Latest / StartUp Health NOW Podcast / David Feinberg of Google Health @ 2020 StartUp Health Festival

Transcript

David Feinberg of Google Health @ 2020 StartUp Health Festival · StartUp Health NOW Podcast

• 0:00
  Please welcome back to the stage, Startup Health's Howard Krein and Dr. David Feinberg, vice president of Google Health. David, thank you for being here. My pleasure.
• 0:09
  I don't know of how many, um, people can understand for me the excitement of having... I got to speak with Toby Cosgrove this morning and you this afternoon.
• 0:18
  That is a physician's dream because these are two truly, uh, two of the biggest names in healthcare and in innovation in healthcare. So thank you so much for being here. Totally excited. All right.
• 0:33
  So, um, first thing, so as most of you know, David, Stephen had mentioned, um, was running Geisinger. You've been, you've run a couple healthcare systems.
• 0:42
  Um, the first thing I wanna know, okay, and we'll get into Google and everything, how does a company, um, uh, take a physician and s- and entice him to leave healthcare
• 0:55
  that you've sort of grown up in and, and were sort of molding to join a company? Well, the, the transition from the healthcare world to the tech world for me was a really, um, difficult decision.
• 1:11
  I, I needed to convince myself that, or be convinced that a place like Google was serious about health. Yeah. Um, that there could be big impact, but most importantly, um,
• 1:27
  would they allow me and the team that I assemble treat the people that come to us either through Search or YouTube or the docs or nurses that we're building tools for, would they let, would they allow me to treat them as if they were my own patients?
• 1:41
  Yeah. And that to me was absolutely crucial in making a decision because, um, I, I don't think I've changed.
• 1:51
  I think I've spent my professional career, uh, starting out with helping little kids who were having mental health issues and helping their families, and then I had these great opportunities administratively to take care of what I think of, through others, more people.
• 2:07
  And to me, this was just an opportunity when I was finally convinced to be able to really improve people's lives around the world. Right. Yeah. Um, for, with a, with a tremendous platform. Yeah.
• 2:18
  The interesting thing is everybody always says, "Oh, Google, i, it's a tech company." Um, the amazing thing as a physician, you know this, uh, patients come in every day, it, it's like Dr.
• 2:28
  Google, because if you ask a patient, you give a diagnosis, you say, everybody has Googled before they even come into the office for the most ti- most time, um, to get information.
• 2:39
  So Google's a data, a data company basically, right? That's what it, that's what sort of, I should say, um, what it started out as. Um, are they serious about healthcare?
• 2:49
  So first of all, I would say that Google is a consumer company, right? And- Yes... and when we talk about tech getting into healthcare,
• 3:00
  um, well, raise your hand in the audience if you've seen your primary care doctor or seen a specialist, and I, I, you don't have to if you don't want to give me this personally, in the last 60 days.
• 3:11
  So we've maybe 10% of people. Raise your hand if you've gone to Google in the last 60 days to look up a medication, a symptom, a diagnosis, way finding, Walgreens, where the CVS is. Yeah. Right?
• 3:24
  So whether we say we're getting into healthcare or not- Right... uh, 80% of people before they make an ER visit, within 24 hours of that ER visit, do a Google search. When you get a new diagnosis, you go to Google.
• 3:38
  Um, this sometimes freaks out lay audiences, but doctors, particularly surgeons, before they do procedures on you, go to YouTube, right? And they wanna check to make sure they're doing the paracentesis the right way.
• 3:51
  So what we need to do on that consumer piece in particular is we have to make that information better. We have to get the dangerous stuff off.
• 4:01
  We have to make sure that the quacky stuff is five or six pages down and hard to find, and we need to create, through partnerships, an ecosystem that gets more authoritative information, so when you do go to Google, you get those three right things that you should then bring in to discuss with your doctor, not stuff that's gonna send you in a wild goose chase.
• 4:21
  So we're really committed to improving authoritative search, improving, uh, information, and, and getting rid of the misinformation. Yeah. Amazing.
• 4:30
  And, and that I think is, is a big part of it, is because, right, there is, there's a lot of misinformation on there too, and how do you figure out what...
• 4:37
  And we, we see that with our patients sometimes, is they'll come in with things and be like, "This has nothing to do with you" or, "This is an ad."
• 4:43
  Um, as far as w- where does Google Health, now that they're, they're, they're, they're coming out and putting a lot of effort and energy into Google Health, where does that fall in sort of the Google family?
• 4:54
  Is that pa- part of the Google AI? Is it... Where does it sit within the Google platform? Yeah, so, um, Google Health is, this round of Google Health is new.
