本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
大家好，欢迎回到《医疗创新播客》。
Hey, everyone, and welcome back to the Medical Innovation Podcast.
今天的嘉宾，做了一件医疗行业几乎没人能做到的事。
Today's guest has done something that almost no one in health care has managed to do.
他亲身深入过医疗体系的每一个关键环节。
He's led from inside all four corners of the health care system.
他当过内科医生，也做过预防心脏病专家。
He's practiced as an internist and preventative cardiologist.
在蓝十字蓝盾担任首席医疗官时，他起草了公司首个按价值付费的合同。
As Chief Medical Officer at Blue Cross Blue Shield, he wrote the company's first value based care contracts.
在奥克纳健康担任高级副总裁期间，他领导了全美前25名的医疗责任组织之一。
As EVP at Ochsner Health, he has led one of the top 25 ACOs in the country.
在沃尔玛担任医疗交付高级副总裁时，他主导了沃尔玛健康诊所网络的建设。
As SVP of Healthcare Delivery at Walmart, he led the Walmart Health Clinic Network.
现在他在Lumeris担任高级副总裁兼首席医疗官。
And now at Lumeris, he is EVP and chief medical officer.
大卫·卡穆什医生，非常高兴您能来参加我们的播客。
Doctor David Carmouche, it's so great to have you on the podcast.
欢迎。
Welcome.
是的。
Yeah.
谢谢邀请我。
Thanks for having me.
首先，我们特别想听听您的经历。
So first, we'd love to start off by hearing more about your background.
您能说说，是什么造就了今天的您吗？
Tell us, you know, what made you who you are today?
您是怎么走到今天的这一步的？
How'd you get to where you are?
天啊。
Wow.
不过，我们大概有多少时间？
But, yeah, how long do we have?
对吧？
Right?
从小我就被我妈说将来一定会当医生，她是个老师。
I mean, you know, from the time I was a kid, my mom, who was a school teacher, said I was gonna be a doctor.
说实话，我这辈子从来没想过自己会做别的职业。
So I don't know there's not been a point in my life where that wasn't reality for me.
我在学校成绩不错，父母也一直支持我，运气挺好，后来就上了医学院，选了内科。
And, you know, I was good in school and and had a lot of support from my parents and, you know, was blessed in that right and and then ended up going to medical school and and chose internal medicine.
我这个人特别喜欢解谜。
I was a guy who really liked puzzles.
享受解决问题的过程。
I liked solving things.
特别爱看福尔摩斯的故事。
I like Sherlock Holmes stories.
小时候，我觉得内科就是无穷无尽的谜题。
When I was a kid, you know, internal medicine was the infinite set of puzzles.
对吧？
Right?
每个病人都是一个待解的谜题。
Every patient had a set of puzzles.
后来我在阿拉巴马大学伯明翰分校接受培训，那可是顶尖的学术医疗中心，还被选为住院总医师。
And got to train at University of Alabama Birmingham, a really, really top notch academic medical center, got chosen to be a chief resident.
所以刚出来行医时，我做的正是我心目中医疗事业的全部模样。
And so coming out into into into practice, I mean, I was doing everything that I thought I was gonna ever do in in health care.
我在治愈病人。
I was healing.
我在建立关系。
I was building relationships.
我经营出了不错的诊所，有很棒的搭档，一干就是十五年。
I built a great practice, had great partners, and did that for fifteen years.
而且说实话，我那时候挺关注这事儿的。
And and, you know, I was interested.
我一直都在观察医疗体系的运作。
I was an observer of of health care.
我知道，虽然我所在的机构声誉很好，但那全靠口碑，没什么数据支撑。
I I was aware that although our the organization I was part of in practice had a really great reputation, it was all based on reputation.
根本找不到任何事实依据。
There was no facts to support it.
我根本说不上来我们机构在控制血压方面到底做得怎么样。
I I couldn't have told you how well do we treat blood pressure at at the organization.
那时候我们还在用纸质病历。
And then we were in a paper record environment.
想搞质量改进，就得雇护士一页页翻病历。
And so trying to do quality improvement projects meant hiring nurses to comb through charts.
后来，正好赶上电子病历开始兴起，也就是《高技术法案》那个年代，我就慢慢介入进来了。
And, anyway, you know, got involved with thinking about, you know, during the the time when EMRs were becoming a thing, high-tech act era.
你知道，我说服了我们机构的110位医生。
You know, we I convinced our org, a 110 docs.
那是一家私营企业，本质上是个合伙制，我成功让这些医生相信，这是一笔值得的投资。
It was a private, you know, company, partnership essentially, convinced those docs that it a good investment.
我们非做不可。
We needed to do it.
于是他们把这项任务交给了我来牵头。
And so they handed it to me to lead it.
太好了。
Like, great.
你去负责把电子病历系统落地到诊所里。
You go you go implement the EHR for the practice.
当时我们想要用Epic系统，但规模不够大，没法直接和Epic签合同。
And we weren't we wanted Epic at the time, and we weren't big enough to contract with Epic.
所以我们只能想个特别的办法：跟另外两家医疗机构合作，成立一个共享服务组织，用这个新实体去和Epic签约，然后在三家机构里共用同一套Epic系统。
And so we had to do something really crazy, which was partner with two other medical organizations, create a shared services organization that could contract with Epic, and then deploy Epic, one instance, across three organizations.
所以决策和治理权的复杂性，还有怎么配置Epic系统，难度直接翻了三倍，但我反而挺享受的。
And so the complexity of decision making, governance rights, how are we gonna configure Epic, was three times as complicated, and I actually liked it.
简直疯了。
It was crazy.
我简直就是自找苦吃。
Like, I was a glutton for punishment.
我有这个耐心去应对这一切。
I I I had patience for it.
我的大多数合伙人觉得我疯了。
Most of my partners were like, you're nuts.
就在那段时间，除了忙这些事，还得应付诊所的日常，我顺便还给路易斯安那州蓝十字保险公司的CEO看病，他是当地最大的保险公司老板。
And and so during that time, in addition to all that and and being in the throes of practice, I also happened to be able to take care of the CEO of Blue Cross of Louisiana, who was the biggest insurer in in town.
每次见到他，我一边给他看病，一边跟他讲他们公司哪里做得不对，怎么才能把医疗体系改得更好，怎么付费才合理。
And so every time I would see him, I would I would see him and take care of him, then I'd also tell him everything they were doing wrong, like how they could make health care better, how they could pay for it.
所以长话短说，在《平价医疗法案》通过之后，蓝十字的商业模式一夜之间变了——从以前跟医疗机构对立、压价、控制经纪人、靠精算赚钱，变成了必须跟医疗机构合作，共同管理医疗成本，因为现在必须接受所有投保人，包括有既往病史的。
And so ultimately, to make the very long story not so long, he convinced me after the Affordable Care Act was passed, and it changed the business model overnight for Blue Cross from one of an adversary with its provider network where they were negotiators and and extracted the lowest reimbursement from that network and owned the brokers and had great underwriters to one where they were gonna have to collaborate with their provider organizations to manage care in a world where guaranteed issue and preexisting illness coverage and all those things were part of their life.
所以他意识到他们毫无公信力，于是说服我放弃照顾几千名患者，转而去为路易斯安那州两百万人口提供服务。
And so he recognized that they had no credibility, and he convinced me to come stop taking care of several thousand people and come take care of 2,000,000 people in the state of Louisiana.
这成了我职业生涯中期的关键转折点，直接带我走向了你刚才提到的那些精彩经历。
And and that was the pivotal career move for me mid career that took me to all the great experiences that you just mentioned in my lead up.
当时，我完全不知道下一步该怎么做——一辈子都以为当医生就是照顾单个病人，现在却要放弃这一切，去干点别的。
And so and and and at that moment, not knowing what to do after your whole life thinking that being a doctor means taking care of individual patients, I was faced with this opportunity to give that all up and go do something else.
我记得曾跟一位我很敬重的资深医生谈过这件事。
And I remember talking to a very smart person who I respected, who was senior to me in in in medicine.
我把这个机会告诉了他。
And I told him about this opportunity.
我问他：我该怎么办？
I said, what do I do?
他说：如果你觉得医疗体系出了问题，那你不是在帮倒忙，就是参与解决。
And he said, well, if you believe health care is broken, then you can only be part of the problem or part of the solution.
你不可能两边都占。
You can't be both.
他说，我想成为解决问题的一方。
And he goes I said, well, I wanna be part of the solution.
他说，你得去蓝十字，因为从那个位置上，你能对医疗体系的变革产生更大的影响。
He said, you gotta go to Blue Cross because you're gonna have a bigger impact on changing health care from that seat.
于是，这开启了我过去十五年的一连串管理岗位，从医保方到大型综合医疗系统，主导价值医疗和基层医疗，一路走到沃尔玛——美国最大的公司，致力于把社区基层医疗引入门店，把超市、药房和医疗服务整合成一种独特的模式。
And so that launched me into a series of management roles over the last fifteen years that took me from payer to large integrated health system leading value based care and primary care to Walmart, the biggest company in The United States, trying to bring community based primary care into stores and combine grocery and pharmacy and health care delivery into a unique offering.
到现在，说实话，我正处在人工智能和你们这个领域的交汇点上，比在我自己的领域还多，学到了大量关于技术的知识，同时也带着我这些丰富的经验在思考：
To now, really, you know, sitting at the intersection of AI and your world probably more than mine and learning so much around the technology, but bringing this rich experience to, hey.
我们该怎么用这项技术？
What do we do with this technology?
我们该怎么负责任地使用它？
How do we do it responsibly?
我们怎样才能真正带来改变？
How do we make a difference?
诸如此类的问题。
And and etcetera.
这段经历真是跌宕起伏。
So it's been a a crazy journey.
说来话长，但特别重要，因为它塑造了我对当下技术使用的看法。
Long story, but it's really important for a variety of reasons because it shapes my view of how we're using technology today.
不管你对人生未来有什么设想，能猜中三成就不错了。
But, also, it's just a crazy reality that whatever you think's ahead of you in your life, you probably have it about 30 maybe 30 right if you're lucky.
剩下的路，都是慢慢走出来的，机会来了就抓住。
And the rest of it, it just evolves and, you know, you take advantage of opportunities as they come.
是的。
Yeah.
作为医生，我经常思考这个问题。现在还在培训阶段，未来想从事内科，亲身看病时，你直接照顾病人；但当你把临床经验跟系统思维结合起来，就能影响更广的范围——就像你说的，不只是看诊的病人，而是整个医疗体系。
It's really interesting too as a physician where I've I've thought about this a lot being in training now and and someone that does wanna do internal medicine and and practice of the perspective of, like, you you provide care directly to people in as a physician when you're providing clinical care, but then you also have the skill set when you integrate it with other you know, your understanding of the system to to span that and, like you mentioned, helping the entire state versus, like, the patients that you're seeing directly.
但这种影响确实没那么直观。
But it does feel like that impact is a little bit harder to feel.
当你面对面给病人看病时，能立刻看到自己的付出带来改变；可当你站在更大的层面，这种感觉就弱多了。
Like, when you're talk when you're talking about how it is with a physician, you know, you're taking care of the patients, you see the the the care that you're providing versus when you're looking at this broader scale.
也许这种影响没那么容易经常看到。
Maybe it's harder to to necessarily see what that that impact is as regularly.
是的。
Yeah.
