Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Google's DeepMind to peek at NHS eye scans for disease analysis (bbc.co.uk)
40 points by jsingleton on July 5, 2016 | hide | past | favorite | 7 comments


"[A]nonymised eye scans" sounds about as realistic as "anonymised fingerprints."

My interpretation of anonymity includes the test, "If I have the data and the person in the same room, is it super-duper easy to match them up?"


Anonymity, especially with healthcare, is usually treated as a spectrum, with a (somewhat arbitrary) cutoff.

For instance, let's say you want to study how much a variety of lifestyle factors influence some rare disease. Is the fact that a person has this disease and lives in a certain zip code enough to identify them? Depends how rare the disease is and how populated the zip code is. How about if you add in age? Ethnicity? Is narrowing it down to 10 or so people still anonymous enough? Etc.

I guess my larger point is there probably isn't a simple practical one-sentence test for anonymity. And giving google eye scans, diagnosis info, and maybe some basic demographics and no other PII seems fine by me, personally.


I was eye scanned on entry to the US at Atlanta Airport a few years ago.

I have also been fingerprinted this month scanned at the US embassy in London for my Visa, to go with my photograph.

The US has more biometric data of mine than the UK govt!

I don't know what point I'm trying to make.


Is it super-duper easy to give someone an eye scan?


I have much less of a problem with databases of eye scans as i do with fingerprints add you can't inadvertantly leave your eye scan at a crime scene.


I don't think is a bad thing on any level, I just hope they are required to publish the research as part of the deal so hospitals and researchers across the world can replicate the results


here's a bit more insight from Mustafa Suleyman- https://www.youtube.com/watch?v=0E121gukglE




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: