In this episode, Griffin Jones interviews Eduardo Hariton about the current and future state of AI in the fertility field. Eduardo Hariton received his undergraduate degree at the University of Florida, followed by a combined MD / MBA at Harvard Medical School and Harvard Business School. He is currently a clinical fellow in Reproductive Endocrinology and Infertility at the University of California, San Francisco. He has published extensively on both clinical and nonclinical high impact journals on topics that range from fertility and reproductive surgery to technology and medical education
Artificial intelligence is changing the landscape in our industry and will continue to affect fertility clinics in the near future. How much and in what ways, are the real questions that should be on your mind. That’s why I brought in Eduardo who is on his way to becoming the leading expert in Artificial Intelligence in the Fertility Industry.
Some topics we cover include:
What is keeping AI from progressing faster
What should we keep under human control vs AI?
Future of IVF Cycle prices
Data rights and privacy issues.
Eduardo Hariton Info:
Facebook: https://www.facebook.com/pg/haritonmd/about/
LinkedIn: https://www.linkedin.com/in/eduardo-hariton-6687ab63/
Twitter: https://twitter.com/eduhariton
Instagram: https://www.instagram.com/haritonmd/
Website: https://obgyn.ucsf.edu/eduardo-hariton-md
To learn more about our Goal and Competitive Diagnostic Click Here.
Transcript
Eduardo Hariton: [00:00:00] People should look at their agreements and see who owns what data and your ability to use it too. I think anytime you enter into some agreement where there will be data sharing, You know, people define very clearly, like who owns that relationship, who owns the IP that comes out of, you know, any insights from this data.
Narrator: [00:00:20] Welcome to Inside Reproductive Health. The shop talk of the fertility field here. You'll hear we are authentic and unscripted conversations about practice management, patient relations, and business development from the most forward-thinking experts in our field wall street and Silicon Valley both want your patients, but there is a plan.
If you are willing to take action, visit fertilitybridge.com to learn about the first piece of building a fertility marketing system, the goal and competitive diagnostic. Now here's the founder of fertility bridge and the host of inside reproductive health, Griffin Jones,
Griffin Jones: [00:00:59] On today's episode, I've got Dr. Eduardo Hariton back with me. We talk about artificial intelligence to a lot of the degree that I've just had Dr. Bob Stillman on and others that I've talked about artificial intelligence with, but we get into the specifics of what needs to come down in terms of the walls that bar certain technologies and platforms from talking to each other, which is actually, what's going to accelerate the progress from artificial intelligence as it applies to outcomes, as it applies to clinical operations, definitely the human impact.
Dr. Hariton has a strong opinion on, and we try to break down the nuance of what that is. And then we really disagree on something. Before I talk about what we disagree with, today's shout out is going to go to Dr. Pietro Bortoletto haven't spoken to him in while I don't know if he still listens to the show, but hopefully he does because he's another one of the rising stars in the field and would love to just get an email from him with this conversation with Dr.
Hariton we talk about the cost of IVF and what's going to happen. In the next five years, I say, it's not coming down. Eduardo says it is. It depends on, I guess, how we're phrasing that. And he and I are making a wager, we're still disagreeing on the terms of our wager because of course, both of us want to be right.
We're both trying to phrase it in a way that I'm hedging to where I am going to be. Right. And he thinks he is, I think I'm right on this. So I gotta get with my legal ease because I want him donating to my charity, not the other way around. And I'll let you decide. I would love to hear what you think about this.
Argument debate discussion that I have with my good friend, Eduardo. I think Dr. Hariton is one of the brightest minds coming up in the field, as much as it pains me to say that, because I'm jealous that he's got both that crazy business mind, as well as your ultra clinical mind. And so you get to hear that discussion being unpacked today, and I hope you enjoy it.
Eduardo, welcome back to Inside Reproductive Health.
Eduardo Hariton: [00:02:55] Thank you so much for having me again, Griffin.
Griffin Jones: [00:02:58] You're back not just because you're my friend, but because I think you're in kind of a unique position. You are a second year REI about to be a third. Your REI fellow at UCSF. And you are also you're in the mix where you're looking at your career thing of practicing medicine.
Also looking at what the future of the field is going to be like. All of these companies and ventures in technologies that are going to impacted. And so I want to center our conversation around artificial intelligence, mainly because if you're not the guy qualified to talk about it right now, Eduardo, I think you're going to be the guy qualified to talk about it in 20 years.
So why not just have you on early and explore some of these thoughts? I want to start with what are you paying attention to right now with regard to AI technology?
Eduardo Hariton: [00:03:56] Thank you so much. I think that's like an overstatement of an introduction, but I'll take it.
Griffin Jones: [00:04:00] But listen, you might be miserably underqualified right now. Maybe you should just turn it off because here's some fellow talking about the future of the field right now.
So I'm not blowing sunshine too hard. I just think that you are going to be the guy. So I will put my. I will bet my ponies on that.
Eduardo Hariton: [00:04:18] I appreciate that. I can tell you what I'm paying attention to is the massive amounts of investment coming into the field in order to bring technologies that we use in other areas of medicine.
Into reproductive medicine. I think when you look at what's happening in not only other areas of healthcare, but around different industries, automation is big. There are a computing power is becoming less and less expensive. () And the ability to draw insights from really complex sets of data is growing and becoming more powerful.
And we see that applied throughout healthcare in diagnostics, where people are using AI to find different targets for therapies to bring the cost of drug development down. You see it across healthcare systems that are applying AI in order to move patients through that system in a more efficient and effective way.
