How accurate are the algorithms compared to doctors? A lot of times algorithms can be significantly more accurate than a doctor. The ability to analyze more data in less time can’t be denied. We aren’t at that point yet where computers can do full analysis of human bodies. And there’s still a need to view edge cases like evaluating a diet and possibly the environment a person has grown up in that can factor into a diagnosis that a computer might be unbiased towards.
Is deep learning in Health Care ethical? I believe so. For this I think we should really look at what there is to be gained from deep learning. Again looking at the greater good for what can be achieved. By using deep learning, the information that takes 8 or so years to teach doctors, would take a couple minutes to download to a computer. Training it could take longer but by training from many computers in many areas and using centralized data, I believe that we can progress a lot faster. If the issue is containing the data I don’t even think it should be considered an issue. Most medical data is already kept and persistent, not in paper documents but through sites that make it accessible online like MyChart. But, if the issue is biased data based on the data sets, I feel we just need to refine the algorithms. I don’t know if we’re at a point where we can use artificial intelligence to diagnose everything, there will always be a human element needed even if just to run the machines and enter the data. But, I don’t think that makes it “unethical”.