• 5:05
  It's about a year old, and it was bringing together three different groups that were already working within health- Mm-hmm... at Google into one what we call product area.
• 5:14
  And the three groups were the teams working, uh, in London, the DeepMind Health group, which done amazing work in predicting kidney injury before it happens, um, the group that was part of the Nest acquisition that was working at health in the home, and the part of the fo- the folks that were working in brain or research specifically around health.
• 5:34
  So it became a separate product area, and then your question about was Google serious, um, I think I'm the only product area lead. Products at Google are things like YouTube and Search and Android and, um, uh,
• 5:50
  Geo, which is Maps and Waze. I'm the only product person lead at Google that's not an engineer. So part of the seriousness was, um-They needed somebody who knew healthcare- Mm-hmm...
• 6:05
  because I think there was a, an understanding that this is not simply a technological fix like Gmail. Right. Right. Right.
• 6:13
  So now that Goo- we know Google's serious, and they're, the, and they're bringing in the right people.
• 6:18
  'Cause again, it is all about the, you know, the theme of, one of the themes of our, uh, festival is collaboration, right? So y- they, they recognize they can't do it alone.
• 6:26
  They need to bring in physicians and other people who can help collaborate.
• 6:30
  Um, Toby had an interesting, uh, comment this morning that said, uh, "Innovation is done at the sort of, uh, intersection between two separate, uh, yeah, um, verticals," like technology and healthcare, and that's where the most exciting, uh, innovation happens.
• 6:46
  Do you agree with that? Oh, absolutely. Um, um, you know, when we think about innovation now, uh, from this side, and Eduardo from Ascension said this earlier this morning, you know, is
• 6:59
  healthcare getting technology, or is technology starting to understand healthcare? Like, you know, w- who's moving faster in which direction? Yeah. It's interesting. Or, or is it both? Is it- Right...
• 7:08
  right, it, it, it really is a combination. So now that we know, um, Google's serious, and technology and healthcare are starting to merge, can, do you think that they can have...
• 7:16
  What kind of impact do you think it's gonna have on our healthcare, on our health system, and on the wellbeing of everybody?
• 7:22
  Well, when you think of other technologies that have come into our world, and you think how that's made life more helpful, whether that's with knowledge, or success, or health, or happiness, I think healthcare is a perfect example of one that hasn't made those jumps.
• 7:36
  So we just, [clears throat] um, published a paper in Nature, uh, the beginning of this year [coughs] on our mammography work, and I think it's a perfect example.
• 7:45
  So our mammography work was, what if we do machine learning on, uh, m- mammograms, [clears throat] and we compare that to doctors from six academic medical centers in the United States?
• 7:57
  But we wanna give the doctors a head start, so we also gave the doctors those women's previous, um, mammograms. We gave them access to the electronic health record.
• 8:06
  They had genetic information, and in some cases they had MRIs of the breast. We just had today's mammography, and the results were stunning. You know, a 9% decrease in false positives, a 5% decrease in false negatives.
• 8:20
  What does that mean? That means that women that are being told... You know, in the United States, in a 10-year period, 50% of women will have a false positive mammography.
• 8:33
  You know, in the UK and the US, there's about 43 million mammographies a year. This is an individual woman being told, "Your mammography's bad. You gotta come back for a biopsy." Right.
• 8:44
  And our computer said, "No, you don't," and we're significantly better than the docs. And the flip side, where the doctors say, "Hey, your mammography looks fine.
• 8:53
  You're okay," but it missed Stage 1 cancer, and we caught it. Yeah. But when you really look at the study, to get back to your collaboration, the best result, 'cause the docs made mistakes- Yeah...
• 9:05
  the computer made mistakes. The best results is the combo.
• 9:08
  So the impact we can have is if we bring these types of technology into workflow to, in this case, [clears throat] decrease a lot of anxiety, and ca- catch a lot of cancer, or according to your moon shots, at Stage 1, where we can do something about it, instead of Stage 44.
• 9:24
  So I think... And, and we can go deep in a lot of these places, where I think we can have that level of impact. Yeah. I think you hit it on the nose.
• 9:32
  The, th- well, this isn't actually a new, it's an, it's not a new idea, right? W- when I first started and we would get EKGs off the EKG machine, they didn't have anything at the time.
• 9:41
  They'd have the patient's information, and then you'd have to read it. And now, I don't think I've seen an EKG in five years that doesn't have the reading from the computer first, and then it takes the doc, right?