To monitor patients in the ICU to recognize that they are not going to do well before. That actually happens inclinations to in order to early intervene. And I think people are realizing that in IVF being a very costly unexpensive treatment with a growing market and increasing the man that outstrips the supply.
There's a lot of opportunity for investment. So everywhere from predictive analytics to the way we stimulate patients, to the way we, follow gametes in the embryology lab on selects, which wants to transfer. I think there's people looking at all of these parts of our fertility journey. I'm trying to apply artificially intelligence solutions in order to improve our process.
Griffin Jones: [00:06:07] So when keeping with other specialties of healthcare, how prevalent is this technology is it's still very nascent. Give us a couple of examples where it is proven and adopted at scale to improve either clinical outcomes or just efficiencies and process. Where is AI being used? Were a couple of specific examples that are pretty established.
Eduardo Hariton: [00:06:36] Well, I think we're not very established anywhere. There's not one company that's taken over one solution that is used everywhere. I think we're going to get there, but I think we're still in the nascent stage. I think when you. AS for examples the are in markets. There are companies that are looking at vision.
So convolutional neural networks, looking at embryos to try to not only grade them and replicate what embryologists can do, but to select embryos with a higher implantation potential. And there's companies that are trying to bring those products to market or have those products to market. They have not been widely adopted.
And I think that one of the challenges of artificial intelligence technologies is that the models can show benefit. But I do still think that needs to be replicated perspectively and you need to actually show, you know, not that it predicts that it will improve outcomes, but that actually does. And I think those studies are ongoing in several parts of the world, but that is kind of the technology that is furthest ahead at this moment.
It's the use of computer vision to select embryos with the highest implementation potential also to select eggs that are more likely to create pregnancies.
Griffin Jones: [00:07:55] So that's with regard to clinical outcomes. Where else are we going to see AI being applied within a practice setting?
Eduardo Hariton: [00:08:07] Well, I think, you know, to the first part of the process, which is understanding a patient's prognosis right now, we have some tools.
We know their age, they're vary in reserve. We have start data, we have studies, so we are able to look at some of these tools, which are relatively crude compared to personalized AI. And we're able to say for someone like you with your diagnosis, your part in their semen analysis, based on your age, I predict that your per cycle light birth rate is 22%.
Right. Whereas when we take AI and we say, you incorporate all of that data, that that patient has where they are in their journey, and you can get a much more personalized prediction. I think that's helpful for patients because it allows them to decide, do I want to go through IUI? Do I want to go straight to IVF?
What are the pros and cons? What is my own individual expected success rates and what does it cost? And they can make that calculation themselves. I think it's helpful for the clinic because they can use their own data to try to drive some of these predictions. I lot of clinics are offering resharing models where they are allowing the patient to sharing the risk of their journey.
They subsidized part of the treatment. Some of them have guarantees and it's much easier to feel comfortable and have your final finance folks feel comfortable with a risk sharing model. If we have a very personalized prediction, As to what that success rate is rather than a rather crude measure. So on the predictive side, we're seeing some folks working on that, that in my mind is something that we could do before with more linear prediction models, but they use of machine learning and more complex algorithm makes them better.
Griffin Jones: [00:10:01] In order to really personalize prognosis technologies would have to talk to each other wouldn't they, in order to have in order to have better data, meaning EMR and fitness apps and all the way down to the smart technologies that will appear in the home. So what technologies are starting to talk to each other now, or if you.
Don't necessarily know the answer to that. What needs to be able to talk to each other?
Eduardo Hariton: [00:10:32] Well, the answer is not enough because no one talks to each other. And I think you hear of one of the big challenges for the artificial intelligence community. And for people trying to work on this, there is no one heterogeneous data set that these models can be trained on.
These models are. Need to be trained on very large amounts of data in order for predictions to be good, you know, training them on, our data UCSA for data of a single institution, even very large institutions. It's not enough because one in the magnitude of AI, it's not enough data for it to be really good.
And then in, if I take an algorithm that's built in the East coast and they bring it to the West coast, or I take it to Europe or China, It's not going to work in the same way. So what we really need to do is we need to build data sets that have patients from all over the world and, you know, different ethnicities, races, weights ages, and see how they do so that the algorithm can know how to react to different situations and weight those to your point.
Those do not exist, and there are some initiatives to create them, but. You bring up another good point, which is not everything that matters to your fertility. You talk about in your initial visits, you know how much you walk everyday, what you drink, what you eat, how you sleep, certainly can have some effects.
I think that probably we are further from incorporating those into our datasets that we use. I think m ost likely the initial models will be more clinically based, based on what a clinic can aggregate. And hopefully, what clinic collects in one area will be. Easy to homogenize with what a clinic collects somewhere else.
Another challenge, most databases don't look the same, even for people who use the same EMR. So there's going to be a lot of work upfront and creating this large data sets. But I do really feel like it will pay off in the long run.
Griffin Jones: [00:12:34] So even before we create the large dataset, let's go down this rabbit hole.
That have technologies that don't talk to each other, let explore the reasons why I can think of a couple that I hypothesize. One of which being eventually you a massive privacy concerns we're already dealing with second is that everyone wants their data. The EMR companies want us to be able to sell their data and people should be.
Paying them for their data and the genetics companies think that people should be paying them for their data. So everyone wants to keep theirs so that they're the ones able to sell it. So I see those as two reasons, privacy concerns being one at a global level, two, being everyone wants to monetize what they have and not give what they have for free and they want to get more of it.
What reasons do you see for technologies not talking to each other yet.
