When the algorithm misdiagnoses, it stores the sound signature...

When the algorithm misdiagnoses, it stores the sound signature and adds it to its data recognition base. The mere fact that the % correct diagnosis of children cleared by GP's significantly increased can only be presented as proof. Do you consider this to be some fluke? I can tell you, as someone that has studied advanced statistics, the sample sizes are more than large enough to prove the algorithm had improved between samples.