Ariel Beery: "Catching Cancer Using AI-Enabled Smartphones" | Talks at Google"
So. It's my pleasure to introduce Arielle. Berry who is the founder, and CEO of mobile, ODT, he's, here to talk about catching. Cancer using, AI enabled. Smartphones, a round of applause for him please. Thank. You very much thank, you hi. Everybody and thank you so much for joining, me today and for inviting me thank you Elana for this wonderful opportunity to share, with you a little bit about what we're doing and my, goal for for today is to. Share with you a, bit, of what is possible. Using today's, technologies. Not. So much from the technical, side but more from the structural. Side I'm happy to speak afterwards, during the QA if you have want to go a little bit deeper but, but what I would love to speak with you about today. Specifically. Is. About healthcare and and. On a fundamental. Level. What. Healthcare requires. More. Than anything else is, information. So. When. I started, mobile. UT together with my co-founder David, Levitz one of the things that we, were. Struggling, with was, we. Had on one hand a technology, which, was. Understanding. That consumer, electronic, devices were getting so much better thanks. To the rise in mobile, phones and. We. Had a need, out there, we. Had identified that cervical, cancer which we'll, speak about in a bit is a leading cause of death for women in rural areas around the world and it. Was so easily, treatable, but. Connecting, the two was the question that we faced and what, we realized, this as part of the the conversation, that we had with some of the world's leading public. Health organizations around, the world was that in the end on the most fundamental. Level, healthcare. And the provision of good health care depends, on data. And, that's something that all of you and definitely Google knows a lot about, so. If I can leave you with two things at the end of this presentation, the. First one, is that health care depends. On. Information. On data and then, the second, thing is that mobile, computing, will, make health care accessible to. Every, person on the planet so, with those two particular, things, I'd, love to jump into the presentation share a little bit about what we do and then hopefully, engage, with you in conversation, about how you. Might be able to use your superpowers. In terms of your abilities as engineers, as businesspeople, as operations. People to, save as many lives as quickly, as possible. So. I'll step back my, name is Arielle berry I have the absolute pleasure to be here today to represent a, company called mobile OD t-mobile, optical, detection technologies, and what. We do on the most fundamental. Level is use the power of mobile phones, to. Address disease, now. There are a lot of different ways that mobile phones can be used to address disease some. Companies, use apps. That, consumers, will use to track their blood sugar or their blood pressure and intervene. On a daily basis, what, what we believe as. A company, and where we're seeking to contribute, is in, prevention so. Good healthcare is best, suited, to. Preventable. Diseases, right if you can catch a disease almost, any disease while it's early enough to treat you. Can clearly. And quickly treat, it before a person gets too. Ill and so. You can save years of person lives and thousands, of dollars and families going into debt just by identifying, disease, early, enough and we. Believe that the people who are best suited, to currently. Empower, with, such technologies, our health care workers and our mission our, mission as a company is to equip health, care workers the nurses, the midwives, technicians. The clinicians, that, are out there serving. The majority, of humanity on a daily basis, to. Equip them with the tools that they need to be, able to provide better, care and. If. We. Look. At that as a mission then the question becomes, as a company where can we make the, most impact and so we, decided to, take dedicated.
