Bart Weetjens 談如何訓練巨鼠嗅出地雷所在


==========================
講者:Bart Weetjens
2010年6月演講,2010年12月在TEDxRotterdam上線
MyOOPS開放式課程
翻譯: 劉契良
編輯:洪曉慧
簡繁轉換 :趙弘
後制:劉契良
字幕影片後制:謝旻均
--------------------------
Bart Weetjens 談如何訓練巨鼠嗅出地雷所在
今天我要和各位分享一趟神奇的旅程,確切地來說,是趟大有所獲的旅程,那段旅程將我帶到訓練巨鼠的工作上,目的是要拯救人命,方法是偵測地雷以及結核病,還是小童時,我對兩事充滿熱情,一是齧齒目動物,我搜集各種的鼠類,家鼠、倉鼠、沙鼠、松鼠,應有盡有,我加以繁殖並賣回寵物店(笑聲);另一項熱情是非洲,我在多元文化的環境中長大,我們家有寄宿的非洲學生,我聆聽他們的故事,那麼不同的背景,依賴進口的技能、商品和服務,生氣勃勃的多元文化,非洲真讓我著迷,我後來成為一名工業工程師,負責產品開發,我專注在合宜的偵測技術,確切地說是第一套合宜的技術,可應用於開發中國家,我加入那項產業,但並不開心,因為我所做的全獻給一個物質消費的社會,且是以一種線性、榨取和生產的模式進行,我辭去工作,轉向真正的世界問題,地雷。
那時是 1995 年,戴安娜王妃在電視宣告地雷形成一道結構性障礙,阻隔任何發展,而那是事實,只要那些地雷還在或仍有受其威脅的疑慮,我們就進不了那片土地,事實上,有個全球性的訴願,主旨是開發新偵測器,使環境永續,以利繁衍所需,主要是在開發中的世界,我們選擇巨鼠,為何選擇巨鼠?牠們不是害蟲嗎?嗯,事實上,巨鼠是,並非大多數人所想的那樣,巨鼠是極具社交性的生物,事實上,我們的產品,如螢幕所示,那裡有個標的物,影片中是一位受過訓練的非洲技工,他的巨鼠當先鋒,牠東找西找,找到一顆地雷了,牠開始挖土,然後牠便回頭討賞,極簡單的流程且有利環境永續發展,巨鼠得賞,大功告成,極簡單的流程。但為何要用巨鼠?鼠類從上個世紀 50 年代開始即被用於各種實驗中,鼠類基因中有更多的嗅覺天賦,對比於任何哺乳類物種,牠們的嗅覺極為敏銳,此外,牠們還具有比對這些味道的機制並能加以溝通,我們又如何與巨鼠溝通呢?我們並不會說鼠語,但我們有個溝通用的發聲器,一套標準的動物訓練方法,如圖所示,發聲器會發出特別的聲音,強化一些特定的行為,首先,我們將發出的聲音與食物連結,那可是美味的香蕉泥加花生及蜜糖,一旦牠們習慣該種聲音=食物、聲音=食物、聲音=食物、聲音=食物,我們便將牠們放入一個有洞的籠子,牠們真的學會將鼻子靠進洞裡,在那裡我們設了一種氣味標的,過程持續五秒,五秒,對牠們就足夠了,一旦讓牠們學會,我們便將任務的困難度提高,牠們學會找出氣味標的,在一個鑽有數個洞的籠子,洞可多達 10 個,之後,牠們還會學習戴著鏈圈走出戶外找出標的物,下一步,牠們學會在真實地雷區中找出真的地雷,牠們都受到檢測與認證,依據國際鼠類行動標準,就像狗要通過檢測一樣,範圍廣達 400 平方公尺,多顆地雷被隨意地埋置,而訓練師與他們的巨鼠必需找出所有的標的物,如果牠們通過檢測,便會得到一張證書,成為認證過的動物,能真槍實彈上場,就像狗一樣,唯一的差別是訓練巨鼠的成本僅是訓練掃雷犬的五分之一。
這是我們在莫三比克的團隊,一位坦尚尼亞訓練師將技能轉移給這三位莫三比克的同事,請注意看他們眼中泛出的自豪,他們擁有一項技能,而那能讓他們較不需依賴外援,另外,這支小團隊,當然,還是需要重機和一個人工的後續掃雷動作,但這項花在老鼠能耐上的小投資,我們已在莫三比克證實,每平方公尺的成本能降低60 %,對比於現時的行情,那是每平方公尺兩塊美金,我們只需 1.18 塊,我們還能再降,端看需求量,如果能獲得更多的巨鼠,我們便能將產能再提高,我們在莫三比克有個示範站,11 個非洲政府已看到他們能減少依賴,方法是善用這個技術,他們已在大湖區簽定了和約與協定,他們贊同讓「巨鼠英雄軍」去掃除共同邊界上的地雷,但我要指出一個非常不同的問題,去年約有 