“So then they understand: ‘If I smell TB, I get food’.”
February 25, 2016 10:03 AM   Subscribe

The rats who sniff out tuberculosis. by Emma Young [The Guardian] The African giant pouched rat can be trained to sniff out tuberculosis more accurately than most lab tests. So why is the medical profession still sceptical?
In a small, hot room in a compound located in Tanzania’s lush southern highlands, one day in mid-December, were three white-clad technicians, a glass-and-metal chamber and a large brown rat named Charles. After being gently dropped into the chamber, Charles aimed his long snout towards the first of a series of 10 sliding metal plates in the chamber’s base. A technician swiftly opened it, revealing a small hole. Charles sniffed at it … and moved on. The hole was closed, and there was a clink as the next plate was yanked back. This time, Charles was gripped. He sniffed hard, scratching at the metal, the five claws on each of his paws splayed with the pressure. The technician called out: “Two!”

Over by the window, her colleague held a chart, which he kept raised so the others could not see it. He inserted a tick. I glanced over. The chart was a grid of small boxes, 10 across by 10 down, each marked with a code. Two of the boxes in each line were shaded grey. The tick had been placed in a white box. It is highly possible that Charles had just saved a person’s life.
posted by Fizz (24 comments total) 20 users marked this as a favorite
 
A rat, which costs $6,500 to train, can rattle through 100 samples in 20 minutes. A GeneXpert device, which costs $17,000, takes around two hours to analyse a single sample. It costs about $1 to screen a sample using a rat, compared with $10 for GeneXpert. GeneXpert requires a stable electricity supply and controlled temperatures; the rats require food and water and play cages.

That's pretty awesome.
posted by redsparkler at 10:50 AM on February 25, 2016 [7 favorites]


Rats are awesome. It turns out the African giant pouched rat can also be trained to sniff land mines.
posted by blnkfrnk at 11:01 AM on February 25, 2016 [2 favorites]


Ah, yes, that's why the deja vu.
posted by one weird trick at 11:04 AM on February 25, 2016 [1 favorite]


I suppose part of the skepticism comes from our expectation that the GeneXpert will cost a tenth of that in ten years, while the rat cost will stay more-or-less the same.

And that's not a ridiculous expectation. I was chatting with a guy last night who's working on improving the accuracy of a little USB DNA sequencer that was/is being used to track the evolution of the Ebola outbreak in West Africa. Ten years ago, you needed a machine the size of a couch to do the same thing in the same time.
posted by clawsoon at 11:29 AM on February 25, 2016 [4 favorites]


Ah, yes, that's why the deja vu.
posted by one weird trick at 2:04 PM on February 25 [+] [!] No other comments.


You might even say that African giant pouched rats know more than one weird trick.

(I'll see myself out ...)
posted by maudlin at 11:29 AM on February 25, 2016 [9 favorites]


Super-weird that the landmine rat and the TB rat cost exactly the same to train: $6,500. Like sketchy sourcing weird.

All the rat-training I've previously seen was super-labby and isolated: push button, get treat. Presumably because of the scale these things are super-manual and therefore expensive. Big factory + computer vision + lots of rats + samples de jour = ALL RAT POLICE STATE. I blame the Vandals for an early desire to control an all-rat army, rabid or no. I may be contented knowing you can put rats on leashes. Packs of rats on leashes. "That's right, we're out here walkin'."
posted by Ogre Lawless at 11:50 AM on February 25, 2016


This may be a good time to listen to this Reply All episode. Or not.
posted by maudlin at 11:57 AM on February 25, 2016 [2 favorites]


To follow up my previous comment, now that I've read the article: On the other hand, the rats are saving lives right now.
posted by clawsoon at 11:59 AM on February 25, 2016 [1 favorite]


Super-weird that the landmine rat and the TB rat cost exactly the same to train: $6,500. Like sketchy sourcing weird.

So which rat do you value more? The white- collar rat that performs medical research or the blue-collar rat that is out there risking his neck trying to disarm land-mines?
posted by Fizz at 11:59 AM on February 25, 2016 [9 favorites]


Whenever something like this comes out, I'm always a little worried. After all, we have to assume that the animals are smart enough to learn, but dumb enough to not learn to lie.

At a certain point, might they get lazy and just pretend they've found a positive for their reward? How can we be certain?
posted by explosion at 12:00 PM on February 25, 2016 [1 favorite]


At a certain point, might they get lazy and just pretend they've found a positive for their reward? How can we be certain?
posted by explosion at 3:00 PM on February 25 [+] [!]

Well it's not like they're taking the rat's word for it.

"Any sample indicated as positive by any of the rats then goes for checking with a more sophisticated, more accurate – and more expensive – microscope technique than the one used in the clinics."
posted by ZaphodB at 12:10 PM on February 25, 2016 [8 favorites]


Ogre Lawless: Super-weird that the landmine rat and the TB rat cost exactly the same to train: $6,500. Like sketchy sourcing weird.

I would guess it is a rough estimate, and from the description, the training is exactly the same, just using different scent samples. Though I suppose the mine rats may need a small amount of additional leash training, as I recall they fence off a search area and let them run around on tethers. But all the main effort in training is making them human-friendly and teaching them to alert on a specific scent, and that's the same either way.
posted by tavella at 12:53 PM on February 25, 2016 [2 favorites]


At a certain point, might they get lazy and just pretend they've found a positive for their reward? How can we be certain?