Care Givers to provide. Them with mobile, compute, labs and to, help them, make. An evolutionary, jump in their, ability to provide health, care quality, particularly. In cervical, cancer, so. Many, of you already know this but I just want to make sure that we're all on the same page cervical. Cancer is a leading cause of death for, women in rural areas, around the, world it's. A cause, of death that is in many ways the. Most tragic because the most preventable I'll, show you in a little bit how cervical, cancer, develops. But just, than on one foot cervical. Cancer can take up to 20 years to kill a patient it comes. From an infectious, disease the, HPV the human papilloma, virus and if you catch it within the first five years you. Can treat it on the, spot with. Basically, the equipment, you have in the kitchen for, less than 28 bucks in. Less than 20 minutes so. How can it be that a disease, that is easier, to cure them the flu is killing. Millions. Of women a decade. Mother's. Pillars, of their community how, is that possible, well it's, it's actually one of those fascinating, things it's because women, aren't screened. So. Every, woman every, single woman of reproductive age, should. Be screened, for cervical cancer, regularly. Whether it be every three years or every five years depending on the technologies, that we have available and yet. The majority of women around the world, aren't. And. All. It has to do is that one moment, of screening, and that one moment of identification, a disease in which a woman's, life can go from the death column, to, the life column. Now. What's fascinating is that despite the. Investment, in maternal and child health since 2015, since recently cervical. Cancer deaths have become, more. Prevalent, than deaths from maternal and child complications, so it's one of those things where just a little bit of investment in regular. Screening, can, mean millions, of women alive. To take care of their families. And. That's not a problem just around the world it's a problem also in America almost. Every, person reproductive, age in this room for sure and many, of the people who'll be watching online, will. At some point in their life contract. The HPV arrives human. Papilloma virus is one of the more popular viruses. Today. And 79. Million Americans, are living with it at this, moment and what, that virus does, is it infects the cells this. Cells that, are infected with high-risk strains of that virus will start to become inflamed, and then, they will develop a disease and if, you catch it early treat it and if the body's immune system can, can flush it out it'll flush it out just like we flush out viruses. Every day but. For some Americans, and for many women around the world it. Kills them, now. The, way it works is as follows and, apologies, for the graphic images but I think it's important to recognize because it is a visually, accessible disease, so, a normal, cervix, the cervix, just for everyone who I, I learned, this when I started at, mobility the cervix is where the baby comes out through right, so, it's it's it's part of the woman's anatomy and the, cervix, a normal cervix, is a is. Fleshy. And has. Squamous. Cell and and, column or cells, as. The virus infects those. Cells they start, to become inflamed, and visibly. On the surface, of the cervix, you, start to see signs. That that infection, is becoming is beginning to grow right. At that, stage at precancerous, stage. You can still just cut, out those cells, literally. Freeze them out or burn them out and a procedure that can take a few minutes using, very, very simple, technology that can be done anywhere. Literally. Anywhere. Around the world and that. Woman will go from the death column to the life column but. If you let it persist, that infection, continues. And those, cells invade. Then. You get cancer now that stage it is almost, impossible to, save a woman's life. So. You have this very, long process that can take up to 20, years where. From a normal cervix to a precancerous, cervix, you have a five year window, and even, during that period of free cancer you can still treat.
Easily. Now. What's. Fascinating is, that. Screening. Is only part, of the problem so, anyone. Who looks at graphs will generally, look at this graph and say oh wait a second the trends are going down but I want what I want you to notice is that from around the, year 2000. Until today, despite. The fact that we've introduced, tremendous. Technological, leaps in cervical, cancer screening. Liquid. Based eye ecology, HPV. Testing, and most, importantly the vaccine. The. Number of deaths and the number of new cases have basically flatlined. Right. They have not gone to zero which, is what you'd imagine right all of a sudden we're vaccinating, people and we're screening with better technologies, and we're intervening when we can know, we're still seeing thousands, of women developed. A disease and thousands. Of women die from the disease in America. Forget. Outside of the United States where those rates are growing every year because infection, rates are growing. So the question then becomes all right so we have all these wonderful technologies. Liquid-based, cytology. And HPV testing and vaccinations and hopefully, everyone in this everyone, listening here is vaccinated. Why. Is this technology. Why are we not yet able to stop such. As simply, stoppable. Disease well, has. To do with lack of experts, so. I like showing the map of the United States as opposed to the map of India or China or, Africa because, it's so striking. The. Red is where. There are zero. Gynaecologists. These. Are counties, of the United States of America, one of arguably, the most developed, and economically. Successful countries. In the world and. That. Is the map of lack, of expertise. So. If you catch the disease where. You gonna go. Let. Me tell you something that's actually fascinating, when I learned it I blew my mind. 25%. Of women 1/4. Of women one out of four women who, were called, up and told you. Have a positive. High-grade, lesion, from, your pap smear you need to come in and get, treated. Do. Not. Return. Half. Of women who are called up and said you have a high grade, HPV. Test, high-risk. HPV test you've got to come in for your next four, your next appointment. Don't. Come in and. The question is why because. The distance and so. Then. Again this is an America where you have great roads and you have Amtrak. And you have buses and you have the ability to get places in theory right forget rural. Africa. India rule, Southeast Asia so, the the. Problem, the challenge, is. The. Connection of the patient and the expert and. That's, that's what we sought to address so this is the enhanced visual assessment system this is Eva Eva, is. Built, around the, technology that all of us know and love. The. Android mobile phone but. The, amazing, thing about the technology that this, company that Google has contributed so much towards, and that we as consumers, have supported. In its growth is that it's become so. Powerful in, its, ability to capture images. Process. And analyze, those images and communicate. With the world of expertise, that is on the internet right, the things that we know and love from everyday, interactions. Are actually critical to health care provision and what that means is that those nurses, those midwives those technicians that do see the majority of humanity by.
Using A device a medical, device of fda-cleared. See, marked, medical device that was enhances. The power of the mobile phone can. Now provide, expert. Treatment. Wherever. That woman is so, anywhere that you can take your phone you. Can take an eva system but, obviously it's not just the, mobile phone and it's. Not just the device it's, that what Eva does and devices. Like it connected devices like it is it, connects the cloud to the patient, at the point of care and it, connects the patient of the point of care and the provider, the health care provider at the point of care to. The cloud and, that's. What's so powerful about, the. Moment that we're in that's why there's, such a moment, of potential, for all of us to contribute because. Once we recognize, that in essence. Solving. Health care challenges is a solving. A data problem it means, that every. Time. That we impact, healthcare. We. Can literally, transform, the way that health care services. Are provided. Through, connecting. Dots through, connecting, pieces of information, now. We've, learned this and I'm. Gonna stop the the mobile Aditi segment for a second we've learned this by working in. 29. Countries around the world, a lot in the United States but around, the world as well in the African continent, in areas of the world in which cellphones are relatively. New but thanks, to the unbelievable. Power, of mobile devices basically. Can be taken anywhere and. What. We've learned in all of those countries whether, it be in the most advanced, cities in the United States and and in India or whether it be in in the most remote areas across the African continent, or in. Latin America. What. We've learned. Rather. The question that we've asked though is is how, can mobile computing, really. Transform. Healthcare and what we've learned is that it can do so in three different ways, so. Specifically. What. I'm gonna pause. It here today and the, the, requests, that I'm gonna end up making at the end of this that, specifically, mobile computing, can help.
Three, Means of intervention, for, healthcare, it, can help in one identifying. Patients, at risk in our, case because we're still youth cervical cancer as our as our case, study here it can and if I women at risk to. It, can ensure, that appropriate. Treatment, is applied. And three. Is it can ensure, adherence, it can make sure that the woman who is told that she's positive, is treated correctly and then it's followed up with correctly so that cancer doesn't continue to become an issue, so. Let's go into those three quickly. First. It. Cani mobile. Computer can help identify women, at risk this is a scatter plot of data that, was, was. Analyzed, as part of a Kaggle hosted competition, that, Genentech. A large pharmaceutical company made, available using, Symphony's, data set symphony is a large data, aggregator. Here in the United States and what. The, competition, requested, was for data science teams to look at this large public data set and identify, based, on this large public, data set which, women, are at risk for cervical cancer and, they, found an accuracy, of I think it's over 90%, and identifying, high-risk women, just from public, data sets now. Why is this important, well, cuz the United States spends about, 6.6. Billion, dollars, every year screening. Women. For cervical cancer six. Point six billion dollars. The five million women are screened. Every year and. Yet ten million women are not and, out, of those fifty five million women only. Thirty million women should actually be, screened. So. You're basically screening. Twenty five million women who, do not need to be screened and you're. Not screening ten million women at highest, risk that. Do need to be screened and that's, why African, American and Hispanic women in the United States have rates the cervical cancer that are almost, equivalent in some places higher, than, women, in Africa and women in Latin America. Because. We're not screening the women that need to be screened and we're. Over screening the women that don't need to be screened that's, just a data issue it's just the data problem, it's something that could be rather, straightforward, solved. And once you have that information and you can reallocate. Resources accordingly. And you ensure. That the, right care is going to the right patients. Then. You can make the right decisions, as to who needs to be taken care of so. That first one is just, identifying. The women that need to be screened the. Second one is applying the appropriate treatment this, is the basic, workflow for, how cervical, cancer screening, works in the United States today and it's the same thing with the pap smear, as well as the HPV test so let's go through it rather, quickly woman. Comes into a doctor's office she, lies on the table the, provider, scraped, cells off the cervix, puts a little bucket closes that bucket gives it a medical assistant medical assistant takes it and brings it into the front corner they call it just six company logistics company comes in picks up the buckets takes them to a lab lab then unloads it brings it to the technician technician then takes it stains, it after, they stain it they then take it and put it under a microscope they look under the microscope or they look under it as part of an assay to look whether there's presence, of HPV DNA they then identify, whether or not that woman has had something that is high-risk they then write that down that, then gets oftentimes, faxed, back to, the provider the medical assistant takes that fax looks at the medical record types in the information into the medical record if it's positive then goes and talks to the doctor the, doctor then needs to make a decision hmm maybe, I'll call back that patient and then the medical assistant calls up the duck the patient and the patient three weeks later after she, had, that test taken, needs.
To Be rescheduled. A little, bit complicated right. So. That tyre period, of time for. Those three weeks after a patient has come in and gotten, a test taken, she. Doesn't know and. Oftentimes. Wage. Workers, rural. Americans. Single. Parents, have, a hard time of scheduling, that next fall they just moved mountains to be able to have their annual physical, so. They get a call and they say - and and the medical assistant says listen ma'am you've gotten, a positive pap smear and the woman, says all right I'll call, you back when I can reschedule on that work now and then. She has a million things that she needs to do when she forgets and. That's why out of the 6.6, billion dollars that we spend every year screening. Women in the United States 1.2. Billion of that are due, to non adherence to, women that are lost to follow-up because she wasn't able to schedule that next visit, in time and then she comes back the next year and she's positive, again and she says oh god you told me last year I was positive, and then they need to schedule again but. They don't all come and so, that long process. Leads to women calling from the life column, to. The death column now. What. We're doing at mobile Aditi is what we found is that you, can replace that whole process of collecting, the sample sending it to the lab and then bringing the woman back for the next step through. Applying, pretty. Fundamental, machine learning algorithms, to the images that you capture using a, culpa. Scope so. There's some fascinating groundbreaking. Work that will be published pretty, shortly that I'm not allowed to talk about too much but I'll try to share the basics of in which, the National Cancer Institute, in the National Library of Medicine have done extraordinary. Work to prove that our partners, of global good have been able to identify cervical. Cancer with an accuracy, pretty. High a, little. Over 90% just. On the single, image alone and that. Is extraordinary, because. What that means is that while the woman's in the, clinic, without. Our going and going through that entire process in, 0.6. Seconds, she. Can go from the death column to the life column, and. That's. Just the beginning right because what's fascinating is, that these algorithms. Because. Remember cervical, cancer just like many other diseases is visually accessible these, algorithms, we already know and love than deploy, on almost everyday basis, many, of you in this room probably write, those algorithms. And they're, based on cafe, on, tensorflow, now we're moving towards tensorflow light they're, simple, straightforward, machine, learning problems interestingly, enough computer, vision didn't really cut it there but, when you build. A large enough data set and currently, mobile. ADT is proud to have large, data sets in the world on, cervical. Cancer you can use, that to identify disease, at its earliest stages using nothing more than, a visual image so. That that has to do with applying appropriate treatment. Because if you could apply that treatment on the point of care at that moment when the woman's in the room you. Can do so, much more for that woman than anything else but then once you apply that treatment you need to make sure that. She's. Actually. Taken care of right so the patient examination which most medical, devices do. Pretty well is only, a very small link, in the chain to ensure that patients, are kept healthy because, after, that woman is examined, we. Need to communicate with, that woman what, it was that was seen and what, it is that she needs to do and if the appropriate treatment isn't there if it's, not for cervical cancer it could be for any other type of disease she, needs to be referred and once. She is referred, then. We need to make sure that we update, her record because if we're not expressing. To the other person, what it is that you saw then. They, might miss it because in the end clinicians. Are only human right and they only know what to look for if they're told what to look for again it's a problem with data and once, that patient record is updated, and the provider provides, the appropriate treatment then. You want to take that information and, you want to warehouse, it, because, you want to have information about how disease spreads within a population, remember. Cervical. Cancer is a communicable. Disease it comes from the human papilloma, virus, and there's so many other diseases like that if you can find trends you, can know how to allocate resources effectively, and then, once you have that data you can start identifying patients, and then.