我的看法是，在临床实践中，你的影响是立竿见影的。
I I mean, my my perspective on it is just that in in practice, your impact is felt immediately.
你能直接看到效果。
Like, you see it.
你跟病人互动。
You you you engage with a patient.
你解决了一个问题。
You solve a problem.
你做出了诊断。
You make a diagnosis.
而在管理或更宏观的商业岗位上，你得用一两年的时间来衡量成败。
And in management or in business roles that are bigger, you you you you you measure success over, like, a year or two years.
你会问，我带领团队做了哪些工作？
You you say, you know, what's the work that I've helped lead the team I've built?
我们整体上达成了什么？
What have we collectively done?
过了一两年回头看，你会惊讶地发现。
And you look back after a year or two, you're like, oh my gosh.
我们真的产生了很大的影响。
Like, we've really made a big difference.
所以，如果你需要快速看到成果来获得成就感，那每天直接照顾病人绝对是最好的方式。
So it's just a matter of, like, if you need the quick hitter kind of kind of, you know, kind of validation for what you're doing, then, you know, taking care of people daily is is an awesome way to do it.
如果你愿意下长远功夫，建立体系，产生更广泛的影响，并且有这份耐心，那管理岗位可能更适合你。
If you're willing to play a longer game and build systems and impact bigger, you know, kinda you know, and have the patience for that, then, you know, management role's probably the way to go.
是的。
Yeah.
这完全说得通。
That that makes complete sense.
说到现在，还有你在Lumeris的职位，我们很想多了解一下这家公司，毕竟它经历过好几次转型，现在的样子和刚成立时已经大不一样了，能再讲讲它的背景吗？
And then as we're talking about kind of the present and your role at Lumeris, we'd love to hear a little bit more about, you know, what Lumeris is and kind of I know the company itself has has gone over multiple evolutions, and and the the latest version of it is is different than it was initially, but, a little bit more about the the context of the company.
是的。
Yeah.
这家公司挺有意思的。
I mean, it's an interesting company.
它是一家风险投资支持的公司，已经存在十五年了。
It's a venture backed company that's been around for fifteen years.
大多数风投支持的公司，要么倒闭，要么被收购，要么上市，要么并入别的企业。
Most venture companies are, you know, either they they either get shut down or or they're or they're sold or go public or get merged into something else.
但这家公司不同，是一家由圣路易斯的医生们创办的资深风投企业。
So this is a it's a senior venture backed company started by physicians in St.
圣路易斯。
Louis.
一群内科医生在圣路易斯。
So a group of internal medicine doctors in St.
圣路易斯的一群内科医生，特别用心地照顾病人，把自己的电话号码给病人，上门出诊，去医院查房，完全是老派全科医生的做法。他们特别反感按服务收费的商业模式，觉得应该追求价值导向。
Louis who really took good care of their patients, would give their phone numbers to their patients, would make house calls on their patients, would round on them in the hospital, like old school primary care, and just hated fee for service as the economic model for that and felt like they wanted value.
他们想要的是总额预付。
They wanted capitation.
他们想要的是预付款模式。
They wanted prepayment.
他们和市场上主流的支付方打得不可开交，联合健康就是其中之一。
And they were at war with the prevailing payers in the market, United being one of them.
于是他们决定，必须自己成立一个支付方，来和他们协同合作。
And so they they decided they needed their own payer to collaborate with them.
汤姆·多尔就是创始人之一。
And so they Tom Doerr, Thomas Doerr was one of the founders.
他的哥哥是约翰·多尔，那位风投大佬，凯鹏华盈的创始人，靠投资谷歌、苹果和亚马逊赚了大钱。
His brother was John Doerr, the venture capitalist, Kleiner Perkins founder, who had made his money investing in Google and Apple and Amazon.
汤姆给哥哥打电话，说：嘿。
And Tom called his brother and said, hey.
你能帮我们买一份医疗保险优势计划吗？
Can you help us buy a Medicare Advantage plan?
他回答说，当然可以。
And he said, sure.
我就当是为我弟弟做这件事。
I'll do it for my brother.
于是他们推出了一款计划，主要是面向中西部地区的。
And so they built a plan, and and it was a Midwestern plan.
直到今天，这个计划还在运行。
Still we still have it today.
它叫Essence Healthcare。
It's called Essence Healthcare.
当时它只是一款HMO类型的计划。
And it's an a it was an HMO only plan at the time.
每个病人都有位全科医生，而大多数全科医生都承担了总额预付的风险。
So every patient had a primary care physician, and most of those primary care physicians took capitated risk.
于是你拥有一群非常优秀的医生，他们把星级评分做得很高，因此这个计划的收入也很高。
And so you got a really good group of great doctors, and and what happened is they managed the star ratings high, so the revenue of the plan was high.
他们还把成本控制得很低。
They managed the cost down.
病人的体验非常好。
Patients had great experiences.
医生们赚了不少钱，整个计划也非常成功。
The doctors made plenty of money, and the plan was successful.
所以他们觉得自己找到了秘诀。
So they felt like they had the secret.
天啊，这简直就是制胜法宝。
Like, man, this is the secret sauce.
保险公司和一群有责任心的医生紧密合作，效果简直神奇。
A collaborating payer with a group of well intended clinicians is magic.
于是我们创立了Lumeris这家公司，要把这套方法商业化，教给全世界怎么做。
Let's build a company, Lumeris, to commercialize that IP and teach the rest of the world how to do that.
于是，Lumeris 就这样成立了。
And so Lumeris was born.
Lumeris 自己经历了多次迭代，最近几年它的核心角色是与大型医疗系统和医生集团合作，通过技术驱动的服务，帮助管理多个基于价值的群体医疗合同。
Lumeris itself has gone through many iterations, and its most recent iteration for the last many years has been a collaborating partner with large health systems and large medical groups, bringing technology enabled services to bear to help manage multiple population value based care contracts.
比如联邦医疗保险、委托的医保优势风险合同、商业绩效付费或共享节余合同，还有医疗救助绩效合同。
So Medicare, delegated Medicare advantage risk, commercial pay for performance or shared savings contracts, Medicaid performance.
他们一直在这块做得不错，但这个领域确实很难。
So so that's what they've been doing successfully, and it's been a that's a tough space.
对吧？
Right?
所以很少有公司敢在这个领域深耕，尤其是和那些 notoriously 难合作的医疗系统搭伙。
And so not a lot of companies have staked that spot out, especially with health system partners who are notoriously difficult to partner with.
我刚加入 Lumeris 时，最让我惊讶的是，通过这些工作，他们已经掌握了两件对人工智能发展至关重要的事。
When I got to Lumeris, what was striking to me was that through that work, they had mastered two really important things that were gonna be important in the AI journey.
第一，他们真正懂基层医疗的工作流程。
Number one is they were really they really understood workflows in primary care.
他们长期深耕基层医疗领域，也熟悉如何与医疗机构打交道——这对很多人来说简直像黑箱，还得应对医院复杂的治理结构、决策流程和漫长的销售周期。
They had worked in and around primary care doctors forever, and they also knew how to call on and call into health systems, which is opaque for many people, and and deal with the governance structures and decision making and sales cycles of health systems.
医疗机构特别重要，因为全美超过百分之五十五的全科医生都是他们雇用的。
And health systems are important because they employ over 55% of all the primary care physicians in The United States.
所以如果你想在基层医疗领域做点什么，这无疑是个绝佳的切入点。
So if you're gonna go do something in primary care, that's a pretty good place to start.
第二点，他们学会了整合零散的数据。
The second thing is they learned how to aggregate disparate data.
想想看，医疗机构里有多少杂乱的数据孤岛：一个或多个电子病历系统、多个医保理赔数据流、药房信息、医疗信息交换平台、患者登记与转诊数据，还有各种可整合的消费者数据。
Think about all the dirty silos of data that come into a health system, one or more EHRs, multiple payer claims feeds, pharmacy, HIE, ADT feeds, consumer data that could be aggregated.
他们搭建了一个平台，把所有数据清洗、整合、标准化，用来驱动分析、生成洞察，再把结果交给人工去处理那些名单。
And they had built a platform of cleansed, aggregated, normalized data that was used to power analytics or create insights that they would then hand to humans to go do things with those lists.
所以我去的时候，他们正想利用生成式AI，尤其是代理型AI，思考一个问题：我们能不能不再只是把名单交给人，
And so when I got there, you know, they were interested in leveraging generative AI and specifically agentic AI ultimately to think about how do we not just hand list to humans.
而是直接让AI自己采取行动？
How do we build AI to actually take action?
咱们就这么干吧。
Like, let's do it.
我当时就想，这主意不错，但光在价值医疗里做，或者只针对想和我们签风险合同的医疗系统，市场太小了。
And I was like, that's awesome, but I wouldn't just do that in value based care, and I wouldn't just do it in health systems who wanna partner with us for risk contracts because the market for that's too small.
为什么不把目标扩大到所有在整合初级医疗资源的机构呢？
Why don't we go after anyone as a customer who's aggregating primary care?
现在初级医疗的人手严重不足，而且未来十年只会越来越紧张。
Because there's a workforce shortage that, you know, is here today and getting worse over the next decade.
对初级医疗的需求非常大。
The need for primary care is great.
医疗系统可以通过初级医疗关系获得后续收益，因此他们有经济动力去投资软件，帮助员工服务更多患者。
Health systems can monetize primary care relationships downstream, and so they can build a economic model to to justify investments in software that can help their their humans take care of more people.
所以我大约一年半前加入公司时，我们迅速调整了方向，从用AI自动化价值医疗，转向更广泛地支持初级医疗。
And so so when I got to the company about eighteen, nineteen months ago, the quick pivot was away from value based care automation with AI to broader support for primary care.
过去这一年半，我们的平台就是沿着这个方向发展的，汤姆——这个名字是为了纪念汤姆·多尔，我之前提到过，他是公司的联合创始人，六年前在上班前游泳时不幸离世。
And that's kind of been the journey over the last eighteen months with our platform, Tom, named after Tom Doerr, who I mentioned was the cofounder of the company who tragically passed away swimming before work six years ago.
所以，为了纪念汤姆，也因为他一直相信科技将是实现规模化、高质量初级医疗的关键，只是他没能活到看到足够强大的技术真正实现这一天。
And so kind of as an homage to Tom and also because he believed that technology was going to be really important part of delivering scale, high quality primary care, just never live long enough to see technology that was powerful enough to do it.
我们把他的名字留在这项事业里，算是一种传承。
We've kind of gotten kinda attached his name to that as kind of a legacy.
这就是公司这些年来逐渐演变到现在这个方向的过程，你知道的，就是如今的样子。
So that's kinda how the company has evolved to what it's now doing, you know, today.
是的。
Yeah.
谢谢你分享这些。
Thank you for sharing that.
这真是一段引人入胜的历程。
It's a fascinating journey.
而且，正如阿尔温德之前说的，公司已经发生了重大转变，现在聚焦在汤姆的理念上，把人工智能融入初级医疗，提升全球范围内的可及性，不再局限于价值医疗，也不再只盯着像你当年在蓝十字蓝盾时那样的单一市场。
And, definitely, as Arvind said earlier, the company has sort of evolved and pivoted a significant amount to now be working on sort of Tom and focus on integrating AI into just improving primary care, making primary care more accessible around the world, not just focused on value based care, not just focused on one market like you were doing back when you were at Blue Cross Blue Shield, for example.
