Tackling Dementia with AI Powered Eye Care

around 200 years ago

london became the largest city in the

world this was an exciting time

the first railways were introduced the

national gallery was built

and the city established itself as a

global hub of trade

however it was a place tormented by

frequent epidemics

some of these were well recognized

infections of the day

like smallpox or cholera but one of them

had doctors scratching their heads

hundreds of patients were being admitted

with unexplained headaches

vomiting seizures and swelling of the

legs

some of them strangely were also going

blind

often hours before their death now it

might surprise you to hear

that the culprit of this was no

infection

rather it was something that we couldn’t

yet detect

but affect many of us to some extent

today

high blood pressure what dr richard

bright revealed to us

in these descriptions was perhaps one of

the earliest hints

that the eyes could harbor diseases

outside of the eyes

the problem was that technology hadn’t

kept up with scientific understanding

and we had to wait several decades

before the invention of

this the ophthalmoscope when we could

reliably distinguish what was causing

vision loss in these patients

so if you were to look at the back of my

eye the retina

this is what you’d see a circular pink

structure

of the optic nerve from which emanate

the major blood vessels of the retina

what might richard bright have seen if

he was able to examine his patients eyes

well probably something like this now on

first glance

these images might appear quite similar

but look closer

and you’ll notice red blots suggestive

of bleeding

fluffy yellow areas indicating swelling

or blocked nerves and severe cases

a lack of clarity of the border of the

optic nerve

suggestive of swelling of the brain fast

forward almost 200 years

and we know that our eyes show much more

than just the state of our blood

pressure

changes in the retina can reflect a

multitude of diseases

from common conditions like diabetes

to uncommon infections like tuberculosis

in fact you’d be hard-pressed to think

of a chronic disease

that doesn’t have a manifestation

somewhere within the eye

now it doesn’t take an expert to look at

these two images

and realize that they’re different but

not all

changes are as clearly discernible to us

what if i were to instead show you

another pair

of seemingly normal images but tell you

that one of these patients

will develop a stroke within the next

three years and that this prediction can

be made through the eye

it’s unlikely that any of us would be

much more accurate than tossing a coin

one option for identifying structures

that may not be obvious to us

is deep learning a subfield of

artificial intelligence

where patterns are discovered using

large datasets

with this tool we’ve realized that

images of the retina don’t just tell us

about disease

they can tell us about ourselves

in the last few years we’ve seen that

eyes can show

our age our sex whether we smoke

whether we’re anemic and give us an

indication of our risk of future events

like heart attack or stroke all from the

single picture

of the back of the eye now the focus on

heart attack and stroke is deliberate

collectively these so-called

cardiovascular diseases

are the leading cause of mortality

worldwide

but many of us will have personal

experience of a new villain

that has emerged as the number one cause

of death in the united kingdom

and that is dementia dementia is rising

