或将颠覆未来医学的创新科技来了！

发布于 2023-08-15 · 时长 54:40

字幕摘录

时间	英文	中文
0:03	Good evening, everyone, especially to our distinguished speaker, Professor Coveney.	各位,特别是尊敬的演讲者科维尼教授
0:10	Welcome all of you to join us on this Zoom talk on Digital Twins, the virtual future	欢迎大家加入我们数字双胞胎的Zoom演讲虚拟未来
0:15	of medicine, co-hosted by Princeton University Press, China Office, and also Yale Center	由普林斯顿大学出版社、中国办事处和耶鲁中心共同主办
0:21	Beijing.	北京报.
0:22	My name is Joyce Wang, the deputy director of Yale Center Beijing.	我是耶鲁北京中心副主任王乔伊丝
0:26	First, let me also quickly introduce our speaker today, Peter Coveney.	首先,让我也迅速介绍今天的发言者彼得·科维尼。
0:31	He's an adjunct professor at the Yale School of Medicine, as well as director of the Center	他是耶鲁医学院的副教授兼中心主任
0:37	for Computational Science at the University College London.	伦敦大学学院计算科学。
0:43	He's also a professor at the Informatics Institute at the University of Amsterdam.	他也是阿姆斯特丹大学信息学研究所的教授.
0:50	Today Professor Coveney will share his research and insights on his latest book called	今天,科维尼教授将分享他关于他最新著作的研究成果和见解。

展开字幕全文（733 条）

序号	英文	中文
1	Good evening, everyone, especially to our distinguished speaker, Professor Coveney.	各位,特别是尊敬的演讲者科维尼教授
2	Welcome all of you to join us on this Zoom talk on Digital Twins, the virtual future	欢迎大家加入我们数字双胞胎的Zoom演讲虚拟未来
3	of medicine, co-hosted by Princeton University Press, China Office, and also Yale Center	由普林斯顿大学出版社、中国办事处和耶鲁中心共同主办
4	Beijing.	北京报.
5	My name is Joyce Wang, the deputy director of Yale Center Beijing.	我是耶鲁北京中心副主任王乔伊丝
6	First, let me also quickly introduce our speaker today, Peter Coveney.	首先,让我也迅速介绍今天的发言者彼得·科维尼。
7	He's an adjunct professor at the Yale School of Medicine, as well as director of the Center	他是耶鲁医学院的副教授兼中心主任
8	for Computational Science at the University College London.	伦敦大学学院计算科学。
9	He's also a professor at the Informatics Institute at the University of Amsterdam.	他也是阿姆斯特丹大学信息学研究所的教授.
10	Today Professor Coveney will share his research and insights on his latest book called	今天,科维尼教授将分享他关于他最新著作的研究成果和见解。
11	Virtual You, how building your digital twin will revolutionize medicine and change	虚拟你,如何建设你的数字双胞胎会革命医学和改变
12	ネ
13	We are also very grateful to co-host this talk with Princeton University Press, China	我们还非常感谢与中国普林斯顿大学出版社共同主持这次谈话。
14	Office.	办公室
15	They will kindly provide three copies of Professor's book, which we will give out to the	他们将提供三本教授的书,我们将将其交给联合国。
16	first three audiences who post questions during the Q&A session later.	前三位观众在后期的QQA会中发问.
17	Now I'd like to invite Lingxi Li, the director from the Princeton University Press, China	现在,我想邀请中国普林斯顿大学出版社的院长李玲茜
18	Office, and to give a very quick introduction of the entity and also Professor Coveney's	以及科维尼教授的
19	book.	书。
20	Lingxi, please.	灵芝请.
21	Thank you very much, Joyce, and thanks everyone for joining this event on the Sunday evening,	非常感谢,乔伊斯,谢谢大家参加这个星期天晚上的活动,
22	and thanks for Professor Coveney to offer these stimulating talks later.	感谢科维尼教授稍后提供这些刺激性演讲。
23	So I will be very brief in introducing a little bit about Princeton University Press in China	因此,我将非常简短地介绍一下中国普林斯顿大学出版社
24	and also about this book, the future talks of today.	还有这本书,今天的未来演讲
25	Before then, I want to say Princeton POP has a profound relationship with Yale in disseminating	在此之前,我想说普林斯顿 POP 与耶鲁有深刻的关系在传播
26	scholarly conversation because for the past 100 years in our publishing history, POP Princeton	学术对话因为在过去的100年我们的出版史,POP普林斯顿
27	University Press has published more than 200 Yale-affiliated scholars across disciplines.	大学出版社出版了200多份耶鲁大学附属学科的学者.
28	So I very much hope to extend the gratitude to Carol Joyce and the WebVC team for your	因此,我非常希望感谢卡罗尔·乔伊斯和WebVC团队的帮助
29	continuous support in making every event possible.	持续支持使每一活动成为可能。
30	So before introducing Professor Coveney and his new book, I want to briefly talk about	所以在介绍科维尼教授和他的新书之前我想简单谈谈
31	what we're doing in China because Princeton University Press published peer-reviewed books	因为普林斯顿大学出版社出版了同行评审的书
32	that connect authors and readers across the spheres of knowledge to advance the global	将作者和读者连接到知识领域,以推动全球
33	conversation.	谈话。
34	So we are the first and so far the only American university press opening an office in China	所以,我们是第一个,也是迄今唯一一个在中国开设办事处的美国大学报刊
35	because POP has a global mission and the very heart of our editorial strategy that is to	因为我们的编辑策略的核心就是
36	the leading scholars around the world and bring the best ideas to the global readers.	并给全球读者带来最佳的想法。
37	This includes Chinese scholarships and Chinese readers.	这包括中国奖学金和中国读者.
38	So Joyce just introduced Professor Coveney, who is the director of the Center for Computational	乔伊斯刚介绍科维尼教授他就是计算中心主任
39	Science at University College London, Professor at the Informatics Institute, University of	伦敦大学学院科学系教授,
40	Amsterdam, and Professor at the Yale School of Medicine.	阿姆斯特丹,耶鲁医学院教授.
41	His new book, Virtue You, is a comprehensive account of efforts by scientists around the	他的新书《道德的你》全面叙述了科学家们在
42	world to build virtual twins of human beings from cells to tissues to organs and whole	世界将人类从细胞到组织到器官乃至整个的虚拟双胞胎
43	bodies.	尸体,还有尸体。
44	In this very deeply illuminating book, Professor Coveney and his co-author reveal the visionary	科维尼教授和他的共同作者在这本非常具有启发性的书中揭示了远见卓识
45	science behind the digital human twins that will enhance our health and our future while	将增进我们的健康和未来
46	also addressing the emerging ethical challenges accompanying with it.	还应对随之而来的新出现的道德挑战。
47	We believe this informative book will provoke thought among a diverse readership, including	我们认为,这本内容丰富的书将引起不同读者的思考,其中包括:
48	future research leaders and pioneers who will drive this field forward.	未来的研究领袖和先锋将推动这个领域前进。
49	So this is a remarkable book with both intellectual depth and ability to engage a broader audience.	因此,这是一本了不起的书,既具有智力的深度,又能够吸引更广泛的受众。
50	So I think this is a privilege to have Professor Coveney with us today.	因此,我认为今天科维尼教授与我们在一起是一种荣幸。
51	And I believe it is a very stimulating event.	我认为这是一个非常刺激的事件。
52	So I will hand over the floor to you, Professor.	所以我会把地板交给你教授
53	Thank you so much indeed for that introduction.	非常感谢你的介绍。
54	And it's an honor to be here.	亦复如来.
55	This is a genuinely multinational event.	这是一个真正的多国活动。
56	I'm sitting with a window behind me, which is actually of what's called the Portico at	我坐在我身后一个窗户, 这实际上是所谓的Portico at
57	University College London, but I'm actually in Beijing at this moment.	伦敦大学学院,但此刻我其实在北京.
58	And as you say, I'm a professor adjunct at Yale School of Medicine.	正如你所说我是耶鲁医学院的副教授
59	So there's a really interesting link there, sort of Anglo-American, Sino-Anglo-American	因此,有一个非常有趣的链接, 排序英美,中英美
60	angle on on this talk that there are many things to try and convey here.	说到这个话题有很多事情需要尝试和传达
