At Univfy, Mylene Yao Is Making IVF More Predictable and Affordable

Harry Glorikian: Hello. I’m Harry Glorikian, and this is The Harry Glorikian Show, where we explore how technology is changing everything we know about healthcare.  

Last month on the show, I brought you an interview with David Sable, an investor and a  gynecologist who’s one of the best-informed people I know when it comes to progress in assistive reproduction technology, especially in vitro fertilization or IVF. 

Every year about half a million babies are born through IVF around the world.  

But David believes that number would be about 40 times higher if IVF were cheaper and more accessible. 

Making that happen would mean transforming IVF from an artisanal craft into something more like a modern automated factory, with AI helping  doctors and technicians make faster and better decisions at every step. 

And that’s exactly what this week’s guest Mylene Yao is doing.  

Dr. Mylene Yao is a physician who completed her residence in OB/GYN medicine, then went on to study fertility and embryo genetics at Stanford, and then launched her own company called Univfy.  

That’s spelled U-N-I-V-F-Y. 

Mylene Yao on The Harry Glorikian Show

The company helps patients with two particular aspects of the IVF process. 

The first is helping them assess the odds of success before they decide whether to invest in one or more cycles of fertility treatments. 

Dr. Yao did research in the 2000s showing that the advice many couples were getting about their chances of success with IVF wasn’t very good. 

The mother’s age is the biggest predictor of whether IVF will lead to a successful  pregnancy.  

But using early forms of machine learning, Dr. Yao showed that age alone explains only 50% of the variation in IVF success rates.  

The other 50% depends on a whole variety of factors.  

And one of the services Univfy provides is to collect more diagnostic data and run it through machine learning algorithms, to provide a more personalized prediction about whether a round of IVF will succeed. 

That way, patients can make better decisions about whether to spend the money.  

And we’re talking about a lot of money. A single cycle of IVF services can cost as much as $30,000. 

That’s why the other aspect that Univfy helps with is financing.  

Univfy works with a bank called Lightstream to provide up to $100,000 in financing for up to three rounds of IVF, with a large refund as part of the deal if the treatments don’t result in a baby.  

Univy’s machine learning models help there too, by determining who’s eligible for financing and making it less risky for Lightstream to offer  the refunds. 

In my mind, the only way we’re going to realize David Sable’s vision to democratize access to IVF is if more innovators like Dr.  rYao get into the business, and start to meld progress in medicine with ideas from the worlds of technology and finance.  

So it was a real pleasure to talk with Dr. Yao and hear how things are going at Univfy. 

Here’s our full conversation.  

Harry Glorikian: Mylene, welcome to the show. 

Mylene Yao: Harry, I’m so excited to be here. Thanks for having me. 

Harry Glorikian: Yeah, I was you know, I was reading about the company and just fascinated by a lot of the stuff that you guys are doing. Now. But I want to sort of, you know, set the stage for everybody and just say like, so the company Univfy offers sort of an AI based counseling service that can provide useful predictions to patients who would like to conceive through in vitro in vitro fertilization.

But before we even get into how you developed the technology, I’d love to step back and talk about the the real fundamental stuff because listeners may not understand the basic challenges involved in IVF unless, you know, they’ve been through it through one or more IVF cycles themselves.

So maybe we can just start by explaining some of the the basics like. I don’t know what is the size of the overall market? How many women or couples who want children find they can’t conceive naturally. I mean, just starting at the basics and then we’ll work our way up. 

Mylene Yao: Mm hmm. Yeah, sure. So I think, you know, I came into this really as a clinician originally, so I come from a clinical background. I was OBGYN, training to be fertility specialist, completed my training. And through many of those years in the fertility clinics in different hospitals, I would see like patients that are being counseled.

IVF was the best treatment, most effective treatment for many of them. And so now today it is a treatment that’s been in place for over three decades. Tried and true. It’s the most effective treatment for most couples wanting to have a family. However, you know, you’re asking what are the challenges?

There are many.

But the first big challenge is it is very much underutilized. In the US, there are estimated to be about 7 million couples that need fertility care to have a family, but that the need for fertility care is way beyond that now because individuals and couples from LGBTQ community, many of them want to use fertility care and need fertility care to have a family.