Eduardo Hariton: [00:13:33] I mean, privacy concerns are real, but the reality is that when you use these apps or use this products, you are agreeing to, for them to sell your, the anonymized data. So it's kind of happening anyways, for the most part. And you usually can request your data.
So that's something that you could pull. I think the reality is that when you participate in, you know, wearing a ring or a smart watch or have some of these products, Part of their value proposition to their investors is that they're collecting a lot of data that they're going to use to drive insights and create the higher value for their investors, for their consumers and grow their own individual products.
So I think that's right. There are very few incentives for those companies to share their data outside of their companies, unless it's in a symbiotic type partnership. And. And that creates a challenge for the data sharing and for incorporating some of these really large data sets that, you know, may help.
But we don't know because we have not explored that or incorporated it into the data that we do have.
Griffin Jones: [00:14:43] And so who has the most leverage then? Who gets to say no, our data is worth the most. What do they have to do in order to aggregate the most. Data or have it as to make their database uniform. Who's got the leverage?
Eduardo Hariton: [00:15:04] I would say that the leverage is whoever can derive the most value for their consumers. So if you are able to create a large dataset, you have some leverage there. If you're able to drive insight from that data set and share it with. Your consumers, then more people are going to come because they're going to want your insights.
They're going to want to learn from what their peers are learning from the device that they think it's really cool from an AI perspective. You know, if you start with a large dataset or you have some sort of relationship or value that you can give someone else for them to share data with you. Then you can get other people to perhaps share anonymized data with you.
Another thing that is extremely interesting is the use of blockchain based technologies to share data. You know, one of the challenges is that for HIPAA reasons or because you don't want to give away your data for free to someone else, you don't share it. And there are ways on the blockchain to be able to aggregate databases.
Without a centralizing institution so that every participating party can contribute data, but can also use other people's vietnamized data without actually owning it and taking it over so that you can train some of these models in broader data sets. But at the same time still be able to own your own data without, you know, openly sharing it or sending it outside your servers.
So those are some approaches that might be able to give us the heterogeneous data sets that we need. But again, to your original point, you know, not all columns are going to align. All rows are going to be the same, not all clean it's code in the same way, or feel the fields in the same way. So it still takes a lot of data clinic and that is incredibly time intensive and manual process that we will have to overcome before we can drive. What I think are the most valuable insights.
Griffin Jones: [00:17:07] Well, there's a rabbit hole question, but it's too tangential that maybe we can get to it. It's about actually aggregating and making that data uniform, but that lack of uniformity might be the reason why.
EMR's aren't the direct answer to that last question who has the most leverage? Because I know as you're talking, I'm thinking, well, isn't the answer. The EMR's because they have the most information, but there are so many ways to query so much different information store, different information EMR. Is that the reason why they might not have the most leverage right now?
Because. Yeah, they've got a lot of data, but it lives in a lot of different places and looks like a number of different fields.
Eduardo Hariton: [00:17:55] I guess the question is that depends the EMR and what their user agreements are and who owns that data. Because just because I use EMR, X doesn't mean that EMR X can just pull up my patient data and then aggregate it and use it for profit.
You know, some agreements might be like that, but some agreements, the data's owned by the individual user. And yes, it's nice that everybody uses the same one. And perhaps there is something there where the EMR says, I'm going to build a product based on all of you guys as data clinic, a, B, C, and D. And then I'm going to give it to you for free because your data helped me build it.
And I'm going to sell it to other places or I'm going to. Use it as a reason for new clinics would be then participate in the original content creation to come into our network of EMR clinics. That being said, it really depends on how that data is shared and organized. And I don't think, at least to my understanding that EMR companies can just pull them profit from their clinics data.
Griffin Jones: [00:19:01] Would we advise Inside Reproductive Health listeners to check those service agreements upon signing to see who owns the data?
Eduardo Hariton: [00:19:12] I think that, you know, in the 21st century, what we do with data and the insights that we drive for them are going to be hugely valuable, not only for clinics, but for our ability to make better decisions.
So. Yes. You know, my guess is most people do. But if you haven't looked at who owns the data that you're creating in your clinic, certainly something worth looking into and making sure that, you know, who's using your data on who's allowed to use your data.
Griffin Jones: [00:19:44] So is that's true for almost any service agreement tha t not just CMRs, but should people be looking at that for the genetics companies with carrier screening companies with. The pharmaceutical companies they buy from, or I guess anyone that would have their data.
Eduardo Hariton: [00:20:03] Yeah. And we didn't say we know with Fertility Bridge also. I bet you people should look at their agreements and see who owns what data and your ability to use it too. I think anytime you enter into some agreement where there will be data sharing, You know, people define very clearly, like who owns that relationship, who owns the IP that comes out of, you know, any insights from this data.
And that should be very clear upfront and something that people should be paying attention because in many of these cases, I expect that some of these companies and some of these algorithms will be quite valuable. And you want to make sure that if you are contributing to an algorithm or you're contributing data, you get to be part of their rewards and fruits of that data know most importantly, I think we're all in it to help patients, but your ability to help patients will also be.
Better and increased if you have the financial means to do well.
Griffin Jones: [00:21:08] What if the service agreement doesn't say anything right now, fertility bridge agreements. Don't say anything about data with regard to centers and we get some, one of the things that we do whenever possible is. We do have people agree to not give us protected health information.
And so from a marketing lens, we only get any kind of patient information after the patient signs, a HIPAA authorization in which case is no longer Phi. So we don't have any of that kind of information. We do get numbers on we, we do track volumes and because we want to know if we're driving IVF volumes or egg freezes or recruiting the donors that we're supposed to be recruiting a new patient volumes. So we do have that sort of stuff. It all lives pretty archaically right now in spreadsheets. So it's not like we have a machine to go out and monetize, but how would that look? Like? What would that, what does it look like when there's no agreement?
