All statistics are dehumanizing. I'm no data scientist but I'd imagine including the whole human when doing statistics would produce a pretty low signal to noise ratio.
Statistics are dehumanizing, you are correct. We do not judge individuals based on the statistics of their categorical groups for this reason. Doing so under the obfuscation of a ML model doesn't change that.
Health is very personal. Family history is usually pretty relevant. I am not sure what we gain by pretending we arent talking about actual people, or that by removing your name from all your specific health details it somehow means you cant still be personally targeted. I know Google gains the benefits of less regulations, though.
Health is very personal. Family history is usually pretty relevant. I am not sure what we gain by pretending we arent talking about actual people, or that by removing your name from all your specific health details it somehow means you cant still be personally targeted. I know Google gains the benefits of less regulations, though.