Once You identify, those patients you need to examine them and that continuum. Is what. Makes health, care and, that. Continuum. Is a, data problem it's. A communication, problem it's a problem that we've solved, in social networks but we haven't yet solved, and, health care but. We can. Now. So. For that there are three things, that, to. Close, off the more frontal part of this and to start our conversation there, are three things that I would ask of you of all. Of you here, and all of you that, you know that. Have those superpowers that you're applying at Google, and that you're applying in all of your work outside. Of it, first. Is let's. Let's. Help us let's help the healthcare system identify, patients, at risk right. Let's, take public. Datasets and identify, how we, can more clearly map, vectors, and then, Express that information, to healthcare organizations, so that they can deploy the proper, resources. In the most effective, and efficient manner so. That those healthcare, organizations. Can, do what they do best and care for patients. Let's. Also apply the appropriate treatment, devices. Have, a huge, amount of data exhaust and obviously. If you have garbage coming in you have garbage coming out and that's why for mobile OTT we, believe that the critical component is ensuring, that you have control. Over the type of information, that comes in so, you can apply the appropriate. Information. That. Comes out for analysis so. Let's. Connect. More devices. To. The cloud, do. It with all the proper, and necessary security. Precautions, to ensure the patient privacy is kept but you can't identify a woman from her cervix, nor, can you identify a, man, from from his or any old cancer or or. Any person, from their from. Their skin disease let's, use that information so. That any provider, anywhere, any nurse any midwife any technician, that can take a mobile phone and bring it into the into the field let's, make sure that every, member. Of humanity. Has. The same access, to the same high, level diagnostic. Accuracy, just. By ensuring, that they all have access to the right information and, then. Finally. Let's. Ensure adherence, we've. Learned so much about social dynamics, over the past decade, alone and, yet. We're not doing a very good job of ensuring, that that woman that gets called talented, told that she has. High-grade. Pre-cancer. We're. Not doing good job bring her back, how. Is it that were able to get her to order stuff on on online. Shops, but we're not able to get her to come back to a clinic or get the clinic to her, we. Understand, the social cost of having mother's died and they're in there in their most important ages, let's. Intervene, so. Figuring, out how to do that that's that's a data problem and, so, the the, key, the. Key here is to do the examination the, analysis. And enable. The action. Immediately. At. The point of care to, do that calculation fast, to do it wherever it is that that. Happens. That the patient is to, bring health care to wherever the. Sick are and, that's. Where mobile devices, are so powerful because. They go wherever we are and almost. Every person, on the planet now has one. Now. We're, doing our part Mobile ADT by taking the same, technology. That sits. At the heart of this device. Enabling. A mobile phone to be turned into a medical, device and applying, it in other diseases that are visually, accessible, anything. That you can identify with. Your eye that a clinician identifies, from endoscopic, procedures, to the skin.
You. Should be able to do with a mobile device you just need to tweak it so using, the right optics the right illumination, the right structures and so on but. There's so many other ways to do it and we're, hoping that every, device. Company that every technology. Provider, will start thinking about how they can use their, core resources, to, ensure that their. Technologies. Are being used to save lives at. The point of care and. Just. The last thought once again is that data is critical and it's. The mobile, element. That, enables, that data to be actionable. So. All of, us here, have. That native, understanding, just because we. Use our phones every day so, let's use them for something that ends. Up saving a life that might be otherwise lost. Thank. You all very much and I. Hope that you join us in making health care accessible, to every person on the planet I welcome. Your questions. So. There are two I was asked to say there are two microphones they're, illuminated, by the spotlights, if you have any question at all I'll be happy to answer them and. And. No question is that amounts. Hi. Thanks a lot for the talk so. I have two questions, the, first is you mentioned the accuracy, is around like a, little. Bit higher than 90% can, you talk. About like maybe the breakdown, of that in the sense of are there false positives, or false negatives, absolutely. So. There. Are a number of papers that that, I could refer you to the seminal, paper that will be coming out soon, will, break, down a lot further what, I can say about it is that what's. Fascinating, about, the. Cervix is that, it's a. Pretty. Clear. Example. Of squamous cell carcinoma which, means that as the cells start to proliferate they start to show visibility. Now, you will get false negatives, but, the sensitivity and specificity are both above 90 which means that you, will know that you're seeing something and you, will more, or less know that that thing is cancer, right, now, to, compare that to the pap smear to HPV. HPV. Catches HPV testing catches, a ton of people who have the high-risk strain that will never develop into cancer because only 5% of the people who, have a high-risk strain of HPV actually.
Will End up developing the cancer and the test, just looks for the strain of HPV DNA right, what. A visual, examination. Does. Is it looks for when the cells are already, growing and that's, when you want to you, want to you want to nip, the bud right you want to take away those cells where the infection, occurs and in, as. Far as the literature shows now around 97%, of cases it'll never reinfect. The. Second question this is I'm. So. The common I. Guess barrier, with healthcare, is. Sort. Of gaining access to data primarily. Because of privacy, issues. Can. You like you mentioned that there's like a. Large, database. That you're probably using to train the algorithm, can you talk. Maybe about like what has been done in that end, like how what's. The current, status in, like. How are the. Healthcare. Providers. Like responding. To, to. A need for you know technology. Companies, to have access, to that data so that we can help them yeah so first of all it's it's a fascinating challenge, I, just came we one of the one. Of the organizational. Networks that we work with are the Centers for Disease Control the CDC and. I just came from a conference, in Atlanta, the, CDC's informatics, conference, and, one. Of the things that I learned there was that even within the. Healthcare system even within the CDC, one organization. To have significant. Challenges sharing. Information right. And it's, it's it's one of the ironies, of the day where. We, can track almost every human on the planet we, know where, they're going and what they're eating and we're you know what they're doing more or less but, we can't keep, them healthy, because the information that we get from health. Care providers, this is all locked, up, right the. Way we address this just by building our own database so every time that somebody uses our system, we, have that information and it's de-identified. And we structure our database in silos and so any one of those pieces of information every one of those tables has, a different lookup function and so you you, can't really get you, know their information about who the patient is with information what the provider thinks with the member who the provider is with information about what, the resulting diagnosis is right so you have the ability, to structure, it so any one of those pieces is. De-identified. By. Nature and and. What we're hoping is, that and I know that Google's involved in this and a number of other organizations is what we're hoping is that these standards, will, start to propagate, that is every. Piece, of equipment that is somehow digitally. Connected should. Enable. Others. To learn from it and to integrate it and to build through its solutions, and and, I I've been I've, been very. Surprised. For the good and for, the bad about, how. That works, within different healthcare systems, so. You know you have some of the, or, so. We work in we work in countries where, you know the basically. There are no landlines, you know everything, is mobile, phones and and and there are countries like that, that. That. Some of them are very comfortable, with the cloud and some of them even though their data center is literally, you know an, open Shack. With, that that, sometimes gets rained on they want things to be hosted, locally but, thanks the continued ization you can do that right if you build a proper micro-services architecture you can you can make it work so.
I Think that we're gonna get there but, we just need to get more people to sign on to interoperability, standards, and and we need to ensure that people recognize, that this literally. Has to do with saving lives and so, delaying, is, costing. Lines i guess. Is a follow up so. What if. You are collecting, that information there's. Always that fear of, your, data being misused or, being identified, so. I guess even for me if i were to go through this it would be like alright like how do i know that this you know no one that doesn't need to see this data will, not will, not see this data so on, your end what do you do to protect the data that, you're collecting so. First of all it's what's called secondary use data by nature because. No. One other. Than your provider, has the information of who you are and what the examination. Is so. We. Cannot be identified, by, our, cervix. Or by. Image well it depends on what images of our skin but you know you, cannot be identified by it it's just not there aren't public datasets available to, ensure that kind of identification of a person, based on one mole or. On a cervix or an oral cavity even and. So because of that you, know the image and the annotation. Is. D identified by nature so, there's no risk from, that perspective that a woman would be identified, but. Yeah we. Need to structure correctly, and so when we structured our database from the very beginning we made sure that there wasn't that you literally needed to yeah it had, a salt and ash lookup table between the different tables to make sure that in any which case you'd have to kind of connect, so many different pieces of data in order to re-identify, someone. But, you just don't give people the keys so you don't make that available. Thank. You of course absolutely. Yeah. I, thanks. For the talk so the question might be out of scope of this talk but still do, you know of any efforts, where, machine. Learning is used, to. Diagnose, cancer are the diseases. Based. On other kinds of images like MRI scans or these games, what, are those yeah, absolutely so first of all there's a there's, now a tremendous, amount of information. Being, made publicly, available around, the world thanks. To national governments in large healthcare systems so I just heard that the, National Institutes of Health just published a huge data set of MRI based images I know I just heard from a friend of mine here that, you can all speak with after that his father-in-law is doing some exceptional, work down. At Hopkins, and identifying, cancer in its earliest stages in, colorectal, or, pancreatic. Pancreatic, cancer so, there's, more and more information becoming, available right. And and, that's critical, so I think that the the, real goal, has and, this is where I'm gonna speak for my thesis so, so, so my thesis is, that, smart, devices, are gonna make the difference so, data. Sets are great but as all, of you who are engineers no, data. Set is only good so far as the source of that data is reproducible, so if the data set if. You have a data set but the way of collecting the data changes, then. The type, of analysis, that you had the classifier, you built based on the earlier data may not be relevant for a future for.
