I was a handsome 12-year-old boy.

我是一個英俊的 12 歲男孩。

I was in a drama competition for national schools in a play written by my best friend.

我參加了全國學校的戲劇比賽，參演的是我最好的朋友寫的劇本。

In that play I was a detective trying to solve a mystery of who robbed a fictional hotel.

在那部劇中，我是一名偵探，試圖解開誰搶劫了一家虛構酒店的謎團。

The hotel was called Hotel El Cheapo Nance Depot.

該酒店名為 El Cheapo Nance Depot 酒店。

I was also a boy with a stammer or speech impediment and I was desperately trying to both remember and say my lines.

我也是一個有口吃或語言障礙的男孩，我拼命地記臺詞、說臺詞。

During that play I spent 70% of my time collecting information to solve this mystery of who robbed the hotel.

在演出期間，我花了 70% 的時間收集資訊，以解開誰搶劫了酒店這個謎團。

So fast forward 26 years and not much has changed.

一晃 26 年過去了，變化不大。

I'm now working as a medical consultant in the hospital for half of my time and as a senior lecturer in applied clinical data analytics in the university for the other half.

我現在一半時間在醫院擔任醫療顧問，另一半時間在大學擔任應用臨床數據分析高級講師。

The context has changed from a detective solving a mystery of who robbed a hotel to a doctor solving a mystery what is the cause of illness in this patient in front of me.

現在的背景已經從偵探破解誰搶劫了一家酒店的謎團，變成了醫生破解我面前這位病人的病因是什麼的謎團。

I still spend 70% of my time collecting information about the patient and only 30% of my time making decisions on that information and communicating with the patient.

我仍然用 70% 的時間收集病人的資訊，只有 30% 的時間根據這些資訊做出決定並與病人溝通。

So the information or data that I collect takes many forms.

是以，我收集的資訊或數據有多種形式。

Patients blood pressure, their medical history, blood test results and this imbalance in how healthcare is delivered this 70-30 is well known all over the world and in many specialties it's been made worse by technology with the introduction of the electronic health record for example that was designed for collecting information about billing and not designed to make healthcare more efficient.

病人的血壓、病史、血液化驗結果，以及醫療服務中的這種不平衡，這種 70-30 的比例在全世界都是眾所周知的，而在許多專科中，隨著技術的發展，這種不平衡變得更加嚴重，例如電子病歷的引入，其目的是為了收集有關賬單的資訊，而不是為了提高醫療服務的效率。

So this extra administrative workload that doctors have to do reduces the face time that they have with patients.

是以，醫生必須承擔的這些額外的行政工作減少了他們與病人面對面的時間。

The face time that we were fundamentally trained to do and the face time that the patients want.

我們從根本上接受了面對面的培訓，而面對面的時間正是病人所需要的。

So the idea worth spreading that I'm going to share tonight is a potential solution to this.

是以，我今晚要分享的這個值得傳播的想法，就是解決這個問題的潛在辦法。

We've all heard a lot about the risks of artificial intelligence but I want to introduce a new perspective.

我們都聽過很多關於人工智能風險的說法，但我想介紹一個新的視角。

One where the responsible use of medical AI could help to solve some of these problems and I'm going to introduce a new type of AI called multimodal AI.

其中，負責任地使用醫療人工智能有助於解決其中的一些問題，我將介紹一種新型人工智能，稱為多模態人工智能。

Multimodal AI is AI that takes data in many different forms, text, images, numbers.

多模態人工智能是一種以文本、影像、數字等多種不同形式獲取數據的人工智能。

When I work as a doctor in the hospital I'm talking to the patient, I'm listening to the patient, I'm listening to their chest with a stethoscope, I'm palpating their abdomen, I'm looking at their blood test results.

我在醫院當醫生時，會與病人交談，傾聽病人的聲音，用聽診器聽他們的胸部，觸診他們的腹部，查看他們的血液化驗結果。

This is multimodal human intelligence.

這就是多模態人類智能。

Multimodal meaning lots of different types of data.

多模態是指許多不同類型的數據。

So in November last year the media coverage of AI really exploded with the release of chat GPT by open AI.

