3 ms·

I’ve discussed this with both doctors and radiologists, and they all seem to back the exact same mantra. So while that might just be anecdotal evidence, I have

I’ve discussed this with both doctors and radiologists, and they all seem to back the exact same mantra. So while that might just be anecdotal evidence, I have a difficult time seeing how your blank dismissal has any value in the discussion. Do you have any data, evidence or even just arguments to put forth to support you oppinions?

dm319 5y ago
Well I work day to day as a doctor and I know for myself what the job entails, and I don't know where to start with a comment like this. I've just finished a clinic this morning. Interestingly there was a relevant case to this discussion. I have someone being treated for an incurable lymphoma with a tablet chemotherapy. Has been on it for nearly 5 years. She had a scan recently which has shown some pulmonary fibrosis. She has some new respiratory symptoms. The question is, has this drug caused her lung fibrosis? Is it even fibrosis at all - it can look similar to cardiac failure, and she has that too. There are around 5 CT scans over the course of several years, I had to bring them up and cross-reference it to the times of various treatments (NB I'm not a radiologist) to assess whether I thought the progression of it fitted with the drug. To complicate matters I know this lady has especially bad lymphoma which relapsed quickly after her first treatment, so if I stop her current treatment, she would probably progress and die of the disease, though I'm unsure what timeframe that would be (it depends on the individual disease). So I'm reluctant to stop it. I might take her back to our MDT where I can discuss it with our radiologist, who will pore over the scans and try to see whether we can blame the drug, and whether it could be explained moreso by the heart failure. This was a single 10min follow up patient in a busy clinic, and I only write this out to give a glimpse of the thought processes that go through decision making for one follow up patient in a clinic of 30. This isn't rote learning or rule-based treatment. It is very much patient-centred and very specific for this individual. It requires discussion with the patient - what they are happy to tolerate, dealing with the unknowns of what would happen if we continued it or stopped. Personally I would love for machine learning to be part of what I use to help guide decisions, there are particular situations where I know it would have helped. But it is irritating to read your comment and to not only be accused of being obstructive for AI adoption, but also to be misrepresented as some kind of robot that looks up every disease in a medical textbook in my head and blindly follows some protocol for deciding what to do. Most people can be trained to drive fairly quickly, how about we figure out AI driving first before attempting to simulate a job that has taken me 15 years of training to get to a position where I can take these decisions in clinic?

dm319 5y ago

Well I work day to day as a doctor and I know for myself what the job entails, and I don't know where to start with a comment like this. I've just finished a clinic this morning. Interestingly there was a relevant case to this discussion. I have someone being treated for an incurable lymphoma with a tablet chemotherapy. Has been on it for nearly 5 years. She had a scan recently which has shown some pulmonary fibrosis. She has some new respiratory symptoms. The question is, has this drug caused her lung fibrosis? Is it even fibrosis at all - it can look similar to cardiac failure, and she has that too. There are around 5 CT scans over the course of several years, I had to bring them up and cross-reference it to the times of various treatments (NB I'm not a radiologist) to assess whether I thought the progression of it fitted with the drug. To complicate matters I know this lady has especially bad lymphoma which relapsed quickly after her first treatment, so if I stop her current treatment, she would probably progress and die of the disease, though I'm unsure what timeframe that would be (it depends on the individual disease). So I'm reluctant to stop it. I might take her back to our MDT where I can discuss it with our radiologist, who will pore over the scans and try to see whether we can blame the drug, and whether it could be explained moreso by the heart failure. This was a single 10min follow up patient in a busy clinic, and I only write this out to give a glimpse of the thought processes that go through decision making for one follow up patient in a clinic of 30. This isn't rote learning or rule-based treatment. It is very much patient-centred and very specific for this individual. It requires discussion with the patient - what they are happy to tolerate, dealing with the unknowns of what would happen if we continued it or stopped. Personally I would love for machine learning to be part of what I use to help guide decisions, there are particular situations where I know it would have helped. But it is irritating to read your comment and to not only be accused of being obstructive for AI adoption, but also to be misrepresented as some kind of robot that looks up every disease in a medical textbook in my head and blindly follows some protocol for deciding what to do. Most people can be trained to drive fairly quickly, how about we figure out AI driving first before attempting to simulate a job that has taken me 15 years of training to get to a position where I can take these decisions in clinic?