Daniella Gilboa is an embryologist in Israel working to bring the power of AI and machine learning to the embryo selection phase of IVF treatment. She explains how her new startup aims to automate this error-ridden process, raising efficiency and lowering the overall cost of IVF.
Doctors helping couples conceive through in-vitro fertilization typically must screen multiple fertilized embryos to select one embryo for implantation—but the process is fraught with risk and subjectivity. from In 2018 Gilboa and her colleagues Daniel Seidman and Eyal Schiff co-founded AIVF, an Israel-based startup developing decision support tools that use deep learning and computer vision to lower the risk by identifying the most promising embryos for intrauterine implantation.
The company's technology takes the place of old-fashioned visual evaluation of embryos by humans, instead of capturing time-lapse video of embryos from the moment of conception to the fifth day after conception, at multiple focal planes. "It's an obscene amount of data," Gilboa says. "Instead of looking at the embryo once a day under the microscope, we have tons of images to annotate and look for the biological features that we know are correlated with success."
Proprietary machine learning algorithms use the video data, together with patients' health history and genomic data, to predict which embryos have the highest chance of developing into a healthy newborn. In theory, the technology will lower failure rates, decreasing the number of fertility cycles required for conception and therefore lowering the overall cost of IVF treatment.
"Many people don't get to fulfill their dream of having a child, and this is really heartbreaking for me," Gilboa tells Harry. "This is what really drives me as an embryologist to be able to provide a new, next-generation IVF treatment that would be accessible, that wouldn't be so expensive."
Check out the full show notes for this episode and other MoneyBall Medicine episodes on our website. For more on how data is transforming reproductive medicine, listen to Harry's interview with Alan Copperman.
Please rate and review MoneyBall Medicine on Apple Podcasts! Here's how to do that from an iPhone, iPad, or iPod touch:
1. Open the Podcasts app on your iPhone, iPad, or Mac.
2. Navigate to the page of the MoneyBall Medicine podcast. You can find it by searching for it or selecting it from your library. Just note that you'll have to go to the series page which shows all the episodes, not just the page for a single episode.
3.Scroll down to find the subhead titled "Ratings & Reviews."
4.Under one of the highlighted reviews, select "Write a Review."
5.Next, select a star rating at the top — you have the option of choosing between one and five stars.
6.Using the text box at the top, write a title for your review. Then, in the lower text box, write your review. Your review can be up to 300 words long.
7.Once you've finished, select "Send" or "Save" in the top-right corner.
8.If you've never left a podcast review before, enter a nickname. Your nickname will be displayed next to any reviews you leave from here on out.
9.After selecting a nickname, tap OK. Your review may not be immediately visible.
MoneyBall Medicine - Daniella Gilboa
Harry Glorikian: Hello, I'm Harry Glorikian. And this is Moneyball medicine. The show where we meet executives, entrepreneurs, physicians, and scientists using the power of data to reinvent healthcare from machine learning to genomics, to personalized medicine. We look at the biggest trends in patient care and healthcare management.
And we talked to people behind the trends to find out where data is making the biggest difference.
After speaking to Dr. Allen Copperman and understanding more about IVF and how it has evolved to embrace data. I decided to dig a little deeper. I spoke to my good friend, David Sable in New York, who is my go-to guy on everything at the intersection of investing and reproduction. And he was kind enough to point me towards our next guest.
For those of you new to this discussion. Let me explain IVF or in vitro fertilization is the process of combining a female, egg and male sperm outside the human body and implanting the resulting embryo into the uterus. This is the most common method today to tackle fertility issues. However, it's not necessarily the ideal solution, IVF treatments, global success rates range between say 35 and 40%.
And the costs can easily reach a hundred thousand dollars not to mention stressful conception cycles that can exhaust the couples both emotionally and financially. My next guest who is truly passionate about this area is Daniela Gilboa. She started a company in Israel last year with professor Daniel sidemen and professor AOL Schiff designed to provide embryologists with an end-to-end decision support tool, using the power of deep learning and computer vision algorithms.
Daniella, welcome to the show.
Daniela Gilboa: Hi, thanks for inviting me.
Harry Glorikian: Daniella, tell us a little bit about yourself and how you ended up at, uh, AI, VF.
Daniela Gilboa: Okay. Um, I just wanted to start off with, uh, thank you so much for inviting me in, and it's such an honor, um, to be here and thank you, David. Um, we, we met at, in New York.