at an alarming rate

in the context of the aging population

and even more concerning is that it’s

estimated

that around 50 percent of cases are

undiagnosed globally

so it’s important that we develop

effective strategies

for identifying those at risk modern

scanning technology

has not only redefined how we treat eye

diseases

but they’ve helped further our

understanding of the links

between the eye and the brain

the retina is derived from the same

tissue as the brain

and the circuitry connecting the two

means that changes in one

are often reflected in the other for

example

one landmark study from the netherlands

showed that thinner nerves

were found in the retinas of people with

dementia

and also several years before developing

any symptoms of it they were able to

show this with less than 100 patients

but applying deep learning to this

problem requires much larger data sets

and that’s why we spent the last couple

years at moorefields

the largest eye hospital in europe and

north america

linking our retinal scans with national

hospital admissions data our outside

cohort comprises more than

5 million retinal scans and

15 000 patients in our cohort have

developed dementia

the prospect of eye scans for dementia

screening is attractive

given the challenge of diagnosis this

typically relies on symptoms

which are a late manifestation of the

disease

all the concerns of a friend or family

member

and awareness is only the beginning

along this arduous path

individuals first need to consult their

general practitioner

complete a non-specific questionnaire

be referred to a specialist and have

blood tests to rule out

secondary causes of dementia finally

they need to undergo scans

which are often expensive time consuming

and potentially invasive all before

revisiting their specialist

to discuss the results all this

invariably leads to frustration

and anxiety but high resolution eye

scanning technology

can show us even subtler changes with

microscope-like resolution

in a non-invasive risk-free approach

that takes seconds to acquire

and much like the field of genetics

where sequencing a person’s genome

is a fraction of the cost it was 10

years ago

each season sees greater affordability

in retinal scanning technology to put

this into perspective

at more fields we did around 25 000 eye

scans

in the 12 months of 2008 we now do more

than that

each month and this technology isn’t

just limited

to the hospital or clinic it’s

increasingly available in the community

for example specsavers one of the

world’s largest optician businesses

now has high resolution scanning in

nearly 700 of its branches in the united

kingdom

but if eye scans are a potential

solution

how do we actively encourage the public

to consider having them

well vision is precious surveys

consistently show that we treasure our

sight

above any other sense some

are even willing to trade five years of

life

to remain with their sight intact and

this is not just an opinion

this powerful sentiment translates into

our health-seeking behavior

and here’s an example when you turn 40

chances are that you’ll be invited to a

general health check

where among other things your risk of

having a heart attack or stroke

will be estimated but how many of us

actually take that up

between 2009 and 2013

in the united kingdom it was around nine

percent

but what if we consider a similar age

group

similar time but instead rephrase the

question

how many do you think reports seeing

