This winter vacation, the kids are out a lot skating on the lake. A new way of experiencing paediatric diabetes management, and wrestling with the challenge of over-time learning….
A few days ago, our five year old with type 1 diabetes, Sacha, began a skating session with his sister and mom. Before starting, my wife tested his blood sugar at 80, and so gave him a snack bar of 22 carbs, with no bolus, expecting he would burn a lot of it off. Sure enough, about 30 minutes later, she checked him again, and he had dropped precipitously to the 50s! Well into the danger zone, as Sacha’s recommended range is 120-180.
A few days later, our caregiver, K., took the kids out again. Conscious of how quickly his blood sugars had dropped last time, we gave him an extra juice box before he went out. About 45 minutes later, we walked over to the lake to check on him; K. told us his blood sugar had just tested at 201. Fair enough, we thought.
Forty minutes later, we brought him back to the house, and re-tested again. 481! What??!! Rechecked once more, and it was 501! Checked the insertion site – all was well. We waited twenty minutes, in case the meter had been screwed up by the intense cold. Again, in the high four hundreds.
Just goes to show. You think you can learn from experience 1 to experience 2, and in a logical world, you could. But diabetes is far too complicated for that kind of simple, over-time learning. Whatever the difference was between the two experiences, separated by only a few days, we’d missed it…..And perhaps it wasn’t ever one we could figure out.
A good lesson for stock analysts, social scientists, or anyone else in the human prediction business. As I tell my graduate students, you may think that situation A and B are similar, but more than likely, you’ve missed some crucial difference. Which means that a successful policy for A may well not work for B….