61	And I want to keep this as understandable and comprehensive as possible.	我想尽可能地保持这种理解和全面。
62	So I prepared to show a video initially.	所以我准备展示一段视频。
63	This video was prepared itself a few years ago.	这段录像是几年前准备的。
64	It's an IMAX film.	这是一部IMAX电影.
65	So it's really at its most stunning on an IMAX screen, which we can't	所以在IMAX的屏幕上它真的非常惊人我们不能
66	obviously run today, but I'll just I'll just put this on first to set the context and	很明显,今天跑, 但我会只是把这个首先设置的背景和
67	then I'll take off with my talk.	然后,我开始说话。
68	So initially, I'm going to share my screen and try to locate where the film is.	因此,我首先要分享我的屏幕, 并尝试找到电影的位置。
69	Maybe we should be opening that film, not that we think this one, but that's it.	也许我们该开拍这部电影, 而不是我们想的,但就是这样。
70	Yeah, I need to share that.	是的,我需要分享这一点。
71	Let me know.	告诉我
72	Can you hear me still?	你还能听见我吗?
73	Yeah, the screen should be black because I'm about to launch a film which has music and	是啊,屏幕应该是黑色的因为我即将推出一部电影有音乐和
74	stunning videos and animations in it.	里面有惊人的视频和动画
75	Let's see if we can get it going right.	让我们看看能不能把事情搞好
76	Very nice.	很不错
77	There's sound.	有声有声.
78	Our bodies are exquisitely complex machine.	我们的身体非常复杂
79	Hello, Professor.	你好,教授,你好吗?
80	Sorry to interrupt.	抱歉打断你
81	Is this video have sound?	这视频有声音吗?
82	We cannot hear any sound on the video.	我们听不到录像上的任何声音。
83	Well, obviously, mastering drug delivery means controlling the body's main transport	嗯,很明显,掌握毒品运送意味着控制身体的主要运输
84	network, your circulatory system by simulating the movement of red blood cells and other	网络,你的循环系统通过模拟红血球和其他细胞的运动
85	cells.	细胞。
86	We can understand important protective processes like those that prevent blood loss	我们可以理解重要的保护过程比如防止失血
87	after an injury.	伤后.
88	Virtual humans could help doctors to plan risky surgery, too.	虚拟人类也可以帮助医生们计划危险的手术.
89	They could be used to work out how to reach an aneurysm deep in the brain that is at risk	可以用来研究如何到达大脑深处的动脉瘤
90	of rupture, which could cause a stroke.	破裂,可能引发中风。
91	Surgeons could then try out the best treatment or implant to suit the location and shape	然后外科医生可以尝试最好的治疗或植入以适应位置和形状
92	of that particular aneurysm.	那个动脉瘤
93	They could even double check that the implant would not cause problems such as	他们甚至可以加倍检查植入不会造成问题,例如
94	before they try it out on you.	在他们尝试你之前。
95	Physics, chemistry and biology intertwine in your circulatory system to drive the most	物理、化学和生物学在你的循环系统中相互交织驱动最多的
96	remarkable pump that evolution has ever created, the heart.	进化所创造的非凡的泵心脏
97	To create a virtual heart, processes have to be modelled at multiple scales, from the	要创建虚拟心脏,就必须在多个尺度上模拟进程,从
98	contraction of muscle to the blood flowing through its chambers to the movement of charged	肌肉向血液的收缩,血液流经其内室至电荷运动
99	atoms during the contraction.	收缩期间原子.
100	Virtual hearts already beat within supercomputers where we can test the effect of different	虚拟心脏已经在超级计算机中跳动我们可以测试不同的效果
101	drugs or pacemakers, but also more fundamentally to understand the way an individual's	药物或起搏器,但更根本的是理解个人的
102	heart works.	心脏工作。
103	Your virtual avatar not only be built in your image, it will also move	你的虚拟形象不仅在图像中构建,它也会移动
104	like you.	喜欢你
105	We can use it to calculate the forces and the mechanical stresses that are	我们可以用它来计算力和机械压力
106	constantly induced on your bones and predict the risk you will suffer a fracture.	并预测你可能骨折
107	Doctors can also use virtual humans to work out how diet, exercise or drugs can help	医生也可以用虚拟人类来研究饮食、运动或药物如何帮助
108	those with virtual hearts.	那些有虚拟心脏的人。
109	And what if we go even further than treating disease, using virtual humans to prevent	如果我们更进一步比治疗疾病, 用虚拟人类预防
110	illness?	生病了吗?
111	Within a supercomputer, many virtual versions of you can explore small changes	在超级计算机中,很多虚拟版本的您可以探索小的变化
112	in your lifestyle and how they affect your health, ageing and quality of life.	在你的生活方式以及它们如何影响你的健康, 老龄化和生活质量。
113	Virtual humans, helping you to figure out the very best that you can be.	虚拟人类,帮助你找出你最好的。
114	So that was the video.	这就是视频。
115	It seemed like we lost the sound for a period, but did you hear the rest of it clearly?	看来我们失去了一段时期的声音,但你听清楚了吗?
116	Yes, yeah, the beginning, maybe first one minute and there's no sound.	是的,是的,开始, 也许第一分钟没有声音。
117	Right, so the purpose of that was to show, they're not computer animations in the sense	对,所以它的目的就是显示, 他们不是计算机动画的意义
118	of, say, a Disney Hollywood movie.	比如迪士尼好莱坞电影
119	These are patient-specific simulations run on supercomputers in many parts of the world,	这些是在世界许多地方超级计算机上运行的针对患者的模拟,
120	in the US, particularly in Europe, and they are indications of the level of fidelity that	在美国,特别是在欧洲,它们表明忠诚的程度
121	we can perform such studies now of people on.	我们可以进行这样的研究现在的人。
122	And the main message that I'm trying to convey in the talk that I'm going to launch from	和主要的信息,我试图传达在谈话中,我要推出
123	here is our ability to predict things ahead of time in medicine now, for the first time,	我们有能力在医学领域提前预测事情
124	because we have enough understanding of mechanisms and our ability to model using powerful	因为我们有足够的理解机制和我们的能力建模使用强大的
125	computers.	电脑。
126	The computers are essential in order to get answers in a hurry.	计算机对于迅速得到答案至关重要。
127	And a very good analogy, which everyone is effectively familiar with, although they may	和非常好的类比, 每个人都很熟悉, 虽然他们可能
128	not realise it, is what happens today when people make a weather forecast.	没有意识到,就是今天人们做天气预报时会发生什么。
129	They use supercomputers, they run predictions of what tomorrow's weather the next few days is,	他们使用超级计算机预测未来几天的天气
130	like you were worried about a typhoon approaching, and the probability of it causing	就像你担心台风即将来临以及它导致的概率
131	damage, flooding and devastation is based on predictions of the same sort that I'm going to	破坏,洪水和破坏是基于同样的预测我要去
132	be talking about in the slides that I want to show now.	现在我要在幻灯片里讲
133	So let me see if I can get hold of those and share them next.	所以,让我看看,如果我能得到这些并分享下。
134	Can you tell me if you can see anything there yet?	你能告诉我,如果你能看到什么吗?
135	Yeah, it's on.	是啊,这是在。
136	I'm trying to see where it is on my own screen.	我想看看它在我的屏幕上。
137	Yeah, a lot of fires.	是啊,很多火灾。
138	Shall I enlarge it?	要我放大吗?
139	Yeah.	对
140	Yeah, looks perfect.	是啊,看起来很完美。
141	OK, so I will talk through these slides and I will then end with another video, which is	好吧,所以我会通过这些幻灯片谈然后我会结束另一视频,即
142	shorter than that one, which is about the next pandemic, meaning what happens after Covid, if	比那个要短,也就是下一个大流行, 意思是Covid之后会发生什么,如果
143	we have one of those to deal with.	我们有其中之一要对付。
144	How will supercomputers play a role in that, as they have done in this?	超级计算机如何在这方面发挥作用?
145	So as the introduction so nicely spelled out, this talk is sort of based around a book that I	所以,正如引言如此精心阐述, 这段话有点围绕一本书,我
146	published recently with my co-author Roger Highfield.	最近与我的合著者罗杰·海菲尔德出版
147	It was published in English language on the 28th of March this year.	今年3月28日以英语出版.
148	Roger Highfield is a longstanding friend and colleague of mine.	罗杰·海菲尔德是我长期的朋友和同事
149	He's a distinguished journalist who specialises in making accessible, complicated ideas.	他是一位杰出的记者,他特意让大家了解复杂的想法.
150	So this is the third book I've written with him and the work we do together is complementary in	这是我和他一起写的第三本书我们一起做的工作是相辅相成的