When one rat in the batch gives a false positive, all the OTHER rats get rewarded... with a food ration of fresh rat meat. They catch on fast.
posted by FatherDagon at 12:59 PM on February 25, 2016 [4 favorites]


I presume they're estimating the cost of rat training as cost of rat housing + # of human hours needed to train them. And some bananas. I can certainly believe that number is similar between land mines and TB samples.
posted by telepanda at 1:00 PM on February 25, 2016


OK, now I sort of feel like maybe I need a small horde of these rats in my life, so that they can constantly be checking me for TB, cancer, whatever else I might contract, and also protecting me against errant explosives. Just to be safe, you understand.
posted by aramaic at 1:11 PM on February 25, 2016 [13 favorites]


The cost comparison feels like cost vs. price - what the rats actually cost to use vs. how the genetic assay is priced on the market. Not really a fair comparison. There is a lot of overhead factored into a price that isn't always factored into cost estimates.

Also, once you have genetics equipment and personnel you can test for all kinds of things.
posted by Mitrovarr at 1:33 PM on February 25, 2016


I, for one, welcome our rat overlords saviours.
posted by mandolin conspiracy at 1:35 PM on February 25, 2016 [1 favorite]


You can sponsor one of these little guys here. I keep saying I'm going to, but I can't pick a good name: Scabbers? Rizzo? Dangerous Beans?
posted by nonasuch at 3:18 PM on February 25, 2016 [3 favorites]


Nicodemus, clearly.
posted by Lazlo Hollyfeld at 3:53 PM on February 25, 2016 [3 favorites]


The GeneXpert's accuracy kind of sucks. It's everywhere in the third world (and TB endemic areas) because it is highly, highly, subsidized.

Also, TB is actually rather difficult to diagnose, especially when its latent. Very difficult to determine if someone has a latent infection or if they've previously had an infection but are bacteria-free now.
posted by porpoise at 5:01 PM on February 25, 2016 [2 favorites]


The GeneXpert's accuracy kind of sucks. It's everywhere in the third world (and TB endemic areas) because it is highly, highly, subsidized.

Hmm. That sucks. It is hard to do qPCR accurately and reliably without an expert on-hand - I work to develop protocols for detection of some agriculturally-relevant pathogens so I some experience with that.
posted by Mitrovarr at 5:34 PM on February 25, 2016 [1 favorite]


What's the deal with the post framing here? Although it does mention "convinc(ing) the remaining sceptics about the usefulness of the rats", the closest the article comes to mentioning resistance to the idea of this method is the description of a standard testing validation trial. There's nobody saying, "Oh come on, now, surely not rats."
posted by gingerest at 6:03 PM on February 25, 2016 [1 favorite]


It's not just rats, either. Dogs can smell cancer. If you're worried about the cost, why not train humans to smell things? I'm pretty sure if you started them young and didn't swamp their tender nasal passages with soap, perfume and diesel, humans could smell some of these things. I personally know kids who would do this for treats. Lord knows there are enough of them around.

Another possibility would be to train horses to smell stuff for us. Yes, they're big and eat a lot, but you can ride them to the smellatorium, and that's a pretty big bonus right there.
posted by sneebler at 6:15 PM on February 25, 2016 [1 favorite]


It is hard to do qPCR accurately and reliably without an expert on-hand

Yup, big ol' expensive automated machine with expensive disposable cartridges and even still technician skill plays a part.

The qPCR stuff for TB detection (and SNP detection for suspected antibiotic resistances) can actually be really really good in the lab (I used to design qPCR TB/resistance detection) with limits of detection/quantification down to three or four copies from patient specimens and able to capture >80% of resistant isolates for rifampicin, isoniazid, and the fluoroquinolones in multiplex.

Field/clinical testing: it works great - as long as the sample prep was done well. In addition to low numbers of bacteria, especially in latent or smear-negative cases, TB has a particularly tough membrane to crack. The sample extraction part of the disposable cartridge isn't great. The mainland Chinese hospitals, that I was working with, doing in-house molecular TB testing mainly used terribly ineffective sample extraction procedures, and it wasn't because of fears of cross-contamination or infecting the technicians. It was baffling. Using the extraction kit that I developed, sensitivity was increased by just shy of 100x. It cost $5 more per test (probably $2 if we could source one item in the kit locally), with less technician time, so they ultimately wasn't interested.

I also suspect that health care providers aren't trained well enough to explain to the patient/client on how to give a good sample and most of it ends up being saliva instead of sputum. Especially in non-English speaking countries and lower-income countries where the health care providers might not have the time to be trained well or have the time to educate the patient/client.

This was published after I left the field.

"...lack of multi-language settings for the GeneXpert Dx...
"...proper support for procurement, infrastructure and supervision..."
"...errors generally arose from inadequate sputum collection or specimen preparation..."
"...invalid results which seem to be caused by problems with the quality of the sputum sample and proper mouth cleansing techniques. Better coaching from staff may be able to help in these instances..."
"...No result outcomes are often linked to power supply problems..."


Sounds like the rats here can detect TB from a patient/client's breath without needing reliable electricity, and that's fantastic.
posted by porpoise at 8:00 PM on February 25, 2016 [4 favorites]


« Older Pulp Fiction: The Internet Archive's "If" sci-fi...   |   #EddieWouldGo Newer »


This thread has been archived and is closed to new comments