm going to design a clinical study, how should I engage to you ...?
Ilias Tagkopoulos:
Yeah, in a perfect world, before you even think about the clinical study, right, thinking of trial. First of all, let's see what the goals are, right? I mean, why are you going after that disease, for example? All that, who is your audience? Who is the target client? And let's do, create and make sure that this business plan is very solid. What we have seen is that many people don't take the data and analytics into account when this is happening. They are more traditional in the way, and a more sometimes gut feeling, sometimes more evidence-based. But the work has changed over the past decade, and people who are being left in the, they may be amazing, you know, ten years ago, but now with all the data analytics we have, the people need to engage those companies, partners or associates within the same company early on. And the design going to the final product, what you want to achieve will need to have this AI component at least discussed throughout the whole continuum of the product development. Again, companies, they have a very clear goal in terms of, hey, this is the market we want to hit. We want a new, for example, pill for people with diabetes that they have also premature arthritis, right? Okay, how do we do that? So that is when we start asking that question. This is a time for us to be engaged and maximize the potential.
Emmanuel Fombu:
Perfect, I really like that, and I think that's something that I think in the life sciences industry in general, I think that's something that we should take into consideration. And I'll tell you that initially, we designed some research, an anti-... research, and then go back and say, oh, I wish I did it this way, I wish I did this way. And so if you are listening to this and you are part of a clinical research business in the life science industry, I think that what Ilias is talking about, definitely reach out to him, make sure we have very efficient kind of studies, right? And we have the right results and we have right products to make to the market that would benefit patients that depend on this, right? So what has been the biggest roadblock so far in industry, in adopting technology? And you have the background on this and so, you're an expert going through roadblocks, so what has been the biggest roadblock in the healthcare industry to adapt all the possibilities that AI brings to us?
Ilias Tagkopoulos:
Yes, thank you, it's always, I, it's good to remember how it was and see where you go, because you don't have a set goal nor, if you don't know where you're sailing, with this ... was, ... was saying. The rulebooks have changed over the years, .... was mentality and especially more in the nutritional space and CPG space rather than healthcare, although healthcare also has its own issues. If you looked at medical informatics ten years ago, there was no AI, there was no machine learning. The closer you could actually go, it was like standard statistics and a lot of statistics. So mentality has been an issue early on, not so much now in terms of whether AI is a hype or not. And don't get me wrong, AI has been hyped in many encounters with many people, but the potential, now people see it because they see actually, they start seeing results. Then the availability, you need two things to happen, to having a successful product, even if you have an amazing AI company or capability. One is having the right data, so both availability and also the quality control, and quality of the data and the metadata involved for the problem we are trying to solve. And the second is, once we have solved the problem, meaning that you have a predictor, you have a product, then you have to have adoption, and then you have to be able to get people to buy in what you're doing for the right reasons. In both, those have been challenges in the past and even now. So that, I would say, is raising this gap is the key to success.
Emmanuel Fombu:
Which is also very interesting, in all things AI force or apply AI .... things, that's right? And something in healthcare ... development, I think what we consider medical health data is changing a lot, right? I'm sure environmental things, for example, factors, the things that don't go necessarily in your EHR rec system, right? It's part of health, that, right, ... other people interact on Facebook or Instagram or whatever they interact on. Some of those things are becoming datasets that you can actually capture from the real-world, piece of it. So as things go, what are your thoughts around data privacy, especially if it comes to healthcare?
Ilias Tagkopoulos:
A very important topic, and especially when you are having something as potent and game-changing as AI. We need data privacy and we need it even more going forward and we need not to be afraid of it, this should not be fear, although what we'll need to do is having a very honest and frank discussion about the danger of not having enough data privacy, especially in a connected world. Because it's not only ..., it's other countries like China and India and others, who are also playing the same arena. So it's not be a barrier, it's be just one of the things to discuss and make the right, put the right safeguards going forward.
Emmanuel Fombu:
Thanks a lot for that. I think it's very important to bring this up to say, yes, we need to be concerned and have some guidelines around it, but not to be scared of it, right? I think that's, that's the way I was approaching this. So with that being said, I think it's a great point for what I need to ask you. What are your accomplishments and what are your goals over the next three months, the next year? What do you aim to accomplish?
Ilias Tagkopoulos:
For the three months, we have so many projects that we and our clients, we want to make sure that we bring value to them, so that's definitely one. Next year, it's a, well, I mean, the reason that we have created this company and the reason we do what we do is because we really want to have impact, an impact in the life of others and making sure that what we do we are proud of. So, you know, I would like to, in three years from now, I would like to point at a type of disease, or it can be a drug or it can be people, right? And say like, look, I mean, I did something good for these people. We really made a dent, even if it is very small to the universe as scientists, engineers, and entrepreneurs.
Emmanuel Fombu:
Thanks a lot for that. And I think, anyone who's listening, I think as you reach out definitely now to Ilias, we'll have your company to mention definitely, below this podcast ...., when we go out there. But I think another great vision of ... is also to grow, make sure you get more involved with life science companies, and I think any listener here should definitely reach out to learn more about what you're joining, what you're doing here to make, you might ... better. So thanks for joining us today, Ilias, and I hope to have you again on the show sometime soon.
Ilias Tagkopoulos:
Thank you, Manny. It was a pleasure.
Emmanuel Fombu:
Thank you. Thank you very much.
Emmanuel Fombu:
Thank you for listening to Bite the Orange. If you want to change healthcare with us, please contact us at info@EmmanuelFombu.com or you can visit us at EmmanuelFombu.com or BiteTheOrange.com. If you liked this episode and want more information about us, you can also visit us at EmmanuelFombu.com.
Sonix is the world’s most advanced automated transcription, translation, and subtitling platform. Fast, accurate, and affordable.
Automatically convert your mp3 files to text (txt file), Microsoft Word (docx file), and SubRip Subtitle (srt file) in minutes.
Sonix has many features that you'd love including share transcripts, automated subtitles, transcribe multiple languages, advanced search, and easily transcribe your Zoom meetings. Try Sonix for free today.
About Dr. Ilias Tagkopoulose:
Dr. Tagkopoulos is a leading expert in computational science, with a focus on biomedical and business applications. He is a professor at the Computer Science and Genome Center at the University of California, Davis, and the director of the USDA/NSF AI Institute for Next-Generation Food Systems (AIFS). Prior to joining UC Davis, Dr. Ilias worked at fixed-income derivatives models at Credit Suisse, NY. Dr. Ilias holds a Diploma from the University of Patras, an MSc from Columbia University, and a Ph.D. from Princeton University, all in Electrical and Computer Engineering.
Things You’ll Learn:
One-third of the world’s population has obstacles and conflicts to secure food.
One-third of the United States has diabetes, other comorbidities related to food, and even a combination of the two.
AI can connect for better supply chain, production, processing, and food consumption.
Diabetes and food scarcity are conditions that don’t consider any age bracket.
Today's agricultural methods are not sustainable for nature or the food they produce.
Having the right data will always enhance the use of AI and machine learning, therefore improving the outcomes of using these technologies.
PIPA can predict the success or failure of a clinical trial process by knowing its inclusion-exclusion criteria.
Once you have a solution, its adoption becomes the next problem to tackle.
AI should not be feared, but there needs to have a lot of data privacy boundaries.