151	so far as I'm very keen on getting the science described.	我非常渴望得到科学的描述
152	He wants to make sure it's accessible to the largest number of people.	他想确保它能被尽可能多的人所利用.
153	And that's what the product is in the case of this book.	而此书中之产也.
154	So let me move on to what we talk, what we mean by virtual humans.	因此,让我继续我们所说的,我们所说的虚拟人类。
155	And in the current era, in fact, you know, there's become a great deal of interest in what are	在当今的时代,事实上,你知道, 人们开始对什么很感兴趣
156	called digital twins and digital twins are what is the concept that dates back to engineering	叫做数码双胞胎和数码双胞胎的概念可以追溯到工程
157	contexts, maybe of the order of 20 years or so ago.	大约20年前的事
158	The idea is that you can design increasingly complicated manufacturing plants.	这个想法是你可以设计越来越复杂的制造厂.
159	It might be aerospace, automotive, you could be designing a rail line as has happened recently in	可能是航空航天汽车你可以设计一条铁路线就像最近发生的
160	London, there's something called the Elizabeth line.	伦敦,有东西叫伊丽莎白线.
161	You build the models purely on a computer in the first instance, but seek to make them as high	首先你把模型建在电脑上, 但试图把它们做得这么高
162	fidelity as possible.	忠诚尽可能。
163	And as you build the system in reality, which is the twin, as it were, the analog, you have a very	当你在现实中构建这个系统, 也就是双胞胎, 就像它,模拟, 你有一个非常
164	close interaction between the two, such that they learn from each other where the digital version	二者之间的密切互动,以便相互学习数字版本
165	needs correction from the reality as you build, you correct it.	需要从现实中校正当你构建,你纠正它。
166	And of course, reality is being designed through the digital version.	当然,现实正在通过数字版本来设计。
167	So that's what we're talking about.	故知此义.
168	The exciting thing about human digital twins or virtual humans is that we're trying to do the same	人类数码双胞胎或虚拟人类的激动人心之处在于,我们试图做同样的事情
169	thing now in the context of medicine, looking at the human body in ultimately great detail and	以医学为背景仔细观察人体
170	mechanistic detail so that we understand how it works and we can model and simulate it and	机械细节,这样我们才能理解它是如何运作的, 我们可以模拟它,
171	interact with it.	与它互动。
172	And the real version, that is one, you know, individual humans are the twins for this type of	而真实的版本,就是,你知道, 个体人类是双胞胎对于这种类型的
173	body.	躯体。
174	And in detail, in terms of modeling, these are complicated, multi-scale and	具体来说,在模型制作方面,这些是复杂的,多尺度的和
175	multi-physics representations of the human body, because the human body is very complicated	人体的多物理表现,因为人体非常复杂
176	from the genome level, essentially, which is a molecular one around DNA, which we saw in the	从基因组水平,本质上, 这是一个分子围绕DNA, 我们在其中看到
177	film, up to the whole human scale, but not just single humans.	电影,直到整个人类尺度,但不只是单一人类。
178	Something I want to emphasize here, but potentially and in fact, many humans, because	我想在这里强调一件事, 但潜在的,事实上,许多人类,因为
179	if we're dealing with an infectious disease, COVID is the latest dreadful example of that.	如果我们在处理传染病,COVID 是最新的可怕例子。
180	Then, of course, there are interactions between humans that propagate infections that need to be	当然,人类之间的相互作用传播感染需要
181	dealt with.	处理。
182	But there are also other things, namely population health and the use of these	但还有其他问题, 即人口健康和使用
183	models for in silico trials.	硅酸盐试验模型。
184	They can be clinical trials that are going to be run purely on computers with a lot of	它们可能是临床试验将完全运行在计算机上
185	benefits.	福利。
186	So this is just holding out a vision for a moment of the things that are coming down	因此,这只是保持一个愿景片刻的东西即将下来
187	the line and why it's important to know about them and look at what medicine is today by	线条和为什么重要要了解他们看看今天的药是什么
188	comparison.	比较。
189	And essentially, I'm being a little provocative here, but I can certainly	基本上,我在这里有点挑衅, 但我可以肯定
190	justify it.	为它辩护。
191	We can say typically current medicine is not predictive in the sense that I'm	我们可以说,典型的当前医学不预测,因为我是
192	referring to it.	提到这一点。
193	And as I mentioned, the weather is because it's based on examples in the	正如我提到的那样,天气是因为它是基于
194	past of doctors seeing a patient that who looks and like others they've seen before and	过去,医生看到一个病人,谁看起来像其他人,他们以前见过,
195	therefore should respond in a similar way.	因此,应该以类似方式作出反应。
196	And we should be mindful of that type of approach to	我们应当注意这种处理方式。
197	medicine is really a statistical inference methodology, which isn't ever going to be	医学真的是一种统计推论方法这永远不会是
198	satisfactory, because if you try to aim for the average behavior of the population to predict how	满意,因为如果你试图瞄准人群的平均行为预测如何
199	the next person is going to respond, it's very likely that there are such things as average	接下来的人会回应很有可能是普通人
200	people.	人们
201	They will always behave differently.	他们的行为总是不一样
202	And if your data sets are unrepresentative of the	如果你的数据集不能代表
203	type of people you're trying to predict medical responses to, it will fail.	那种你试图预测医学反应的人, 它会失败。
204	So that's a	这么说吧
205	problem.	问题。
206	That's also going to be a problem with methods like artificial intelligence and AI,	人工智能和AI等方法也会有问题
207	which are based on data sets of other people against whom your predictions are made.	以其他预测对象的数据集为基础
208	So that's	这么说吧
209	a little warning about how we have to handle AI in these contexts.	警告一下我们如何在这样的背景下处理AI
210	So the development of the	因此,发展
211	virtual human and to make it truly predictive has to be something that integrates all of this data	虚拟人类并让它真正具有预测性必须是整合所有这些数据的东西
212	and is firmly personalized.	并坚定的个性化。
213	And to get the context motivated even better by what's happening in	并且让背景的动力更好的发生。
214	terms of studying complex systems today.	今天研究复杂系统的条件。
215	I mentioned the weather earlier, the weather is a	我之前提到天气, 天气是一个
216	challenge, but we do a reasonably good job at predicting weather into the near term using	但我们在预测近期天气方面做得相当好
217	computer models and feedback with what's actually happening with sensors that are built and measure	计算机模型和反馈实际发生与传感器
218	data around the world.	全世界的数据。
219	So this picture is just showing you an example of the planet, but there	所以这张照片只是给你们展示一个地球的例子, 但是在那里
220	are bigger challenges than just weather forecasting, which is a matter of only a few days	不仅仅是天气预报,只是几天而已
221	ahead.	前进。
222	Typically, we all know that we live with a challenge called climate change.	通常,我们都知道,我们生活在一种被称为气候变化的挑战中。
223	And climate	还有气候
224	change is something that has to be dealt with over decades to centuries and beyond.	变革是必须在几十年到几百年以及以后处理的问题。
225	And the	还有
226	challenge there is similar to what I'm getting at with medicine, how well can we predict the	挑战与我对医学的看法相似,
227	future such as the interesting long term course, long time course or longitudinal behavior of	未来,例如有趣的长期路线,长期路线或纵向行为
228	humans in a medical sense using modeling and simulation?	医学意义上的人类使用模型和模拟?
229	How do they do climate modeling?	他们如何做气候模型?
230	And they have digital twins, probably the most sophisticated ones that exist today are ones in	他们有数码双胞胎可能是今天最精密的双胞胎
231	which people try to capture the behavior of the world and its climate in computers by building in	人们试图捕捉世界的行为及其气候在计算机中通过建设
232	as much data as they can, accurate models of the processes that are involved, which are	尽可能多的数据, 准确的模型涉及的过程,即
233	complicated, maybe not as complicated as in the human body, but certainly very complicated.	复杂,也许不像人体那么复杂,但肯定非常复杂。
234	And	还有
235	then they have strong feedback between what's happening in the physical world, which is subject	然后,他们有强烈的反馈在物理世界正在发生, 这是主题
236	to measurements and observations, and what the digital twin in the computer is doing.	和计算机中的数码双胞胎在做什么。
237	And the	还有
238	interaction between those is what we really mean by digital twins.	这些互动是数字双胞胎的真正含义。
239	And that's what I'm getting at	这就是我要得到的
240	in the biomedical case.	在生物医学的案子。
241	And I want to stress this because areas like climate are often seen as, you	我想强调这一点,因为像气候这样的地区经常被视为,你
242	know, non biological scenarios.	你知道,非生物情景。
243	And we understand in the physical world, we can model things.	在物理世界中,我们可以模拟事物。
244	And	还有
245	I'm trying to make sure we all recognize that it's now possible to model and simulate humans in a	我试图确保我们都认识到,现在有可能模拟和模拟人类在一个
246	similar way, not to say it's too complicated, we cannot do it.	类似的方式,不是说它太复杂, 我们不能做到这一点。
247	So just briefly, the benefits in	简而言之,好处在于
248	climate change are enormous, because you can see the role of things in the natural world influencing the	气候变化是巨大的,因为你可以看到自然世界中事物的作用影响着人类
249	predictions.	预言。
250	And of course, the predictions feedback into what's going on in terms of the	当然,这些预测会反馈正在发生什么
251	evolution into the future.	进化到未来。
252	And it's an interesting fact that in the most sophisticated digital twins	一个有趣的事实是, 在最复杂的数字双胞胎
253	of planet Earth, and there's one very large European project in that space that I'm thinking of,	地球, 有一个非常大的欧洲项目在这个空间,我想,