Um, I was just there. Uh, visiting Cornell. And so we had a really good discussion about, um, IVF and, and you know, the market and where it's evolving. Um, I, myself, I'm an embryologist for many years. Um, actually I have two hats, uh, embryologist, a clinical embryologist and a biostatistician. And so, um, uh, I was always interested in, in the data side of, um, of IVF and how we could leverage data to really, um, do IVF differently.
Um, and so I was working, um, I've been working in, um, IVF labs, um, in Israel. Um, Israel is, um, considered, um, as a powerhouse of, of IVF. Um, since, um, the, the treatments here are subsidized. We do a lot of, um, IVF cycles. The patients here are entitled to, um, most, uh, I think, um, endless, uh, number of cycles. And so, um, most of the patients who most, most of the people here that, that, um, I meet.
And so during IVF, during the IVF process, most of them, and that was the baby. So. So this is, you know, this is where I come from. Um, I was just in, in New York and I met with, um, an attorney that, that is, you know, I was, I wanted to think with her about the market and you know, where, where things are going, um, in IVF.
And she told me her story. Um, she tried IVF for many years and it didn't work out. And then, you know, she, she. Um, she divorced and she's now, um, in her mid fifties with no children and these kinds of stories, um, for me as an Israeli, you don't know, really get to hear. And so it really hits me that, um, some people, many people, um, don't get to fulfill their dream of having a child. And this is really heartbreaking for me. The, this is what really drives me, um, as an embryologist to be able to provide, um, uh, a new, I would say next generation IVF treatment, that that would be accessible. That wouldn't be so expensive. That would be that people could, you know, people could have children.
Harry Glorikian: So let me, let me, let me ask the question where you're not actually changing the process necessarily, but you're, you're utilizing data to add insight and, and change the odds by picking the right embryo or, uh, you know, walk me through,
Daniela Gilboa: Well, I'm optimizing the process. I'm optimizing the process. That's what I do. As it is today, it's kind of like a trial and error because you start IVF and you wouldn't know, you don't know how long it will take. You. And, uh, where would you be, um, where you, you will find yourself in the end.
Harry Glorikian: So, tell me why, what is the data that's, you're bringing to the table that isn't necessarily easily available in other places, right?
That. Is allowing you to do this?
Daniela Gilboa: So when I, you know, when I started, uh, doing embryology, we would check the embryos. Um, every day, once a day, we would take out the dish from the incubator, look at the embryos, um, under a microscope, um, count the cells, uh, try to figure out how many fragments there are and put it back in the incubator and that that's what you would get for embryo evaluations. So after three days or after four days, you would get like, um, uh, a picture, I would say, uh, you know, how the embryo looks the, we evaluated the embryo once a day and then the chronology started, um, going into the lab and, and now we have incubators that actually, um, take, pick, um, uh, take a video of the developing embryo.
And so you get time-lapse video from the very beginning where you, um, just minutes after you inject the sperm inside the egg. And it takes, uh, any films, the embryo, all throughout the development until the minute you, uh, transfer the embryo back to mommy's uterus. And so it's, uh, it's about five days. And you would you get, you get so much, um, uh, data, um, from different focal planes because the, this, these incubators, um, the camera is taking pictures every, um, about 10 minutes from different focal planes.
So it's like you get a 3d, um, uh, embryo, uh, for the whole five days. So it's just, I would say an obscene amount of data. And so what the embryologist now have to do instead of looking at the embryo once a day under the microscope, we have tons of images to actually annotate and look for the biological features that we know are correlated with success.
And this is just undoable, um, for know, irregular IDF setting.
Harry Glorikian: So is it just, the data is just from the images. Or is there EMR data? Is there information coming from the patient? Is that, you know, are you looking at
Daniela Gilboa: Yeah, we’re looking at, yeah, we're looking at everything because basically, um, it's what we believe in a holistic approach. It's not only embryos. You've got the mother, you've got the embryos, you've got, you know, the history, the medical history of the mother. You've got everything. Um, so you have to look at everything. So we, um, we gathered, uh, data, um, everything that is related to the mother, everything that is related to the embryo, we looked at it.
Harry Glorikian: So that's possible. I mean, let's face it now. It's not every place where you can do that. I mean, Israel has a centralized EMR system, which lends itself to data capture and making that data available to be able to be utilized in this sort of situation.
Daniela Gilboa: Right. Right. But the thing is when we started out, um, we said something really needs to change and you've got so much data.