• 9:52
  You have to sort of agree with it and say, "Okay." You still have to read it, um, and then you can compare. But, um, so it's not a new idea, but certainly the potential for pattern recognition in- Yeah...
• 10:05
  all of these areas, radiology, pathology, um, is, is mind-boggling. Yeah. In, in the ophthalmology, so I think it's an amazing story.
• 10:13
  So we're able to do diabetic retinopathy screening better than, than eye docs, and we've already launched this in Thailand and India.
• 10:21
  In places where you can't get that kind of screening, we've been able to do the screening and get those folks in for treatment to prevent blindness.
• 10:28
  Um, but the, the mind-blowing part is we have this Google intern who comes, and we're trying to teach her machine learning. And we say, "Put some variables in, um, to see how this stuff works.
• 10:39
  So put things in that you can't tell from the eye- Uh-huh... like gender." And she comes back and she says, "The computer's able to differentiate between male and females in the retina."
• 10:52
  And we said, "No, there's no difference. Go do it again." And she said, "Well, now it's at, from 70% it's at 90%." So she was right.
• 10:59
  And so we're able to look in the picture of the eye and not only tell gender, which may not be very practical, because when I would do eye exams, I usually knew people's sex before I got to their eye.
• 11:10
  [laughs] But the, but we're able to, to identify hemoglobin A1C levels, age, blood pressure, cardiovascular risk factors, uh, risk factors of cognitive issues like Alzheimer's.
• 11:20
  And when you combine that retina picture with EHR information, with genetic information, you have even better ways to improve diagnostic accuracy and figure things out.
• 11:32
  So there's magic in those pictures, in those pixels, that, um, really, we think would free doctors and nurses and other healthcare professionals up to spend time with patients and talk, "Now what do we do that we have this information?"
• 11:45
  instead of doing the road stuff that actually humans aren't very good at. Right. And, and I think that the, the power, like you said, is, and the magic, is also in the data.
• 11:53
  Because the more data you have, the more we do this, the longer we're doing this, the better the predictions are gonna become, right? Yeah. We're actually pretty good at, um, uh, trying to do it with as little......
• 12:05
  data as possible, so as opposed to more data. So we were able to get really good results on the, on the mammography without, uh, you know, what I would call lots of data.
• 12:13
  So we're really trying to see, how can we do this with the least amount to get the positive results? But there is incredible power in, uh, the ability, not only with the data, but combining the different data sets. Okay.
• 12:24
  That's, uh, it's amazing. So, uh, you, you, you, you, um, men- mentioned the Ascension, uh, yeah, partnership. Can you tell us a little bit about that? Yeah. Super, super proud of our work with Ascension.
• 12:36
  Ascension, uh, to me, is one of the best examples in the United States of an organization incredibly focused on the poor and vulnerable. I think last year, $2 billion of care to the poor and vulnerable.
• 12:49
  So for when we talk at Google about we wanna build helpful products for everyone- Mm... um, when we say everyone, that is code word for the rich and the poor, right?
• 13:00
  For those that are online and those that are not online. For those in the developing world and what we call the next billion users in the devel- in the developing countries.
• 13:09
  So very, very tight alignment with Ascension around their focus on caring for everybody. Mm. Um, and then the work we did with Ascension, despite what they say in the newspaper, i- we're super proud of.
• 13:21
  So a couple of things. Um, Ascension chose us as a cloud provider for their records. A lot of healthcare companies are moving their records to the cloud. We have cloud services.
• 13:33
  Uh, Amazon has cloud services, Microsoft, and others. Um, and in our cloud services, the information is encrypted in transit. It's encrypted at rest. We have no access to the information. We cannot see...
• 13:47
  I can't tell you how many medical records there are, 'cause we actually charge for storage space, right? So think of it as a warehouse. The only one that has the key to that record is Ascension, right?
• 13:58
  We are literally, uh, uh, a more secure way and a safer way. Ascension also decided to use our, um, G Suite products or Gmail and Docs and all that kinda stuff, which is great.
• 14:08
  And then they asked us in two hospitals to do something special. Two hospitals, not three quarters of the United States- Yeah... as the Wall Street Journal said.
• 14:16
  In two hospitals, they said, "Could you help us organize the health record?" So as a practicing clinician, if you go into these health systems,
• 14:27
  it, it's like finding a needle in a haystack to try to find a piece of information. Yeah. And you really have a view just for Epic or Cerner.
• 14:34
  You don't necessarily have a view on that patient for the other parts of their record that aren't in the Epic and Cerner. Yeah.
• 14:40
  So we brought the Google technology of Search and created a unified platform for doctors and nurses at two hospitals to search their patient's individual record.