没错。
Exactly.
但在深入之前，我想先退一步，聊聊你加入Lumeris的经过。
But before going a little bit deeper, I'd love to take a step back and just discuss your transition to Lumeris.
就像你之前提到的，你之前在沃尔玛，当时他们正进入医疗领域，建立实体和线上诊所，主要服务医疗资源匮乏的地区，目标和你现在做的其实很相似。
Like, as you mentioned before this, you were at Walmart as they've entered into the health care space, building out brick and mortar and virtual clinics for primary care, especially in underserved areas, a very similar mission, sort of same goals as you're working on right now.
我们都知道，我相信很多观众也知道，沃尔玛的医疗项目最终没能成功。
As we know, and I'm sure a lot of our audience knows, Walmart's health initiative unfortunately did not work out.
他们后来把这项业务关闭了。
They ended up shutting that down.
我想知道，从这段经历里，你学到了哪些东西，现在用在了你的新工作中？
I'd love to know what did you learn from that experience that you're working into your current role?
另外，这个问题的第二部分是，是什么让你特别选择了Lumeris？
And, also, as a second part of that question, what attracted you to Lumeris specifically?
我猜你肯定也考察过其他公司，那些在AI和价值医疗、AI和初级医疗交叉领域做事情的企业。
I'm sure you explored your options, looked at various companies doing, you know, intersection of AI and value based care, AI and primary care.
在这么多选择里，到底是什么让Lumeris脱颖而出，让你觉得非来不可？
What about Lumeris stood out to you that made me you decide that I have to go to this company?
是的。
Yeah.
太棒了。
Awesome.
问得真好。
Awesome question.
沃尔玛健康项目是个绝佳的机会。
So Walmart Health was such an amazing opportunity.
虽然最终令人失望，但当初加入沃尔玛健康的人，几乎都对传统医疗体系感到不满，怀揣着改变的使命感。
It ended with disappointment because of how mission oriented all the people that everyone who came to work at Walmart Health was in some ways frustrated with traditional health care.
现实是，传统医疗体系的诸多问题，并不全是因为激励机制错位或实体诊所的固定成本造成的。
And the reality was that, you know, not all of the ills of traditional health care are related to mixed incentives or financial realities of the fixed cost of bricks and mortar.
这确实是原因之一。
I mean, that's part of it.
但问题远不止这些。当时有不少有想法的人，坚信沃尔玛有足够的规模和资源，能另辟蹊径，真正解决问题。
It just turns out that, you know, there's more to it, but but there was a lot of malcontents who really had great ideas and thought that Walmart had the scale and the resources to figure it out and do it a different way.
所以那是一群充满理想的人。
And so it was a bunch of idealistic folks.
我的意思是，很多人真的有医疗行业的经验。
I mean, a lot of people had real health care experience.
我特别喜欢这家公司的地方，就是能和一群真正想改变世界的人一起工作。
So I what I loved about the company was just working around really mission oriented people who wanted to make a difference in the world.
我第二喜欢这家公司的地方，其实跟医疗一点关系都没有。
The second thing I loved about the company was really nothing about health care.
我只是特别想了解，一家即将达到万亿营收的公司，到底是怎么运作的。
I just loved learning about how do you run a company that is, you know, on its way to being a trillion dollar in revenue company.
比如，两千两百万名员工，全美五千多家超级中心，还有电商业务。
Like like, millions you know, 2,200,000 employees, you know, 5,000 supercenters across the country, the an ecommerce business.
能去参加一些高管会议，学习零售业是怎么做的，真的特别有意思。
Like, it was just fascinating to learn the retail business to sit in some of the leadership meetings.
我学到的一点是，沃尔玛拥有一项非常宝贵的经验，而医疗行业恰恰缺少这个——这也是我最重要的收获。
And and what I learned was that there was a really valuable lesson that that I think Walmart had that health care doesn't quite have, and it and this is the most important learning.
其实，我在沃尔玛学到的所有事情里，接下来这一条最重要。
Actually, every of all the things I learned at Walmart, this next thing is the most important.
我来给你讲一个让我彻底明白这一点的故事。
And I'll give you the story that crystallized this.
我加入沃尔玛之前最后一次开会，是在奥克纳医院。
The last meet before I went to Walmart, was at Ochsner.
疫情期间，我参加的最后一次奥克纳会议，是一场管理层会议。
So during COVID, right, the last meeting at Ochsner that I was at was a leadership meeting.
当时三千多名管理者齐聚一栋大楼里。
We had 3,000 leaders altogether in a in a in a, you know, big building.
我们有五条核心价值观，其中一条就是‘患者至上’。
And we all of our values were like we had five, you know, corporate values, one of which was we put, you know, patients first was was one.
每次管理层会议开始前，我们都会分享一个患者的故事。这次讲的是一个十几岁的女孩，她感染了新冠，自己却不知道病毒已经影响了心脏。
And so we would always start a leadership meeting with a patient story, and we had the story of this young woman who, a girl, really, a teenager, who had, had COVID, had, unbeknownst to her, it was impacting her heart.
她在学校操场上突然心脏骤停，被抢救过来后，先送到合作的社区医院稳定病情，接着由空中救护转运到总部，住了三周ICU，用了ECMO，最后康复出院，和父母一起走上舞台，全场起立鼓掌，很多人当场落泪。
She was on the playground at school, had a cardiac arrest, was resuscitated, taken to our partnering community hospital, stabilized a bit, flown by Life Flight to our main campus and spent, you know, three weeks in the ICU, was on ECMO, and survived and walked out on the stage with her parents and, you know, standing ovation, lots of tears, everyone.
那真是个特别动人的时刻，我们的首席执行官说，这就是我们存在的意义。
It was a very emotional moment, and our CEO was like, this is why we exist.
这就是所谓‘患者至上’的真正体现。
We you know, this is what patients first looks like.
嗯。
K.
几周后，我参加了我在沃尔玛的第一次管理层会议。
I get to my first leadership meeting at at Walmart a few weeks later.
六百名高管齐聚一堂，全是沃尔玛的各级负责人，他们讲的也是‘顾客第一’。
600 executives for, you know, all the leaders of Walmart in in one room, and they talk about they put the customer first.
道理其实很像。
So similar.
对吧？
Right?
患者第一，顾客第一。
Patient first, customer first.
他们开会一开始，就放了一段客户访谈，大家在谈现在在美国生活有多难，汽油价格太高了。
They start their meeting with a series of interviews of the customers who were basically talking about how hard it is to live in America these days, The cost of gas.
那时候取暖油价格飙升，有一位年长的女士说，她只能把恒温器设在63度，不敢调高，不然付不起账单，得在取暖和买药之间做选择。
At the time, heating oil costs were high, and there was a lady who talked about having to only an older lady who could only afford to keep her thermostat set at 63, not warmer, because otherwise, she couldn't afford the bill and had to choose between that and her medicines.
有一位女士根本买不起胰岛素。
A woman who couldn't afford insulin.
这些就是实实在在的问题。
And and it was the problems.
沃尔玛想表达的是，我们存在的意义，就是帮客户解决这些问题，你们必须真正理解他们的困境。
And so what Walmart was basically saying is, like, we exist as a company to solve those problems for our customers, and you need to understand their problems.
如果按奥克纳的做法，不该是庆祝救了一个14岁少年的英雄事迹。
The corollary to Ochsner wouldn't have been to celebrate the modern miracle of saving a 14 year old, you know, heroically.
而应该是，开会一开始就放所有患者抱怨：我根本约不到医生。
It would be as if they started the meeting with all the patients who complained about I can't get in to see my doctor.
我打多少电话都没人回。
I can't get anyone to call me back.
我付不起药费。
I can't afford my prescriptions.
我不明白我的权利是什么。
I don't understand my right?
如果健康体系的讨论就从这些真实困境开始，而医疗系统也说，我们的存在就是为了解决这些问题，那美国的医疗状况现在肯定要好得多。
If that was the starting point for a health system conversation and and health system said, we exist to solve those problems, man, health care would be way further ahead in this country than it is.
所以对我来说，我从沃尔玛的经历中汲取的，并不是那个失败的商业模式——它确实失败了。
And so for me, I think what I will take from the Walmart time is not the failure of the business model, which was was a failure.
而是这样一种理念：如果我们能真正把患者放在首位，打造一个真正为他们服务的体系，那大家都会受益。
It was this notion of, like, if we can figure out how to embed that notion of putting patients actually first in health care and and building a system that works for them, that that we will all win.
因此，我在跟Lumeris讲这个故事时，部分灵感就来自这一点。
And and so for me, the the story to Lumeris in part is is informed by by that.
对吧？
Right?
最后我想说说沃尔玛的这段经历：就像所有大公司一样，沃尔玛在大体系里想做点小规模的创新，也特别困难。
The last thing I'll just say about Walmart and the experience was that Walmart just struggled like any large company to do something subscale inside a scaled company.
我刚去的时候，沃尔玛健康业务已经有20家健康中心了。
So Walmart Health had, when I got there, 20 health centers.
我离开的时候，我们正朝着80家迈进，但其实我们需要的是几百家。
We were on our way to eighty three years later when when I left, but we needed to get to, you know, hundreds.
这事儿耗时太长，成本太高，一路上肯定要亏不少钱。
And it was gonna take a long time, and it was gonna be expensive, and we're gonna lose a lot of money along the way.
而且，沃尔玛在医疗领域根本没什么品牌认知。
And, you know, plus Walmart didn't really have a brand in health care.
尽管沃尔玛有海量顾客，但人们并不愿意去沃尔玛的诊所脱衣服看病。
And so even though Walmart had lots of consumers, they didn't necessarily wanna take their shirt off in a clinic at Walmart.
所以，它最终没成功的原因有很多，但我对这家公司依然充满敬意。
And so there was just a lot to why it didn't ultimately work, but I know a lot of respect for the organization.
我到现在还有很多朋友在那里。
I still have lots of friends there.
至于Lumeris是怎么回事的，你说得对，我确实挺幸运，得到了不少机会。
The way Lumeris actually happened, you're right, and I was blessed to have a lot of opportunity.
我在沃尔玛期间，认识了一位非常友善的女士，叫艾丽斯·沃尔顿。
One of the things that happened at Walmart when I was there is I got to meet a delightful woman named Alice Walton.
她是山姆·沃尔顿的女儿，也是世界最富有的女性，热心公益，特别关注医疗健康，正在阿肯色州西北部筹建一所医学院，不仅真心想改善那里的医疗，还愿意自己掏钱推动变革。
So Sam Walton's daughter, richest woman in the world, philanthropist who takes a real interest in health care, is building a medical built a medical school in Northwest Arkansas and has a lot of interest and and and puts her own money where her mouth is when comes to making health care better in the Northwestern Arkansas.
我刚去的时候，她就想见见我。
She wanted to meet me when I got there.
她以前住在新奥尔良。
She had lived in New Orleans.
我来自路易斯安那州。
I'm from Louisiana.
她早年在新奥尔良生活了大约十五年，一直很想更深入了解医疗体系。
She had lived in New Orleans for about fifteen years earlier in her life, and she was just really wanted to understand health care more.
她请我吃了好几次饭，慢慢就熟络起来了。
And so took me to dinner, few times got to know her.
通过这段关系，她把我介绍给了约翰·多尔。
Through that relationship, she introduced me to John Doerr.