Um, let's try to look at the data we started out with, with looking at the, you know, the, the, the time-lapse images, um, of the embryos and, um, What I believe in really believe in is that, uh, if you really want to, um, uh, change IVF, you really, really need to understand IVFs. Um, you really, you need to understand the IVF from the inside and out and insight again.
So. Since we're deeply involved in, in, you know, IVF myself and, and, um, Danny and they are the two physicians that we, uh, that are, um, co-founded this with me. Um, since we're so deeply involved, we, we, we know people, we know other clinics, we know, you know, we know the, the ecosystem of IVFs. So, um, so we reached out.
To, um, um, I think talk, not the best clinics in the world. And we said, join our mission in doing the new, next generation IVF treatment. So we approached, um, Stanford and Cornell and, um, um, IVI, um, in Spain and, and, uh, um, LWC inLondon and embryo lab increased. These clinics really change the way we would do IVF.
And so we've got, um, we're collaborating with, with, um, other clinics where they provide us with data and IVF knowledge. And we put in the, you know, the AI knowledge that in Israel is. Uh, you can find it. Yeah. It's like, everyone's doing AI here. So, uh, so we're in the process of really not only research collaboration, but, but also, uh, designing the product that would be suit for, for the patient and for the physician and for the embryologist.
And, you know, I think all stakeholders, um, should talk, should be able, um, To get involved in this process. And so, um, so we have, I think now we have the, the most, um, reach, uh, the largest database, uh, IVF database in the world.
Harry Glorikian: So Daniela, how are you, how are you guys applying AI and, and you know, what are you trying to create?
But let's start with the first one. Like how, how are you employing an AI, I, you know, that, that word drives me crazy. Right. Because it's just a bunch of different tools. Right. So I'm sure you're applying different technological approaches. And so the question is, is how are you using AI in what you're doing?
Daniela Gilboa: Um, okay. So, um, the, the videos, the embryos, which is really the most beautiful thing you could, uh, look at, it's just a, it's a developing human embryo, um, from the, from the one cell stage to the blastocyst stage about five days, um, just before, um, it gets transferred in the uterus. You know, our platform or what we developed the platform with, um, softwares that would be, um, that would fit in each, um, decision points you have during the process, our platform would enable you to, um, make data-driven decision rather than subjective human analysis.
And so, for example, let's say the hardest decision you could, um, take and in the, in the process is which embryo transfer back to mommy's theater that will result in pregnancy. Uh, this is the $1 million question that you have. You have like 10 embryos, one patient, 10 embryos, which one would you know which one to choose?
So you could look at the embryos and devaluate with, uh, different, um, methods that were, you know, accumulated throughout the years. Or you could have a decision support tool, a software that would look at the embryo and maybe find features that are not even seen, but the human eye and maybe find features that we know, but we can, we never, we never modeled them.
And find, oh, I don't know millions of people features millions of features that are correlated with success, which is implantation and would tell you exactly, um, which one to choose as in, you would get as an embryologist, you would get, um, a software that would tell you the probability of success for each embryo.
And so this way you would have data-driven um, you would get a data driven decision as an embryologist, rather rather look at the embryo and say like, like the gold standard grading as it is today, which is like, uh, eight B or something like that. You would get a probability of success. This is one example.
Another example is that you would have the patient. Being able to be much more involved in the process. She's now the patient that the couple, um, their voice is not heard enough. They're they're not part of the decision-making. Um, so she would be able to log in, uh, to the lab, to our platform, see the embryos, understand what's going on.
Um, uh, talk to the physician, talk to the lab. Everything would be. And um, on one, uh, um, one platform. And the only way to do that is by looking at the, taking the embryos, the videos, and adding the, the, all the history, the medical history of the mother, everything, every parameters you could think of and, um, and something that is not, you're not able to do the human mind is not able to.
To do in one specific moment at the lab.
Harry Glorikian: Well, um, I'm actually dying to take a look at it, right? Because it sounds, it, it sounds not just fascinating from the aspect that you're standing, that you're explaining, but just the, the interface. I can almost see it being used in, in different situations, not just the one that you're articulating, if that makes sense.
Daniela Gilboa: Yeah. Yeah. And I know this is only one, one, um, uh, example, there's another example of, of genetic testing that is now, um, an invasive procedure and we would do it, um, non-invasive with, you know, just, um, AI, deep learning neural net that would beable to visualize the embryo. And, uh, find features that are correlated with, uh, genetic abnormalities or chromosomal statues or something like that.