There are men and women with medical conditions such as cancer, where they really need fertility preservation, which also involves some of the procedures that are used in IVF to preserve their fertility. 

Mylene Yao: And also egg freezing is not experimental anymore. It’s tried and true, there are long term results coming from that now. So single women actually do have an option to preserve their fertility. So there is however, the cost is still quite prohibitive for most people in the US, but also around the world.

And so this is really a US and a global problem. And then the problem is not just that it’s expensive, it’s actually not that expensive compared to many other medical treatments and surgical treatments that we take for granted. If someone goes skiing and tears their meniscus, you get a repair, you go to see your orthopedic surgeon.

No one even blinks an eye. IVF is not much more expensive than that. However, for most people, they don’t have insurance coverage. Medicare doesn’t cover it. So you have a big, it is, first of all, a medical problem that has a huge emotional aspect to it, of course, social aspect, but also it’s also tied in with financial challenges. So it is such a complex problem to navigate for many of these patients. 

Harry Glorikian: Yeah, I if I’m not mistaken. I mean, I was trying to look all this up, but. But it’s somewhere between the $10,000 to $20,000 range, right, for a single round. 

Mylene Yao: Right. Right. So for the first, the IVF costs varies across the country depending on local cost of living. And on average, the estimate is around $12,000 to $15,000 per cycle per treatment. But really, if you’re talking about on both coasts, it could be as high as $30,000 or over. Yeah.

And, and that actually, so, just so that’s the basic information. However, to make this even more challenging, each and IVF treatment may not result in a baby even if. 

Harry Glorikian: Right. 

Mylene Yao: And in most situations doctors are doing an excellent job. You know, embryology labs are doing an excellent job. Everybody is doing an excellent job. And you may still not have a baby and it is a probability game.

So, you know, if you think of it as, instead of tossing a coin where, you know for certainty is 50/50, heads or tails, in IVF, depending on the woman’s and her partner’s health data, that probability may be 70?30 or it may be 30/70 or 20/80. Right. And that’s the challenge.

And that that personalization, you know, having insight into what that means is very important and a lot of times not available or not possible without AI and machine learning. 

Harry Glorikian: How many how an average how how many rounds does a let’s, you know, on average somebody have to go through before they conceive. 

Mylene Yao: Right. So most couples have a very high chance of having baby if they could do up to three cycles now. So so the good news is IVF is very effective for most patients.

The bad news is most people can’t afford to do three cycles. Right. And they may not even be able to afford one cycle. But even if they could, they may not be able to afford multiple treatments and also even if they could afford multiple treatments if if after the first failed cycle, it is so emotionally and financially devastating.

And even if someone has employer’s coverage or they have the resources, it is still so emotionally, so emotionally devastated that it may be difficult for the patients to even think through How am I going to do more treatment and put my body through this again.

There are also side effects as well, so it is not an easy treatment to get through. So those are those are the challenges. But yeah, but this is not unique in medicine. So like for example, I think you don’t need to be an oncologist to know, oh, a patient has a certain cancer, they may need chemo.

A course of chemo may require going in six times. Nobody tells the patient, Oh, why don’t you just try one session, see if you get remission? And if you don’t get remission? Great news. You save some money. If you get remission, uh oh.  that’s when you consider paying more money for more advanced chemo. I mean, no one does that! 

Harry Glorikian: And if and if it did work in one session, we would charge a lot more for that one session. 

Mylene Yao: Right! We’re, our society is not viewing, no matter what everybody says, we are not really as supportive of family building as we all should be. And, you know, so because we all understand a patient with cancer, they should be entitled to get this course of chemo. But why aren’t patients navigating fertility care be having the same access? 

Harry Glorikian: So this sort of, you opened the door a little earlier by a couple of the comments you made. But this let’s sort of open some questions about how do women or couples sort of make decisions about IVF? So Univfy is built around the notion, if I’m not mistaken, that it’s useful to have sort of reliable estimate of the likelihood that IVF will lead to a successful pregnancy.

And you mentioned that earlier, but. You know, I don’t know. Why are those predictions so important? What was the state of art in IVF prediction before Univfy came along? I mean. Can you explain why predictions based on age alone may be inaccurate? I mean, these are some of my questions that might help people understand what you guys do. 