254	humans are actually being modeled in these systems.	人类实际上是在这些系统中的模型。
255	Is that a surprise?	这是惊喜吗?
256	No, because climate change	不,因为气候变化
257	is largely being caused by human behavior, maybe over several centuries.	大部分是由人类行为引起的, 可能是几个世纪了。
258	And the response to it,	和对此的反应,
259	and the impact that it has on humans is an important aspect of how we handle it.	以及它对人类的影响是我们处理它的一个重要方面。
260	So these	所以这些
261	models have humans in the simple ones.	模型有人类在简单的。
262	So there's human modeling already going on in those things.	所以那些东西中已经出现了人类的模特。
263	And the benefits start to become clear as a result.	而利益由此开始变得清晰.
264	I'm just going to touch on some important	我只是要去接触一些重要的
265	aspects of modeling, which people must understand in the life of medical sciences, because	在医学生活中人们必须理解的建模方面,因为
266	historically, they are not used to doing modeling and simulation, let alone using it to predict	历史上,他们不习惯做模型和模拟,更不用说用它来预测
267	outcomes before events occur.	事件发生前的结果。
268	But that's exactly what we have to be able to do.	但这正是我们必须做的。
269	If we're to	如果我们要
270	implement personalized medicine, the idea is we don't sort of observe a patient and discover	执行个性化的医学, 想法是,我们不排序观察病人和发现
271	they're in a bad way, and perhaps dead, and try to do something about it.	他们的方式不好, 也许死了, 并试图做一些事情。
272	We anticipate	我们预计
273	their state of health, and we address it.	他们的健康状况,我们来处理。
274	But in modeling, we encounter problems of the sort that	但是在做模特时我们遇到的问题
275	this slide shows.	此幻灯片显示。
276	And it's a very well known one in the physical sciences, and in particular,	它在物理科学上是众所周知的,特别是,
277	the big challenge for weather and climate prediction.	对天气和气候预测的巨大挑战。
278	That is the role of uncertainty	这就是不确定性的作用
279	in the models.	在模型中。
280	All models have uncertainty in them.	所有模型都有不确定性。
281	There's a lot of parameters we don't know	有很多我们不知道的参数
282	exactly.	没错
283	And in a biological human model, the parameters will be patient-specific.	而在生物人类模型中,参数将是针对病人的.
284	And there	还有
285	are aspects of the dynamics that are also uncertain.	这些都是动态的方面也是不确定的。
286	In particular, there's this concept	特别是这个概念
287	that we've known about for at least 30 years or so.	我们认识至少30年了
288	I wrote a previous book on that.	我之前写过一本书
289	In fact,	事实上,
290	The Hour of Time, I wrote in Frontiers of Complexity, dwell heavily on this.	我在《复杂之境》中写到的《时间的小时》大量关注这个问题。
291	That is that	就这样
292	unless you were to know the initial conditions for a simulation with absolute precision,	除非你知道模拟的初始条件绝对精确,
293	you have uncertainty perpetuating.	您的不确定性将永远存在。
294	So we are making probabilistic predictions,	所以我们在做概率预测
295	and that's what we will be doing in the medical case as well.	这也是我们在医疗病例中要做的
296	We can't know with absolute certainty	我们绝对不能肯定
297	what tomorrow's weather is like, but we can talk about what the probabilities are of it happening.	明天的天气是怎样的, 但我们可以谈论它发生的概率。
298	And that leads to actionable predictions in weather forecasts.	这导致了天气预报中的可操作预测。
299	When I'm in London, I hear	当我在伦敦,我听到
300	prediction that tomorrow it's going to rain.	预言明天会下雨
301	I get my umbrella ready, and I take it out.	我准备好伞,然后拿出来
302	You see where this is going to go with the biomedical case, which is much more a matter	你看这与生物医学的案子有什么关系这更是一个问题
303	of life and death in some cases.	在某些情况下,生死攸关。
304	So this is a situation you get predictions of weather	所以,这是你得到天气预测的情况
305	forecasts.	预测。
306	They tend to work very well for the next three or four days.	在接下来的三、四天里,它们的工作往往非常好。
307	And then they become	然后他们变成了
308	increasingly less accurate over time, over a few more days.	随着时间的推移,越来越不准确,多过几天。
309	But they can be rectified and improved	但是,它们可以纠正和改进
310	by continually taking into account new data from observations at the current time,	不断考虑到目前观测中的新数据,
311	rather than relying on the stuff that you had to make the predictions at the beginning of the	而不是依赖你不得不做的预测在开始
312	week in this case.	每周,在这种情况下。
313	And this video, and I hope you can see it, is an example of	这段视频,我希望你能看到, 就是一个例子,
314	Tropical Storm Ian that was simulated ahead of the impact of that hurricane	热带风暴伊恩在飓风影响前被模拟
315	in mainland USA, where it would impact the coastline in, I should say, southeast of the USA,	在美国大陆,它会影响海岸线, 我想说,在美国东南部,
316	and the impact that it would have at its strongest point in terms of damage.	以及它在损害方面最强烈的影响。
317	And as a result of these	由于这些
318	sorts of predictions, actions were taken that are known to have saved several hundred lives.	各种预测和行动已经拯救了几百人的生命
319	So it shows the power of prediction, even of events which are as complicated as these,	因此,它显示了预测的力量, 即使事件是复杂的,
320	if you can run the simulations with sufficient fidelity.	如果你可以运行模拟有足够的忠诚。
321	That's again the message I'm trying to	这又是我想传达的信息
322	convey here.	在这里传达。
323	So now at last I turn to the human body and I ask the question of how realistic is	所以现在我终于转而谈谈人体, 我问一个问题,多现实
324	it for us to do these things?	我们做这些事?
325	And it's easy to try and claim it's not going to work.	且试而称不可行,亦易之.
326	But the point	但重点
327	about the physical sciences, modeling and simulation using techniques from computer science, for sure,	关于物理科学,模型和模拟利用计算机科学的技术,当然,
328	AI, machine learning, is that no matter the complexity of the systems, it's possible to make	AI,机器学习,是,不管系统的复杂性如何,都有可能使
329	predictions that have quantitative value.	具有数量价值的预测。
330	We don't always have all the information, but we can	我们并不总是有所有的信息, 但我们可以
331	quantify the uncertainty in the predictions, such that we know within limits how reliable they're	量化预测中的不确定性,这样我们就可以在极限内知道它们有多可靠
332	going to be.	将会是。
333	This is all part and parcel of the state of science in the modern period.	这是现代科学状况的一部分。
334	So just to	所以就这样
335	talk a little bit briefly about us and the virtual human, you know, I've been talking about this for	简单说说我们和虚拟人类
336	quite a while with various people in the public eye as well as scientifically.	在公众和科学的眼中和不同的人在一起
337	It's just an example	这只是一个例子
338	of a picture from a big meeting that we ran at the Science Museum in London back in 2017.	2017年在伦敦科学博物馆举行的大型会议的照片
339	In fact,	事实上,
340	when that film I just showed you, the beginning was premiered and I'm on the second from the right	我刚刚给你们看的那部电影首映的时候我从右边开始第二部
341	and the guy in the center is Roger Highfield who's my co-author here.	中心那个人是罗杰·海菲尔德他是我的合作作者
342	The others are colleagues of mine	其他人都是我的同事,他们是我的同事
343	who work in the areas I'm discussing today.	在今天我讨论的领域工作的人
344	So how do you build digital twins from multi-scale data?	如何从多尺度数据构建数码双胞胎?
345	You need a lot of data and it comes at all types of length and time scales.	你需要很多数据,它来自各种类型的长度和时间尺度。
346	So make no mistake,	所以不要犯错
347	I'm not trying to claim we would build a virtual human that's reliable purely out of the genome	我并不想声称我们会建立一个虚拟人类完全从基因组中产生
348	level.	级别。
349	I know there are some people who might try and claim, give me your genome in its entirety and	我知道有些人可能会试图并声称, 给我你的基因组的全部,
350	I will predict your life course from birth to death, along the way all the things that were	我会预测你的人生历程从出生到死亡, 沿途的一切
351	as a result of disease and so on.	由诸病等故.