是以，去年 11 月，隨著開放式人工智能（open AI）發佈哈拉 GPT，媒體對人工智能的報道真正爆炸了。

So chat GPT is a type of AI called a large language model or generative AI.

是以，哈拉 GPT 是一種人工智能，被稱為大型語言模型或生成式人工智能。

But it's not the only type of AI.

但人工智能並非只有這一種。

There's other types of AI that are less familiar to the general public like machine learning, computer vision, natural language processing.

還有其他類型的人工智能，如機器學習、計算機視覺、自然語言處理等，不太為大眾所熟悉。

And those AI's mostly take in a single type of data and we call this single modal AI.

這些人工智能主要接收單一類型的數據，我們稱之為單一模式人工智能。

So I'm going to give three cutting-edge examples of single modal AI in healthcare.

是以，我將列舉三個醫療領域單一模式人工智能的前沿案例。

The image behind me is an image of a chest x-ray with the heart in the center, surrounded by the lungs, the ribs going across the shoulders at the top.

我身後的圖片是一張胸部 X 光片，中間是心臟，周圍是肺部，肋骨從頂部穿過肩膀。

AI has become really good at distinguishing normal from abnormal.

人工智能已經非常擅長區分正常與異常。

We call this triaging.

我們稱之為分流。

And most x-rays done in the world are actually normal.

而實際上，世界上大多數 X 光片都是正常的。

So this software called ChestLink by a company called OxyPit, it's a medical AI triage system.

這款名為 ChestLink 的軟件由一家名為 OxyPit 的公司開發，是一款醫療人工智能分診系統。

And it's the first system that's received regulatory or CE approval to be used in a fully autonomous way to report on chest x-rays.

這也是首個獲得監管或 CE 認證的系統，可以完全自主地報告胸部 X 光片。

So what OxyPit does is it looks for 75 abnormalities on the chest x-ray.

是以，OxyPit 的作用就是在胸部 X 光片上尋找 75 個異常點。

And if it doesn't find any of those abnormalities, it reports the x-ray as normal without any human involvement.

如果沒有發現任何異常，它就會報告 X 光片正常，無需人工參與。

If it does find an abnormality, it passes the x-ray back to the human radiologist to report.

如果發現異常，它就會將 X 光片傳回人類放射科醫生進行報告。

This is an example of task sharing between the AI and the human radiologist.

這是人工智能與人類放射科醫生分擔任務的一個例子。

The picture behind me is a retina.

我身後的照片是視網膜。

This is the tissue at the back of your eye.

這是眼球后部的組織。

If you've ever been for an eye test, this is what the optician sees.

如果您曾經做過眼科檢查，這就是驗光師看到的情況。

The optician is looking for reversible causes of blindness like macular degeneration.

驗光師正在尋找可逆的致盲原因，如黃斑變性。

A group of researchers in University College London developed an AI model trained on 1.6 million pictures of the retina.

倫敦大學學院的一組研究人員根據 160 萬張視網膜圖片開發了一個人工智能模型。

This model is able to diagnose eye disease and predict outcomes from eye conditions like macular degeneration.

該模型能夠診斷眼部疾病，並預測黃斑變性等眼部疾病的預後。

And that's very impressive that it can do what most non-specialist doctors find hard, but it doesn't stop there.

令人印象深刻的是，它能做到大多數非專業醫生都難以做到的事情，但它並不止於此。

If we think about a condition like Parkinson's disease, we don't think about the back of the eye.

如果我們想到帕金森病這樣的疾病，我們不會想到眼球后部。

Parkinson's disease affects your movement, causes a tremor, affects your walking.

帕金森病會影響你的運動，導致顫抖，影響你的行走。

The same AI model can look at the back of your eye and predict Parkinson's disease years before patients develop symptoms.

同樣的人工智能模型可以通過觀察眼球后部，在患者出現症狀前幾年就預測出帕金森病。

So now not only can it see what the human can see, but it can see things that the human can't see.

是以，現在它不僅能看到人類能看到的東西，還能看到人類看不到的東西。

However, this model will never diagnose Parkinson's disease, and it definitely will never give compassionate care for Parkinson's disease.