• 14:53
  So think of it as a Google search bar, and you would write in anticoagulation, and it knows, 'cause we know when you do search, what you're thinking, that you're really talking about hemorrhage.
• 15:04
  Or did this patient have an MRI? Boom, it pops up. Has this patient had a Pneumovax? Well, there was a fax that came in that's in the record that Pneumovax is misspelled in, but boom, it comes up.
• 15:15
  The same way we've developed search, search technology to organize the world's information. So we're gonna organize this one patient's information for their doctor using this technology.
• 15:26
  Now, to do that, there may be times where Google employees are exposed to personal health information, because if I'm one of those, uh, folks on our team that's working with that doctor- Mm... we wanna make sure that Mr.
• 15:40
  Smith's labs are Mr. Smith's x-rays. So to do that, you have to have a business associates agreement with Ascension to say, "We're gonna do something for you." They've asked us to do this.
• 15:51
  We're gonna help organize this record. So those Google folks are trained in HIPAA. It's through a business associates agreement. Ascension has 600 business associates agreement.
• 16:00
  When you go to any doctor's office- Yeah...
• 16:02
  any hospital, they give you a privacy form that says, "We're gonna keep your information private, but we have third parties that we work with, like Epic, Cerner, LabCorp, maybe Mayo Clinic s- you know, Specialty Diagnostics Lab" Insurance companies.
• 16:18
  Insurance companies, billing companies, that are gonna have to see part of your record to do what we've asked them to do. We had a- been asked, and the only thing we did was organize the record.
• 16:29
  We just put it in a way that it was searchable. That's all we did. So very proud that they picked us for Cloud. We have no access to the da- data. H- happy that they're using our, you know, G Suite tools.
• 16:42
  And we think actually organizing records where doctors now are free to spend time talking to patients and organizing records so they don't repeat the test 'cause they can't find it- Yeah... is actually really powerful.
• 16:55
  The other thing that we have in that is auto-complete.
• 16:58
  So if you're a clinician and you see a patient in the ER like me, who's 57, you type 57-year-old white male in the ER complaining of chest pain for four days, um, now with shortness of breath. Yeah. Right?
• 17:09
  You, how many times have you done that? Yeah. If you do it on our tool and that search bar was on me, you would write 57 and it auto-completes that sentence for you. Based on your information. On my record. It read it.
• 17:21
  That is amazing. It's amazing. Yeah. It's amazing when you're in healthcare and you say it. Oh, yeah. When people outside healthcare, they're like, "Wait a second, that's how everything works."
• 17:27
  No, it doesn't work like that in healthcare. It doesn't. So- And if you know how, you can't understand how many times you have to write it sometimes- The same... and, or type it the same times.
• 17:34
  So the press has made this into something that it's not. Yeah. This is not us mining somebody's records to sell ads, to learn it from it to do machine learning, to develop products.
• 17:46
  This is, we developed this on de-identified data. We brought this to Ascension. We're piloting with them.
• 17:51
  Every doctor that's using it, of course, is trained, 'cause they see a Google search bar, uh, on their patient's record, and it wasn't in secret. And so we're actually really, really proud of the work.
• 18:01
  Um, and we hope that we can demonstrate a real improvement in care, uh, less physician burnout, and, um, a, a more joy to taking care of people. Absolutely.
• 18:11
  And I think part of it is getting people comfortable with it, right? It's, I think that, um-Uh, very often fear is used to sort of quell innovation, right? Oh, it's gonna be...
• 18:21
  The, the, the people are gonna have access to your records. Nothing's gonna be secure. And it creates this fear that, f- that, that people go, "Whoa, whoa, whoa, slow down."
• 18:30
  But it really is more hurtful to innovation than it is helpful to secure data or anything. Because as you mentioned, we do this now.
• 18:38
  It's n- it really is no different than when you go to your doctor's office, and, you know, the, the, the, the people checking you in, checking you out, the insurance companies have the same access.
• 18:48
  And they're just supposed to look at the record for what they're supposed to do, and that's what we were just brought in to organize. We're pretty good, we're pretty good at organizing information. Yeah.
• 18:55
  Well, as, as a practicing physician- Yeah... I'm excited- Yeah... for you to roll this out to- Yeah... all, all the hospital- Yeah... systems, so we can a- we can all ben- benefit from it.
• 19:02
  Another, another area that I think is a good example of us bringing Google technology into healthcare is making predictions. So we can predict, um, where the flood's gonna be, where the fires are gonna be.
• 19:14
  So how can we make predictions in healthcare? So our team in London developed an app that, um, the rapid response team, the team in the hospital that gets called when someone's sick- Sure... um, uses.