6,000 人誤踩地雷,但去年全球有幾近 190 萬人死於主要由結核病引起的感染,特別是在非洲,當地結核病與 HIV 具有明顯關聯,那是一個普遍的大問題,顯微鏡檢查是世衛的標準程序,可靠度僅達 40 到 60 個百分比,在坦尚尼亞,數字不說謊,45 %的人,結核病人在死前被診出患有結核病,意即,如果一人患有結核病,更大的可能將是診斷不出來,只會因為結核病併發症等而死,但如果及早偵測、診斷出來,便能開始治療,就算是 HIV 陽性,也同理可證,結核病可治癒,就算是 HIV 陽性也能,荷蘭語的結核病講法是Tering,從詞源來看,那意指柏油味,古時的中國人及希臘人,如希波克拉底(古希臘名醫)即已寫到,記載結核病能藉由病人發散出的揮發氣味診測出來,我們所做的是,收集一些樣本,檢測方法一致,樣品由醫院提供,我們用以訓練巨鼠,觀察是否可行,令人驚奇的是,我們的巨鼠達到 89%的靈敏度,86%的鑑別度,而且是多隻巨鼠共同進行,影片秀出訓練過程。
這真是一項共性技術,我們目前僅將其應用於爆炸物與結核病,但試想,其實任何應用都有可能,但如何訓練?將 10 種氣味樣本放入卡匣中,立即將這 10 種樣本放入籠中,巨鼠僅需二百分之一秒即能分辨出不同的氣味,過程極速,影片中,牠已來到第三個樣本,那即是標的樣本,牠在聽到給賞的聲音便前去領賞,這樣神速的過程,我們可將其應用成另一種選項以檢視病人是陽性,還是陰性,要指出的是,顯微鏡師能處理的工作量是每天 40 個病例,而一隻巨鼠處理相同份量的時間僅需七分鐘,在像這樣的籠子裡(掌聲),在像這樣的籠子裡,只要擁有巨鼠,而我們現時擁有25 隻「結核鼠」,在像這樣的籠子裡,牠們一整天的工作量可處理 1,680 個病例,試想潛在的未來應用,環境偵測、土壤污染、客製應用、偵測貨櫃中非法貨物等,但目前先針對結核病,我要簡短地點出,表中藍色柱狀圖是靠顯微鏡所得僅五間診所的百分比,地點是達累斯薩拉姆(坦桑尼亞共和國首都),人口是 50 萬,其中,據報有 15,000 人已完成檢測,顯微鏡檢出了 1,800 位病人,但只要再次將樣本拿給巨鼠嗅聞,再看一次結果,我們能將病例檢出率提高到超過 30 個百分比,去年一整年,但端視所取間隔,我們已持續提高病例檢出率,地點就在達累斯薩拉姆的五間醫院中,百分比是 30% 到 40%,所以那是很可觀的比例,要知道的是顯微鏡只要漏測一人,那就會每年再使另外 15 人受到感染,健康的人將受感染,可以確定的是,我們已拯救了眾多性命,至少是我們的「巨鼠英雄軍」已拯救了眾多性命,我們目前的願景是標準化這項技術,一些小細節,像是,我們在嗅聞洞中裝個小鐳射,讓牠們能嗅上五秒,將這些標準化,同時也是標準化蕉泥球,食物獎賞,使其半自動化,以符合較大規模的運營,並拯救更多人的性命。
總結,還有很多層面的應用,這是我們的「攝影鼠」原型,讓巨鼠穿上鼠包,附裝一支鏡頭,讓牠可以爬進瓦礫堆偵測受難者,應用於像地震之類的情況,現在還只是原型階段,還未形成運作系統,最後總結來說,各位可能會以為這些專案是有關巨鼠,但其實是有關人們,這是有關賦予脆弱社群力量以對抗困難、昂貴和危險的人道偵測任務,同時,使用的是當地的資源,源源不絕的資源,完全不同的是,持續挑戰你的認知,關於周圍的資源,不論那是環境、技術、動物或人類,尊敬地與其和諧共處,以促進永續的志業,感謝聆聽。
(掌聲)
============================
以下為系統擷取之英文原文
About this talk
At TEDxRotterdam, Bart Weetjens talks about his extraordinary project: training rats to sniff out land mines. He shows clips of his "hero rats" in action, and previews his work's next phase: teaching them to turn up tuberculosis in the lab.