然而，這種模式永遠無法診斷帕金森病，也絕對無法為帕金森病提供體貼入微的護理。

AI like this must be used in conjunction with highly trained healthcare professionals.

這樣的人工智能必須與訓練有素的醫療保健專業人員配合使用。

I'm going to move away from computer vision towards large language models.

我將從計算機視覺轉向大型語言模型。

In December of last year, Google released a medical large language model called MedPAM.

去年 12 月，谷歌發佈了一個名為 MedPAM 的醫學大型語言模型。

They trained their generic large language model called PAM to perform medical question answering.

他們訓練了名為 PAM 的通用大型語言模型來進行醫學問題解答。

And this is the first time ever that a computer or an AI model has passed a US medical licensing exam with a passing score of 67%.

這也是計算機或人工智能模型有史以來第一次以 67% 的及格分數通過美國醫學執照考試。

And then only three months later, MedPAM2, the next version, got a score of 86%.

僅僅三個月後，下一個版本 MedPAM2 的得分率就達到了 86%。

This is expert level on that exam.

這是該考試的專家級水準。

If you have a smartphone in your pocket, multimodal AI is available to you right now.

如果你的口袋裡有一部智能手機，你現在就可以使用多模態人工智能。

Four weeks ago, OpenAI released a multimodal version of ChatGPT.

四周前，OpenAI 發佈了多模態版本的 ChatGPT。

And this is an example I gave us last week, where I pass in a picture of an ECG.

這是我上週給大家舉的一個例子，我把一張心電圖的圖片傳了進去。

This is the electrical activity of the heart, a very little scenario. 60-year-old male presented with palpitations.

這就是心臟的電活動，一個非常小的場景。60 歲男性出現心悸。

That's a sensation of your heart beating in your chest.

那是一種心臟在胸腔中跳動的感覺。

He could feel his heart skipping beats.

他能感覺到自己的心跳在加速。

No past medical history, not currently on medications.

無既往病史，目前未服用藥物。

The attached picture is ECG.

附圖為心電圖。

What is the next step for this patient?

病人下一步該怎麼辦？

Now, very unhelpful for this presentation, ChatGPT told me that it's a machine learning model and not a device.

ChatGPT 告訴我，它是一個機器學習模型，而不是一個設備。

So I asked it to help a friend out and told it I'm doing a TEDx talk about multimodal AI and please play along.

所以我讓它幫我朋友的忙，告訴它我正在做一個關於多模態人工智能的 TEDx 演講，請一起玩。

And it did exactly that.

它確實做到了這一點。

Now, although the ECG analysis wasn't perfect, it was very, very close.

現在，雖然心電圖分析並不完美，但已經非常非常接近了。

And the follow-up advice that it gave was perfect.

它給出的後續建議也非常完美。

But this gets even better when multimodal large language models are trained on medical tasks.

但是，當多模態大型語言模型在醫療任務中進行訓練時，效果會更好。

And the best example of this is MedPamM, M for multimodal, released by Google in July of this year.

最好的例子就是谷歌今年 7 月發佈的 MedPamM，M 代表多模態。

MedPamM takes multiple different types of input, pictures of the skin, pictures of chest x-rays, pictures of pathology, texts from radiology images, and performs multiple medical tasks.

MedPamM 可接收多種不同類型的輸入，包括皮膚圖片、胸部 X 光片圖片、病理圖片、放射影像文本，並執行多種醫療任務。

So it's not perfect, but the radiology report that generated from MedPamM was compared to a human radiologist report.

是以它並不完美，但 MedPamM 生成的放射學報告與人類放射科醫生的報告進行了比較。

And the blinded assessors preferred the MedPamM report in 40% of cases.

在 40% 的病例中，盲法評估員更傾向於使用 MedPamM 報告。

So the things we need to implement multimodal AI safely are trust, explainability, and randomized clinical trials.

是以，要安全實施多模態人工智能，我們需要信任、可解釋性和隨機臨床試驗。

In relation to trust, there was a survey done in the United States, and over half of the respondents would feel anxious if they knew their healthcare worker was relying on AI for their assistance, feared that doctors were going to integrate AI too quickly before understanding the risks to patients.