Future Data and so, one, of the things that we we. Believe is a companies that tying the data to, the device and ensuring, that as a full solution. You can control, the way you bring data in the, way you annotate, that data and the way you analyze that data and that becomes, a. Virtuous. Cycle then. You can really provide care and so more and more, companies. Are starting to recognize that as well ge has a huge initiative around it I know that others also and. I think that as people start thinking about. Medicine. As the, application, of expertise, and device. Is an extension, of the ability, to provide medicine, then we'll, see more and more of that so there is scanned, I mean I think Google did a project. On the eyes. There, is MRI. Scan CT scans and others and and, what I find very fascinating, is that, that. For. The past 20 years there have been different approaches that mainly, use computer, vision but. Really where the breakthrough, is panas and applying really standard, off-the-shelf. Models, and I'll give you an example that might, be a little crude but is interesting, because it, was just fascinating so, it took us a, pretty long time to develop an, algorithm, that can kind, of identify the cervix, and then. Some of you will know that there's a there's a show an HBO show called Silicon Valley that open. Source or, at least gave a walkthrough of how they created, an algorithm for a hot dog not hot dog we. Took that hot dog not hot dog instruction. And within. Three days we. Built an algorithm, that was cervix not a cervix. In. Three days it. Took us like a year to do it before and then with the new tools that literally, just became open, it took three days and that exponential. Logarithmic. Shift, in terms of speed and potential, is astronaut. It's just it's crazy right, and it also makes sense and this is one of the things that you know that that's why the you know your superpowers. As engineers are so, critical, here because what. What is fascinating, is that, you. Have here, a moment. In which, we. Knew. For, for, all of history we, knew that. Human, beings have the ability to perceive, information and to, use our brains in order to find patterns and that information and then to make decisions, right I mean that's what medicine is medicine. Is where you have a person, who is you know a magician, so to speak a doctor, who goes through many years of school in order to learn all the different things they need to learn they, get to that point of care and they make, very, educated, guesses as to what to look for based, on what they see they, reference the information that they've learned to make certain decisions and then based on the decisions they make the person either lives or dies right. Well. Algorithms. Can see tens, of thousands, of more patients, than doctors ever could, just, because they can literally review tens of thousand where patients so now we have the ability to actually do what the doctor if doctors have done in. A way that we simply, before yeah. It's. The. Anonymized. Data. Collected. By a to. Mobile or duty. For. People to download. Is. That proprietary so, we do have a data set that we made available through Kaggle in, a competition that Intel sponsored. On type of cervix which is critical for treatment we're, gonna be making more and more data available to the research community and we're very excited to work with folks so please feel, free to contact me at any time and I'd be happy to to share information. And. On a similar note have you thought of using. This device as, a means to collect, collect. Data on all other, sorts of data. You, use an image yes. So. We already have the ability to do endoscopic procedure is oral skin absolutely.
Which Is critical absolutely critical thank, you yeah thank you yeah. Hi. I. Just a little clarity on how exactly the product works absolutely, I saw a photo with what looked like an attachment sort, of scopes but for cervical cancer specifically. Are. You taking, images from, the pap smear swab no, so I usually. Don't show it cuz it's a little more graphic. For people who aren't aware a woman, will lie on the table and something, called a speculum, is inserted, just like inserted, today to take a pap smear or, a colposcopy, and our, device, which is classified as a culpa scope which is a long-range microscope, to look at the cervix the vagina and the vulva is placed, at a distance it never touches the patient and at, a distance it images, the cervix and based, on what it sees and what the provider sees they make certain decisions so, at this moment what, the providers are doing is using this as a long-range microscope, to look at the patient and based, on what they see they make decisions and over. Time as we start to provide more and more guideline, interventions, and workflow interventions, it'll also guide them to do more and more and so one of the things that we've so, for example we. Have a wonderful, partner in the Rotary Club in San. Diego and they, built a partnership between the. Jalalabad. Hospital and Afghanistan, and Scripps Medical Center in San Diego and the first cervical cancer screening program in Afghanistan, and was started because Eva systems out in Afghanistan were being used by midwives and a, gynecologist, in San Diego was watching. What they were doing and providing them with tips on how to actually. Perform the examination, thank. You absolute, pleasure. Thanks. Again for the talk have you found cultural, barriers in, terms of using the actual device and if so how do you overcome those how do you do education, that the, women, who are receiving care feel comfortable, with the idea of a, more. Complex device being used sure, so, first of all what's been I. Found. Amazing. Universality. In certain. Elements, that we had not previously, expected. So, the first there, are cultural, differences in terms of insertion, of a speculum.
