00:00My name's Rob Harris, I'm an operational canine specialist and today you're going to see Mark,
00:04one of our trainers, working Hetty, Jodie and Willow to detect bowel cancer. Now what that
00:10means is when we're in the room there are a number of samples are set out, so in this case LB4,
00:16some of those contain normal urine, maybe from someone that's unhealthy but doesn't have bowel
00:21cancer and one will contain bowel cancer. So Mark will task the dog down the line and as the dog
00:27sniffs down each one of the samples when it comes across the right target, in this case bowel
00:30cancer, it will sit down and then Mark uses a device called a clicker which gives an audible
00:35signal to the dog to say it's correct, that the behaviour that they've displayed, which is a sit,
00:40is correct and the dog will then come back for a reward. So the project that you'll see him
00:44working today on bowel cancer is really really innovative, so it's the first time canines have
00:49been used to show that the odour of bowel cancer is available in a sample of urine.
00:54Our clinician Mr Ian Hunter from Hull University Hospital was really interested in the work of the
00:59dog because he carries out surgery that's organ preservation when he detects bowel cancer
01:04and what that means is he will use something called targeted radiotherapy to try and help
01:08the treatment of bowel cancer but that can unfortunately have a recurrence rate so he's
01:13interested in whether dogs are able to support in the detection of bowel cancer in that remission
01:18stage. Medical Detection Dogs charity was founded in 2008 and since that time we've worked on all
01:24sorts of different diseases, these include bladder cancer, prostate cancer, colorectal
01:30cancer as we're talking about today, also things called pseudomonas, malaria and parkinson's
01:35disease. When we're looking for a recruit to become a biodetection dog for the charity
01:40it's really important for us that dog is a problem solver, it really likes to use its brain and it's
01:44motivated by toys, food or by praise by the handler because this is a challenging role for the dog
01:50it requires a lot of problem solving skills. We don't know what this odour is so we are reliant
01:55on the dog being able to discriminate between what is a healthy person what is an unhealthy person
02:00initially but then it has to move into if I've got a problem with a certain organ like my prostate
02:05there are other conditions that are not cancerous the dog needs to be able to say no that's not
02:09the right thing but yes you do have cancer. On average one of our recruits will spend six months
02:14to a year training to do this fantastic work and that's because the challenge they have in having
02:19to use their nose to be able to detect disease is difficult it's not an easy thing and we have to
02:24step through a lot of different procedures to be able to get the dog there so one of the things we
02:28have to teach it to do is learn what the clicker is. So that clicker is the audible signal the dog
02:34is going to hear when it knows it's done something correct and we start things quite simple when the
02:38dog is sitting, when it's downing, when it recalls and then we actually progress things to searching
02:43for a toy for instance and then we may cut that toy up and hide it and make it more difficult to
02:47find and every time the dog is successful we'll hear this audible sound and that will then say to
02:51the to the dog and help the handler know when to actually give the reward. So over a number of
02:56years medical detection dogs has been looking at the way that the dog interacts with the sample
03:02so the stands you'll see working today are a collaboration with the Open University
03:06and they are taking readings from the dog of how much it's interested in a sample so they have a
03:11pressure sensor on them. Also they're able to measure the length of time the dog stays with
03:15that sample so we're able to use that as a signal to let us know how confident the dog is with that
03:21one particular sample. What these pieces of machinery do are linked back to a laptop that
03:25takes an independent reading so eventually over time we're hoping to develop an AI model that's
03:30able to interpret the dog's responses and in the future who knows we might be able to attach that
03:35to an automatic rewarder so the dog actually gets its reward from machine rather than a person.
03:41you
Comments