352	But frankly that's not taken seriously because there are	但坦率地说,这不是认真对待的因为有
353	many things in human biology that are controlled like the genome expression is controlled by events	人类生物学中很多被控制的东西就像基因组表达被事件控制
354	at much higher levels in the human body and also by the environment.	在人体和环境的更高层次。
355	And as soon as I talk about	我一谈起
356	the environment, you're thinking about climate change and other things that has nothing to do	环境,你正在思考气候变化和其他没有事情做
357	with what was written down in the genome at the beginning.	和基因组最初记录的东西
358	So it's far more complicated and we	所以事情要复杂得多我们
359	need all this other information on phenotypical data as well.	还需要其他所有关于苯二氧基数据的信息。
360	And then there's the role of	然后还有角色
361	things like artificial intelligence.	就像人工智能一样
362	So if we do stay down at the molecular level for a moment	所以如果我们在分子水平上保持低调
363	here, some of the things we're interested in doing are to predict let's say drugs that will work	在这里,一些我们感兴趣的事情是预测,让我们假设药物会有效
364	properly effectively for individuals.	对个人来说是有效的。
365	You'll be aware probably that in the farmer industry,	你可能会知道在农民行业,
366	we don't have a very convincing picture.	我们没有非常有说服力的照片。
367	It takes on average 10 years to produce a drug.	生产一种药物平均需要10年。
368	It costs three billion dollars.	耗资三十亿.
369	When it works and when it works at best, it seldom works for more	当它工作,当它工作最多, 它很少工作更多
370	than 50% of the population because there are all these effects that are person specific,	超过50%的人口,因为所有这些影响都是针对个人的,
371	particularly the genome in this case, that are not being considered.	特别是本案中的基因组没有被考虑
372	Now, if you have a method	现在,如果你有办法
373	就像DeepMind的AlphaFold,它现在是Google的一部分, 这很有趣,因为它
374	purports to take the primary sequence of the proteins which are programmed by the DNA sequence	意图取取由DNA序列编程的蛋白质的主要序列.
375	and it will predict their three-dimensional structures.	它将预测它们的三维结构。
376	Why do we need the 3D structures?	为什么我们需要三维结构?
377	Well, if we're doing drug design, we absolutely must know that information.	如果我们在做药物设计我们绝对要知道这些信息
378	And historically,	历史上,
379	it takes a long time to acquire it.	获得它需要很长时间。
380	But if we can just write down the sequence of the protein that	但是如果我们能记录下蛋白质的序列
381	we've sequenced, let's say it's the targets from COVID or some other new disease, then we can predict	假设是COVID或其他新疾病的目标然后我们可以预测
382	the 3D structures very quickly and we're away beginning to do drug discovery and we can	3D结构非常快,我们离开开始做药物发现,我们可以
383	potentially do it on a person specific level as well.	也有可能在某个人身上做
384	That's one of the things we're interested in.	这是我们感兴趣的事情之一。
385	So AI is always going to have something very important to say here, but it does have limitations	因此,AI总是会有非常重要的东西在这里说, 但确实有限制
386	I will come a bit back to those.	我会回来的
387	But at the moment, I want to just emphasize this interaction	但此刻,我想强调这种互动
388	with supercomputers that was mentioned at the beginning and in the video.	与超级计算机在开头和视频中提及。
389	Because what we're	因为我们是什么
390	trying to do here is to understand mechanistically what's happening with individual patients,	试图在这里做的是机械地理解发生在个体患者身上,
391	not just doing this statistical inference.	不只是做这个统计推论。
392	I've seen a bunch of previous patients,	我见过很多以前的病人
393	therefore I've got a machine learning model and I can tell what's going to happen to you.	因此,我有一个机器学习模型我可以告诉你会发生什么。
394	That does require a lot of computing to train the models, but once it's in place,	这确实需要大量的计算来训练模型,但一旦它到位,
395	you would think it's instantaneous.	你会认为这是瞬间。
396	I hinted at why that may not be reliable	我暗示了为什么这不可靠
397	and can come back to that later.	以后再说吧
398	So if we do the mechanistic modeling here,	如果我们在这里做机械模型
399	we absolutely must use supercomputers.	我们必须使用超级计算机。
400	And today's supercomputers are very powerful ones.	而今天的超级计算机非常强大。
401	We've just	我们刚刚
402	sort of breached the exascale, which is 10 to the 18 floating point numbers per second.	以每秒10到18个浮点数表示。
403	And certainly we have an official exascale computer that I use at Oak Bridge National Lab	当然,我们有一个正式的 exascale 计算机,我在橡树桥国家实验室使用
404	in Tennessee.	在田纳西州
405	It's a Department of Energy supercomputer called Frontier.	这是能源部的超级计算机,叫做Frontier.
406	But there are	但还有
407	some interesting Chinese machines that are particularly relevant to this talk in this	一些有趣的中国机器与这个演讲特别相关
408	country, which are at or around the exascale already.	国家,它们已经处于或正在接近规模。
409	And I mentioned exascale.	我提到夸张。
410	At that scale,	在这个规模上,
411	you can begin to do really interesting large scale simulations of many organ systems,	你可以开始做非常有趣的大规模模拟许多器官系统,
412	large parts of the drug discovery platform can be included on these machines.	药物发现平台的大部分部分可以包含在这些机器上.
413	And the purpose of	和目的为
414	this is to do or exploit IT to do predictive studies prior to doing experiments and only	这是在进行实验前进行或利用信息技术进行预测性研究,并且
415	start doing the experiments once you've sorted out the best candidates, let's say for a drug	等你们找到最好的人选开始做实验吧
416	discovery problem or for a problem in another area of healthcare.	发现问题或者另一个保健领域的问题。
417	And I have emphasis on what	我强调什么
418	we call VVUQ, which is validation, verification and uncertainty quantification.	我们称之为VVUQ,即验证、核查和不确定性量化。
419	I've already used	我已经用过了
420	the words uncertainty quantification in this talk three or four times.	本演讲中的不确定性量化词三四次。
421	And the importance is make	这一点很重要
422	your predictions actionable, trustworthy, convincing so that in this case, the medics and the clinicians	您的预言是可行的,可信的,令人信服的因此在这种情况下,医生和临床医生
423	realize that they had better take notice of what the predictions are rather than saying,	他们知道自己所说的话,与其说,不如说:
424	I'd rather wait to see what the outcome.	我宁可等着看结果如何
425	And as we get to more and more powerful machines,	而当我们进入越来越多的强大的机器,
426	this becomes increasingly feasible.	这越来越可行。
427	So what we do today in this area, things like virtual	所以我们今天在这个领域做什么比如虚拟的
428	drug testing, coming back to what I mentioned earlier, you know, we've done a number of studies	药物检测,回到我前面提到的, 你知道,我们已经做了一些研究
429	here that have an impact on how you can treat individual patients by looking at their genome	影响你如何通过观察病人的基因组来治疗他们
430	sequences.	顺序。
431	Maybe there are mutations in infectious diseases that are different for different patients,	也许传染性疾病的突变对不同的病人来说是不同的
432	and they will respond differently to different drugs.	他们将对不同的药物做出不同的反应。
433	But you can do the studies that are	但你可以做一些研究
434	determined how strong the candidate compounds will be with those patients' target proteins	确定候选化合物与患者目标蛋白质的强度
435	in the computer prior to actually administering.	在实际管理前的计算机中。
436	And so you can build up a large work flow around	这样你就能在周围积累大量的工作
437	drug development and the personalized medicine on these big machines that can cover a lot of	药物研发和这些大机器上的个性化药物可以覆盖很多
438	candidates, patient-specific protein sequences concurrently on the machines.	考生,机器上同时出现患者特有的蛋白质序列.