约翰·多尔，你知道吧，我说这话是出于善意的。
John Doerr you know, I mean, it turns out and and I I say this kindly.
我觉得这个国家的亿万富翁圈子其实挺小的，大家多多少少都认识。
I think the billionaires in this country, that club's small enough where they they kinda know each other.
所以爱丽丝很慷慨地抽出时间，把我介绍给了约翰·多尔，我也就慢慢和他熟了起来。
And so so Alice, generous with her time, introduced me to John Doerr, who I got to know a a bit.
当沃尔玛健康业务关闭的时候，约翰和Lumeris，还有他参与的部分，很早就主动来找我了。
And so when Walmart Health shut down, I mean, John and and Lumeris and his involvement, they were really early to kinda come to me.
我记得跟约翰聊过，说我很希望你能看看这家公司——Lumeris。
And and I remember having a conversation with John saying, you know, I'd really like you to look at this company, Lumeris.
我当时说：我知道Lumeris，也了解你们在做的价值医疗，真的很欣赏你们的工作，但那不是我重点想提的。
And I'm like, well, you know, I know Lumeris, and I've done the value based peer thing, and I love what you're doing, but I that's not really what I mentioned.
他当时有点儿不太明白。
And it was a little bit like, no.
不是这么回事。
No.
不行。
No.
你认识萨姆·阿尔特曼吗？
Do you know Sam Altman?
你听说过OpenAI是做什么的吗？
Like, have you heard of what OpenAI is?
那时候刚推出不久，没过多久，就席卷了整个世界。
Like, are you know, this was shortly, like, not too long after, you know, that was unleashed on the world.
他说，他觉得我们有机会用这项技术在医疗领域做点不一样的事，但真的需要医疗行业的人帮我们弄清楚怎么负责任地用，怎么把公司定位好才能成功。
And and he's like, I really think we have the opportunity to to leverage this technology to do something different in health care, but we're gonna really need health care people to help us figure out how to do it responsibly and and how to position it as a company to make to make it successful.
如果你愿意把之前积累的丰富经验带进来，学学这项技术，一起想想怎么用，我觉得这事儿特别有吸引力。
And if you're interested in taking that breadth and wealth of experience that you've developed and learning this technology and helping us think about it, I think this could be really exciting.
当一个像他这样有分量的人给你这样的机会时，真的很难拒绝——我原以为自己在你说的那些领域已经把职业发展到头了。
And and it just you know, it's hard when someone like that with that much experience gives you that opportunity to I thought I evolved my career about as much as I could, right, across the domains you mentioned.
现在，我要在AI已经登场的世界里，当一个医疗科技创业者，这机会实在太诱人了，我根本没法错过。
Well, now I add health tech health tech entrepreneur in a world where AI is on the scene is just was too hard for me to pass up.
那里的人都很友善，我也大概有了点想法，觉得我们或许能一起做点什么，最后我就决定试一试。
And it was a lot of really nice people there, and, I I had an idea of of of what we might be able to do together, and, ultimately, I took a shot at that.
是的。
Yeah.
这段经历特别丰富，听你讲职业发展的各个阶段，还有普拉纳夫之前提到的那些经历，真的很有意思。
It's super, super rich, and it's it's so interesting to hear the the evolution of your career and, like, the stages that Pranav mentioned earlier of the exposures that you're getting.
比如在沃尔玛，你们关注的是要解决哪些问题；而医疗系统呢，往往只是自我表扬，说他们做了哪些好事——这当然也不是坏事。
And that that point of, like, the the the company, like, for example, at Walmart pointing out the problems that are gonna be solved versus the health systems kind of patting themselves on the back for the good things that they're doing, which isn't a bad thing.
当然，他们确实应该这么做。
Like, you should definitely Oh, they are.
你的认可很重要。
Your acceptance.
是的。
Yeah.
你当然该这么做，但同时，这也定下了你真正想达成的目标的基调。
You should definitely do that, but at the same time, it's like setting the tone for what are you trying to accomplish.
这确实是个完全不同的环境，但你已经接触到了方方面面。
And it's a it's definitely a different different environment, but you've been able to get exposed to all of it.
接下来的这个阶段，我和普拉纳夫现在都深度投入到人工智能领域了，和这些在该领域打拼的公司交流，真的特别有意思。
This this next evolution, you know, I think Pranav and I are, like, like, super deep into the AI stuff right now, and it's it's been really interesting to talk to all these companies that are building in this space.
我想更深入地聊聊汤姆，跟你探讨一下，因为从表面上看，它目标非常宏大——你们在打造的是一种智能代理系统，有些公司只做日程安排这类行政辅助功能。
And I wanted to dive a little bit more into Tom and talk to you about talk to you about it because on paper, sounds incredibly ambitious, you know, like in terms of the agentic system that you're building, you know, some companies are building agentic scheduling assistance where they're focusing on the admin stuff.
有些人专注于临床参考系统。
Some people are building clinical reference.
还有些人在做语音转录服务。
Some building people are building scribing.
但在我看来，汤姆似乎不只是安排日程，还要提供基于循证医学的患者资料。
But from what it sounds like to me, it sounds like Tom is is is intended to do the scheduling while also providing patient materials while which have to be grounded in evidence.
所以它其实是把其他公司各自做的零散功能整合在了一起。
So it's this multifaceted thing that's kind of integrating what some companies are just doing entirely, you know, the small parts.
所以我想问的是，你们是怎么一步步推动汤姆这个产品演进的？
So I guess my question is is how do you do you evolve how Tom as a product?
那最初是怎么解决的？
Like, what was the initial solve?
然后，没错。
And then and then Yeah.
你们现在还想往哪些方向继续推进？
Where are places that you're still trying to go with it?
你们真是做足了功课，一针见血。
Man, you you guys have done your homework, and you have nailed it.
对吧？
Right?
我觉得，就拿我们现在的Tom版本来说，它实际上已经相当于40家公司的功能总和。
I think we believe, like, if you look at just even our current version of of Tom, it probably represents the equivalent of, like, 40 individual companies.
对吧？
Right?
这确实是个大胆的投入。
I mean, it it's it's a big ambitious bet.
这背后有几个原因。
A couple things go into that.
我倒希望说是经过周密规划的。
I'd like to say it was really well planned out.
但其实不是，我说过，我刚介入的时候，情况是这样的。
It it it wasn't I mean, the way it started was, again, I I told you, you know, I arrive on the scene.
我看到他们已经找到了聚合数据的方法。
I see that they figured out how to aggregate data.
对我而言，我一直是个平台派。
And so if you you know, for me, like, I'm always platform guy.
就算我不是技术专家，我也知道，要大规模产生影响，单点解决方案通常做不到。
Without even being a tech guy, I just know that to create impact at scale, like, point solutions rarely do it.
少数几个能行，但不多。
There's a few, but not many.
大多数根本改变不了大局。
Most most won't move the needle enough.
我早些时候跟一些医疗系统的数字化负责人聊过，他们说，是的。
And I had some early experiences with health system digital leaders saying things like, yeah.
我们给所有放射科医生都部署了一套AI辅助诊断工具，使用率达到了百分之百。
You know, we put a a an AI radiology, you know, assist tool in for all of our radiologists, and we had a 100% adoption.
整个医疗系统里的放射科医生都在用，但影像阅片的处理量却一点都没提高。
Like, every radiologist in our health system uses it, and there's been zero change in throughput of reads for our images.
我当时就明白，那些放射科医生其实是把AI当成了第二意见，用来核对自己的判断，但还没到完全信任它、彻底改变工作流程的地步。
And I'm like he's like, so, like, what what's happening is in that world, the radiologists were using the AI tools of second opinion, and they basically tried to check their own interpretation against it, but they weren't trusting it enough to totally change their own workflows.
所以这位负责人告诉我：如果这些AI工具不能真正创造价值，我们根本负担不起一个接一个地买点状解决方案。
And so this this leader was telling me like, man, we are not gonna be able to afford AI point solution after point solution after point solution if they don't create value.
这件事也一直在我脑子里盘旋。
So that was in my head a little bit as well.
另外，要想在基层医疗上产生重大影响，
And so, the other thing is to make a big difference in primary care.
可不只是靠预约排班、就诊提醒或者健康宣教这些事。
It's not just about scheduling or appointment reminders or patient education.
全科医疗涉及的内容太多了。
There's so much that goes into primary care.
我记得几年前看过一篇文章，说一个普通全科医生管理2500名患者，如果要完全按照循证指南提供所有护理，每天得花26.7个小时。
And I just remember reading the article a few years ago that came out that said, for an average primary care doctor who took care of 2,500 patients, if you were to give all the evidence based care and guideline driven care to that panel, it would take twenty six point seven hours a day to do it.
但如果你组建一个大型医护团队，这个时间能缩短到大约9个半小时。
And you could make that nine and a half hours or so if you would add a big care team.
对吧？
Right?
比如加上护理协调员、健康教练、护士执业者、医疗助理，你懂的？
If you would add care coordinators, health coaches, nurse practitioners, MAs, you you know?
所以我心想，好吧。
And so I I thought, okay.
如果我们想用人工智能来改善全科医疗，那到底要解决什么问题？
So if we're gonna use AI to improve primary care, what problem are we solving?
对吧？
Right?
对我们来说，我们关注的是，这个国家有整整一亿人无法获得基本医疗照护。
And so for us, we indexed on this notion that there's a 100,000,000 people in this country that don't have access to primary care.
有时候，所谓缺乏接触，就是他们住在偏远农村，根本找不到任何全科医生。
I mean, sometimes it's access meaning, like, there is no primary care doctor that they can see in their community if they live in a rural community.
还有些人住在人口密集的城市，但所有诊所都不接收新病人。
Others, it's they live in a crowded city, but there's no one taking new patients.
还有一些人根本不愿意去看全科医生。
Some, it's they don't choose to go to primary care.
他们年轻，有事就去急诊或者诊所应付一下。
They're younger, they go to urgent care or whatever.
但事实是，如果你看看长期全科医疗的价值——高质量、医生真正了解你、作为首诊的第一站——有这种关系的人，健康状况明显更好，长期医疗成本还能降低大约三成。
But the the reality is if you look at the value of longitudinal primary care, right, high quality, someone who gets to know you, first stop shop, about, you know, definitely better health outcomes if you have it versus not, about a 30% lower cost over time if you have a a primary care relationship versus not.
你大概能明白为什么。
And you can imagine why.
毕竟，医疗体系这么复杂，要是有个了解你情况的人帮你协调，事情会容易得多。
I mean, just maneuvering the ecosystem of primary of health care is hard if you've got someone who knows you who can help you.
这是个非常重要的问题。
So really important problem.
能接触到这种服务的人太少了。
Not enough people have access to it.
那我们该怎么来构建这样的体系呢？
So how do we how do we build it?
对吧？
Right?
所以我一开始就在想，究竟是哪些工作造成了这26.7个小时的负担？
And so I started with thinking about what are the jobs that created that twenty six point seven hours?
包括预防、健康管理、慢性病管理、照护协调，还有急性护理。
Prevention, wellness, chronic disease management, care coordination, and acute care.
对吧？
Right?
然后我就想，好吧。
So then I was like, okay.
所以我们必须打造AI来协助所有这些环节，因为单靠哪一部分都不足以提升可及性。
So we gotta build AI to help with all of it because no one piece of it will be enough to improve access.