在信任方面，美國曾做過一項調查，超過半數的受訪者如果知道他們的醫護人員正在依賴人工智能提供幫助，就會感到焦慮，擔心醫生在瞭解對患者的風險之前就過快地整合人工智能。

So we have a lot of work to do to bridge this gap.

是以，要彌補這一差距，我們還有很多工作要做。

The second thing is explainability.

第二點是可解釋性。

Explainable AI opens up the black box to tell us why it met the output.

可解釋的人工智能打開了黑箱，告訴我們它為什麼會達到輸出結果。

So in our research, what we're interested in is why did the AI model pick a particular blood pressure medication for high blood pressure?

是以，在我們的研究中，我們感興趣的是，為什麼人工智能模型會選擇特定的降壓藥來治療高血壓？

Should we just go along with what the model says, or do we want to know why it got to that conclusion?

我們是應該順著模型的思路走，還是想知道它為什麼會得出這樣的結論？

If the output from the model agrees with our worldview, then we might just go along with it and not question it.

如果模型的輸出結果與我們的世界觀一致，那麼我們可能就會順從它，而不會質疑它。

And that's a very risky area in medicine called confirmation bias.

這在醫學上是一個非常危險的領域，叫做確認偏差。

The third thing we need is randomized clinical trials.

我們需要的第三件事是隨機臨床試驗。

AI models must be the same way that we test medicines.

人工智能模型必須與我們測試藥物的方法相同。

For medicines, we use randomized control trials.

在藥品方面，我們採用隨機對照試驗。

This is the peak of evidence in medicine.

這是醫學證據的頂峰。

One group receiving the AI model, another group not receiving the AI model, and follow them up to see who does better.

一組接受人工智能模型，另一組不接受人工智能模型，並跟蹤觀察誰做得更好。

So what's the missing piece?

那麼，缺少的是什麼呢？

Where does the art of first look at the patient?

病人的第一眼藝術在哪裡？

You never interpret a result, an X-ray, an ECG, without knowing the context of the patient.

在不瞭解病人背景的情況下，你永遠無法解釋一個結果、X 光片或心電圖。

We often call this the eyeball test.

我們通常稱之為 "眼球測試"。

And this has been tested where patients coming to emergency departments, nurses would triage them as red, yellow, or green from just looking at the patient.

這種方法已經在急診科進行過測試，護士只需看一眼病人，就能將其分為紅色、黃色或綠色。

And this was shown to be more accurate than sophisticated models.

事實證明，這比複雜的模型更準確。

So in the future, I see a world where a picture or a video of the patient is also fed into the multimodal model.

是以，在未來，我認為病人的照片或視頻也會被輸入多模態模型。

So now, looking back at myself, my 12-year-old self, standing nervously on that stage, I can see the parallels between then and now.

現在，回想我自己，12 歲的自己，緊張地站在舞臺上，我可以看到當時和現在的相似之處。

Back then, I was looking for data to solve the mystery of who robbed the fictional hotel.

當時，我正在尋找數據，以解開是誰搶劫了這家虛構酒店之謎。

But now we're looking for data for a more profound reason, and that's to make healthcare more efficient, personalized, and accessible.

但現在，我們尋找數據有了更深層次的原因，那就是讓醫療保健更高效、更個性化、更便捷。

Imagine a world where remote corners of low- and middle-income countries that have no access to specialized care can gain insights from these models.

想象一下，在低收入和中等收入國家的偏遠角落，那些無法獲得專業護理的人可以從這些模式中獲得啟示。

That's a world the medical AI, and especially multimodal medical AI, can help us create.

醫療人工智能，尤其是多模態醫療人工智能，可以幫助我們創造這樣一個世界。

So as we look to the future, we have to prioritize compassion and understanding.

是以，展望未來，我們必須把同情和理解放在首位。

We have to build this relationship between AI and the humans to allow the doctors more time to spend with the patients, to understand them, and give them a better chance at health and happiness.

我們必須在人工智能和人類之間建立這種關係，讓醫生有更多的時間與病人相處，瞭解他們，讓他們有更好的機會獲得健康和幸福。

Thank you.

謝謝。