Introducing the meter into the training sequence

I decided that Senator is ready to start retrieving the meter for me after he alerts.  Ideally, he will alert me and then get the meter from a place that I now keep it where he can reach it.  I will test to confirm that he is right and then get the treats out.

I had been working with him on the retrieving part alone and he has been doing very well.  I had an old meter that I would put on a certain small table in the living room.  I would say "get the meter" and he would get it and bring it to me.  I am doing the same thing with a decoy "snack" that I will incorporate into the sequence later.

I asked how others do this on a facebook page that I had found called Diabetes Alert Dog Talk.  I got some pointers and decided to try it yesterday.

I had a low after a walk and Senator alerted me.  I told him to go get the meter.  He looked at me like he had no idea what I was talking about.  I tried to get him to do it a few times, but then have up and got it myself because I was uncomfortably low.

I tried the same thing again later that afternoon with the same result.  I started thinking about what is different between these situations.  It is a different meter.  The first one had the meter in it, but no test strip container or lancets, so I suppose it was smaller.  I didn't have the treat bag on me as I had in the other times while I was in training mode.

I tried it again in training mode and he did just fine.  I'll do it a few more times with the treat bag and then try to wean it away.  I wouldn't have thought it would make such a big difference.  The little stinker.

Cruise is off for his final training now.  We miss him very much, but I think Senator's training will go more smoothly.  Cruise was very aware of everything that I did and would tell Senator when to alert.  He would also beat Senator to the meter every time.  So this is a nice opportunity to move ahead with Senator now.

Generalizing the alert

Senator seems to have the alert down well.  I can say that he can definitely recognize the scent of both the high and the low blood sugar.

He is learning to differentiate between them.  He is not as accurate on this as I would like him to be and we are working on that.  I've been using training games and lots of repetition during his alerts to get him to be more accurate.  He is better than 50%, but not up to the 85% which is the goal.

Now that I'm feeling better and the weather is cooperating, we are working on doing the alert in various situations.  I put the scent pads into test strip containers and take them with me when we go places.  He has been very good in indoor situations and fair in outdoor ones.  This is understandable with wind currents and so many other interesting scents.  At this point, if he doesn't get it, I'll let him see the container just to allow him to be successful.  I think its good to let him have this experience to learn to filter the other scents and tune into the BG scent.  I'll know that he "has it" the first time he alerts me during a walk.  That is the most common time for me to experience a low, so its important that he masters this.

I've also been working on socialization.  I had to put this mostly on hold this winter and I can tell that he has lost some skills.  So, we are working on leash walking and public access.  He has seemed very ill at ease in some situations.  He can't seem to relax and lie down under a table.

Senator did have one very nice alert lately.  I was busy looking at something with Gary and he came up from behind me to paw me to tell me that I was low.  Its the first time that he has approached me from behind, so I was very pleased.  Sometimes, he'll actually wait for me to move so that he can paw me from the front which is not a good thing.  I appreciate his good manners as to not interrupting me, which seems to be what he is thinking, but
in a low blood sugar situation, waiting is not a good thing.

I had a wonderful experience last week with Senator.  I took him into the school to talk about diabetes, my transplant and diabetes alert dogs.  He was of course the star of the show. He behaved very well while I was talking.  I let him wander around the room and visit with the kids which they all enjoyed, mostly Senator.  He performed a nice alert demonstration too.  The kids who were in 5th graders in the gifted program asked some really great questions which I really appreciated.

Training games with two dogs

I've been adding some scent games to our training lately.  Senator has seemed a little complacent lately and I thought that this might perk him up.  I have been hiding the scent containers in different rooms throughout the house.  We go room to room looking for the scent that is hidden there and Senator will offer the correct alert behavior for the high or low scent as he finds them.

I decided to include Cruise, the chocolate lab that we are fostering as a service dog, in this activity.  It was either that, or kennel him while we were having all the fun, and that just didn't feel right.  I've given him a few clues as to what we are doing.  I have been having him smell the scent container after Senator finds it and then treating him.  He has found it first a few times, but I think by sight.

Its interesting to watch the two dogs searching for the scent containers.  Senators "sniffing" is very obvious.  His exhaling through his nose is obvious.  When Senator finds the scent, he immediately comes to me to offer the alert.  Cruise will pick it up and bring it to me.

I've also noticed that Senator seems to be able to detect the alert when it is hidden on my body quicker (or easier?) than when its hidden in the containers.  I'm not sure what to make of this.  Sometimes, while he is searching for the container, he will walk right by it, sometimes with his nose just inches from the hidden container.  But, when the container is concealed in my pocket, or if it is a real high or low alert, he gets it quicker and from a greater distance.  I'm not sure what to think about this.