Speculum. Insertion is problematic. In certain, places for. Obvious reasons. Once. You get past that hurdle which has nothing to do with us then it's pretty straightforward it's, just visualization. And any place in which speculum, is inserted they're used to a woman visualizing. Them. What's. Been fascinating actually, culturally, is that what. We noticed first in Kenya, proved. To be correct also. Around the world which is that women loved, seeing an image of their cervix, because, women, had. Never seen, their cervix and remember, in emerging. Markets in rural areas and roars, even America certain, cancers a leading cause of death so they will know a woman, who has died of cervical cancer, so they hear again and again that this is something they should be worried about but they don't know what it is and so to, be able to turn, around and say this, is what I saw, look right here you have to go and get this treated, was, huge and as, part of the interviews one of the women called it I saw my cervical selfie right they had this this moment where it, was it was very empowering, for, her to see what she looks like on the inside and and and in doing so reduce that that fear, barrier, so I think that's critical I think it's critical across all of healthcare that is you, know in the end, health. Care has. To do with the. Person the, patient right, so, we a many. Times medical device companies, think. Of health care in a very abstract form, and has do with providers, and payers and, institutions. And networks it has to do about that. Woman that one woman and that one woman needs to own her health because if she doesn't take care of herself no matter how much the health system is developed, and robust she'll. Die. Thanks. Very much absolute. Pleasure. So. We'll take one more question and then we'll start to wrap up yeah. So. My question is more on a marketing, standpoint so, who is your target that, you're going towards, and trying to sell this to like. Are you allowed to say how much it costs or is that something. That you're keeping under wraps right now but. Just more of a marketing, where are you trying to go to is it hospitals, or different, organizations stuff. Sure that so, mobile, ADT works primarily with healthcare, networks, and Hospital networks although. Obviously the person who uses it as a nurse. Or a midwife or a gynecologist. Their, procurement. Officer is generally the person who purchases it in, the United States we sell them for four, thousand eight hundred dollars each for for, this generation which. Is about 20% of the price of a traditional, unit, and the, goal is for price to never be a barrier, right we want to make sure that every provider in, every corner of America and, the world pricing, obviously, is lower globally. We'll be able to provide, the care that they're. Called to provide when they provide it and and, there was a there was a slide here that I wanted I'll show you in a second which is important for us from a marketing. Perspective. The. Woman here the the icon here that's, that's, who we call Eva just like the products called Eva that's Eva and Eva. Is who we engage with Eva is the health care worker the nurse the Midwife the technician, that woman, that serves day and night to serve the majority of humanity and we seek to work with her so from. A from from a basic, service, perspective our goal, is to provide, that person with a tool she needs and will work with her healthcare system, with, her hospital network to give her the tools she, needs to provide care to our patients. So. I'll just I'll wrap up on. A very basic, request. From all of you so on. You. Know when, when, I was given the opportunity to come. Here and Thank, You Alana again for making it possible the. Thing that kept on going through my mind was how, to. Use a lack, of a better term how pregnant with potential. This organization Google, and companies. Like it are for. Impacting. The, very fundamentals, of health care, lessons. That we have learned over the past decade, and how, to identify, individuals, support, those individuals, through choices ensure. That they get to the right information at the right time with the right people are lessons, that are immediately, applicable to. Every person on the planet and mobile. OTT can only do its part in it and hopefully will, continue to strengthen. The, way that providers, provide health care. Organizations. Such as yours and others, in your peer group have, so much opportunity to. Be able to fundamentally.
Affect The human condition on whether. A person, lives another year or just not and so, I hope. That you could use your superpowers, whether it be from marketing, to business, to engineering to, ensure. That no. Person. No woman no child no father no man. Dies. Of an easily, preventable. Disease, again. Thank. You very much. You.