Mylene Yao: Yeah, Yeah, that’s a great question. So that is really the fundamental breakthrough that we had. And this goes back to my time when I was faculty at Stanford. I was assistant professor there. I led and I funded research projects. I had my own independent lab on the tenure track and it was amazing that I got to team up with my company’s co-founder, Professor Wing Wong, who at the time was my collaborator.

And we brought our teams together because, you know, I saw these problems happening in the IVF clinic that people, you know, every patient wanted to know what are my chances to have a baby from this treatment, if that’s what the doctors are recommending? Is this going to work for me?

And I felt like, wow, as doctors, we really want it to give people this information and personalize it because first, you know, we really doctors believe in giving accurate prognosis in order to support decision making. And patients should be informed of their chances of having a successful treatment, whatever the treatment is, before making that decision.

So that’s a fundamental responsibility as doctors. But then in this situation, knowing the probability and having an accurate prediction. We’ll be able to tell the patient how many cycles might you need, and that can help the patients have the financial planning to be able to afford the care or to be able to make their personal trade offs. 

Mylene Yao: The best path is not the same for everybody. This is a personal, subjective decision. How much money you are willing to spend on something like this. But it’s not possible to make the personal trade offs or make this decision if you did not know how many treatments might you need and how might how much might it cost in the end?

So I am a big believer in empowering women and couples, but here particularly women, because they’re the ones putting their bodies through it with this information and transparency so that they can make the best decisions for themselves.

What is really terrible that you hear all the time is people saying, had I known, had I known this, my chances were so low, I would have done this other thing, or had I known my chances were pretty good, I would have given it a shot and now it’s too late, you know?

Or now I chose this other alternative, this other treatment that just because it was cheaper and I didn’t realize I would have had a better shot with that other treatment and now it’s too late. I think those types of regrets should not be necessary. Not everybody may come out with a baby, but everybody should be informed. What are their options? 

Harry Glorikian: So how does the Univfy algorithm predict a higher probability of success compared to just age based predictions alone. 

Mylene Yao: Right. So traditionally in medicine, and this is not just fertility, it’s all areas of medicine, if you look back at all these research papers, the approach had always been hypothesis driven. And there are there’s a wealth of knowledge. So it’s very important.

There’s a wealth of knowledge from decades of medical research that have led doctors to already have databases of the IVF data and outcomes. So all this data is already there, which is great news. And we already knew that women’s age is very important and is a very big, or probably the biggest predictor of fertility success, no matter what treatment you pursue, and especially IVF treatment success.

However, because it was so doctrinaire and when we started doing this research, I thought, you know what? Why don’t we mix it up? Why don’t we pretend we don’t know anything? Because by thinking we know everything, we have not helped everybody. So let’s kind of cast our knowledge aside and let’s pretend, what if age was not that important? I don’t think so. But what if? And as doctors, we really were very tied to these clinical diagnosis.

Oh, this patient has this ovulation disorder. This patient, this couple has this male factor. Well, all those diagnoses are really a result of the available diagnostic test. We don’t know what we don’t have a test for.

And if those diagnoses were the only problems that these couples had, everybody would have had a baby after one IVF cycle because the whole treatment of IVF is designed is very smart treatment. And all the doctors that, you know, had figured this out, all the research and doctors figured out were amazing. 

Mylene Yao: They were trailblazers to have figured all this out. And that treatment is designed to overcome every problem that a couple could have. But if that’s the case, why is everyone still not, not everybody is successful. It’s probably because there are some problems we don’t have diagnostic tests for.

So we probably don’t capture all the data that we want to have. But given that, why don’t we mix it up? And it turns out I mean, this was a while back, so at the time nobody, and maybe I shouldn’t say how long, because that kind of dates me. But, you know, at the time, nobody even even in the medical research field, nobody was talking about AI in health care.

Nobody was talking about machine learning. In fact, when I learned the term machine learning and we started using it, it was so intimidating to doctors. We had to change it because people were looking at us like AI, machine learning, that we’ree aliens and we have three heads or something.

So we’re like, Oh my gosh, this wording is just not helping with the communication, it’s scaring everybody. Let’s just use regular English words.