Eduardo Hariton: [00:22:12] That one that I'm going to defer to my lawyer friends. Cause I actually don't know the answer and one pretend to know, but you know, my only advice is, you know, get a good lawyer and make sure you understand what you're signing, which is probably good in life.
Griffin Jones: [00:22:27] Yeah. Well, in the meantime, we, I don't, we don't have so much data to, to really worry about, but I think ultimately even client services firms like mine will have to get into.
The data game to some degree. And I think it's a lot like how software has been the last 20 years, where in the beginning, there were a lot of people creating proprietary softwares and some of them really needed it. And very often a lot of people found that they were much better off just using an off the shelf software or some SAS company that already existed and applying it to.
Either clients or themselves. And so I don't see us as builders, but I do see us. I see even client services firm like mine, having to just review the insights that come from data. When we put together averages right now, they're pretty rudimentary. It's not the same accuracy that one would have if they were all aggregated so
other than you making people scared to do business with fertility bridge when we're like number 190 seventh on the 197,000 down the list of people that are actually trying to get data who is trying to get the data. You don't have to say particular companies, but in the direction of, who's really trying to both aggregate and ultimately monetize, data from patients in clinics?
Eduardo Hariton: [00:24:07] I think who's trying to get data from you is literally everybody who you touch online, Google, Facebook, Apple, literally every single interaction that you have is recorded and a, and studied and used to monetize and sell your stuff or understand you better or serve you better products so that you spend more time.
So, you know, You know, broader level, every interaction that you have in the digital world is. Studied and likely monetize to some degree. I think on the clinic level, you know, without mentioning companies, there are companies that are trying to aggregate data. There are academic institutions that are trying to create consortiums to aggregate data in order to drive these solutions.
Like I mentioned, I think they're still in the early stages. I think the data sets that are being built are on the smaller side. They're usually single center or a few small centers and the projects that are coming out are more on the proof of concept side. So there are people trying to show. Yes, we can predict pretty well how people are going to do, or yes, we can, you know, help make better decisions in the stimulation process in order to.
You know, make outcomes better or, you know, remove physicians from part of the process or at system at the list to make better decisions or in the lab, we can help embryologist create embryos or pick embryos so that we get patients pregnant faster. So we're seeing some of these projects happening, fertility and sterility is seeing more and more publications regarding AI and just had a whole, you know, aim monthly.
Journal dedicated to AI in reproductive medicine. So I think we're at the early stage. I do think that over the next five to 10 years, we're going to see a lot larger databases and perhaps more heterogeneous databases come out and. Prospective projects where you not only build an algorithm, but actually test it and compare it to physicians or make a prediction and then see what happens after.
And that will help validate this concept. And perhaps some of those will come to market and become widely adopted. But I don't know if it's going to be six months or six years. You know, we are terribly bad at predicting timelines, but I do think that in my life then as an REI, the decisions that I'm going to be actively involved with in a day to day basis are going to be incredibly different than some of the decisions that the people who trained me were involved that their beginning of their careers.
Griffin Jones: [00:26:51] I would be a bad fertility doctor because I only want to take on the cases that I know are going to be successful. I only want people to say these sorts of things about me and my company, like Greg in Chicago,
"Our resources are not endless. And I think that with fertility bridge there's a much deeper dive."
or Dr. Young in Iowa,
Narrator: [00:27:14] "I've gotten more positive feedback from patients from anything in the last 30 years of practice"
or Brad in Seattle,
"You have multiple experts on your team and for, you know, a very small price to get that level of consulting for just a couple hours"
Griffin Jones: [00:27:33] Would be really valuable.
Okay, you get the idea. So this is how we set you up. So you are 100% guaranteed to be successful in your goal over time. It's not a magic wand until you do this, do not pass. Go do not collect $200 indefinitely. Do not get in any long-term commitments or launch initiatives, you sign up for the goal and competitive diagnostic at fertilitybridge.com.
You fill out your business needs profile. We establish your benchmarks and desired outcomes. Then we meet for our 90 minute consult. We provide you with business Intel revenue estimates, and a competitive overview of the field to facilitate the prioritization of your goals between your partners and leadership team.
Then we have a 30 minute up. We tell you exactly what you need to audit and strategize to build your plan. I'll also give you one big marketing idea that will make you say, damn, that's good. If we failed to do any of these things, we give you your money back because it's only five 97. And because I need you to be successful because I need you to say all those really sweet things about me and my company.
Maybe even a gem like this one from Holly and Dr. Hutchison from Arizona.
Narrator: [00:28:41] "I have, we didn't have fertility bridge, honestly. I think we would be getting close to retiring."
Griffin Jones: [00:28:47] There's no long-term commitment whatsoever and there's a hundred percent money back guarantee. Send your manager to fertility bridge.com, have them sign up for the golden competitive diagnostic.
And I will see you and your partners on zoom.
Well question in terms of what the timeline will be like in our inability to predict it. I see the same trajectory happening with broadband and voiceover internet protocol voip, where it sucked for years for fricking year in 1999, we're like this, everybody's going to have broadband.
We're going to be able to download movies in a second. And we're going to be able to have conversations like we're having on zoom right tomorrow. And then 2005 came, it still sucked. And then 2010 came and it still sucked. And all of a sudden, I don't know if it was. 2017, but all of a sudden it was like, Oh, we've all we all have.