So these are all these developments are old. Um, we could actually develop these, um, products now, 2020 or 2019, um, because of the, the, um, you know, AI abilities that. There that are seeing basically all around us.
Harry Glorikian: Oh, I'm, I'm, I'm one of these days. I'm I would love if you, if, you know, if we can schedule it and make the time zones work, I'd love to see a demo of it.
But, you know, assuming you know where you're going and what you're doing in your experience, where, where do you see, where do you see the, the field in the future?
Daniela Gilboa: Yeah, it's, it's, uh, One of the, I think that the most beautiful thing you could see and with our, um, um, AI software, you you're actually, um, able to see the embryo and what the, the, um, system, um, found to be that the features that the system found to be correlated with success and uh, explaining to you why this embryo is, um, um, good or bad. Uh, it's, it's an, it's really unbelievable unthinkable, but it's, it's just happening.
Harry Glorikian: So, but when you say good or bad, I think we're, we're framing it as. Something that has a higher probability of
Daniela Gilboa: Exactly viable embryo viable, healthy embryo. That's that's it
Harry Glorikian: So interesting. Right. I think about this with all the other pieces that are coming on, like you said, you know, genetic testing and even crispers, and it's sort of, uh, interesting, uh, You know, shape of how things are changing.
Daniela Gilboa: I agree. CRISPR is, you know, something that we, we can leave aside, although-
Harry Glorikian: Not for long, from what I can tell
Daniela Gilboa: It's out there. It's out there and the technology, I guess, is not ripe yet, but it will be
Harry Glorikian: Well I'm, I'm also looking at this compared to something like, you know, baby seek that, uh, you know, Robert Green is, uh, is doing in his lab. And so, you know, where you can sequence based on every child that's born.
Right. So you know, it's there, there's such a constellation of, um, Technology's coming together. It's converging into this area where, um, you can see in the next five or 10 years, it has a profound impact on human life.
Daniela Gilboa: I agree. I, you know, I agree. And I think IVF is really becoming something different now, nowadays, because it's not, you don't do IVF only under medical indication.
You would do IVF because you know, it might be, if you want to plan your future, fertility, IVF would just be there if it's optimized and the treatments are not so, um, hard and tough as it is now, if you would be able, um, you know, to go to get to the physician or log into any system and, um, get the medication needed.
And you would have, uh, I don't know, like five, six, um, eggs. And then after five days, one of the embryos would be implanted back and the four left would be frozen. And then you would get pregnant by a month. I mean, the time to pregnancy would be a month or two. So all these, um, so you wouldn't need, um, medical indication to do IVF, you know, people, um, freeze their eggs.
Now people delay childbirth birth. So it's, it's, it's there, it's happening. It's part of planning, future fertility. And so I think IVF will be um, the demand for IVF will just go grow bigger and bigger and, um, the only way to actually do it correctly is by using AI technologies.
Harry Glorikian: So how do you see this lowering the cost?
I mean, is it, uh, you know, 50% reduction? Is it, what do you see this driving to?
Daniela Gilboa: Well, you know, this is, this is an interesting, um, question because, uh, it's, it's, it's complicated basically, but for the patient, you know, for, for, uh, in the U S um, the calculation is about 20, about $20,000 per cycle. And since you know, you don't know how many cycles you would need to have a baby on average, you need about five cycles.
So that translates into about $100,000. That's insane. Uh, in Israel, we do very good IVF for $2,500. It's not 20,000 but 2,500.
Harry Glorikian: We should all move. We should move there when we want to conceive, I guess.
Daniela Gilboa: So that's the thing. So, so, so basically we, you know, the world or, or a science know how to do good idea.
So it's not the money, but I think that the way, you know, AI technologies would, um, would change the market is, uh, for the patient. You would know your time to pregnancy It wouldn't be like, uh, as it is today, you start and you don't know how much, how long it will take you. So with something that is all data-driven AI driven, then you would know you would know your chances.
You would know how, how long it will take you. Um, you would know whether you should start IVF or maybe go straight to, um, egg donation. These, um, things that will really change the markets.
Harry Glorikian: Right
Daniella Gilboa: That's the optimization I was talking about.
Harry Glorikian: yes, no, I can, I can see a lot of it happening in my mind now. I'm actually, you know, in the back of my mind, I'm wondering why we haven't done it already, but, um, you know, you're a startup, where are you now in the product development and when do you anticipate, you know, making this available for.
You know, prime time?