So this is a while back. So we, you know, so it was amazing that we got a team of clinicians, embryologists team of really people that are really expert in AI and machine learning and then biologists all looking at this with different perspectives, clinical, developmental biology, kind of perspective, the machine learning angle.

And then we said, okay, let’s just put all the data that there is in this database. And we’re very lucky that the Stanford database was very high quality and let’s just see what comes out of it. 

Mylene Yao: And we didn’t even think there was going to be a prediction model. We were just thinking in very simplistic terms, let’s just classify patients into, you know, maybe they fall into different buckets, not the way we see. And it turns out that, of course, this is actually fundamental to machine learning.

If you do the classification correctly and is validated and is repeatable, you’re going to end up having a pretty powerful prediction model. And out of that we realized, Wow. Age is the most important predictor, but it only explained about 50% of the prediction and that’s what we learned.

And now fast forward, we have now, you know, performed that kind of modeling and analysis for many other fertility centers, validated prediction models. And that finding has been so reproducible around the world. We have now worked with IVF centers, you know, all over the US from not just academic centers.

So a very big part of what we’ve done from the early research days to today, having a commercial, highly scalable, commercialized platform is that we’ve worked with centers that are very small in the suburbs to very large academic centers, to centers in metro areas all over the US, also in Europe, in different countries with different, you know, support for the medicine.

You know, a self page is very socialized medicine and we’re really finding the same thing. Age is the most important predictor, but only explains 50%. And it is so important, therefore, that we get the rest of the data to make up for the other 50% so we can give patients a very accurate and personalized prediction. 

Harry Glorikian: Yeah. And and not to date you because I know you were trying to do that, but you’ve been working on this IVF prediction modeling since 2005. If my notes are correct. 

Mylene Yao: Oh my gosh, you are doing way too extensive research and unfortunately, your research is correct. 

Harry Glorikian: But but now, if my notes are also correct, you and your co-founder, Professor Wong, started the company in 2009. Now. You know, if I heard you earlier, you said you had a tenured lab and you decided to go off and do something in the commercial space. Why? 

Mylene Yao: Right. And when I did, I mean, today this is quite common because many people know, oh, we need to be innovative. We need to be entrepreneurial. It is very much a culture now, which is great. But at that time, it was not, it was something very different. People looked at me like, What is wrong with you? What are you doing?

You’re not a business person. You don’t know business. So, I mean, so it’s not like so at the time. The honest story is it’s not like I grew up with an entrepreneurial spirit and I was, you know, I’m going to be an innovator and I’m going to start a company. A lot of young people have that opportunity to think that way today.

I really didn’t know that. And it’s not like I woke up one day and said, I’m going to start a company. It really was just it came out of pragmatism. And as a researcher, academic researcher, and this also ties in. It is complex. Women’s health, not well funded by the NIH. And, you know, this is is a big, it’s a big barrier.

This is not just for fertility, you know, just overall all areas of women’s health. And we did submit grant applications. Now, any grant application is hard to get through. But I had already been successful with other grant applications and say to the NIH that you want to use AI to study IVF data. 

Mylene Yao: It’s almost like, whoa, like we don’t really have a study section to review this type of application. Today, that would be a different story, right? So then, okay, to do that you need funding. So very exciting, very fortunate. We got funding from Stanford Coulter Foundation and this is a partnership between Stanford and Coulter Foundation.

And their mission was to fund academic findings that have significant potential impact immediately if disseminated into the market. And they really believe in bringing technology to the marketplace as soon as possible to benefit patients. And we got such a grant that was great. And that I mean, it was a very small grant for $100,000 that kind of change our path.

And in order to meet all the requirements of that grant, we actually had to write a business plan, We had to do market research. And I was like, Whoa, what is all this? But by the time we got through it, we felt like,

Wow, we need to bring this technology to patients and doctors. This is not useful. We could publish papers and act like we’re smart, but it’s not going to help anybody. This needs to be in the marketplace. So that’s kind of how it got started. 

Harry Glorikian: Now. From 2009 to 2013, you guys were mostly in R&D mode, right? So. 

Mylene Yao: So yeah. 

Harry Glorikian: I’m trying to think back then to machine learning, right. You know, you had to have a lot of labeled training data at that time.