所以，我们想解决的问题——让所有美国人，甚至全世界的人，都能获得高质量且负担得起的初级医疗——本身就是一个足够大的挑战，要达成这个目标，你必须解决很多问题。
And so so part of the answer was just the problem we're trying to solve, which is to get really high quality affordable primary care to all Americans or maybe all people in the world, is a big enough problem that, by definition, you have to solve a lot to do that.
我们董事会里有约翰·多尔这样的投资人，他们愿意耐心等待这样的长期投入。
So and we had a investor in John Doerr and others on our board who have the patience for that.
他们非常认同这个愿景。
They love the vision of that.
市面上有不少AI公司进入医疗领域，只做单一环节的解决方案，虽然能快速获得市场认可，但影响范围非常有限。
There were other AI companies that are coming into health care with point solutions and getting market traction, but with very narrow impact.
这让我们不得不反复问自己：我们该不该去追那种短期机会？
And and it took a lot of discipline to basically say, should we go chase that?
我们该不该因为怕被排除在外，就去选一个狭窄的方案，先冲进市场？
Should we are we gonna get locked out, or should we take a narrow solution and just flood the market?
但我们最终决定不这么做。
And we chose not to do that.
展开剩余字幕（还有 314 条）
我们选择定义一个大问题，花时间打造一个全面的解决方案，宁愿慢一点进入市场，慢一点实现销售，我觉得现在正是这样在发生。
We chose to define a big problem and take our time in building a big solution and and and being slower to the market and being slower to sales, and and I think that's what's happening.
不过我得说，我们很快就意识到，我们的想法在市场和产品上契合得非常好。
Although I will say we very rapidly knew that our idea had amazing market product fit.
对吧？
Right?
我们跟每一个医疗系统谈的时候，他们都说：只要你能搞定这件事，我们立马就要。
I mean, every health system we talked to is like, if you can figure out how to do that, we we're gonna want it.
我们根本招不到足够的全科医生。
Like, we can't hire enough primary care doctors.
根本没法让足够多的人接受全科诊疗。
We can't get enough people into primary care.
因为全科医疗负担太重，病人都流向了专科诊疗。
And because primary care is so overburdened, patients have seeped into specialty care.
现在连专科的就诊通道也被搞乱了。
Now they've mucked up access there.
你知道吗，明年我30%的心脏科预约都是高血压患者。
I can't you know, thirty percent of my cardiology appointments over the next year are for hypertension.
我需要全科医生来管高血压，这样我才能让心脏科医生专心处理缺血性血管病患者，做研究。
Like, I need primary care to manage hypertension so my cardiologist can see people with ischemic vascular disease and do studies.
所以产品就这样慢慢发展成了一个包含很多模块的平台。
So that's kinda how the product evolved into being, yeah, a platform that had a lot of components to it.
并不是说在某个瞬间突然成型的。
It wasn't really, like like, like, it wasn't like a single moment.
更像是，好吧，我们得一步步来。
It was just more of kinda like, okay.
如果真想解决这个大问题，我们就得靠大量人工智能才行。
If we're gonna solve this big problem, we're gonna need a lot of AI to do it.
不对。
No.
这确实很有道理。
That that makes a lot of sense.
我觉得在当前以价值为导向的医疗体系下，关注预防性护理、全面看待患者比以往按服务收费的模式重要得多。医疗系统整体上确实应该更注重预防，确保患者长期健康，而不是只解决眼前问题，这才是正确的方向。
And I think looking at preventative care, looking at sort of the whole patient is a lot more important in this era of value based care compared to the traditional fee for service model where we actually knee well, health systems as a whole need to sort of focus on preventative care and making sure the patient is healthy long term, not just fixing short term solutions, which is the right direction.
我很高兴我们正在朝这个方向走。
I'm glad we're moving in that direction.
是的。
Yeah.
对。
Yeah.
我也这么认为。
I I think so.
但你想想，AI 到底能在哪些方面真正帮上忙呢？
But, I mean, if you think about, like, what so then, you know, so then what are the things that I think AI here can actually do maybe to help?
对吧？
Right?
你说是不是？
You know?
所以，关于我们解决方案的思路，我认为，无论现在还是未来很长一段时间，甚至可能永远，人类都是必不可少的。
So so there's a so the way our our the way we think about our solution, like, I do think that for today and for a long time and maybe forever, we still need humans.
也许有一天，技术会变得极其先进，几代人从小就在这种环境中长大，他们会自然接受技术取代医生。
I I I maybe we'll get to a point one day where technology is so amazing, and we have so many generations who've just grown up around this technology that they accept that as a replacement for for physicians.
但就目前而言，考虑到技术的发展阶段，我们依然认为医生的角色至关重要。
But I think for the time being, and probably because of the state of the art of the technology, we still think that the role of the human physician's important.
因此，我们把这套系统设计成医疗团队中的一名智能代理成员。
So we we have deployed this as a, essentially, an agentic AI member of the care team.
想象一下，如果我现在是一名临床医生，身边有一支住院医师团队，全天候关注每一位患者的状况，每天都在思考：我们能主动做些什么来帮助他们？
Think of a if I was a practicing physician today, if I had a team of residents who just were always on thinking about every one of my patients every day and thinking, what can we do proactively to help every one of those patients every day?
这种事在现实中根本不可能实现。
Of course, not possible in a human world.
对吧？
Right?
但在人工智能的世界里，这恰恰是我们正在努力实现的目标：针对一组患者，整合足够的数据，理解他们的健康趋势、风险和走向，就能理论上找出当天最该做的干预措施，让AI替诊所执行，或者至少先迈出第一步。
But in an AI world, that actually is kinda what we're trying to do, which is to say for for a panel of patients, if we can assemble the right amount of data and understand their trajectory and risk and where they're moving, then we could theoretically identify the best thing we could do from that day and have AI just kind of do it on behalf of the practice or at least have a start.
所以，Tom 系统中有一个面向临床实践的环节，主要是这样：
So there's a from the practice out component of Tom, which is largely, hey.
该做结肠癌筛查了。
You know, it's, you know, time for colon cancer screening.
AI 可以主动联系你，跟你聊聊，告诉你有哪些选择，然后帮你处理任何疑问或特殊情况，最后直接帮你预约。
AI can reach out to you, have a conversation, give you your options, and then ultimately handle any exceptions you may have or concerns you may have and then schedule it for you.
对吧？
Right?
医生根本不需要插手这个流程。
And the physician doesn't really need to touch that workflow.
一切都能自动完成。
That can just happen.
第二天晚上，病人因为高血压急症被送到急诊，这时候第二天该做的事，跟结肠癌筛查完全不一样。
Patient shows up at the emergency department the next night with a hypertensive emergency, the thing that needs to happen the next day is very different than colon cancer screening.
他可能需要回来看全科医生，重新测血压，调整用药方案。
It's it's they may need to get back in to see their primary care doctor to have their blood pressure rechecked and and and get them on on better medications.
所以要做到这一点，你得能实时掌握病人的动态情况。
So to do this, right, you have to be able to assemble a pretty dynamic view of patients.
不能是静态的。
It can't be static.
不能隔太久才去了解病人现在的状况。
You can't go too long between understanding digitally what a patient, you know, looks like.
几乎每天都要更新一次。
Almost every day, you have to do this.
所以第一个难题是：我们能不能把足够多的数据整合成一份病人的综合摘要，让多个人工智能系统能分析出，从预防保健、慢性病管理、照护协调到急性治疗，到底该做哪些事？
So the first challenge was, can we aggregate enough data into a summary of that patient and an understanding of them such that we could run it through a multi LLM engine or brain to identify across prevention and wellness and chronic disease and care coordination and acute care?
有哪些可能的干预措施？
What are all the things we could do?
然后，我们又该优先处理哪一件？
And then how do we prioritize the first thing we wanna do?
这就是‘从临床出发’这个产品视角的核心思路。
So that's that's kind of the view one of the product is from the practice out.
当你用语音或文字与患者互动时，会从中获得大量新的洞察。
When you do that and you interact with patients with voice conversation or or text, you generate a lot of other insights.
这些洞察会作为数据点反馈回模型中。
Those become data points back into the model.
随着时间推移，你会逐渐用这种通过AI与患者互动产生的经验性理解，来丰富临床数据。
And over time, you just start to enrich clinical data with this experiential understanding that is generated through the interaction between AI and a patient.
对吧？
Right?
这个循环不断优化我们对患者的认知，从而挖掘出更多洞察，发现那些真正该去做的事。
So this this loop just kind of evolves the knowledge that we have of patients, and that allows us to unlock more insights and discern maybe things we should do.
比如，如果数据提示某位糖尿病或高血压患者没去续药，AI就可以主动介入，说：嘿。
Like, if we have a signal from the data that someone hasn't refilled their medications and we know they have diabetes or hypertension, well, AI could be triggered to say, hey.
你是不是在拿药上遇到什么困难了？
You know, are you having problem, you know, getting your medicines?
有没有什么副作用？
Do have any side effects?
你有什么担心的地方吗，之类的？
Is there something you're concerned about, etcetera?
这是一部分。
So that's one piece.
接下来，一旦患者参与进来，他们可以把对话带到任何方向。
The next piece is once you have the patient, they could take the conversation wherever they want.
所以，系统能评估症状，或者那些你可能没想到的事，比如——
So the ability to assess symptoms or things that, you know, hey.
我很高兴你今天打电话来谈结肠癌筛查，但我过去三天都发烧卧床不起。
I I'm glad you're calling me about colon cancer screening today, but, you know, I've been in bed sick with fever for the last three days.
你可不希望汤姆只说一句：祝你好运，
You don't want Tom to just say, well, good luck with that.
我们几周后再联系你。
We'll call you back in a couple weeks.
然后你得问问：你的症状具体是怎样的？
Then there's, like, you know, tell me about your symptoms.
所以这里有个后台代理负责处理，整个过程患者是察觉不到的。
And so there's an agent for that, and, you know, it's kind of a handoff to another agent behind the scenes.
患者只觉得这是一次自然的对话，之后这些信息会自动同步到诊疗系统中。
The patient just views it as a pre flowing conversation and then connecting those insights back into the practice.
接下来就是电子病历系统的对接。
So then there's EMR integration.
对吧？
Right?
因为现在关键问题是，这些信息到底该往哪儿放？
Because now we gotta where do we push this information?
是放进诊疗记录里？
Do we push it into an encounter?
还是发到待办消息箱？
Do we push it into an in basket message?
或者在电子健康档案里设置一个提醒？
Do we push it into an alert in the EHR?
我们要不要把信息转给另一个工作人员处理？
Do we push it to a work queue for another human?
你得仔细考虑这些问题。
And so you have to think through that.
最后，也是我最期待的部分，是我们一旦掌握了患者的数据——比如检验结果、健康状况、用药情况，还有他们的感受和身体状态，这些信息可能来自远程监测，比如汤姆从你的血压和血糖数据里发现了什么。
And then, ultimately, the last piece and the piece that I'm most excited about is once we understand someone's data, their labs, their conditions, what drugs they're on, and how they're feeling and how they're feeling and how they're doing, and that could be remote monitoring and, you know, what what has Tom learned about your blood pressure, your blood sugars.
在临床背景下，把这类证据呈现给医生，能极大提升全科医疗的效率。
Now presenting clinical evidence in that context of the clinical data to a clinician will massively upscale the performance of primary care.