Daniela Gilboa: Um, well the, the first one, um, product would, um, go into clinics and a few months then, um, basically beginning of 2020, um, we'll, we'll start clinical trials with all our collaboration, collaborating our clinics. Um, and so this is pretty amazing, uh, uh, prospective trial where you would get the patient uh, consent to having an algorithm, um, choosing for embryos.
Harry Glorikian: So would it be the algorithm doing it or would it be the algorithm suggesting it to the
Daniela Gilboa: Suggesting it's a decision support tool. So you would, you always have the physician or the embryologist or, you know, the clinic, the clinician basically involved, but you know, you wouldn't have the, the embryologist uh, annotating the embryo for hours and hours and deciding whether this one or this one or this one or this one, just something that would be much more accurate. And, um, well-defined and also, you know, for the patient. Um, it's, it's, it's funny, but the way we explain. The, the embryo, uh, to the patient is something like you have a nice embryo.
You have a cute embryo. Your embryos are really cute. We would transfer the one that the cutest that's what we say. And when she asks, um, so what are my chances? We didn't know what to answer. It's like, it's, it's, uh, like I would say a guesstimation. And having such a data driven process. Now we would be able to explain to her what, um, how we, um, um, validated the, the, the, the embryos.
Uh, what does it mean that the, uh, the, uh, the embryo is, uh, 77%, uh, Why we choose to transfer the, the one that is, um, 85% and three is the ones that are maybe 95%. You know, we would be able to explain all these things to her. And the thing is that she signs a consent that she understand everything.
But she cannot understand everything. If, you know, if the way we explain her embryology is by using terminology as nice and cute. So it's, it's a, it's a whole, you know, it's a different game.
Harry Glorikian: Well, and it's interesting because if you have historical images, I know you said five days, but if the system can actually understand the most important features, it might even be able to tell you earlier.
Daniela Gilboa: Yeah, exactly, exactly. Exactly. We would be able to choose the Embryo after 24 hours and not after five days, we wait for five for five days because we really don't know which one to choose. So exactly everything would be optimized.
The lab work, the lab flow would be optimized because we would know which one to choose to transfer, which one, which ones to, um, freeze. We wouldn't freeze everything. We would freeze just the ones that are, that would be able to, um, succeed after throwing. But we would know for sure that these are the embryos.
Um, so no more guesstimation, everything would just be that's what I said before optimization.
Harry Glorikian: What you're describing is multiple services from one product capability,
Daniela Gilboa: Right? One platform, a platform with few products, but a platform and this platform would, would be able to connect with the physician, connect with the embryologist and connect with the patient.
And this is tricky because each one of these stakeholders, want different answers. The, the, so yeah
Harry Glorikian: So my assumption is this is going to be a cloud-based.
Daniella Gilboa: Yeah. Yeah.
Harry Glorikian: Because I have to believe that there's differences in different populations that you need to take into consideration.
Daniela Gilboa: Exactly. The cloud-based.
Yeah. Yeah. This is, uh, this is why Israel has an important role in this game because the data here is such so heterogeneous. We have, uh, patients, you know, older patients, young, younger patients, the patients that have, uh, Um, medical history of, of, you know, many IVF cycles. And so these kinds of things, you don't really get to see a lot in Europe and in the U S because it's not subsidized.
So, and so here we in Israel, we, um, we are collaborating with, um, with, uh, sit-up medical center, which is the leading, um, one of the living clinics in, in Europe and in Israel. Because he does, um, in terms of volume, it does about 15,000 IVF cycles per year, which is, which is unbelievable, you know? Only one clinic,
Harry Glorikian: Well it sounds very exciting. Um, I I'm sure that my children cause my days of doing this or over, but, uh, my children are, uh, will, uh, may, may look forward to this sort of technology that men might make. Um, Their life easier. You know, I, uh, I look forward to keeping in touch. I would love to see a demo of the, um, of the technology just so I can broaden my imagination in the space.
And, uh, thank you for the, for the time. And, um, I look forward to keeping in touch.
Daniela Gilboa: Yeah, well, we'll keep in touch and thank you so much for inviting me. It's just a, it's always a pleasure for me to talk embryology. So anyone that, you know, that. Gives me the stage talk embryology I’m there. Excellent.
Harry Glorikian: And that's it for this episode, if you enjoyed Moneyball medicine, please head over to iTunes, to subscribe, rate, and leave a review. It is greatly appreciated. Hope you join us next time until then farewell.