I mean, I’m thinking about some of the stuff we have available now versus what was available then. And so I’m wondering, you know, if part of what you were doing, you know, back in those days was just gathering the data sets and then labeling them and getting them ready for processing by the system. 

Mylene Yao: Right. So even today, we all know EMR data is EMR data. And it’s not, our data is not in a state ready for machine learning, typically. I’m sure there are exceptions. And at that time, I mean, that was even more so. And I think in the first five years from 2005 to 2009, we were really figuring out what is the best way to process the data every time.

And it was already the best database, one of the best in the world. So we were very lucky to even be able to work on that. And that was already a brainchild for many people that had worked on it before.

And but we went through several years of, Oh my gosh, we need to process this. We need to process that. And these are some exceptions. We need to go back and reprocess the data.

However, you know, we believed and being smart, making things scalable, never do the same work twice. So the great thing is our founding team, we just built a whole pipeline of proprietary code and so it got faster and faster. The first data set took us four years to process and to clean to the point where it was machine learning ready.

Because every time we thought it was machine learning ready, we realized something wasn’t quite ready and we had to redo it and meet all these rules. 

Mylene Yao: But once we had formed the company and we had outside clinics collaborating with us, then the next data set took six months instead of four years, and the next one after that took three months, you know, three months, two months, one month. And today it’s so fast, you know, so.

So I think that’s a great thing. We never have to do the same work twice. But yeah, it was quite a learning experience. And also at the time we had questions, a very valid questions from doctors. Doctors are really, they need to guard their patient’s safety because if they were to use every innovation that just came out, I mean, things would be all over the map.

So they asked very good questions. They they want it to know, oh, it’s amazing that you built this prediction model and this validates it from the Stanford data. But I’m not as Stanford. What if it doesn’t work for my data? I am such and such clinic on the East Coast. I am such and such. We’re not an academic center.

We’re in the Midwest. We’re in the South, we’re in Canada. We’re in Europe. You know, we’re in China. What if it doesn’t work for our data? And so and which is a very valid point, and it is still a point that I think many A.I. groups or A.I. companies are still struggling with today. 

Mylene Yao: Right? You build a model in one place. It may not be validated in a different population. And this is the one very big danger.

Now, the great thing, I think, looking back is. By answering the question that the doctors asked of us, we actually avoided all those problems. We said, okay, you’re right. We don’t know that. Let’s build a model for you, for your center, for your patient population, and let’s see if it’s validated.

So we basically did center specific validation population specific, which at the beginning seemed daunting. Oh! Investors said, “You don’t have one model that fits everybody. What kind of business model is that?” But I’m glad we did that. And of course that’s highly scalable now.

We have an entire proprietary platform and is highly automated still, but still with human experts monitoring this work and making sure it’s correct. But we can now really say to every center that we work with, we use their data, which is most relevant for that patient population, validate it to their specific patients.

And this is very important because the when a doctor uses a patient counseling report, such as our Univfy report to counsel patients, the first question the patient’s going to ask and we heard this around from around the world, the first question the patient wants to know? “Is this prediction based on your clinic’s data and outcomes?” Patients are very savvy. 

[musical interlude] 

Harry Glorikian: Let’s pause the conversation for a minute to talk about one small but important thing you can do, to help keep the podcast going. And that’s leave a rating and a review for the show on Apple Podcasts. 

All you have to do is open the Apple Podcasts app on your smartphone, search for The Harry Glorikian Show, and scroll down to the Ratings & Reviews section. Tap the stars to rate the show, and then tap the link that says Write a Review to leave your comments.  

It’ll only take a minute, but you’ll be doing a lot to help other listeners discover the show. 

And one more thing. If you like the interviews we do here on the show I know you’ll like my new book, The Future You: How Artificial Intelligence Can Help You Get Healthier, Stress Less, and Live Longer.  

It’s a friendly and accessible tour of all the ways today’s information technologies are helping us diagnose diseases faster, treat them more precisely, and create personalized diet and exercise programs to prevent them in the first place. 

The book is now available in print and ebook formats. Just go to Amazon or Barnes & Noble and search for The Future You by Harry Glorikian. 

And now, back to the show. 