Perfect voice over internet protocol right now. And good timing too, with a global pandemic happening in March of 2020. But it was like, where is it? Why isn't it here yet? We've been talking about it forever. And then all of a sudden it was just here and that's not a very scientific way to, to anticipate the advent or growth of.
Of artificial intelligence we're past the admin, but I do think that that's, what's going to happen.
Eduardo Hariton: [00:30:15] Yeah. I mean, I don't disagree. I think bill Gates is the one who said that we always overestimate what's going to happen in the next two years, but then underestimate what's going to happen in the next 10 because technology does not
kind of advancing a monotonic linear way. It advances in an exponential way, the cost of technology goes down in an exponential way. So, you know, I agree. I don't know if it's going to be two or 10, but I do really think it's coming on and I'm excited to see the impact that we can have on patient outcomes by using some of these very powerful tools.
Griffin Jones: [00:30:51] We also don't know what. The catalystic events will be to speed it up. And so the example, I knew that we were moving to a virtual. Dominant workforce. It's why in 2014, I started my company. We've been virtual from the beginning. All of my employees live elsewhere in the United States and Canada, as well as do our clients have never had a physical home office other than the office in my home.
And I knew because I knew that's what, the direction that we're going to. And in 2014, it felt like. Starting a digital agency in 1999. Like it was too late to do the brick and mortar type of route, but it was still like early and it was kind of awkward. And I remember our clients in the earliest years, some of them would be like, Oh, she's in Denver and you're in Buffalo.
And your project managers in Tennessee. And. They're your account managers in Florida and people didn't totally get their heads around it. I knew that it was moving to that. I just didn't think that there was going to be a global pandemic that made it the status quo. And so what do you think are potential catalystic events?
And I understand that I'm making you speculate and putting you on the spot to do it, but that would. Accelerate the adoption of artificial intelligence in healthcare fertility, specifically.
Eduardo Hariton: [00:32:26] Yeah, well, unlike you, I also didn't predict COVID and did not invest in soon a year ago. Wish I had, but you know, I can tell you about some trends that I think are definitely going to keep pushing us towards adoption of some of these tools, you know, partly because they improve outcomes, but also because they will improve.
Efficiency and lower costs. I think when you look at the IVF market in the United States, we don't have enough capacity to handle the volume that we need to handle. David Sable gave a very good talk at ASM a couple of years ago. And he said, we're doing somewhere between two 80 300,000 cycles. And when he sizes up the potential market for IVF, based on the infertility cases, we have the, you know, genetic disease prevention, opportunity, egg freezing trends on how fast that growing.
We can easily do up to a million cycles a year. You compare us to places like Israel, where they're doing like, you know, a cycle for every 200 to 150 people, Japan, which is somewhere around 300 Europe, which is under a thousand. We do a cycle per 1600 people. So we're very under-penetrated and we have an opportunity to grow our market.
There was a study by started in 2016 that showed that on average physicians, REI is lead about 130 cycles a year. Some people do none some people do a thousand and you know, I want to meet those people because I'm interested to see what they do. But with about 1300 we need to do around 800 to 900 cycles per person, per REI, to meet that demand of like one to 1.1 million cycles.
Anybody who you ask right now that works, their tail off is doing. 300, 400, you know, 200 is a lot. So we are not designed to accommodate this kind of demand. Yes, we could work nights and weekends and nonstop for 24 hours
Griffin Jones: [00:34:31] And have 15 IVF coordinators and never do an ultrasound.
Eduardo Hariton: [00:34:35] And. You know, but that's a challenge, right?
We need to get more efficient than, yes, we can. We can stop monitoring. We can stop doing, you know, procedures. We can stop doing everything. You still don't have enough hours in the day, you're still going to hit a wall. So the reality is how do we number one become more efficient. So what are some aspects of this process that can be automated?
From our prediction to our stimulation, to our embryology lab in order to make this process more efficient. And that will give the REI opportunities to spend more time with patients. Because I think one thing that does not get talked about enough is that. People are still human. Even if you're taking care of 600 cycles a year, those people want to see your face.
They want to hear from you. They want to call from you. If they're pregnant, if they're not pregnant. And we really have to think very carefully as we redesign the way we take care of patients to not lose that human touch? I think it's important for the patients, but it's also important for the area.
You know, we came to medicine because we like thinking we like being challenged and we like learning. And if you take all the fun out of it, because it gets automated, you're gonna lose REI's as well, because it's not going to be what they signed up for. So it's really important to keep that in mind as we do this.
I think the go ahead. Well, let me finish this. I think the other aspect where this is important is. Part of the reason we're under-penetrated is because IVF is really expensive. Access to care is a real issue. And, you know, art still something where high socioeconomic. Status patients have a much differential aspect and it's not something that's accessible to the lower classes.
And I think that's a real problem by applying some of these technologies by removing some of the human component, which is exceedingly expensive and contributes a big amount to the cost. We will be able to lower the cost of. Not only an IVF cycle, but of the ultimate goal, which is reaching a pregnancy.
Cause our cycles will be a little cheaper and they will be a little better. And hopefully we will get to offer the amazing, you know, the amazing opportunity to start families that we offer some of our patients to everybody who wants to have a cycle or get fertility treatments.
Griffin Jones: [00:37:04] You think that the price of cycles is going to go down?
Eduardo Hariton: [00:37:08] I think that as we incorporate technology, I can remove some of the costly elements that we have now. Yes. I think it will go down. I think there's also increased amount of payers coming in and that's going to put downward pressure on the price that gets paid for cycles. So that's another aspect, unrelated to the AI that will.
you know, will push prices down, but you know, if you want to compete and your payers are pushing what they want to pay you for a cycle down, I way to maintain your margins is to become more efficient than AI is a way that you can do that.