现在我可以直接说：针对这位患有糖尿病、高血压和慢性肾病、正在服用这些药物、血压和血糖值如此的患者，下一步你应该考虑的是，加用这种药、这个剂量，同时停掉另一种。
So now I can say for your patient with diabetes and hypertension and chronic kidney disease on these drugs, with these blood pressures and these blood sugars, the next thing that you should think about doing for them is to add this drug at this dose and stop that one.
这是有充分依据的。
And here's the reasons for it.
另外，我们和沃尔特·克鲁维尔以及UpToDate有合作，他们的临床内容经过向量化和精选，我们可以调用API，在患者的具体情境下，把相关临床信息推送到医生决策的那一刻。
By the way, And we have a partnership with Walter's Kluwer and UpToDate, and they're vectorized curated content where we can ping, you know, an API and bring that clinical content in the context of a patient and put in front of a clinician at the point where they need to make a decision.
天哪。
Woah.
想想看这有多大的力量。
Think about the power of that.
对吧？
Right?
现在那些不想处理心衰的医生，因为太复杂了，都会转给心内科。
Now clinicians who don't wanna manage heart failure because it's complicated, they send it to cardiology.
但他们其实可以开始自己管理了。
They can start managing it.
也许原本要转给风湿科、内分泌科或其他专科的病例，以后都能在全科门诊处理了。
Maybe, you know, things that get sent to a rheumatologist or an endocrinologist or others start being managed in primary care.
这就是Tom背后的理念。
And so that is the the idea behind Tom.
我们正在打造的就是这个。
That's what we're building.
我觉得，等它完全成熟的时候，能真正打破‘医疗只能在诊室里进行’的旧观念。
And, and I think at its when it comes to full fruition, it will allow us to literally move away from this notion where care is delivered inside the confines of office visit.
我觉得，未来在基层医疗中，亲自去诊所看医生会变成少数例外。
I think going to the office, going to the doctor in the future in primary care will be an exception.
我们总想着能随时和医生在一起，而这种陪伴应该是全天候的。
We like to think about being with the doctor, And being with the doctor is gonna be twenty four seven.
通过智能设备和人工智能作为医患之间的桥梁，这种连接就能实现。
It's gonna be through a connected device with AI being the interface between the practice and the patient.
这对医疗资源的分配意义重大——这样一来，医生的时间就能集中用在真正需要帮助的少数人身上，而不是像现在这样，人手和就诊人群严重错配，大部分来看病的其实是没什么大问题的健康人。
And that has massive implications for access, because then the human time can be focused on the small subset of people who are really needing it versus today, there's this giant mismatch between the human time and who's in the office, and it's kind of largely for the worried well or for people who are doing quite well.
但现在医生要维持和患者的联系，只能靠让他们定期来门诊，而我们用人工智能，完全可以打破这种模式。
But just the doctor's only mechanism to stay connected to them is to have them come in periodically and see them, and I think we break that totally apart with our use of AI.
是的。
Yeah.
这真的太令人着迷了。
This is this is absolutely fascinating.
我很感谢你举了这些具体的例子，说明到底用了哪些工具。
And I I'm I'm really thankful that you're giving examples of the agents of, you know, which tools are being used.
我觉得特别难得的是，你作为一名医生，对这些技术在临床中的实际影响、患者的具体流程，都有非常清晰的理解。
I think it's also special that you're a physician and you have, like, very clean understanding of the actual clinical implications of this, of clear workflows of what patients would look like.
尤其是因为汤姆这个系统涉及的方面实在太多了。
I think especially because Tom is, like, so there's so many aspects of it.
如果你能给我们举个例子就好了，比如当一个患者来就诊时，他是怎么和汤姆互动的？
It would be super helpful if you could walk us through maybe an example of, right, if you see a patient that's coming in, how how does a patient interact with Tom?
我问这个问题，是因为很多人倾向于用单一功能的工具，难点就在于怎么把诊所里各种零散的工具连起来，确保所有系统都互通——比如患者用的健康工具、医生端的系统、医保平台等等，而汤姆在这方面做得很好。
And the reason I I asked that too is I think there's the reason a lot of people are towards, point solutions is because part of the difficulty is wiring together all the point solutions that are, that that a clinic might use and making sure everything's connected, whether that's like a patient facing health tool to, like, connecting to the physician side that connects to the payer and, like, all this all this stuff that it seems like Tom is doing well.
你刚才提到了很多应用场景，但如果能具体讲讲，当患者在诊所时，他是怎么和汤姆配合，同时又和诊所本身协作的？
So, you know, you you highlighted all the examples of things that are doing, but if you could talk us through, like, if a patient is at a clinic, how how is how are they working with Tom in addition to the the clinic that they're working at?
是的。
Yeah.
首先，得弄清楚怎么向患者介绍汤姆。
So the well, first of all, it starts with how do you, like, introduce Tom?
也就是说，这个基础该怎么建立？
Like like, what's the basis?
对吧？
Right?
当我们进入一家诊所或医疗系统时，第一件事就是他们会给Tom起个名字，而且不一定是Tom。
So when we go into a clinic or a health system, first thing is they name Tom whatever they're gonna name Tom, and it's not always Tom.
我们有个合作伙伴正在部署这个系统。
I mean, we have a a partner who's implementing it.
他们给它取名叫Mia。
They they named it Mia.
他们希望是个女性的声音、女性的形象。
They wanted a female, female voice, female persona.
所以在那个场景里，他们专门做了一个营销活动，向患者介绍Mia是医疗团队的一员，核心就是让所有患者把Mia添加为手机联系人。
So in that world, it's there's a marketing campaign literally to to introduce the patients in the practice to Mia as a member of the care team, and that means largely getting all of those patients to download me as contact on their phone.
我们很早就决定，不会去开发一个四十万功能的健康App，因为每个医疗系统都有自己现成的数字入口。
So we use we we decided early on that we're not gonna build the 400,000 health app because both health systems have their own digital front door.
他们已经有自己的数字化工具了。
They have their own digital assets.
他们其实并不想要那样。
They don't really want that.
我们只想让沟通变得像平时一样简单。
And we wanna just make it easy for people to communicate how they normally do.
现在几乎人人都有智能手机。
So almost everyone has a, you know, smartphone now.
所以我们让患者把Mia保存为手机联系人，这样他们就能知道，来自医生的来电或短信都是Mia发来的。
And so so we get we get the patients to download Mia as a contact on their phone so that they know when they get an inbound phone call or a text that's coming from their doctor, it's Mia.
你想想，这有多厉害。
You know, think about the power of this.
对吧？
Right?
我们的一个合作伙伴谷歌，对我们这个AI应用场景特别感兴趣，因为我们是在和患者建立长期的AI关系，而不是一次性、零散的互动。
And one of our partners, Google, was super interested in our use case for AI because we're building longitudinal AI relationships with patients, not transactional one offs.
他们认为，我们正走在前沿，探索人们如何与AI建立信任——因为同一个AI形象会持续陪伴患者，而不是换不同的人来安排预约、提醒就诊。
And so they believe, like, we're on the cutting edge of learning how do people build trusting relationships with AI because it's gonna be the same persona longitudinally over time, not a different agent who's scheduling you and a different one who's reminding you of your appointment.
整个过程都是Mia在跟进。
It's it's Mia throughout.
所以第一点是，我们必须让患者明白，Mia是诊所的代表，基于医生授权使用他们的数据，他们完全可以信任AI。
So that's number one is, like like, we have to get people to understand Mia is an agent of the practice, is working from their data under the doctor's authority, and and that they can trust AI.
这不是支离破碎的体验，这就是第一点。
It's not a disconnected experience, and so that's that's one.
当患者与AI互动时，几乎都是通过设备进行的。
So when patients are interacting with AI, it's almost always through a device.
他们在诊所里看到的场景大概是这样的。
The way it shows up for them in an office looks something like this.
Mia或者Tom会给患者打电话，已经和他聊了大概几周了。
Mia or Tom calls the patient and, you know, has been talking to him a cup you know, for for maybe a couple weeks.
他们 upcoming 有个预约，于是就说：
They have an appointment upcoming, and they're like, listen.
我想提醒你一下。
You know, I want to remind you.
你明天要去卡穆什医生的诊所就诊。
You have an appointment tomorrow at doctor Carmouche's office.
明天见卡穆什医生时，有什么特别想聊的吗？
Is there anything you wanna talk about doctor Carmouche tomorrow?
有没有什么特别想确保医生明天帮你处理的事情？
Like, anything you wanna make sure doctor Carmouche addresses tomorrow?
有的。
Yes.
之前两次我拿这个药的时候，药剂师都提醒我，它可能和我其他药有相互作用，我本来想问问医生，可每次都忘了。
You know, the last two times I've filled this prescription, my pharmacist warned me that, you know, it could be interacting with a couple of my other drugs, and my I may wanna talk and I always forget to, you know, I always forget to ask the doctor that.
现在我是佩默里斯医生，正要走进病人的诊室。
Now I'm the doctor, doctor Permeris, getting ready to walk into the patient's room.
我会打开电子健康档案，看看汤姆的总结报告。
I'm gonna look on the EHR, and I'm gonna see a Tom summary.
报告会汇总我和汤姆之间所有的互动内容。
And it's gonna summarize all the interactions that I had that Tom's had with me, the patient.
我走进房间时，不仅明白这一点，还会清楚患者这次见面真正想谈什么。
And I'm gonna walk into the room not only understanding that, I'm gonna also understand the patient's agenda for the for the meeting.
所以我知道，这位患者其实希望我确认这两种药会不会有相互作用。
And so I'm gonna know, oh, this patient actually wants me to address whether these two drugs have a have an interaction.
现在我心里有数了。
And now I I know.
不管患者是忘了提，还是紧张不安，我一进门就会说：‘很高兴见到你，阿尔文。’
So whether the patient forgets or gets anxious or whatever, I'm gonna walk in, and I'm gonna start my conversation with, it's good to see you, you know, Arvin.
我知道你刚和汤姆聊过。
I understand you've been chatting with Tom.
我知道你对用药有些顾虑。
I understand you got some concerns about your medication.
我们直接聊这个，而不是花前十分钟慢慢套信息。
Let's talk about it Versus spending the first ten minutes getting all of that information.
我能直接切入对你重要的事，让这次就诊更贴合你的需求。
I can just get into the things that are gonna be important to you and make the fit the visit meet your needs.
这其实很好地说明了，如何从具备主动性的AI洞察出发。
And that that's just a great example of, like, how you would go from agentic AI insight.
首先得赢得患者的信任，让AI能与患者互动，提炼出有价值的洞察，把这些洞察融入工作流程，呈现在医生面前，这样医生一进门就能直接切入重点。
So first of all, gotta build trust with a patient, have AI be able to engage that patient to create insights, make those insights meaningful to change workflow, put them in the doctor's purview in in his workflow, and then have the doctor know that when he walks in, he can just jump into the encounter.
当然，还有别的思路可以考虑。
I I mean, there's other ways to think about it too.
我们理论上有一天可以提前把就诊信息填入主诉部分。
We could theoretically one day prepopulate the visit in the HPI.
我的意思是，完全可以让人提前做好准备再进门。
I mean, we could actually have someone walk in.