Harry Glorikian: So you launched this report in 2013. It’s your Univfy Pre-IVF report, and it’s still one of your main services or products. Right. But can you explain So people who are listening, can you know, what’s the basics of the report? What what data do you gather from patients and what kind of predictions can you make? 

Mylene Yao: Right. So the types of data, the data fields we use are very basic by design because such as age, body mass index, the couple’s reproductive history, if they’ve had a pregnancy before, if they’ve had a pregnancy loss or so-called miscarriage, if or if they had several losses, have they ever had a live birth? And how many months or years of infertility have they experienced? And also clinical diagnosis.

This is premised upon the couple having completed the diagnostic work, a standard diagnostic workup that is the standard of care that every fertility specialist would order for their patients. So the clinical diagnosis would be available, and the test results from the diagnostic workup would be available, such as blood tests or ultrasound results relating to ovarian reserve.

So some of those for people who know anti-Mulleria  hormone, AMH, at the time many clinics were using day three FSH that test is gradually being phased out. Ultrasound results such as antral follicle count, the number of follicles you can count on ultrasound, which indicates ovarian reserve and of course, semen analysis.

So all of those data fields are used to. We test those to see which ones are valid predictors in a certain patient population, and then those predictors would be used in the model. 

Mylene Yao: Now, I want to go back to a question you asked before, which is really important. What are they going to, what are patients going to get out of this report? And it is the same as, how is the prediction going to be different compared to the age based?

So conventionally, and this is the case in all, not just in our national registry in the US, but in all national and global registries, many countries have its own national registry of reporting. This is kind of like a public health reporting service so that you can have some capture data and be able to tell patients what might be their success rates.

And it’s very necessary. It was amazing that all these national registries have been set up in many, many countries, including the US, and everybody can go to a public website and have free access to this information. This is incredibly vital for vitally important for for public health. However, in a health and a national registry like that, you’re going to see categories that are based on age.

So typically it would be it would say age under 35, 35 to 37, meaning if you are 35 years old all the way up to the day before you turn 38 and then 38 to 40. 41 to 42. And if you’re older, above 42, and then you would have these numbers. What is the average success rate for that age group? I think that was great if machine learning was not available.

However, every patient and every patient, every patient and doctor would know if you have two women and they’re both 34. Just because they’re both 34 does not mean they’re going to have the same chance of success.

Because if one woman has a certain clinical diagnosis, let’s say she doesn’t have regular periods and is something very easily overcome with IVF and another patient has a much more serious condition, they’re not going to have the same probability. 

Mylene Yao: So this is something that you don’t need to be a doctor to know that. So but if you then cannot give a more personalized prognosis based on that patient and that couple specific health data, what’s going to happen is when you give this age based average, first of all, doctors take it with a grain of salt themselves.

Patients take it with a grain of salt. The doctor is trying their best to do this counseling with this number that everybody’s taking with a grain of salt. And it just is not very effective as a counseling tool. And so by using machine learning, we can be a lot more precise based on the patient’s own health data and profile.

So, for example, let’s say I’m going to give you an example for a clinic. This is a real world situation for a clinic that we’ve worked with. And this is very typical across the board. Many clinics are in this situation, let’s say the age based average for a woman under 35 years old is actually, let’s say 60%. So 60% chance of having a baby from the first IVF treatment, including the use of any of the frozen blastocysts.

I say that for the audience that may have more technical knowledge about this. 

Harry Glorikian: All right. 

Mylene Yao: Okay. It turns out when you use machine learning in that group of patients that are under 35, depending on their health data, the probability could be as low as 28% and as high as 82%. So the question is, does that matter? So sometimes people say, does it even matter?

Doesn’t every woman want to do IVF anyway? And I’ve even had the question, isn’t it all emotional anyway? Why does anyone need to know the numbers? Okay, let’s respect that. Women are smart. Let’s respect that woman. Yes, this is a medical and emotional situation.

But women are smart. Women want to know this transparency, this information. And for a couple, depending on their situation, knowing that their chance of success might be 80% versus 28% means the number of treatments they might need to do is different, and the amount of money they might need to spend is different.