Griffin Jones: [00:37:44] So it does push the price down because of what they reimbursed. But they're also bringing so many more people.
If we look at markets where that are really. Progeny heavy and maybe it's character kind by now, but it's employer benefits. When we look at those markets that have a lot of those companies, they're so fricking busy, right? I mean, you live in the Bay, so you know how busy they are and it's not just, Oh, we're busy on the clinic side, but maybe we're not converting enough people to treatment.
There's met mashed in the lab too. And so I don't see prices going down. And what, where are, where's the precedent for that in healthcare of prices going down
Eduardo Hariton: [00:38:29] It's market power. I mean, you see it in places where there is a. You know, someone that controls a large share of the population, they can say, you know, I don't want to pay you, you know, $15,000 a cycle.
I want to pay you $13,000 as cycle. And if I represent 40% of your cycles, you can't lose me. So you will take 13,000
Griffin Jones: [00:38:52] that's the thing they might be. They might be able to lose it because people are getting so busy and as more employers start to offer coverage and more States mandate. Then now it's not just a progeny game.
Now it's United and Aetna getting back in because insurance and who has well, we're losing all of these employers. And so we're not getting a cut of any of this. So they start to get back in whether it's Cared or Kind Body, eventually that. Particular profile becomes a two horse race. And then if you're in a big enough market of busy enough market, you could say, okay, well, these, this group has Facebook, Amazon, Google.
This group has McDonald's LinkedIn and general motors. And this group reimburses 10% higher than the other group. Yeah, we can lose the other group. I don't think that's out of the realm of possibilities.
Eduardo Hariton: [00:39:47] Yeah. I mean, there might be some centers that feel comfortable losing a payer because they don't want to do it.
And you know, that's the art of negotiation. You have to know when to walk away, you have to know when your what's your BATNA or your best next alternative and walk away. But my guess is that as these negotiations play out and as these players start covering more and more cycles, they are going to start reimbursing less, or they're going to start reimbursing for value.
Or it generally is going to drive what they are willing to reimburse down. You know, some clinics might walk away. Some might take the lower reimbursement, hopefully no one's losing money on a cycle, but ultimately the way to create value here is to lower your cost of the cycle, because it's good for you.
It's good for your payer. It's good for your patients. And. And that technology is incredibly scalable. So that's something that will be helpful. I think another part that I didn't touch on where I, you know, AI or machine learning can be helpful is, you know, you might be familiar with this Griffin.
There is an incredible amount of heterogeneity in the way that we practice. There are some standards of care that we follow, but if you go. To my clinic and the clinic next door, and then the cleaning next door to that, we do things three different ways. No one way is better than the other. And we don't know for sure because otherwise we would all do it the same way.
And then within the clinic, Dr. A likes to look at things one way and Dr. B likes to do things another way. So then the lab has to be always on their toes, figuring out which doctor is said, do they want to transfer this day or that day? What kind of, you know, Extra concoction they want on their media. And at the end of the day, that heterogeneity and lack of standardization is incredibly expensive for the labs.
If we apply big data approaches, and if we use AI to standardize, what is the best approach for a given patient or a given clinic, or maybe we realize that it doesn't really matter. We should just pick one so that the lab knows that when those eggs are coming, they're going to be processed in the same way all the time, you know, on, you know, nothing's going to be a hundred percent.
There's always going to be patients that don't fit the mold. So I don't mean to say that AI is going to be a hundred percent better for everybody. We still need our brains. And we still are going to have patients that have receptor mutations or don't respond like we expect. And we're going to have to think them through that's the art of medicine and where.
Our education and all the years we've put in will really matter, but we're going to find that a lot of things we can standardize and that can also lower variability and reduce costs and take that out of the system.
Griffin Jones: [00:42:34] So is AI going to be the hammer of Thor that finally breaks down? At least some of that heterogeneity in that isn't every other REI in EDS, except for the one.
That the given context at any moment, except for maybe their partners or someone else. But it seems to me like everyone I talked to, Eduard, it's pretty amazing, everyone has the best success rates in the country. It's pretty incredible how they're all number one. And they their competitors are idiots that don't know what they're doing.
And I'm. Hyperbolizing a little, but this isn't something that I hear rarely. And so it's also been one of the main challenges in the consolidation that's happened on the private equity side. You have. Standardization and people don't necessarily want to follow them. And there are some groups that could be selling to private equity and haven't and it's because they want to have that say so that, I guess there's a marketplace of ideas happening within that heterogeneity.
How does AI break the tie?
Eduardo Hariton: [00:43:44] Because physicians are committed to giving their patients the best outcomes they know how to give, and they don't want someone else coming and telling them. I know you do it this way, but I want you to do it that way because that's how, you know, the clinic that we acquire in X city does it.
That's not what physicians are going to respond to. They're going to respond to data. They're going to respond to well
Griffin Jones: [00:44:10] . Why isn't the data from the clinic that we acquired in X city sufficient right now. And what's so much more compelling about the data that comes from AI.
Eduardo Hariton: [00:44:20] Because it's going to be much bigger and larger scale.
Like if you come and you tell me Griffin, like, Hey. You know, this clinic in another city does this way and they have 3% better outcomes or 10% better outcomes. I'm going to say, well, look at that patient population. They're three years younger. Their BMI is a little bit lower. It doesn't really apply to me.
I, you know, I can just change my whole protocol based on what someone else does, but when you have, you know, a group that has data from 15 clinics and you aggregate all the data and you say, Hey protocol, you know, doesn't matter beyond these two or the starting, those should be this within this parameters or.