• 19:25
  And the app looked at creatinine, which is a measure of kidney function, to say, "Hey, this person's creatinine went up. Um, go check 'em out."
• 19:34
  Uh, just that, just creatinine, good user interface, and a communication tool in an app, took the time of diagnosing acute kidney injury from four hours to 14 minutes.
• 19:45
  It, um, uh, decreased the cost of care 17% and 30 r- percent less patients had cardiac arrest. It's a wow. But that was just creatinine. Wow.
• 19:54
  Then we said, "Okay, well, what if we did our machine learning and artificial intelligence on it?"
• 19:59
  So partnering with the VA, we took that same algorithm, and instead of looking at creatinine, we looked at the person's entire electronic health record in real time as it happens, so it happened to be about 600,000 variables per patient.
• 20:14
  And now, we were able to predict acute kidney injury, instead of four hours to 14 minutes, negative 48 hours.
• 20:21
  So two days before any clinical signs of any deterioration, change in any, any sign that we would know as a doc, we would say with 90% certainty this patient is gonna end up on dialysis.
• 20:33
  So when you think about that ability to bring those kinda predictions into healthcare, it creates a whole new way of thinking about caring for patients.
• 20:41
  So if I told you, that woman or that man in your waiting room that looks fine is gonna be deathly ill in two days, w- I don't even know we'd know what to do with that, right? Yeah.
• 20:51
  But maybe we'd stop some nephrotoxic meds- Yeah... and we might wanna check fluid balance. But whatever you would do, it, it, it really gets into anticipatory care by bringing those predictions in.
• 21:02
  Now, to do those kinds of things, we need access to your creatinine level. Yeah. We need access to, if we're predicting discharge and readmission, I need to know your discharge date. We gotta be very, very sensitive.
• 21:14
  We not only need to follow the rules, we gotta explain what we're doing here, but the power is unbelievable in how I think we can improve the lives of literally billions of people. A- absolutely.
• 21:26
  Well, it, it's, you know, it's one of the things that you hear over and over for the last 10 years, everybody talks about that medicine now is a reactive system instead of a proactive system, and I think that's really the first step, and a powerful step, to changing the way we're treating patients.
• 21:40
  And as you know, one of our moonshots is driving down, uh, in curing chronic disease. And, you know, certainly if you can, if you can understand what's gonna happen before it happens, that's one of the ways we do it.
• 21:53
  Absolutely. Yeah, that's, that is, that's so, it's so exciting.
• 21:56
  The fact that you can predict, like you said, whether, uh, whether it's an MI or anything before the patients even enter the, enter into the hospital, um, uh, it, magnificent.
• 22:05
  So what, d- coming from healthcare, I just wanna g- go back to sort of your road.
• 22:10
  Coming from healthcare and now m- being in the middle of the, the, the merging of technology and healthcare, what do you think the biggest challenges are, um, both rolling it out and then also, um, having it accepted?
• 22:25
  I think the a- I got the same answer to both. I think it's trust. So I'm convinced that the technology is amazing. It's got issues. We gotta make sure that we don't have bias in it.
• 22:36
  I mean, there's issues around, I don't wanna overplay the techno- the technology though blows me away. I literally, the first- Are you surprised now that you have the inside look?
• 22:44
  The first 90 days of, at Google was, like, the best film festival I've ever been to in my life. Like, so I'd go in, you'd meet these people that are spectacular. Uh, I think we've won the war on talent.
• 22:55
  So we got these amazing people, and they would start telling you what they're working on. And, and it would be a two or three-hour meeting, and I'd do two or three a day, and I did it for th- you know, 90 days.
• 23:06
  And I was like, "Whoa." I mean, just blows me away. But we can't do it by ourselves. There's no way. It's gotta be with partnership.
• 23:16
  And when you talk about frontline caregivers, and when I say caregivers, I don't mean doctors and nurses and social worker, I mean doctors, nurses, and social workers, and people that are caring for their own families at home, right?
• 23:27
  The alpha d- daughter who's caring for everyone in the family, the elderly couple that's taking care of themselves, the, the elderly person who's alone, the, the self-care.
• 23:36
  W- we have to make it easy for these people, all of these people, to become super caregivers, to make it super easy. We have made it super hard in healthcare.
• 23:49
  So the technology is amazing, but if you don't trust us, and we can't bring it in a way that actually makes it easier, that's comprehensible, that's affordable, that's dignified, that's culturally sensitive.
• 24:02
  If we can't make it the way we would give care to, that I would care for my own family, there's n- I don't care how good the AI is, it's not gonna work.