Maybe it does not make a difference to some couples, but it may make a difference to many others and it is only fair to give them that information. And actually the amazing thing is most couples are going to find that their probability is higher than they would have been told based on the age average.

And that was a big motivation for us because we realized many couples were missing out on the opportunity of a very effective treatment because there is a vast underestimate of IVF success in the fertility space, even for the doctors not realizing this.

And every clinic that we’ve analyzed, we found 50 to 80% of patients have a higher probability of success than they would have thought based on age. And that’s just so important to make sure people know this information. 

Harry Glorikian: Yeah, it’s, it’s, you know, I mean, I’m generalizing, but it’s personalized medicine in this area to give people. Right. That are insights into what happens next or what can happen next. Right. But the company is not just about modeling and counseling around fertility. There’s a whole side of your company that is about the finance, right? Right. 

Mylene Yao: Yes. Yes. 

Harry Glorikian: Can you talk a little bit about the refund program you’ve set up for to make IVF more affordable for patients who need to go through multiple cycles? I mean, I don’t know, how do the refunds work? Do you think of your loan program as a fintech company, almost?

I’m you know, I’m just I’m wondering whether your AI based prediction models are what makes these loan programs possible. Right. For instance. Maybe having all that data, that you can give the bank confidence to finance a refund program. 

Mylene Yao: Right. Right. So, yes, so affordability is very important. And we learned this early on. The information itself is not helpful if it doesn’t help people to be able to afford the care. So so but the great thing is that once you have a validated, accurate prediction model, then you can actually set up the right pricing program.

That is a true value based program that gives visibility and transparency into the cost-success relationship. And nobody needs to nobody needs to go do the math. We’ll do that for them. But we have been able to help clinics set up refund warranty programs. The way it works is like this. But first, I want to say I’m not going to take credit for the concept of a refund warranty program.

This is something that’s already in the fertility market way before we started working here. However, without machine learning, refund programs don’t work very well for the clinics or the patients. So by having a very accurate prediction model, we can actually help really cap the costs for for patients and have a way for them to move forward. 

Mylene Yao: The way it works is, let’s say you set up a refund program instead of paying one treatment at a time. And let’s say one treatment costs $25,000 per treatment. And this could be different at different price in different places. But let’s say $25K, a couple goes in, they pay $25,000. Let’s say it’s successful. Great.

They don’t pay again. It’s done. They pay $25,000. But if the couple were not successful, and even for patients that have very high chance of success, they say 80%, that’s very high. That means one out of five of those couples still walk away without a baby and having spent $25,000 and they are devastated.

Now, if they had and the doctor’s going to say, oh, you have really excellent chances, you should do this treatment again. But for the couple, we just lost $25,000. There is no consumer purchase where you pay $25,000 and you don’t get what you paid for. Even though you got great service. Right.

What they paid for was to have a family, to have a baby, not to have a treatment. So for that couple now, they have to pay. If they want to have a treatment again, they need to pay another $25,000. And for most people or for most people, that’s already not possible. The cash is not there.

And for many other people, they could come up with the money, but they’ve lost all confidence. What if we lose $25,000 again? We can’t lose $50,000. So they stop and they don’t try further. Which is terrible.

The overall dropout rate for people that are paying on their own is about 80%. The overall dropout rate, even if they have insurance coverage from employers, is still about 50%. So. 

Mylene Yao: But with a refund warranty program, the way it works is okay, instead of, you know, paying $25,000 each try, it may be a little bit higher and maybe pay $30,000 upfront. But after the first treatment, if it doesn’t work, you don’t pay again. You do the second treatment. And hopefully it works.

If it works, great, you’re done. If it doesn’t work, you get a third try. You go through a third cycle of IVF, and each cycle, each cycle you can use up the losses that you may have extra blastocysts that can be frozen for a retry. It includes all of that. And so by doing that, you give people kind of the peace of mind, like I paid $30K, I know it will not be more than $30K.

That’s it. That’s the most that I would need to pay. And I get three chances. And it’s really more than three, because each try there are multiple blastocysts that they can try.

And if I don’t have a baby after the three treatments, I get a certain amount of money back. And that really makes makes it fair to the patients. And doctors are happy to do that because doctors are, unlike what some people think, fertility specialists, they’re not trying, they don’t want to make, they don’t want to make $50,000 from someone and not have them walk them with a baby.

