5 comments

[ 1.7 ms ] story [ 23.2 ms ] thread
(comment deleted)
Maybe add a reference to DementiaBank as required by the dataset.
From your FAQ: "While this is a beta product, our tests show it to be about 82% accurate."

What does this mean? Is it the true positive rate? The true negative rate? An average of the two? An average weighted by frequency?

Yes, that "82%" figure is so handwavy as to be meaningless. In addition to your questions, I have more: how was this figure arrived at? Where are the studies? How were they conducted? Etc.

Any single study needs to be treated with extreme caution even if it was well-conducted. In this case, we don't even know if a study was conducted and, if it was, whether or not it was of high enough quality to matter.

Is there an actual theory about the link between voice and dementia?