• 24:10
  So we're really, really focused on understanding what people need, especially on the consumer side, and then deep, deep relationships on the provider, provider side, so we're wowing them and making it easy for 'em.
• 24:24
  The technology, it would be a shame on us if we can't get this technology into the real world, because we don't act in a humble way, we don't act in a thoughtful way, we don't appreciate how hard it is to be a caregiver.
• 24:37
  Absolutely. So do you see... It's funny, when I was, when I was sort of preparing and I was looking at all the different, um,
• 24:44
  uh, all the different verticals that Google's in, whether it's transportation and the maps and that, I was thinking, could you, uh, a- a- and I wondered, do you have the same vision that one day, um, uh, it's gonna be, uh, a telemedicine visit with digital health, you literally aren't even going...
• 25:01
  So w- my point first isMedicine's moving from the hospitals back into the community, right? Is Google in line with that? Do you see that as the trend?
• 25:12
  E- every product that we're coming up with, every product, we hold up against kind of a matrix of... Or actually, when we prioritize, should we do this?
• 25:22
  So one, it's gotta be a big enough problem that Google should be involved. It has to be a big enough problem that really somebody else probably couldn't do it, right? So we really want those big things.
• 25:31
  Like, we mapped the world. We're gonna now try to do something that I think- Little, little thing like that. [laughs] I think this next one's even bigger than mapping the world.
• 25:37
  We're gonna actually try to figure out physician directory, but that's a separate issue. [laughs] Um, so, um, i- the, the other thing we say is that by and large, hospitalizations are a failure of outpatient care.
• 25:50
  Outpatient care is, by and large, a failure of home care. Yeah. And home care, by and large, is a failure of caring for communities. So we wanna push as much as we can toward communities. We also wanna up-level everyone.
• 26:03
  And when we say up-level, everyone talks about, you know, advanced practitioners, da, da.
• 26:06
  I wanna up-level people at home that are taking care of themselves, and then prevent them from even entering the healthcare system. Give them the tools- Correct... to, to do it correctly- Correct...
• 26:16
  and live longer, healthy- Correct... healthier lives without intervention, um, more with proactive- Yeah... support. So we want to f- f- f- we wanna fix communities. Esther's here. She's embraced five communities.
• 26:30
  We're, I, w- I'm so thrilled that we recruited, uh, Dr. Karen DeSalvo as our chief innovation officer, who I think, um, it seems like, she won't take, she'll get mad at me, single-handedly brought
• 26:43
  N- New Orleans back, right? We wanna fix the community of the world. Yeah. Right?
• 26:47
  We really wanna do that same kind of understanding the depths of what's going on in these communities, and each one is different, and bring up the level so that everyone can kinda live their best life. Yeah.
• 26:59
  It's a, it's, it's an amazing thing. So w- what are, what are the next steps for you as you go on? So we're, congratulations, your at, your, your, your anniversary.
• 27:08
  Um, I, you know, you already explained how wowed you were on your first 90 days. Yeah. How's, how's the rest of the year been, and what are you most looking forward to over the next, let's say, year to two years?
• 27:20
  So I hope I'm back. I hope you invite me back next year. I hope next year- You heard it. There you go. He's coming back next year. Absolutely.
• 27:25
  Next year, I hope we've fundamentally improved the quality of Search, that you can't find things that make me upset, like drinking bleach, it cures your cancer, right? I, I, I, I, I wanna fundamentally change that.
• 27:39
  I want people to feel like they got more from Search than just information, potentially got to action. I want, um, thousands of doctors
• 27:50
  caring for patients in a better way because they're using our tools, and I want to im- say a year from now, we have fundamentally improved diagnostic accuracy, um, using computer vision around imaging.
• 28:04
  If we can do that in 2020, um, I will be thrilled. Uh, Google, Google thinking is, you know, I don't know, we have 10 companies with a billion users and five with five billion, or so- insane numbers.
• 28:19
  Um, we would love to be one of those, where we can say we've saved billions of years of life this year, and there's billions of people coming to us. Yeah. That's a, it, it, I mean, it's an amazing goal.
• 28:30
  It's, it's your own moonshot, and as, as we know, that's what we're about, is audacious goals that everybody thinks is near impossible, like mapping the entire world- Yeah...
• 28:41
  um, and having the fortitude and the long-term vision to stick by it and say, "We will do it," partner with companies that believe in your moonshot and, um, and, uh, and help you achieve it.
• 28:54
  Um, the, the, a- what I was gonna say before, which I, I th- I was sort of off in my own mind thinking about what the world would look like, um, in, uh, healthcare in 10 years.
