The only data we were able to get at the time was mostly white patients. We talked to many hospitals but many were/are reluctant to share anonymized data for research. I'm not at Google so I'm not sure the status of the project now, but there was a real attempt to try and gather more diverse data. Unfortunately there were a lot of obstacles put up by those who have the data (hospital systems).
Fundamentally, it seems to me like there just aren't as many lung cancer screening scans out there for non-white patients as there are for white patients. How do we get around this? How do we improve on the situation? I fundamentally believe that machine learning in the long term can make medicine more accessible to more diverse groups, but not if we shoot it down out of fearmongering right away.
I agree that bias is a problem, but part of what needs to happen to get more diverse data is simply having more data available. There is real promise in this technology and if we have a one dimensional view of it ("is it or is it not dangerous because of bias/privacy") then we will fail to get past the initial humps related to fear and distrust.
The only data we were able to get at the time was mostly white patients. We talked to many hospitals but many were/are reluctant to share anonymized data for research. I'm not at Google so I'm not sure the status of the project now, but there was a real attempt to try and gather more diverse data. Unfortunately there were a lot of obstacles put up by those who have the data (hospital systems).
Fundamentally, it seems to me like there just aren't as many lung cancer screening scans out there for non-white patients as there are for white patients. How do we get around this? How do we improve on the situation? I fundamentally believe that machine learning in the long term can make medicine more accessible to more diverse groups, but not if we shoot it down out of fearmongering right away.
I agree that bias is a problem, but part of what needs to happen to get more diverse data is simply having more data available. There is real promise in this technology and if we have a one dimensional view of it ("is it or is it not dangerous because of bias/privacy") then we will fail to get past the initial humps related to fear and distrust.