Youโre assuming there is always a good history. Many times we have to call a clinician because the notes arenโt full updated especially in the acute setting. Iโm not arguing that itโs going to help our job. But to sit there and say that the entire field of diagnostic radiology will be made irrelevant because of ai shows your ignorance to how medicine is practiced, the heterogeneity of technology and imaging systems across medicine, reimbursement structure, and the rest. Stay in your lane pal.
@TheDutchRuler@rohanpaul_ai You are assuming that there is always a binary yes no answer to imaging. Itโs very rare that an image is interpreted in that way. The majority of the time the interpretation is nuanced and more blurred than a yes no answer.
Hey Mark! Long time follower of your work. Thanks for all you do. I am a diagnostic radiology resident physician at the University of Washington and am interested in the role that we as radiologists can play in the transition to a healthier America and preventional, precision medicine with AI as a horizontal enabling layer.
Would love to hear how this Full body MRI fits together with the rest of your vision, and I also would love to be a radiologist that works with you to pioneer the role of the preventional, health focused radiologist.
Feel free to shoot me a message so we can chat more! Thanks again for inspiring me and the countless millions across the world to create a more prevention focused healthcare system and healthier world.
Yeah itโs not to say AI cannot get to this point (in fact it technically is already there) but integrating it into the rest of the healthcare stack to achieve that, determining who is paying for it, determining liability, and ensuring monitoring of these systems are the next problems to solve
Radiology requires context. It requires medical knowledge beyond the image. It requires understanding acuity, treatment patterns, etc. it requires communicating information to people at the appropriate time depending on the clinical scenario. It requires protocol ing the study correctly to get the right image. It requires recommending the right next study if appropriate. It requires being gray when necessary in a diagnosis. And by the way, all of these ai algorithms that are pixel based and help with detecting various pathologies are limited to their training data set and the types of scanners they were trained on, which do not match the environment they are released in in the real world and greatly impacts the quality of the algorithm. The people who made these products donโt understand radiology and it is clear to me thatโs the case because they went after a reductionist approach to radiology, thinking all it was was โpattern recognition on imagesโ
@DeryaTR_@OpenAI How do we infuse this intelligence into the software physicians use (both the EMR and the PACS/Radiology software)? Epic and these legacy vendors needs to step out of the way and we need to build a healthcare operating system infused with these new models