• 29:05
  And do you see a world where, um, like I w- I started to say, where a patient has a telemedicine visit, they then have digital health technology where they're actually getting physical, um, vital signs, um, back to either a, a physician or even AI, AI is understanding something, and then you send a, a, self-driving car to pick 'em up to bring 'em in if they need something?
• 29:27
  Yeah. We got those too. Right. I know, right? We've got drones. We got everything.
• 29:29
  That's, I mean, you really, you're covering the whole- Um, I think fundamentally in 10 years, I feel like we'll be successful if we have helped, uh, the healthcare system and communities redesign healthcare so it's actually centered around patients and families.
• 29:46
  I think we fundamentally have built the wrong system, and I think these tools can help us achieve that. How that looks, I'm not sure. Um, but it'll be more affordable. It'll be much easier. The quality'll be better.
• 29:59
  There'll be joy in being a caregiver. Um, and health will be something that you don't have to worry about, because when bad stuff happens, or if we can predict bad stuff, we can nudge you in the right direction.
• 30:11
  And as you mentioned, and this is taking care of everyone. This is not just a, a, a local phenomena where we're doing it in the United States. This is the vision for the world, right? Oh, yeah.
• 30:22
  So we're, we're currently, we have a, in our offices in Palo Alto, this, uh, big TV screen, and we see how many people are being screened each day in India and Thailand with our diabetic retinopathy screening.
• 30:34
  These are people that would go blind, uh, if it were not for our tool and our partnership with the hospitals there that then said, "Okay, if you can find the right people, we can do the surgeries."
• 30:44
  Um, and we said, "We'll find the people for you," right? They could not handle the mass of people because of the screening, so we did the screening, in essence, using our computer vision.
• 30:53
  Now, we need those charts for every disorder in every country in the world. Yeah. What, what an amazing vision. Well, I know we're, we're running out of time a little bit.
• 31:02
  Um, I, I wanted to thank youWe at StartUp Health are so excited about what you're doing.
• 31:07
  Um, we're excited to partner with you and, and, and be available to, to, to help you with your moonshot, and I know you fit into pretty much every one of ours. So we're, we're, we're just excited.
• 31:18
  So, um, we might have time. Is it okay for just... There were... People had s-asked for, uh, you know, if they could ask a question, and we might have time for one or just one or two questions. If we can, uh, grab a mic.
• 31:31
  There we go. Uh, hi there. This is, uh, Jerry Kile from Health4 Labs. Uh, my question ties back to the prior presentation actually a little bit.
• 31:41
  It ended by saying that we need to mind the water between the caregiver and whoever is on the other end.
• 31:48
  And I would argue that this is not environmental air or water or anything like that, but, but it's the interconnectivity, empathy, essentially. So as we're moving more and more towards
• 32:00
  technological and self-help solutions, do you see a way that we can try to maintain this level of empathy that actually gets outcomes and results? How, how could we do that using technology?
• 32:10
  Yeah, I think fundamentally-- I think it's a great question, Jerry. I think fundamentally, um, healthcare is people caring for people.
• 32:19
  I mean, that's the core of this business, whether that's, uh, a family member caring for someone else or professional. It's people caring for people. The technology, to me, can't get in the way of that.
• 32:31
  It has to facilitate that and allow us to scale that. So, um, that, that's how we're thinking about it. So the, the empathy-- Look, I'm a child psychiatrist.
• 32:43
  It, it's about understanding what's going on, meeting people where they are, talking about different options. Um, I haven't seen a computer able to do that yet.
• 32:52
  The computer can help give me the information so you and I can have a more thoughtful discussion, and it can actually level the playing field, which is very helpful 'cause it can give you a lot of information, right?
• 33:00
  If we were in this relationship. And so then to me, it allows us to have that meaningful interaction and discussion, which is where the healing takes place.
• 33:10
  I was gonna say, and that we-- I spoke about that this morning when, um, we're in a system now where doctors are forced-- doctors or healthcare providers are forced to do so much additional work that i-is necessary, but there has to be a better way because we are losing the human connection- Yeah...
• 33:28
  um, at people's most vulnerable times, where we say, "Okay, wait, hold on. Let me just, you know, enter, enter this into the record."
• 33:35
  And if there was a better, more efficient, effective way to do that so we could maintain those connections, I think that's one of the, one of the strongest ways that we can maintain our, our empathy. Um, yep.
• 33:46
  One, one more. Over here. Yep, there you go. Okay. Um, sorry. I have my mobile. Um, yeah.