This is how we should, you know, do XYZ in the lab and it's working well across the system and it's clearly superior. You know, we're all competitive. We want the best for our patients. And we want respond to a suggestion or we won't respond to an example, but we do respond to data. We read the journals and we try to understand.
How do we change our practice in order to provide the best outcomes to patients? We do that every month and every week and every day we continue to incorporate data. And I think what AI is going to do is that it's going to give us data that is a lot more convincing and powerful because it's a heterogeneous.
So from a lot of places, very large and very robust. Another interesting thing that I think will happen is as a field, we have accepted a lot of Adam's therapies. So new medications that make it to market or new therapies or injections, because we want our patients to do better. Sometimes what happens is that these medications make it to market and become available to patients before they're truly studied.
So before you have a randomized controlled trial that can show benefit, what happens then is that it's very hard to do a randomized controlled trial to show benefit when people can go to the clinic next door and get that treatment anyways, because they don't want to get, you know, the sugar pill or the saline shot.
They want to get the medicine they're spending 15, $20,000 and their time to get pregnant is now. So. Doing those studies is quite hard right now when things already made it to market. I think by aggregating data from a lot of places from cycles that look the same, other than the fact that one of them use, you know, growth hormone or some other additive medicine and recognizing, Hey.
This medication really works, but it only works in this subset of patients or there is no patient where these medications showed a difference. We're going to be able to figure out what actually works and hopefully stop using the words that do not.
Griffin Jones: [00:47:08] I want to go back to the human touch part because I've been, it's been cycling around my head because we are at this bottleneck challenge where there's what 11 or 1200 of you in the entire country.
And you talk where maybe we're doing 300,000 cycles. We can be doing a million to me. A million seems like on the conservative side of the estimate if other variables were addressed. And so. You mentioned, well, people still want to see their doctor. The doctors still want to have that human interaction.
They don't want to just be behind a screen and managing dozens of case loads at a time without getting to know people.
But we have a ways to go before we can meet the demand. There ultimately seems to me like even when we address so many of these other elements that can be,
That can be taken care of with technology that will still have a very limited bandwidth for the attention of the REI for any given individual.
So what are the things, as you talked about having to be intentional about how AI comes into play and what we're automating versus what remains human interaction, what. Human interaction. Do we need to safeguard? And I know it's a general question, but try to be as specific as you can.
Eduardo Hariton: [00:48:40] That is a hard question.
And it's something that I, as a believer of AI and as a believer of how fast this market's going to grow and how limited the amount of REI is struggle with on. And I think about it. Day to day think about it in the shower. It's something that I think it's really important to keep top of mind. You know, when I think of some really efficient decisions, you know, you had Doctor Amy on the show, You know, a couple of months ago and she sees hundreds of cycles and has a huge case caseload.
And she goes out of their way to make sure that every patient feels like she's thinking about them all the time. And she has a process that she set up in order to do that. So thinking about that type of process, I think is important. I think understanding which are the interactions where physicians can add the most value to patients face-to-face and which ones can be.
Perhaps delegated, not to a computer, but to another human, to a wonderful nurse. Our nurses are the backbone of our industry. They interact with our patients more than anyone else. So helping build that group of nurses and mid-levels, that can still. Make them give them that human touch without perhaps extending the REI beyond the hours that they have available will be important.
And I think that one thing that we're going to see, you know, related or unrelated to AI is that eventually I think patients are going to segment themselves and figure out how much do they care about seeing the REI and how much are they willing to pay for that? I don't think that. I think that some places, for example, have NPS that are managing fertility preservation cycles and doing those initial visits and for some patients that's okay for some complexity of case, et cetera.
Okay. So rethinking the system and understanding what is the value that we bring is important. And you ask for a specific example. One example that I always think about is. The only time in most places where a physician spends a whole hour with a patient is in the initial visit. That visit is where you take a history and you get information from the patient, but you don't have labs and you don't have testing and you don't have anything concrete to guide them.
You say, well, if the semen analysis is normal, we'll do this. If it's abnormal, we'll do that. If you're a vary service high, we'll do this. If it's low, will you that if your tubes are blocked, we'll do this. If they're not blocked, we'll do that. And then you spend an hour counseling them on generalities, and then you still need to come back and counsel them after the testing.
So why don't we switch that around? We get some information, do some testing and do the counseling two weeks later when the testing is long done two months later, or whenever it is, that is a much better use of physician time. Patients would appreciate it a lot more. And I think. Rethinking this kind of framework where you go from visit to testing, to treatment, to hopefully pregnancy and really white boarding it and rearranging how we spend our time with our patients so that they feel connected to us.
But we also are giving them. The most valuable time that we can give is something that I hope happens again. I don't know if it's going to be two or six or 20 years, but I think we're going to be pushed into doing that sometime in the near future.
Griffin Jones: [00:52:09] Well, that's a good point that those operational changes though, are things that people can do now.
They don't need to wait for AI to come and they only help you as things start to become. Automated. So there's no reason to say, Oh, I'm just going to wait until something established comes down the pipeline, the way that people use software, the way that people manage their operational systems allows people to incorporate these technologies as they change.
And the example that you gave, I don't know that I have enough evidence. To make it our official point of view yet, but we might soon enough because what I'm seeing anecdotally, Eduardo, is that what you described where the initial consult is shorter, those groups, actually, those physician convert more people to treatment because in that initial comment, you make it a half hour.