诊室里的环境录音已经能做得不错了，但像汤姆这样的系统，其实可以在患者进诊室之前就完成这些工作。
An ambient listen does a pretty good job of doing that in the exam room, but, like, Tom could probably do it before the exam room.
对吧？
Right?
所以我们得决定，是不是真有必要开发这个功能。
And so we'll just have to decide if that's a feature that that that we need to build or not.
但最主要的是，医生会显得更了解你的状况，能更高效地利用就诊时间，更好地满足你的需求——这正是我们所期待的。
But but mainly, the the way this comes up is your doctor appears way more aware of what's going on with you and can use the minutes of that visit much more effectively and meet your needs is kinda what the hope is.
没错。
Right.
随着普通门诊的平均时长越来越短，而患者想讨论的问题却越来越多，这一点变得越来越重要。
And this is getting more and more important as sort of the average length of visit for primary care is going down and the number of problems patients have that they wanna discuss is going up.
对。
So Right.
这确实是一个非常重要的问题，我觉得可以分成两个部分来看。
This is definitely a very important I think sort of two part question here.
第一部分是，你正在合作的那些实际使用Tom或Mia的医生，以及和Tom或Mia互动的患者，他们对这个系统反应如何？
One is how are the physicians that you're working with that are actually using Tom or Mia in certain cases and the patients that Tom or Mia are interacting with, how are they responding to this system?
他们的反馈是什么？
What has their feedback been?
第二部分，从长远来看，比如说十年后，你提到过，全科医疗领域存在二十亿小时的人力缺口。
And looking, I guess, long term as part two, let's say, ten years down the road, you've said there's a two billion hour primary care shortage.
那假设十年后，你觉得Tom、Mia这类系统会对整个初级医疗体系产生什么影响？
So let's say ten years down the road, how do you envision Tom or Mia or whatever these systems call it impacting the overall primary care system?
你觉得全科医生的数量会减少吗？
Do you envision less PCPs?
患者去看全科医生的次数会变少吗？
Do you envision less visits by a patient to the PCP?
如果Tom真能如你所预期的那样运作，你觉得整个体系会怎么变化？
How do you envision the system changing if Tom does what you expect it to?
当然。
Sure.
我来给你做个合理的推测吧。
I mean, I'll give you my best guess.
对吧？
Right?
首先，我们已经在多家诊所部署了Tom的某些功能，也收集到了一些反馈。
I I think first thing is I'd say we are just so we've deployed pieces of Tom in in a variety of different clinics, and we have feedback on that.
目前Tom的整体功能，虽然可能永远都不会完全定型，但有一部分已经投入使用，这是第一次以整合方案的形式在医疗系统中落地。
The totality of Tom or at least as Tom exist in his totality today is not it'll never be done probably, but it's it's you know, there's a chunk of Tom functionality available today is being deployed in health systems at as an aggregate offering for the first time now.
我不想夸大其词，真的，这行里头太多夸张的说法了。
So I I don't wanna I'm not I don't wanna you know, there's a lot of hyperbole in the field.
很多人动不动就做出各种不切实际的承诺。
There's a lot of people who make all sorts of claims.
我得保持谨慎，不能乱说。
I wanna be really cautious.
我们测试过的那些Tom功能，首先一点是，患者对这种对话式AI接受度特别高，尤其是那些原生语音功能——没有延迟，对话自然流畅，还能感知情绪，效果真的很惊艳。
The the the components of Tom that we've tested, I think first thing is that patients really take to conversational AI really well, especially with some of the native audio capabilities where, you know, you don't have the latency and it's free flowing conversations and there's sentiment, and it's amazing.
各个年龄段的患者都一样。
And patients of all ages.
我们甚至有92岁的老人，能跟Tom进行非常深入的对话，这确实让人意外。
I mean, we've had patients as old as 92 years old, have a very robust conversation with Tom and, you know, obviously, folks.
所以大家对Tom特别开放，愿意跟它说心里话，讲笑话，听这些通话真的特别有意思。
And so people are really open and willing to talk to Tom and share their deepest fears and tell jokes, and it's really amazing to listen to the calls.
所以我非常欣慰，之前其实我很担心。
And so I'm very gratified, and I was really worried.
我本来以为在医疗领域，人们可能会更保守，但我觉得这个领域，尤其是人工智能，正在快速扩展，你们也一直在关注。
I I I thought maybe in health care, like, people would just be more reticent, but I think this field is just expanding, the the AI field, and you guys follow it.
现在人工智能已经渗透到生活的方方面面，它出现在医疗领域，简直顺理成章，这真是好事。
I mean, we're starting to see it in all walks of life such that it's showing up in health care almost feels like inevitable and natural, so that's good.
医生们确实更谨慎一些，但他们的疑虑并不在技术本身。
I think clinicians have a little bit more skepticism, and I but it's not around the technology per se.
他们担心的是，这项技术会不会让他们能照顾更多病人，却反而让生活更糟。
It's about the notion that they could figure out somehow that this technology can allow them to take care of more patients and not make their life worse.
换句话说，Tom 不停地跟所有患者互动，这种感觉就像是会给我塞进太多信息，多到我根本处理不过来。
So it's more like, man, this notion of Tom constantly interacting with all my patients has the feeling that it could just inundate me with more insights than I can can handle.
对吧？
Right?
这就像是我每天都要给每个患者发消息问：今天有什么我能帮你的吗？
It'd be be like me reaching out to every one of my patients every day saying, is there anything I do for you today?
他们当中有一半人都说好。
And they all and half of them say yes.
我就在想，我什么时候才能开始呢？
I'm like, when am I gonna do it?
他们说，别一下子推给我们，慢慢带我们适应。
And so what they've said is like like, walk us into it.
别逼着我们接受。
Don't run us into it.
有没有办法控制一下功能开启的量？
Like like, you know, is there a way to to gate how much functionality we turn on?
产品至少得有点配置空间。
Make the product configurable to some degree.
总不能一下把所有功能全打开吧。
Like, can't be like we just turn it all on.
所以他们的反馈是，我们确实理解这个愿景，也觉得它会非常棒。
And and so what we've got so so the feedback from them is, like, we we clearly understand the vision, and we think it's gonna be amazing.
但我们觉得，会有一段时间，我们需要共同弄清楚，哪些关键信息应该呈现在医生眼前。
But we do think there's gonna be a period of time where we collectively need to learn what exact insights need to come to the clinician's, like, line of sight.
比如，我们能不能把这一点做好？
Like, is it can we get that right?
能不能调整好，别让太多无用的信息堆到他们面前？
Can we calibrate it where we're not inundating it with meaningless information?
但另一方面，我们也不希望标准定得太高，把一些他们其实想了解的重要信息给漏掉了。
On the other hand, we don't wanna set the the the bar so high that we're leaving out a lot of information that they might have wanted to know.
所以他们的建议是，跟我们紧密合作，认真听我们的反馈，让我们帮你把这种平衡把握好。
And so think what they say is, like, work with us and listen to us and let us help you get that calibration right.
这就是他们的反馈，而且这建议非常到位。
That's the feedback, and that's got and and it's very smart feedback.
我们已经有不少早期采用的医生，愿意迈出这一步，主动尝试并给我们反馈。
And we've got a lot of early adopting physicians who want to who who are willing to take that leap and experiment and give us that feedback.
所以我觉得，无论从哪个角度看，我都持谨慎乐观的态度。
And so I'd say I'm cautiously optimistic on on it on both sides of the equation.
顺便说一下，安全问题和幻觉并没有成为主要困扰。
By the way, what's not been as big of an issue is concerns about safety, hallucination.
我本来以为这会是个大问题。
I thought that was gonna be a big deal.
我觉得部分原因在于，我们这个系统有点像一个庞大的上下文模型。
I think partly it's because the way we we you know, it's almost like this giant rag model.
对吧？
Right?
因为我们是以患者的数据为起点，而这个智能代理是专门设计的。
Because we've got the patient's data as the starting point, and the agentic AI is is purpose built.
它不像那些开放式的大型语言模型，随便把患者丢进去就让它们自由对话。
So it's not like open ended LLMs that we just put in patient in front of patients and say, have a conversation.
这些代理的任务非常明确，而且它们的起点不仅受安全限制，还被患者的具体数据所约束。
Like, the agents have very discrete tasks that they focus on, and their starting point is constrained by not just guardrails, but the patient's data.
所以这方面根本不是问题，两边的反馈都非常积极。
So I think that hasn't been an issue, and so really positive feedback, I think, on on both sides.
至于基层医疗的未来，我觉得大概会是这样，我还写过一篇白皮书讨论这个。
As for the future of primary care, I think it looks something like this, and I wrote a a white paper on it.
我预测的是五年后的样子。
It looked five years out.
我觉得五年内，一名全科医生就能照顾五千名患者，这完全做得到。
I think that that the reality is that it's it's gonna be possible within five years for a primary care physician to take care of 5,000 patients.
以前美国平均每个医生管两千五百个病人。
So the average panel in America used to be said it was 2,500.
天啊，现在这个数字早就缩水了。
Man, I can tell you it's shrunk.
我觉得是因为我们给全科医生加了太多任务，他们根本忙不过来——每名患者都要做编码、写病历、满足质控指标、回应需求，样样都得顾到。
I think because we've put so much work on primary care doctors specifically, they they can't take care of that many patients because it's just too much to do for every patient by the time you put coding and documentation and quality measures and meet the patient's needs and, and all.
事情实在太多了。
It's just a big agenda.
所以我觉得，未来的模式大概会是这样。
So I think in the future, it's gonna look something like this.
未来的全科医生每天上班，大部分时间都会用来处理由AI支持的非实时诊疗服务。
You're gonna have a primary care physician who's gonna start his day, and most of his day is gonna be spent with asynchronous AI enabled care.
在五千名患者中，会有数百个这样的小情况，比如某人的血压还是略高。
Hundreds of encounters across your 5,000 patients, and they're gonna be little simple things like this person's blood pressure is still slightly high.
系统会建议加一种药，医生点一下确认，药物指令就会自动填进电子病历，事情就解决了。
Would recommend add this drug, and a physician's gonna hit click accept, and it's gonna populate the orders in the EHR, and it's just gonna take take care of it.
他们会快速处理上百个类似的情况，大多都是慢性病管理，有充分的临床证据、可靠的生理数据、已经完成的化验，还有分析和建议。
And they're gonna go through hundreds of these things, which are largely gonna be chronic disease focused things where there's lots of evidence, there's good biometrics, labs have been ordered, and there's interpretation, and there's recommendations.
只要还需要医生开药、做诊断，这些信息就必须交给医生处理。
And as long as we need clinicians to order meds and make diagnosis, we'll have to feed that to a clinician.
医生只需要快速点选，处理大量高频率的简单病例。
It'll just be clicking through a bunch of high velocity encounters.
当然，也会有一部分情况必须进行实时沟通，可能是线上问诊，也可能是面诊。
And there'll be a subset of encounters that are either that that require synchronous care, and that could be virtual or it could be in your office.
每天大概有六到八个病人需要花四十五分钟到一小时，因为他们的病情比较复杂。
And so maybe six to eight patients a day will take forty five minutes to an hour each, and they have a lot of complexity to them.
你得亲自盯着他们，或者让他们在你面前，才能好好照护。
And you need to either have eyes on them or you need to have them in your in your presence to take care of them.
我觉得，这就是未来全科医疗的样子。
And that's what the world, I think, of primary care looks like.