About Bart Weetjens

Bart Weet­jens is developing a new approach to the land ­mine detection problem: train­ing giant pouched rats to detect explosives in minute amounts. Full bio and more links

Transcript

I'm here today to share with you an extraordinary journey -- extraordinarily rewarding journey, actually -- which brought me into training rats to save human lives by detecting landmines and tuberculosis. As a child, I had two passions. One was a passion for rodents. I had all kinds of rats, mice, hamsters, gerbils, squirrels. You name it, I bred it, and I sold them to pet shops. (Laughter) I also had a passion for Africa. Growing up in a multicultural environment, we had African students in the house, and I learned about their stories, [such] different backgrounds, dependency on imported know-how, goods, services, exuberant cultural diversity. Africa was truly fascinating for me.

I became an industrial engineer -- engineer in product development -- and I focused on appropriate detection technologies, actually the first the first appropriate technologies for developing countries. I started working in the industry, but I wasn't really happy to contribute to a material consumer society in a linear, extracting and manufacturing mode. I quit my job to focus on the real world problem: landmines. We're talking '95 now. Princess Diana is announcing on TV that landmines form a structural barrier to any development, which is really true. As long as these devices are there, or there is suspicion of landmines, you can't really enter into the land. Actually, there was an appeal worldwide for new detectors sustainable environments where they're needed to produce, which is mainly in the Developing World. We chose rats.

Why would you chose rats? Because, aren't they vermin? Well, actually rats are -- contrary to what most people think about them -- rats are highly sociable creatures. And actually, our product -- what you see here. There's a target somewhere here. You see an operator, a trained African with his rats in front who actually are left and right. There the animal finds a mine. It scratches on the soil. And the animal comes back for a food reward. Very, very simple. Very sustainable in this environment. Here the animal gets its food reward. And that's how it works. Very, very simple.

Now why would you use rats? Rats have been used since the 50's last century in all kinds of experiments. Rats have more genetic material allocated to olfaction than any other mammal species. They're extremely sensitive to smell. Moreover, they have the mechanisms to map all these smells and to communicate about it. How do we communicate with rats? Well don't talk rat, but we have a clicker, a standard method for animal training, which you see there. A clicker, which makes a particular sound with which you can reinforce particular behaviors. First of all, we associate the click sound with a food reward, which is smashed banana and peanuts together in a syringe. Once the animal knows click, food, click, food, click, food -- so click is food -- we bring it in a cage with a hole, and actually the animal learns to stick the nose in the hole under which a target scent is placed, and to do that for five seconds -- five seconds, which is long for a rat. Once the animal knows this, we make the task a bit more difficult. It learns how to find the target smell in a cage with several holes, up to 10 holes.

Then the animal learns to walk on a leash in the open and find targets. In the next step, animals learn to find real mines in real minefields. They are tested and accredited according to international mine action standards, just like dogs have to pass a test. This consists of 400 sq. meters. There's a number of mines placed blindly. And the team of trainer and their rat have to find all the targets. If the animal does that, it gets a license to be an accredited animal to be operational in the field -- just like dogs, by the way. Maybe one slight difference: we can train rats at a fifth of the price of training the mining dog.