their optometrist at least

every four years a very telling

80 percent the current landscape of

community healthcare

and technological development has us at

an exciting crossroads

and not just in europe and north america

but low and middle-income countries

are also undergoing infrastructural

transformation

with the introduction of affordable eye

scanning technology

and regional eye health programs some of

these scans can even be taken

using your standard smartphone by

combining our understanding

of how the eyes change in systemic

diseases

with artificial intelligence the growing

availability of eye scanning technology

and our natural concern for our vision

we can democratize

access to a comprehensive health

evaluation

not through a visit to your primary care

doctor

blood tests or a long wait for that

hospital mri scan

but through a single picture taken in

the blink of an eye

you

大约 200 年前,

伦敦成为世界上最大的城市,

这是一个激动人心的

时刻,第一条铁路被引入,

国家美术馆建成

,这座城市确立了自己作为

全球贸易中心的地位,

但它是一个经常受到流行病折磨的地方,

其中一些 天花或霍乱等当时公认的

感染,

但其中一个

让医生挠头

听到

这件事的罪魁祸首不是感染,您可能会感到惊讶,

而是我们

尚未发现

但在一定程度上影响了我们今天许多人的

高血压,理查德·布莱特博士

在这些描述中向我们揭示的可能是其中之一

最早的迹象

表明眼睛可能在

眼睛之外有疾病

,问题在于技术 没有

跟上科学的理解

,我们不得不等待几十年

才能

发明检眼镜,当我们能够

可靠地区分

导致这些患者

视力丧失的

原因时 是你会看到的视神经的粉红色圆形

结构

,从中散发

出视网膜的主要血管

如果理查德·布莱特

能够很好地检查他的病人的眼睛,他可能会看到

什么 乍一看这些图像可能是这样的

可能看起来非常相似,

但仔细观察

,您会发现红色斑点

表明出血

蓬松的黄色区域表明

神经肿胀或阻塞,严重的

情况 视神经边界不清晰

表明大脑肿胀快

进近 200 年

我们知道,我们的眼睛所显示的

不仅仅是我们

视网膜中血压变化的状态,它还可以反映

多种

疾病 从糖尿病等

常见疾病到肺结核等罕见感染

事实上,您很难

想到一种在眼睛某处没有表现的慢性疾病

现在不需要专家来查看

这两个图像

和 意识到它们是不同的,但

并不是所有的

变化对我们都一样清晰可辨

预测

可以通过眼睛

进行 我们中的任何人都不太可能

比掷硬币更准确

识别

对我们来说可能并不明显的结构的一种选择

是深度学习 人工智能的一个子领域,

其中使用

大型数据

集发现模式 这个工具我们已经意识到,

视网膜的图像不仅可以告诉

我们疾病,

它们还可以告诉我们关于我们自己的信息

。在过去的几年里,我们已经

看到 是的 可以显示

我们的年龄 我们的性别 我们是否吸烟

我们是否贫血 并告诉

我们未来

发生心脏病或中风等事件的风险 所有这些都来自

眼睛后部的单张图片 现在关注

心脏病发作和中风 故意将

这些所谓的

心血管疾病

统称为全球死亡的主要原因

但我们中的许多人都会亲身

经历一个新的恶棍

,该恶棍已成为英国的头号死因

,即痴呆症痴呆症正在上升

在人口老龄化的背景下,一个惊人的速度

,更令人担忧的是,据

估计

,全球约有 50% 的病例

未被诊断出来,

因此我们制定

有效的策略

来识别处于危险中的人非常重要现代

扫描

技术不仅重新定义了我们的治疗方式 眼部

疾病,

但它们帮助我们进一步

了解

了眼睛和大脑之间的联系,而大脑

是视网膜的来源 与

大脑相同的组织

和连接两者的电路

意味着其中一个的变化

经常反映在另一个中,

例如

来自荷兰的一项具有里程碑意义的研究

表明,

在痴呆症患者的视网膜中发现了较细的神经

,而且在发展前几年也发现了神经

任何症状他们都能够在

不到 100 名患者中证明这一点,

但将深度学习应用于这个

问题需要更大的数据集

,这就是为什么我们在过去

几年里

在欧洲和北美最大的眼科医院 moorefields

连接我们的视网膜 使用国家

医院入院数据进行扫描 我们的外部

队列包括超过

500 万次视网膜扫描

,我们队列中的 15,000 名患者患有

痴呆症 鉴于诊断的挑战,通常依赖于晚期症状的眼部扫描用于痴呆症筛查的前景

是有吸引力

疾病的表现

朋友或家人的所有担忧

在这条艰巨的道路上,意识只是开始

个人首先需要咨询他们的

全科医生

完成一份非特定的问卷

被转介给专家并进行

血液检查以排除

痴呆的继发性原因 最后

他们需要接受

通常昂贵的扫描

在重新拜访他们的专家讨论结果之前,这一切都非常耗时且具有潜在的侵入性

,所有这些

总是会导致沮丧

和焦虑,但高分辨率眼部

扫描技术

可以通过非侵入性、无风险的方法向我们展示更细微的变化,

类似于显微镜的分辨率

几秒钟内获得

,就像遗传学领域一样

,对一个人的基因组进行测序

的成本只是 10 年前的一小部分,

每个季节都看到

视网膜扫描技术的可承受性更高,以便

在我们进行了大约 25,000 次眼部扫描的更多领域中看到这一点

在 2008 年的 12 个月里,我们现在每个月都做的比这更多

技术不仅

限于医院或诊所,它

在社区中的应用越来越广泛

,例如 specsavers

世界上最大的眼镜商之一

现在

在其英国近 700 家分支机构拥有高分辨率扫描,

但如果眼部扫描具有潜力

解决方案

我们如何积极鼓励

公众考虑让他们

拥有良好的视力是宝贵的 调查

一致表明,我们珍视我们的

视力胜过任何其他感觉 有些

人甚至愿意牺牲五年的

生命

来保持他们的视力完好,

这不仅仅是一个 认为

这种强烈的情绪会转化为

我们寻求健康的行为

,这里有一个例子,当你年满 40 岁时,你有

机会被邀请参加

一般健康检查

,其中包括估计你

患心脏病或中风的风险,

但如何 我们中的许多人

实际上

在 2009 年至 2013 年间

在英国采取了这一比例,大约为 9

%,

但如果我们反对呢? 类似的年龄

组在

类似的时间,而是重新表述这个

问题

,你认为有多少报告至少每四年看到一次

他们的验光师,

非常有说服力的

80% 当前

社区医疗保健

和技术发展的前景让我们处于

一个令人兴奋的十字路口

,而不仅仅是在 欧洲和北美,

但中低收入国家

也在进行基础设施

转型

,引入了负担得起的眼部

扫描技术

和区域眼部健康计划,结合我们对眼睛如何变化的理解,其中

一些扫描甚至可以

使用您的标准智能手机进行

在具有人工智能的系统性

疾病中

,眼部扫描技术的日益普及

以及我们对视力的自然关注,

我们可以民主化

获得全面的健康

评估,

而不是通过拜访您的初级保健

医生进行

血液检查或长时间等待该

医院的核磁共振扫描,

而是 通过一张图片 眨眼间拍到