439	Because this talk is general, I don't want to go into a lot of detail here, but it's an example,	因为这个话题很笼统,我不想在这里详细讨论,但这只是个例子,
440	in fact, for us of where we combine two methodologies that I've also touched on.	事实上,对于我们来说, 我们结合了两种方法, 我也提到过。
441	Some of it is AI machine learning, where the calculations, once you've done training,	有些是人工智能机器学习, 在那里计算,一旦你完成了训练,
442	can be exceptionally fast, but they're not typically accurate enough.	但是它们通常不够准确
443	They can rank a	他们可以排队
444	set of compounds without doing very severe compute on them.	一组化合物,没有进行非常严重的计算。
445	And then with that rank set	然后,随着这个等级
446	of compounds, you can do more, we call fine-grained physics-based studies to work out exactly what the	在化合物中,你可以做更多的, 我们称之为精细的光线物理研究找出究竟什么
447	ordering of the compounds is.	化合物的订单是。
448	And that can be an iterative loop.	这可以是一个迭代循环。
449	And on a supercomputer-like frontier,	在一个超级计算机般的前沿,
450	which is vast, you can run these calculations concurrently and huge numbers of them.	巨大的,你可以同时运行这些计算和巨大的数量。
451	So just for a moment, thinking ahead of where this is taking us,	因此,只是一瞬间, 思考之前,这是带给我们,
452	I mentioned at the beginning, what does it offer?	我一开始提到它能提供什么?
453	If you recall, I was talking about population	如果你记得,我说的是人口
454	medicine.	药
455	I talked about trying drugs out on different people.	我说的是尝试用毒品对付不同的人
456	And then there's the animal	还有动物
457	testing, which clearly is going to be diminished.	测试,这显然会减少。
458	And it's already being shown by some of the work	有些作品已经展示了
459	we're involved in, that using human data and simulations in silico clinical trials is more	我们参与, 使用人类数据和模拟在硅化物临床试验
460	effective at determining which drugs will work than using animal experiments.	有效确定哪些药物比使用动物实验有效。
461	Today, we issue	今天,我们发行
462	human patient data, which is very gratifying.	人类病人的数据,这是非常令人满意的。
463	And then how we can treat people at risk and	然后我们如何对待处于危险中的人
464	improve their experiences.	改善他们的经验。
465	So briefly, because we're going to run out of time if I go into all of	简而言之,因为如果我们进入所有的时间会用完
466	things in detail, I just spoke about the molecular end of this, particularly in drug discovery	详细来说,我刚刚谈到这个分子端,特别是药物发现
467	and personalized medicine in the sense of treating patients with different compounds and drugs.	以及治疗不同化合物和药物患者的个性化药物。
468	But of course, what's challenging in biology is this hierarchy of structures.	当然,在生物学中,最具有挑战性的是这种结构的层次。
469	And this picture	这张照片
470	is just showing some interesting representations of the most sophisticated cell models that exist	只是展示了一些有趣的表现现存最复杂的细胞模型
471	today, where we've got 3D representations of all the glorious things going on inside a cell	今天,我们得到了 3D的表示所有光荣的事情发生在一个牢房
472	and the representation of how the genome is to some extent controlling the evolution	基因组如何在某种程度上控制进化
473	and the replication of those cells.	复制这些细胞
474	And one of the leading people in this area is somebody called	这个领域的领导者之一是一个叫
475	Marcus Covert, who's at Stanford University, who we worked with quite a bit in writing this chapter.	在斯坦福大学的Marcus Covert 我们和他合作写了一章
476	And he's been able to do some really interesting studies that have made predictions of things	他能够做一些非常有趣的研究对事物作出预测
477	in certain sets of bacterial cells that have been shown to be the case after the predictions	在某些细菌细胞群中,经过预测,已经证明是这种情况。
478	have been made.	已经做了。
479	But I'm not going to dwell on anything more because of the lack of time until	但我不会多谈什么因为没有时间,直到
480	you get to the level of the organ systems.	到达器官系统的水平
481	Because those ones are ultimately, from a	因为这些是最终,从
482	sexual point of view, the things where we look at a human in its entirety or some visually appealing	性的观点,我们看一个人类整体或一些视觉吸引
483	scale and say, well, we're really dealing with some humanity.	规模和说, 好吧,我们真的在处理一些人性。
484	And I just show this example,	我只是举个例子
485	because Yoon Sun is a Korean woman who died relatively young and whose body was donated to	因为尹善是韩国女子,死得相对年轻,尸体被捐给
486	an anatomical library at a company called It Is.	一家叫It Is的公司的一个解剖图书馆。
487	It's a not-for-profit organization	这是一个非营利组织
488	spun out of ETH in Zurich.	苏黎世的ETH出局了
489	They have anatomical libraries.	他们有解剖图书馆。
490	They have all the data on	他们有所有的数据
491	her organ systems and some of the things we're interested in there, like heart and	她的器官系统还有我们感兴趣的一些东西比如心脏
492	the vasculature from head to toe.	从头到脚的阴道
493	And I just want to mention examples here, because some of the work	我只想举个例子因为有些工作
494	I do myself in this area is building computational models that are patient specific that treat	我做我自己在这个领域是建立计算模型具体病人治疗
495	the entirety of the vasculature from head to toe of a human, arteries and veins and the coupling	人类、动脉和静脉的头到脚的全脉动和连结
496	of them between capillary beds and networks that can run on very big supercomputers.	其中,在毛细床和网络之间可以运行在非常大的超级计算机上。
497	So we're now	所以,我们现在
498	able to run whole human blood flow simulations on the frontier machine.	能够运行整个人类血液流模拟在前沿机器。
499	No machine that we've	我们没有机器
500	had access to before that is capable of doing those simulations.	在能够进行这些模拟之前,就已有机会进入。
501	And that takes a lot of effort.	这需要很多努力。
502	You	你们
503	have to develop a code that's, as we say, scalable, because there are of the order of 10,000 nodes	必须开发一个代码,正如我们所说的,可扩展, 因为有大约10,000个节点
504	on frontier.	在边境。
505	And let's say maybe half a million to a million cores.	也许50万到100万个核心
506	But more particularly, there are	但更特别的是,有
507	by some counting, eight GPUs per node there.	通过一些计数,8GPU 每个节点在那里。
508	So you could say this machine has the order of	所以你可以说这台机器有
509	80,000 GPUs on it.	上面有8万个GPU
510	So to be able to harness all of those to a single calculation is what I'm talking	因此,能够利用所有这些一个计算,这就是我说的
511	about here.	大约在这里。
512	And to run it scalably is the challenge.	以可扩展的方式运行是挑战。
513	But we now are able to do that.	但我们现在能够做到这一点。
514	And I'm talking about organ systems.	我说的是器官系统。
515	This is an example of the human heart.	此为人心之例.
516	In some cases,	在某些情况下,
517	the heart being the most advanced in terms of modeling and simulation has software that's	心脏是最先进的建模和模拟软件是
518	being prepared and used by commercial companies to support studies by medics with very kind of	由商业公司准备和使用,以支持医务人员的研究,包括:
519	interfaces.	接口。
520	It's just snapshots from software from, I think this is ANSYS in the US, but that's	这是来自软件的快照, 我想这是美国的ANSYS,
521	很明显,STEM。
522	A French company also offers these things now.	一家法国公司现在也提供这些东西。
523	And you may have seen a simulation	你可能看到一个模拟
524	earlier that we ran for the respiratory system, figuring out what's going to happen to aerosols	之前我们跑到呼吸系统找出气溶胶会怎样
525	when they're sprayed into the nose.	当喷入鼻中.