I do feel that both dogs are enjoying theses sessions and I intend to continue with them.  I'm going to continue to give Cruise some additional cues to see if he can pick up more on what he is actually learning.

Documenting the Alert

Senator and I have been working on his scent training for almost a year now.  He's made some real progress.  He catches most of my highs and lows.  He does it quickly and he is learning to distinguish between them by giving separate alerts.  His public access has been suffering somewhat lately.  I've had some health problems related to my immunosuppression.  I haven't been taking him on many walks or outings lately.  I seem to be over the worst now and am hoping to get back where we were with all of his training.

My concern at the moment, is that he seems to have lost his excitement for some of this.  I've tried switching treats and that seems to help.  I'm planning on playing some scent searching games to up his enthusiasm.  He really enjoyed those when he was first learning to recognize the scent, and I think we will do those same lessons again.  But this time we will also practice the correct alert with them.

I've also been looking for some scientific approaches to teaching DADs and there really aren't any.  The one that I mentioned in my last post (ICAN and Lilly) seems to be the only one.  I've decided to start documenting some of my training.  It will be a good way to mark his progress and to see what we need to be working on more.  I'm measuring how long it takes him to alert and whether he offers the correct behavior for the high or low blood sugar.  He should sit for a high and down for a low after the initial alert which is to paw me.

Date                      Time taken to alert                         Alert give for high or low BG
11/25                           10 sec.                                                 Low for High BG
11/25                           25 sec.                                                 Low for High BG
11/26                           20 sec.                                                 High for High BG
11/26                           15 sec.                                                 Low for Low BG
11/26                           25 sec.                                                 High for High BG
11/27                           15 sec.                                                 High for High BG
11/27                           20 sec.                                                  Low for Low BG
11/27                           15 sec.                                                  High for High BG
11/28                           10 sec.                                                   Low for Low BG
11/28                           15 sec.                                                          ?
11/29                           10 sec.                                                   High for High BG
11/29                           10 sec.                                                   High for High BG
11/30                           10 sec.                                                   High for High BG
11/30                           15 sec.                                                    Low for High BG
12/1                             15 sec.                                                   High for High BG
12/2                             2 min.                                                    Low for Log BG  (old sample)
12/2                             15 sec.                                                    High for High BG (live sample)
12/3                             45 sec.                                                    Low for Low BG
12/3                              5 sec.                                                     Low for Low BG (live sample)
12/4                             15 sec.                                                    High for High BG
12/4                             30 sec.                                                    Low for High BG
12/4                             15 sec.                                                    High for High BG
12/5                             15 sec.                                                    Low for High BG
12/5                              5 sec.                                                    High for High BG

An exciting DAD study and more scent training

Eli Lilly and the Indiana Canine Assistance Network (ICAN) are conducting a study to see exactly what it is that our diabetes alert dogs are sensing when they alert for low blood sugars.  I'm so glad to hear that someone is finally doing this.  If it can be determined what the chemical compound is, then it might be able to be recreated for testing.  The product would be reliable and the trainer could always be sure that they are training with something that the dog can sense.  Sometimes now, I have to wonder about the stability and accuracy of what I am training with.  If the sample isn't exactly "right", they the dog is not getting trained, only confused.  Here is the link to the study.  

I volunteered to send a sample in for this study.  Interestingly, they are using sweat, not saliva samples.  I collected  my sample by rubbing four absorbent cloths over my back and forehead and then blowing on these samples and mailing them in.  There is also a control sample where I did the same thing during a normal blood sugar.

I've been trying some sweat low blood sugar samples.  My low blood sugars usually occur during/after my walks.  I have been using the clothes that I was wearing as the low scent to test Senator, and they work really well.  I keep them in a big plastic bag and they seem good for 3-4 days.  I just put them under my clothes or sometimes just carry them into the room and Senator senses them almost immediately.

I'm continuing to teach him to tell me if I am high or low after the initial alert.  He seems to be doing very well with it.  I've decided to back off of the training with both alerts for awhile.  I'm concerned that I might be confusing him.  For now, I am continuing to do about three alerts per day.  I make sure that I expose him to any natural alert that I have during the day and also do some stored samples to supplement these.  After he paws me for the initial alert, I ask him "what is it".  He does either a down or a sit and I give him a treat.  I repeat this about 5 times and then hug and thank him.  I would estimate that he gets the right alert about 80% of the time.  I tend to have more lows, so he is exposed more to them and tends to default to them (I think).  All in all, I think we are doing well.  I'm trying to get some more public access training experience for him now.