在这种模式下，全科医生要承担的任务比现在多得多。
And I think primary care is taking on way more than it does today in that model.
AI已经让全科医生能够管理心力衰竭、类风湿性关节炎、溃疡性结肠炎，还有许多原本被推给专科医生的疾病。
I think AI has enabled primary care to take care of heart failure, rheumatoid arthritis, and ulcerative colitis, and lots of conditions that today kind of get pushed into specialty care.
专科医疗会变得更偏向于操作和手术。
I think specialty care becomes much more procedural.
那些需要复杂判断的领域，将成为新的全科医疗核心。
I think the cognitive kind of disciplines become the new primary care.
那些基础性的工作，几乎都能由AI完全接手了。
I think the low level stuff almost becomes AI enabled fully.
比如饮食建议、运动指导这些事，未来都会变成作为医疗设备的软件。
Like, nutritional advice, exercise advice, all that stuff, I think, becomes software as a medical device.
我觉得人们现在就像在使用 ChatGPT Health 这类工具，自己就能搞定日常健康问题。
And I think people are self serving just like ChatGPT Health or others are trying to offer them the ability to do day.
我觉得这层就是最前端的体验。
I think that's the front layer.
这大概就是我想象中它该有的样子。
And I think that's what I think that's what it looks like.
要实现这一切，最关键的是技术上的突破。
The only big thing that has to happen, I think that's there's the technology unlock.
Tom 就是个很好的例子，说明平台该怎么设计才能推动这个转变。
I think Tom would be an example of what the platform might look like to make that happen.
支付模式也得跟着改变。
The payment model's gotta evolve.
因为在这种模式下，大部分医疗服务没法再靠现有的诊疗编码和按服务付费来报销。
Because in this case, most of the care is not gonna be compensated through existing E and M codes and existing fee for service.
所以我们需要为初级医疗引入订阅制，才能充分发挥这项技术的潜力。
So we will need subscription type models to primary care to unleash the full power of that technology.
所以我认为，最大的瓶颈根本不是技术。
And so I I think the long pole in the tent is not tech.
而是支付方式。
It's gonna be payment.
我希望支付方能开始想象一下：大规模应用AI驱动的初级医疗会是什么样子。
And I I hope that payers would start to imagine that, man, AI enabled primary care at scale.
如果需要新的支付模式，那就改吧，毕竟高质量的初级医疗能带来巨大的长期回报，这笔投资太值得了。
If it requires a new payment model, let's do it because that's such an amazing investment given the leverage that high quality primary care has downstream.
没错。
Right.
你提到一个关键点，技术其实已经在快速发展了。
Now you bring up a really good point in that the tech is already being developed.
现在有太多解决方案，太多公司在开发很棒的工具，提升患者护理，优化医生的工作流程等等。
There's so many solutions, so many companies building amazing tools that are improving patient care, improving physician workflows, etcetera.
但最终，支付模式、保险公司，还有监管方面也得跟上，适应这个全新的局面。
But eventually, the payment model, the insurance companies, the sort of regulatory aspect also will have to catch up and just catch up and be able to play in this new world.
你们这一代人真的很幸运，能亲眼见证技术这么快地变化。
Well, so so so so, you know, you guys are great you you guys are young enough to see how rapidly this technology is evolving.
对吧？
Right?
简直不可思议。
It's crazy.
历史上从来没出现过这样的情况。
Like, we've never seen anything like it in the world.
想想看，监管体系和支付机制却还停留在原地，两者之间差距太大了——一边是医疗体系里那些陈旧的结构，另一边却是汹涌而来的技术浪潮。
Think about how slow the regulatory environment and the payment it's just there's this giant mismatch between the arcane structures that are important part of health care and then this technology wave.
美国卫生与公共服务部和医保中心刚任命了一个由18人组成的顾问小组。
You know, the United States Department of Health and Human Services and CMS just appointed a 18 member advisory panel.
我很幸运，被选中参与其中。
I'm very blessed to have been picked to to serve.
他们之所以看重我，是因为我长期接触医疗支付体系和医疗系统，清楚它们必须怎么调整，才能跟上即将在医疗科技领域爆发的创新。
And one of the reasons they're interested is because of this breadth of of experience where I've got, like, this knowledge of how payment models and systems of health care need to evolve to facilitate the innovation that we're gonna see in health IT.
所以我觉得，这就是原因，你看。
And so, like, I think that that's why I mean, look.
我不是学者。
I'm not an academic.
我也不是医疗界最有名的人，但他们特别看重一点：我们必须解决怎么付费的问题，才能真正发挥这项技术的价值。
I'm not the biggest name in health care, but they were very keen on we know that we're going to have to solve how do we pay for this in order to get the full value of that.
所以，在我担任这个委员会成员的这两年里，这将是我的核心关注点。
And so, you know, for the two years that I'll be part of that committee at least, this is my area of focus.
我觉得，哪怕我别的什么都不做，至少也要推动这些系统更快跟上步伐——作为全国最大的支付方，CMS完全可以带头，利用你们每天都在接触和研究的这项技术。
I I I think if I do nothing else, it's like, do we bring these systems up to speed more rapidly and CMS can be a leader there as the biggest payer in the in in the country to take advantage of this technology that you guys are seeing every day and studying.
我觉得，这些改变必须比平时更快地发生才行。
I I think, you know, I think we need we need those those changes to happen faster than they typically would on their own.
是的。
Yeah.
太棒了。
That's amazing.
你刚才说的，正好是我心里想的。
I think you just read my mind too.
我正打算问你关于你被任命加入那个顾问委员会的事，谢谢你主动提到这个。
I was about just about to ask you about your appointment to that committee, the advisory committee, so thank you for bringing that up.
是的。
Yeah.
对。
Yeah.
当然。
Sure.
说实话，这真让我挺意外的。
I mean, that was a big surprise.
我的理解是，圈子里有很多非常优秀的人才。
I mean, you know, there was a lot of really talented in my understanding, a lot of really talented people.
能提供帮助的人太多了。
There's so many people who can offer.
我再从头说起吧。
And, again, I'll go back to the beginning.
对吧？
Right?
你要么是问题的一部分，要么是解决办法的一部分。
You can either be part of the problem or part of the solution.
你要想成为解决办法的一部分，就得真正参与进来，投入时间，贡献自己的智慧。
The only way you can be part of the solution is to get in the game and to and to offer your time and your and your and your, hopefully, wisdom.
对我来说，被邀请参与这件事，是职业生涯中的一个重大里程碑，我非常重视。
And so for me, it's a it's a it's a career defining milestone for me to be asked to do that, and I take it very seriously.
我会尽我所能贡献力量。
I and I look forward to contributing best I can.
我很期待看到你未来几年能带来什么成果。
I look forward to seeing what you're able to contribute and what happens in the next couple of years.
别有压力。
No pressure.
我知道时间快到了。
I know we're running out of time.
在结束之前，我最后问一个问题。
Just one last question as we wrap up.
我觉得你是一位真正独特的医生。
I think you are a truly unique physician.
你在临床、保险行业、科技领域，甚至像沃尔玛这样的消费健康领域，都承担了这么多不同的角色。
You've taken on so many different roles working in the clinical space, the insurance industry, working in the tech space, the sort of consumer health space with Walmart a little bit.
我想知道，你有什么建议想分享给像阿尔文这样的医学生，或者像我这样职业早期的医生，还有我们的听众——那些刚起步、希望有所作为、想如你所说成为解决方案的一部分，而不是问题一部分的人？
I'm curious, what advice would you want to share to, whether it's medical students like Arvind or earlier career physicians like me or our listeners who are earlier in their careers and want to make an impact, want to, as you said, be a part of the solution, not the problem.
是的。
Yeah.
我能说的实在太多了。
I mean, there's so many things I could say.
但归根结底，就这一点。
I think it just boils down to this.
第一，眼光要放远。
Number one, think big.
在医疗行业，我们有太多只顾眼前的人了。
Man, in health care, we have so many incrementalists.
太多人只能看到自己鼻子前面那点事，但真正要带来大改变，就得有大构想和大抱负。
We have so many people who can only see right in front of them, and, you know, we need big ideas and big ambitions for big change.
所以第一步，得从大想法开始。
So first of all, it starts with big ideas.
我很欣赏你这么年轻，就能从新一代领导者、新医生，或者想入行并有所作为的人这个角度来思考这个问题。
So I love the fact that young you you frame this around maybe new leaders or peep new physicians or people who wanna get into the game and make a difference.
首先，要敢想敢梦，而现在的技术让我们终于有能力去实现那些过去根本不可能完成的大构想。
So first of all, think big, dream big, and technology allows us to now execute big ideas in a way we never had.
第二点，可能更重要的是——要有勇气。
The second, and maybe more important, is just be courageous.
真正的领导力，归根结底是敢于质疑现状，而当一名医生领导者，需要极大的魄力。
Leadership ultimately is having the courage to question the status quo, And and it takes a lot of fortitude to be a physician leader.
很多医生并不欣赏那些自以为高人一等的同行。
Lots of physicians don't appreciate other physicians who, quote, elevate themselves above others.
这从来不是领导力的本意，医生领导者也不是想这样。
That's never the intent of leadership, and physician leaders aren't trying to do that.
我们只是想用不同的方式解决问题，并且愿意付诸行动。
We're just trying to solve problems a different way, and we're willing to do it.
但你的同行未必能理解，所以你常常会被背后捅刀子，比如有人会说：‘你看大卫，他到底以为自己是谁，居然想出来解决这个问题？’
But but but your peers don't always understand that, and so you get you get stabbed in the back a lot by people who like, you know, look at David, you know, try you know, what who does he think he is trying to go out and solve this issue?
你说是吧？
You know?
所以这确实需要勇气。
So it takes some courage.
对我而言，离开临床岗位、不确定自己能不能胜任管理角色，这本身就需要很大的勇气。
You know, for me, took courage of leaving practice and and not knowing if I could deal with a management role.
所以我想说，要有远大想法，要有勇气，敢于冒险，坚持你认为正确的事。
And so I I'd say just, you know, have big ideas, be courageous, be willing to take a chance, and, you know, stand up for for what you think is right.
对于有才华的人来说，其他事情自然就水到渠成了。
And I I I think the rest kind of flows from there for for talented people.
是的。
Yeah.
这些建议太棒了。
That's incredible advice.
非常感谢您，卡穆什医生。
Thank you so much, doctor Carmouche.
今天特别感谢您抽出时间，分享您的经历。
I really appreciate your time today and and sharing your experiences.
听您讲汤姆的故事，还有像普拉纳夫说的，您之前那段不寻常的经历，以及这些如何最终汇聚成您即将在这里打造的这个平台，真的特别有意思。
It's been really interesting learning about Tom and, like like, Pranav said, your your interesting background before that too and how that's all culminated in in in what you're gonna be able to build here with this, platform.
所以，再次期待Lumeris未来的发展，也再次感谢您的时间。
So, again, really excited, to see where Lumeris goes, and and thank you again for your time.
谢谢你们。
I appreciate you guys.
这场对话我很享受，你提的问题也很有深度。
I really enjoyed the conversation and the thoughtful questions.
等汤姆在全公司更全面地投入使用后，我可能会再回来，跟你们分享一下实际运行的效果。
So, maybe I'll come back at some point when we have Tom deployed more fully across and and let you know kinda how it's how it's working out.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。