They don’t want that. They don’t want to make money that way. They want them to be able to go home with a baby and they want to provide great service, excellent treatment, and be compensated for it. 

Mylene Yao: So we really see the biggest problem in IVF is a marketplace problem. And we’re really having a platform, a platform to facilitate this transaction, if you will. Patients want the treatment, doctors want to offer it. They have a lot to offer. But how do you make this work? It has to be value based.

So and this worked very well. Patients really love it. But now we’re going one step further. Employers. More and more employers want to support family building. That’s great because patients shouldn’t even be paying out of pocket. This shouldn’t even be a discussion. And more and more employers want to support family building.

But we are in a world of their constraints, their economic constraints. How do we help employers support family building while containing costs? We are translating all the math and modeling that was behind this refund program for patients paying out of pocket to employers.

So employers and now we have multiple years of track record and very data driven. So we know by using this cost success kind of transparency, this value based modeling, which ties together IVF success prediction and the financial aspect, we can actually double the chance of success for each, for patients while cutting the cost by a third or more. And this is the kind of value based care that we can bring to employers now. 

Harry Glorikian: Oh, believe me, I’ve been touting value based care for a long time. You know, sensors, data and data analytics. I think we can bend, I don’t think there’s any other way to bend the cost curve because of all the trial and error we tend to do. If the data is there, we can make more.

I mean, I would never run my business by trial and error, right? I would be defunct right by that process. But so I was clicking around the map of, you know, some of your specialists. It looks like you have, you know, quite a few partners in New York, not here in Boston, a few in a bunch in San Francisco, none in L.A.

What are the factors driving geographic expansion or limiting geographic expansion? And is there a way for couples to access the service even if they don’t live near a participating center? 

Mylene Yao: The main limiting factor was just Univfy getting the funding to half the team. And now we have, we’re a Series B company, we have the funding and we’re talking to many more centers than what you are seeing on the website. And so we’re in a really exciting time that doctors are embracing AI and machine learning.

They see that. They see how this could help patients. We’ve had several years of track record to show that when patients are counseled with the unified report, more of them move forward with the recommended treatment and more of them are going to end up having a baby.

The report is not a treatment, so it’s not a therapy, but it’s really bringing the information, bringing the right information that’s personalized and accurate can help people make the best decision for their treatment path and they end up having a baby sooner. And this is now validated in multiple places.

So more doctors are having confidence to use our report. So we’re talking about in a few months there will be many more centers across the country using us. And we’re also now working with clinics in Canada, in Europe.

So and we aim to be global as well. So there are already clinics in APAC, Middle East that are very interested, that are reaching out to us, and we’re talking with them about how to analyze their data. This is a tool that women and couples all around the world should really have access to, and that’s what we really aim to bring to. 

Harry Glorikian: Well, it’s been great having you on the show. I can’t you know, I wish you incredible success because I think, you know, this data approach to making better decisions and helping people, helping people make better decisions is the only way that we can go in the future.

And that’s sort of what the show is about, is using data to understand, you know, what’s happening biologically that then helps you make better decisions going forward. So I wish you incredible success and hope we stay in touch. 

Mylene Yao: Thank you, Harry. Really appreciate being on your show. This has been great. And yeah, and you’re doing an amazing job, really bringing some topics that are difficult to understand for everyone to realize how much A.I. and analytics can help make our health care system better. Really appreciate your podcast. 

Harry Glorikian: Thank you. 

Harry Glorikian: That’s it for this week’s episode.  

You can find a full transcript of this episode as well as the full archive of episodes of The Harry Glorikian Show and MoneyBall Medicine at our website.  

Just go to glorikian.com and click on the tab Podcasts. 

I’d like to thank our listeners for boosting The Harry Glorikian Show into the top two percent of global podcasts. 

To make sure you’ll never miss an episode, just open Apple Podcasts or your favorite podcast player and hit follow or subscribe.  

And don’t forget to leave us a rating and review on Apple Podcasts.  

We always love to hear from listeners on Twitter, where you can find me at hglorikian. 

Thanks for listening, stay healthy, and be sure to tune in two weeks from now for our next interview. 

Related Posts