526	This is very relevant to COVID infectious disease as well.	这与COVID传染病也非常相关。
527	What happens when you inhale aerosols, where the particles end up going.	当你吸入气溶胶, 粒子最后会在哪里。
528	Another thing that	还有一件事
529	we talked about, you'll remember, I hope, was the study of the neuromuscular skeletal system	我们谈过,你会记得,我希望, 是研究神经肌肉骨骼系统
530	and our ability to predict fractures in advance of their occurrence by using modeling and simulation	利用模型和模拟预测断裂的能力
531	and imaging that's patient specific.	和图像,这是病人的具体。
532	And so you can ward off fractures and cause people,	这样你就可以避免骨折并引起人们,
533	you know, save people a lot of problems by predictively advising how they can evade those	你知道,通过预测性地建议人们如何躲避这些问题来挽救很多问题
534	types of difficulties.	困难类型。
535	This talk is not supposed to dwell heavily on the brain.	这段话不应该大量停留在大脑上。
536	I suspect sometimes	我怀疑有时候
537	in these presentations, people want to discuss the brain, and I'm happy to do that.	在这些演讲中,人们想讨论大脑,我很高兴这样做。
538	Here I'm	我来了
539	just talking about networks.	只是谈论网络。
540	We do model blood flow in the brain.	我们做模型血流在大脑。
541	I mentioned that earlier,	我之前说过,
542	to a very high level of precision now, but this one is just an example of what happens	现在的精确度非常高但这个仅仅是一个例子
543	with people who suffer epileptic fits and the drugs that people try to use and administer there	患有癫痫症的人和人们试图在那里使用和服用的药物
544	need to be targeted to the right parts of the brain.	需要瞄准大脑的正确部分。
545	So the pathways that are relevant here	所以这里相关的路径
546	are now enumerated and there's a large sort of clinical trial that's based on modeling and	现在被列举, 并有一个大种类的临床试验基于模型和
547	simulation of the effects of the interactions within the brain here in a French ongoing	模拟大脑内部相互作用的影响在法国进行中
548	clinical study that appears to be having benefits to the health and wellbeing of those who don't	临床研究似乎有利于那些没有健康和福利的人
549	respond to standard drug treatments.	回应标准药物治疗。
550	So just by way of illustrating what goes on when you go beyond	所以只是通过说明当你超越
551	single organ systems, because in the long run, the virtual human will be based on joining	单器官系统,因为从长远来看,虚拟人类将基于加入
552	components together.	组合组件。
553	Organ systems, which are roughly speaking the same physico-chemical length	物理化学长度大致相同的器官系统
554	and time scales like heart, vasculature, lungs, all these sort of things, we need to couple them	以及时间尺度,如心脏,脉搏,肺, 所有这类事情,我们需要夫妇
555	together.	一起来
556	There's the multi-scale aspect from molecule upward that also has to be paid attention	分子上方的多尺度面也需要注意
557	to, but this is just an example of what we do today in coupling the heart models that colleagues	但是,这只是一个例子我们今天做结合的心脏模型同事
558	of mine at Barcelona Supercomputing Centre have developed.	我在巴塞罗那超级计算中心发展了
559	It's called Alia, mentioned at the top	它叫阿丽雅,上面提到过
560	left here, and the code I've developed called heMLB with my team.	和我的团队一起开发的HEMLB代码
561	So I was talking about the	所以我说
562	vasculature and our ability to model blood flow in the arteries and the veins, but what happens	脉搏和血管中血流的模拟能力但会发生什么
563	if we have a very high fidelity model of the heart, which has electrophysiology, the mechanical	如果我们有一个高度忠诚的心脏模型, 它有电生理学,机械
564	beating of the heart, but not only just those things, the fluid dynamics of blood flow in and	心脏跳动, 不只是那些东西, 血液流动的流畅动态
565	out?	出来?
566	That's what Alia provides, and it's arguably one of the most sophisticated and powerful	这是阿丽雅提供的可以说是最精密最强大的
567	and patient-specific heart models that exist.	以及患者特有的心脏模型
568	Suppose we couple that to heMLB.	假设我们和HEMLB是一对
569	Well, the news	新闻
570	here is that since this book was written largely, we're now doing precisely that, so that we can	因为这本书大部分是写下来的我们现在正这样做这样我们才能
571	have a closed system, which is a cardiovascular one, the heart pumping blood around the arteries	心脏在动脉周围抽血
572	and veins.	和脉.
573	And immediately when you do that, you're at a scale that's largely	当你这样做的时候, 你是一个很大比例
574	unconsidered in medicine and certainly scientifically, because it's becoming	不考虑医学,当然科学,因为它正在变得
575	like a global, entire human scale, where we can ask questions about the influence of	像一个全球的,整个人类的尺度, 在那里我们可以问关于影响
576	treatment in one area and the impact on other organ systems and diseases.	以及对其他器官系统和疾病的影响。
577	So rather than looking	所以,不要看着
578	at disease cases as totally separate, we start realising that they're not.	疾病病例完全分开, 我们开始意识到他们不是。
579	There are always	总是有
580	interactions between them, and it's better for us to understand those than to continue to just	他们之间的相互作用对我们来说理解这些比继续
581	treat things in isolation.	单独对待事物。
582	This is the way we're going to be able to do it.	此是诸法能行.
583	And in the long run,	从长远来看,
584	it's the connection of all of these things up to increasingly high scales that we're concerned	我们所关心的就是所有这些东西的关联一直到越来越高的尺度
585	with from, as it were, molecule to whole human.	从分子到整个人类
586	Though I would like to emphasise that we will	虽然我想强调,我们将
587	never be trying to build from the bottom up here.	从来没有尝试从下到上。
588	I mentioned that earlier.	我之前提到过这一点。
589	We want to build at	我们想在
590	different levels to address different medical conditions and eventually stitch the things	不同级别解决不同的医疗并最终缝合的东西
591	together with equal weight being given to each level.	并且对每一级别给予同等的分量。
592	So just because this is interesting science	所以,只是因为这是有趣的科学
593	and technology, well, maybe before I go into the science and technology, just mentioning here	和科技,嗯,也许在我进入科学技术之前, 仅仅在这里提到
594	something I also touched on earlier, that my colleagues of Barcelona Supercomputing Centre	我之前也提到过巴塞罗那超级计算中心的同事
595	and also in Oxford now are doing studies which are virtual patient scale.	牛津大学也正在做虚拟病人规模的研究。
596	So hundreds of patients,	这么多病人
597	thousands being run on these large supercomputers.	成千上万人被这些大型超级计算机运行。
598	Each patient's heart in these cases is represented	这些病人的心脏都有代表
599	by the different parameters that the persons have.	由人的不同参数决定。
600	And so you're able to run virtual trials.	所以你可以进行虚拟试验。
601	You can	你当然可以
602	test the effect of given drugs on the heart.	测试给药对心脏的影响。
603	So this is an example of a toxicology study,	这是毒理学研究的一个例子
604	which is important in the drug development cycle, which is again virtualised and can be done in	药物开发周期的重要性,该周期再次虚拟化,可在
605	silico rather than inflicted on real patients.	而不是对真正的病人施压
606	So it's part of accelerating the drug discovery	所以这是加速药物发现的一部分
607	process.	进程。
608	That's using models at the scale of the heart.	这是用模型在心脏的规模。
609	So this is all starting to happen.	所以这一切都开始发生。
610	This is just to whet your appetite.	这只是为了讨好你的胃口
611	If you're interested in quantum computing,	如果你对量子计算感兴趣
612	and I know there are people on my visit here in China who do quantum computing in a major way,	我知道有人来中国做量子计算
613	and also where I am in the US and UK, why is quantum computing interesting?	为什么量子计算很有趣?
614	Possibly one of	也许是其中之一
615	the messages I've conveyed so far is we can't have enough computing power to solve these problems.	目前我传达的信息是我们没有足够的计算能力来解决这些问题。
616	We will always benefit from more.	我们将永远获益于更多。
617	And the promise and potential of quantum computing	量子计算的前景和潜力
618	is that it can solve some problems that conventional computers, the conventional digital	它可以解决某些问题常规计算机,常规数字
619	ones that I was talking about so far, even the exascale machines cannot solve fast enough.	即使是超大规模机器也解决不了那么快
620	And that's because quantum computers have an interesting form of parallelism that speeds up	这是因为量子计算机具有有趣的平行形式加速
621	calculations in a way you cannot have accelerated on conventional computers.	以某种方式计算,你不可能在常规计算机上加速。
622	This is just an example	这只是一个例子
623	such a machine called Sycamore in Google back in the US.	这种机器在美国谷歌叫西卡摩尔
624	And I think I will just sort of show	我想我会只是那种展示
625	another example of this because this is Zhu Zhang from Professor Peng Zhang's lab in the Chinese	这是彭章教授的中文实验室的朱章
626	Academy of Sciences in Beijing here.	这里的北京科学院.
627	I'm visiting in a couple of days as well.	我几天后也会去拜访你
628	His is an optical	他是一个光学
629	quantum computer that's been shown to solve some problems that are sort of intractable for	量子计算机已经证明可以解决一些难以解决的问题