• 33:54
  So with GDPR, HIPAA, and CCPA and the number of AI initiatives that you're working on, how do you balance confidentiality and patient consent notification requirements with respect to pilots and algorithms, um, algorithms development?
• 34:11
  Yeah, great question. So, um, I'd say very carefully. So the, um, uh, the-- It's pretty clear from an IRB standpoint when we're doing research that, um, you consent people to what we're doing, right?
• 34:27
  So most of our work, most of our work, the vast majority of our work is... Well, when it's on the cloud, we're not looking at anything, and we don't need to, and we're happy to just be in the cloud business.
• 34:38
  When we're doing the research like on the mammography, um, on the diabetic retinopathy work, it's de-identified data, uh, that we always do in partnerships with academic organizations, and I think it's our moral imperative to come up with these discoveries.
• 34:56
  Um, and if people are participating in research and it's identified, you need to have their consent. Um, the different countries, it's, it's a little different, um, about consent when it's around care.
• 35:13
  So in the US, for example, um, your healthcare provider does need other people to help provide care to you, and that's through that business associates agreement.
• 35:25
  And then the consent is at the level of the provider getting it from the patient, but nobody goes over all six hundred business associates agreements.
• 35:34
  We keep thinking about this one, how we can get more agency there and more understanding.
• 35:40
  There, there's-- We do appreciate that we're s- in a somewhat unique position in that we're also a consumer company, and there's, there's worry that we're combining stuff.
• 35:53
  We're not combining stuff, but to build that trust, I think we really have to think about ways we can give people agency to understand what's going on in this particular instance where we may be exposed to some private health information.
• 36:09
  I hope that answers. Okay. I know there were two-- Can we do one more? Sure. I, I know, I know you have to go, too. Sure. Uh, well, where is the mi-- Is there a mic somewhere? It's coming. Yeah. Wait, it's coming.
• 36:27
  Thank you. Outstanding. Uh, I don't think I've ever, uh, tweeted this much during a call, talk, uh, as I have, uh, taking notes from what you said, uh, Dr. Feinberg.
• 36:35
  It seems like this seems, it seems the focus is on identifying patients who need care. Thirty percent of the care delivered in the United States is unnecessary.
• 36:45
  Shouldn't the focus be put more on identifying or helping patients identify that they don't need care for a specific ailment?
• 36:53
  That we-- that Google could play a better role as a triage for unnecessary care rather than a triage for necessary care?
• 37:00
  It would be a better impact on the overall economic well-being of the system, I would imagine.Yeah, I w- I would push back a little. I'd say it's higher than thirty percent.
• 37:10
  So, um, uh, or I guess I'd push back that I'd agree more with you than you're even saying.
• 37:16
  Um, yeah, we think there's ways about digital triage, transparency on pricing and quality, particularly in this country, that we can attack that issue by, uh...
• 37:29
  I mean, in my last organization, Walmart used us as a center of excellence for spines, hearts, and bariatric surgery, and, um, if you were a Walmart associate, a cashier, whatever, and you needed that surgery, they sent you to us, or you had to pay out of pocket full price at any other hospital.
• 37:46
  And these people were scheduled for surgery. Scheduled. That's how the program got triggered. Um, forty percent of the time, we didn't operate. And I just met with the Walmart folks.
• 37:58
  They said, actually, the number now is at fifty percent. So I would call that, uh, in some cases, assault, right?
• 38:06
  So we're, we're providing treatment to patients that are not-- don't have a positive outcome and can definitely cause harm.
• 38:15
  So we do really focus on that, about how we can, um, uh, understand better who needs what and get you to the right level of care.
• 38:24
  We wanna attack that from both sides, both on the consumer side with digital triage, and also on the clinical side in understanding better, um, the, the, the outcomes and the, the who needs what at what level of care.
• 38:38
  So completely agree, completely agree, except you gotta up your percentage. And, and that's what you were sort of talking about when you, when you were saying even with the mammograms, identifying- Right...
• 38:46
  the false positives- Correct... as much as the- Right... the false negatives. Right. Correct. Yeah, absolutely. Well, I ca- I can't thank you enough.
• 38:53
  I, I li-- I feel bad because I would stay up here for at least another hour, and I know that you have places to go, and I know the audience would probably sit here and just continue to ask questions.
• 39:03
  Thank you so much for being so open, sharing your vision, your visions, and I have to tell you, as a physician, as a patient, as a family member, very, very excited to see what the next, uh, few years brings for you and Google.
• 39:16
  So thank you. So excited. Thanks so much. All right. Thanks. [audience clapping]