For example, you just spent it telling them this is what we're going to do next and not go into the contingencies and the variables. The patient is able to digest that information better. I don't know that I have enough evidence to say that that's certain yet, but I'm starting to see more of that. And that's just one example of an operational change that can be made now and among other things that help AI to come in.
I want to, I was reflecting on your answer of why it's so hard to be specific about the human touch answer, what human touch still needs to be. Available my philosophy is that the patient needs to be cared for, but needs to feel cared for bottom line. It doesn't necessarily need to be the physician for something.
Ultimately, the patient decides what feeling cared for means and how much the physician needs to be a part of that. It isn't the physician that necessarily gets to this side. And I think it's important for. People debate well what can our team do to make the patient feel very cared for? But I think the things that we use either you and I could really quantify, have to do with the things that go above the expectation.
And when I first got into the field, I asked my clients if I could talk to some happy patients that just really understand what they liked about the process, what they didn't like. I remember one of our earliest clients. Someone talked to, they just adored his position because he walked her to her car and then there's, that has nothing to do with clinical outcomes.
It doesn't even have to do with how you make them feel cared for in the office. It just makes them feel cared for. And so I, as you mentioned that the struggle to think of something specific, that's why, because it's above the expectation as opposed to being within it. That I think. Feeling cared for.
Eduardo Hariton: [00:55:05] I think it's, that's a very good example. And you hit the nail on the head because it's not the same to every person, you know, so to someone feeling cared for is getting a call after each pregnancy tests for someone feeling cared for is getting their labs. As soon as they resolved for someone feeling cared for is seeing you.
For their monitoring ultrasounds, even though you have a sonographer, so you stopping by, or you're doing it yourself and every patient's different, you know, I always think about how can I capture in my initial visit? What are the things that matter to a patient so that I can go. Above and beyond for that given patient in the way that they want me to go above and beyond.
And so that I'm not calling the person that rather get a text and texting the person that rather get a call. And, you know, I go back to fertility IQ, ask patients in their questionnaires if they like to a blunt doctor or they like. A doctor with a soft touch and they ask questions like that, about what kind of physician or what kind of care do they want to get?
You know, I imagine that there are some questions you can ask a patient in your initial intake to build you some kind of profile so that you can make sure that. When you're going to call that patient, you talk to them in the way that they want to be talked to. You, share information in the way that they want information to be shared.
And that might not mean a lot to you because you're just trying to take care of them and your care won't change. But how your care is received will meaningfully change and your patients will I'm sure. Feel a lot more connected and a lot more satisfied, no matter what the outcome of their treatment is .
Griffin Jones: [00:56:48] That personalization is something we're getting.
A lot more in doing the patient acquisition journey with regard to physician profile, you don't even need to get that preference from the patient. You can share with them. We've got five doctors. Dr. C is not necessarily is is not the warm and fuzzy type of doctor. It's okay to say that Dr. C is very direct.
If you would like someone that has a, more of a social bedside manner. Choose from one of our other four doctors, people will choose Dr. C, they do it. And that's a bit tangential. I want to kind of conclude Eduardo, with when you see IVF prices going down, because he came on the show for the first time. I think two years ago.
It won't be the last time that you're on the show. And I want to know when I can tell you that I was right. And you were wrong. So, and
And when prices don't go down, when do you think that the price, how long is it going to take for the price of an IVF cycle to decrease? Because I say it's not happening in the next five years.
Eduardo Hariton: [00:57:58] Well, I will say this. I don't know when it's going to decrease in absolute dollars or what that's going to look like. I think there's like inflationary pressures that will distort that equation. But I think that in 10 years from now, if you have me on the show again, I think the average American's ability to access an IVF cycle will go down.
You know, if, even if there's no universal healthcare, the ability of an average American to cash pay and IVF cycle will go down and it might not be an IVF cycle with the top doctor, the top clinic, because that might still be concierge lag, but their ability to go through an ovarian stimulation and egg retrieval, and basically go through IVF will be more accessible.
To the average American who does not have coverage. And yeah, you know, you can remind me of this. We can see what happens, but I think that generally the increase access and all of this technology will drive the cost of a cycle down, you know, for the people listening to this and worried that's gonna obliterate our margins.
I think we're going to have a lot more volume. So yes. Prices might go down and reimbursement might, you know, put some downward pressure, but like you mentioned, some of these players, I'm bringing an incredible amount of scale. So as long as we keep up and we're able to handle it by incorporating some of these technologies and becoming more efficient, we will be just fine.
And more importantly, more patients who desperately need access to our services will have access to them. I certainly hope so.
Griffin Jones: [00:59:43] I don't think it's happening in the next five years, but part of the reason is because I think that technology needs to happen before the, because the volume is rising too quickly right now.
And that technology needs to get ahead of that curve, you know, and like even being equal to it would take some time. And so I don't see it happening in the next five years. You said you've given yourself a comfortable time period of 10. I want to be right about this because I think in most things between you and I, you will end up being right, because you're one of the smartest guys that I know in this field.
And I do think that you're one of the rising stars in the field. It's a privilege to have you back on Inside Reproductive Health.
Eduardo Hariton: [01:00:28] Thanks for coming back Eduardo. Thanks for having me, my friend. Good to talk to you, Griffin, and look forward to seeing what happens in five or 10. Sounds good. Bye-bye
Narrator: [01:00:39] You've been listening to the inside reproductive health podcast with Griffin Jones. If you're ready to take action, to make sure that your practice thrives beyond the revolutionary changes that are happening in our field and in society, visit fertility bridge.com to begin the first piece of the fertility marketing system, the goal and competitive diagnostic.
Thank you for listening to Inside Reproductive Health.