630	conventional computers.	常规计算机。
631	They published a paper in Science two or so years ago.	他们大约两年前在科学杂志上发表了一篇论文.
632	It may be difficult	可能很难
633	for you to discern the scale here, but that's roughly the size of what we call a billiard table.	让你看清这里的尺度, 但这大概是台球桌的大小。
634	And I'm interested in working with them in case my problems that I'm interested in could be put	我有兴趣和他们合作以防我感兴趣的问题被提出来
635	such a machine.	这样的机器。
636	And the things that we might be able to do there are actually	而我们也许能做的在那里实际上
637	accelerate aspects of the drug discovery pipeline that I mentioned even more than we do today.	加快我今天提到的药物发现管道的各方面。
638	So that's part of the message here.	所以这是这里的信息的一部分。
639	There's one other amusing thing.	还有一件有趣的事
640	This is the Greek	这是希腊人
641	Antikythera mechanism, just to remind everyone that the Greeks themselves and others, no doubt	安提基太拉机制只是提醒大家希腊人自己和其他人毫无疑问
642	the Chinese thousands of years ago, knew how to use analog computers.	几千年前的中国人懂得使用模拟计算机
643	And analog computers are	模拟计算机是
644	things that we turned away with from, with the development of the digital computer during and	随着数字计算机的发展,我们放弃了这些东西。
645	just after World War Two, because it's easier to program digital computers.	在二战之后, 因为数字计算机更容易编程。
646	But I'm not, and I won't	但我不是,我不会
647	discuss now because there isn't really time here to dwell on the details of this.	现在讨论,因为现在没有时间详谈这个细节。
648	But we've shown	但是我们已经证明
649	in some of our work that floating point numbers, particularly for chaotic systems,	在我们的一些工作中,浮点数, 特别是对于混乱的系统,
650	floating point numbers are not always reliable in predicting behaviour, whereas analog systems are,	浮点数在预测行为上并不总是可靠的,而模拟系统是,
651	and analog machines are ones that can handle the continuum numbers in mathematics, whereas	模拟机器是能够处理数学中连续数的机器,而
652	digital computers by definition cannot.	数字计算机从定义上说不能。
653	And I just want to remind everyone that	我只是想提醒大家,
654	whereas let's say the frontier system requires 30 megawatts of energy to power itself,	边疆系统需要30兆瓦的能源才能发电
655	the human brain needs only 20 kilowatts or so to do it.	人类的大脑只需要20千瓦左右。
656	So there's a huge disparity in the	因此,有一个巨大的差异在
657	energy requirement there, which is, I should say 20 watts here, which is something very desirable,	能量需求,我应该说,这里有20瓦, 这是非常可取的,
658	again, in the context of things like climate change and the amount of power and energy that's	同样,在诸如气候变化以及能量和能量的范畴
659	dissipated in these large machines.	消失在这些大机器中。
660	We want systems that are more environmentally friendly	我们想要更环保的系统
661	in the future.	未来
662	That's one word.	这是一个词。
663	And I think with that, I can stop.	我想这样,我可以停止。
664	This is a slide that indicates	此幻灯片显示
665	a number of people who've worked with me.	一些曾经和我一起工作过的人
666	It's by no means reasonably representative.	这绝对不具有合理代表性.
667	It's just some	这只是一些
668	in recent past and a number of agencies that have funded my work.	最近,一些机构资助了我的工作。
669	But I would like to stop there.	但我想停下来。
670	If I might bail out of this, I'd like to then run the last second movie and then we can open	如果我能逃出这个圈子,我想再看第二部电影然后我们就可以开拍
671	up a discussion.	讨论一下
672	Would that work with you?	那对你有用吗?
673	Great.	不错
674	So let me just do that.	所以,让我这样做。
675	I have to now find	我现在必须找到
676	where that is.	在那边
677	Is it down here?	是这里吗?
678	This one.	这个
679	Yeah, I can share this.	是的,我可以分享这个。
680	So briefly, this	所以,简而言之,这个
681	film is called The Next Pandemic and it looks at what happened with the COVID pandemic and how	影片名为"下一个大流行",它审视了COVID大流行发生的事情以及如何
682	computers started to be used to try and handle problems with it and indicates what might happen	电脑开始用来尝试和处理问题,并表明可能发生什么
683	with the next one that comes along.	和下一个来。
684	So let's see if I can get this running.	所以,让我们看看,如果我可以运行。
685	We make sure we don't lose the sound on the way.	我们保证不会在路上失去声音
686	Let me know if you can hear it and see it.	如果你能听到的话就告诉我
687	Over the past decades, viruses spread by animals have caused severe diseases in people.	在过去几十年里,动物传播的病毒给人们造成了严重的疾病.
688	这就是SARS,MERS,现在的COVID-19的故事.
689	As the global population	作为全球人口
690	surges in coming decades to almost 10 billion people, changes in land use will prime the planet	未来数十年,人口激增至近100亿,土地使用的变化将使地球充满活力
691	for future spillovers of infections.	未来感染的蔓延。
692	The question is not if there will be future pandemics, but	问题不在于今后是否会发生流行病,而是
693	when?	什么时候?
694	And most importantly, will we be ready?	最重要的是,我们准备好吗?
695	We just might.	我们只是可能。
696	There is a new weapon in the fight	战斗中有新武器
697	against global diseases, the supercomputer.	全球疾病,超级计算机。
698	During the global COVID-19 pandemic, we used them to	在全球COVID-19大流行期间,我们利用它们
699	simulate almost everything from the microscopic world of the virus to the many possible ways to	从病毒的显微世界模拟几乎所有可能的方法
700	treat the infection, understand how the virus spreads, or predict the future course of the	治疗感染,了解病毒的传播方式,或预测病毒的未来走向。
701	pandemic.	艾滋病毒/艾滋病。
702	Using supercomputers, we can look at the spread of the virus as it travels through the	利用超级计算机,我们可以观察病毒的传播过程,因为它穿过
703	air from one person to another and study the effectiveness of masks, air conditioning, and	人与人之间的空气,并研究口罩、空调和
704	social distancing, or move down to the microscopic level and simulate the virus in amazing detail	社会疏远,或移动到微缩水平并模拟病毒的惊人细节
705	to understand how it pirates our cells to reproduce.	了解它是如何盗用我们的细胞繁殖。
706	This coronavirus uses spikes to latch on to human	这种冠状病毒是用尖刺来锁住人类的
707	cells, coating them with compounds called glycan chains to make them invisible to our immune system.	细胞,用称为甘氨酸链的化合物涂上它们,使它们对我们的免疫系统看不见.
708	Computer simulations have revealed chinks in this armour, small regions free of glycans that are	电脑模拟显示,中国人在这个装甲, 小区域,没有甘蔗
709	vulnerable to human antibodies.	易感染人体抗体。
710	Enzymes like this one help coronaviruses infect our cells.	这样的酶能帮助冕病毒感染我们的细胞
711	Scientists can simulate in a matter of hours how this and other targets respond to hundreds of	科学家可以在数小时内模拟这个目标和其他目标如何应对数百个
712	drug candidates to find new antiviral drugs, which may work for new variants, even novel	药物候选人寻找新的抗病毒药物, 这可能会为新的变体,甚至新颖的。
713	human coronaviruses.	人体冠状病毒
714	But a promising treatment can sometimes come at a high cost because viruses	但有希望的治疗有时成本很高因为病毒
715	borrow our biochemistry to multiply, they are hard to combat without also damaging infected cells.	用我们的生物化学来增殖, 它们很难战斗而不破坏感染的细胞。
716	The drugs employed to treat the disease can occasionally cause life-threatening side effects,	用于治疗这种疾病的药物偶尔会造成危及生命的副作用,
717	such as disturbances to the rhythm of the heart.	比如心脏的节奏受到干扰
718	Luckily for us, simulations of human cells,	幸运的是模拟人类细胞
719	tissues, and organs are becoming ever more sophisticated, allowing doctors to explore	组织,器官越来越精密让医生探索
720	the effects of an infection on the entire body.	感染对整个身体的影响。
721	We are entering the era of exascale computers,	我们正进入一个大规模计算机时代,
722	machines a thousand times faster than the previous generation.	机器比上一代快一千倍
723	More powerful machines	更强大的机器
724	combined with new knowledge, artificial intelligence, and smarter software will help	与新知识、人工智能和智能软件相结合,将有助于
725	us understand and fight these silent enemies.	我们理解并打击这些沉默的敌人。
726	When the next pandemic comes along, and it will,	当下一场大流行爆发时,
727	the virtual human will be ready to simulate the virus and its effects,	虚拟人类将准备好模拟病毒及其效应,
728	and devise solutions to the pandemic at pandemic speed.	并设计以大流行病速度应对该大流行病的办法。
729	There we go, I'm done.	好了我受够了
730	Thank you.	谢谢
731	Thank you, Professor Khamenei for such a fascinating talk and with a lot of cutting-edge	谢谢你,哈梅内伊教授这么迷人的谈话和很多尖端
732	information, and also the video is really promising.	信息,以及视频是非常有希望的。
733	Thank you.	谢谢

该视频共有字幕 733 条。解锁更多字幕为会员功能，请移动到价格

下载安装