This is our team in Mozambique. One Tanzanian trainer, who transfers the skills to these three Mozambican fellows. And you should see the pride in the eyes of these people. They have a skill, which makes them much less dependent on foreign aid. Moreover, this small team together with, of course, you need the heavy vehicles and the manual deminers to follow-up. But with this small investment in a rat capacity, we have demonstrated in Mozambique that we can reduce the cost-price per sq. meter up to 60 percent. of what is currently normal -- two dollars per sq. meter, we do it at 1.18, and we can still bring that price down. Question of scale. If you can bring in more rats, we can actually make the output even bigger. We have a demonstration site in Mozambique. 11 African governments have seen that they can become less dependent by using this technology. They have signed the pact for peace and treaty in the Great Lakes region. And they endorse hero rats to clear their common borders of landmines.

But let me bring you to a very different problem. And there's about 6,000 people last year that walked on a landmine, but worldwide last year, almost 1.9 million died from tuberculosis as a first cause of infection. Especially in Africa where T.B. and HIV are strongly linked, there is a huge common problem. Microscopy, the standard WHO procedure, reached from 40 to 60 percent reliability. In Tanzania -- the numbers don't lie -- 45 percent of people -- T.B. patients -- get diagnosed with T.B. before they die. It means that, if you have T.B., you have more chance that you won't be detected, but will just die from T.B. secondary infections and so on. And if, however, you are detected very early, diagnosed early, treatment can start. And even in HIV-positives, it makes sense. You can actually cure T.B., even in HIV-positives.

So in our common language, Dutch, the name for T.B. is tering, which, etymologically, refers to the smell of tar. Already the old Chinese and the Greek, Hippocrates, have actually published, documented, that T.B. can be diagnosed based on the volatiles exuding from patients. So what we did is we collected some samples -- just as a way of testing -- from hospitals, trained rats on them and see if this works, and wonder, well, we can reach 89 percent sensitivity, 86 percent specificity using multiple rats in a row. This is how it works. And really, this is a generic technology. We're talking now explosives, tuberculosis, but can you imagine, you can actually put anything under there.

So how does it work? You have a cassette with 10 samples. You put these 10 samples at once in the cage. An animal only needs two-hundredths of a second to discriminate the scent, so it goes extremely fast. Here it's already at the third sample. This is a positive sample. It gets a click sound and comes for the food reward. And by doing so, very fast, we can have like a second line opinion to see which patients are positive, which are negative. Just as an indication, whereas a microscopist can process 40 samples in a day, a rat can process the same amount of samples in seven minutes only. A cage like this -- (Applause) A cage like this -- provided that you have rats, and we have now currently 25 tuberculosis rats -- a cage like this, operating throughout the day, can process 1,680 samples. Can you imagine the potential offspring applications -- environmental detection of pollutants in soils, customs applications, detection of illicit goods in containers and so on.

But let's stick first to tuberculosis. I want to briefly highlight, the blue rods are the scores of microscopy only at the five clinics in Dar es Salaam on a population of 500,000 people, where 15,000 reported to get a test done. Microscopy for 1,800 patients. And by just presenting the samples once more to the rats and looping those results back, we were able to increase case detection rates by over 30 percent. Throughout last year, we've been -- depending on which intervals you take -- we've been consistently increasing case detection rates in five hospitals in Dar es Salaam between 30 and 40 percent. So this is really considerable. Knowing that a missed patient by microscopy infects up to 15 people -- healthy people -- per year, you can be sure that we have saved lots of lives. At least our hero rats have saved lots of lives.

The way forward for us is now to standardize this technology. And there are simple things like, for instance, we have a small laser in the sniffer hole where the animal has to stick for five seconds. So, to standardize this. Also, to standardize the pellets, the food rewards, and to semi-automate this in order to replicate this on a much larger scale and affect the lives of many more people. To conclude, there are also other applications at the horizon. Here is a first prototype of our camera rat, which is a rat with a rat backpack with a camera that can go under rubble to detect for victims after earthquake and so on. This is in a prototype stage. We don't have a working system here yet.

To conclude, I would actually like to say, you may think this is about rats, these projects, but in the end it is about people. It is about empowering vulnerable communities to tackle difficult, expensive and dangerous humanitarian detection tasks, and doing that with a local resource -- plenty available. So something completely different is to keep on challenging your perception about the resources surrounding you, whether they are environmental, technological, animal, or human. And to respectfully harmonize with them in order to foster a sustainable work.

Thank you very much.

(Applause)
--------------------------------------------------------------------------------

No comments: