How are applicants dealing with the uncertainty of machine learning?

[Gadofosveset]

---

Members don't see this ad.

A large part of a financial analyst's job is "pattern recognition & data interpretation." I imagine that AI will be put to use in the financial sector at least in tandem with health care, if not before. The markets will then become almost supremely efficient and no excess profits will be able to be sustained without insider information. Financial analysts will be out of work. Actually, with the exception of manual labor, every knowledge profession will be at risk. And we have robots for manual labor. Every well-paid employee in any profession who works with his or her mind will eventually be "automated," one by one, starting with the most expensive... since in our scenario AI and deep learning is basically a replica of a human mind. We won't have to automate the informatics-worshippers, since they already say the same damn thing over and over. Just give me a single program that can accurately and reliably pick small nodules out of a chest radiograph and I will be embarassingly grateful; since we can't even do that I'm not the least worried.

 

[tco]

Haha I love how you think the math needs to be 100%. In my department when one person leaves the others just work more hours to compensate. One radiologist works at 150%... Why keep the guy working 50 but hey let's pretend I'm wrong and business people will be kind to you

Good luck with retention in that department.

 

[RemoteDoc]

Another aspect that should be considered is the growing need for radiographic interpretations. At stroke centers there are specific turn-around-times for radiographic reads. This consequently has changed the practice of radiology from being an historical "ROAD" speciality (lifestyle friendly), to one where providers now maintain call schedules and work longer hours. Furthermore, medical clinicians historically have used more clinical judgement to diagnose medical conditions, but the future of medicine will expand the need and use of radiographic technology to diagnose disease. I see a short-to-medium term future where machine learning will advance in parallel to the need for radiographic interpretations. Or as the advances of machine learning progress and augment radiology, the clinical protocols to diagnose a disease will become more lenient on when to order medical imaging. An example of this is pediatric providers will routinely diagnose mycoplasma "walking" pneumonia by end-inspiratory wheezing on auscultation. They wont order an x-ray because of added cost to the patient. But as efficiency goes up, the cost-per-interpretation will go down, all at the same time increasing the demand for timely, less costly radiographic reads. History teaches us this: first with the decreasing cost of compensation for radiographic interpretations, radiologies started to employ transcription devices/software. Later macro utilities and pre-build reports were used to augment workflow. The next thing is AI & Machine learning. Ultimately, many years from now, machine learning will replace the radiologist who only reads film, but it will be a while before that happens, and in the time between now and absolute automation, the demand for radiographic interpretation will increase as the cost per read declines.

 

[Cognovi]

Another aspect that should be considered is the growing need for radiographic interpretations. At stroke centers there are specific turn-around-times for radiographic reads. This consequently has changed the practice of radiology from being an historical "ROAD" speciality (lifestyle friendly), to one where providers now maintain call schedules and work longer hours. Furthermore, medical clinicians historically have used more clinical judgement to diagnose medical conditions, but the future of medicine will expand the need and use of radiographic technology to diagnose disease. I see a short-to-medium term future where machine learning will advance in parallel to the need for radiographic interpretations. Or as the advances of machine learning progress and augment radiology, the clinical protocols to diagnose a disease will become more lenient on when to order medical imaging. An example of this is pediatric providers will routinely diagnose mycoplasma "walking" pneumonia by end-inspiratory wheezing on auscultation. They wont order an x-ray because of added cost to the patient. But as efficiency goes up, the cost-per-interpretation will go down, all at the same time increasing the demand for timely, less costly radiographic reads. History teaches us this: first with the decreasing cost of compensation for radiographic interpretations, radiologies started to employ transcription devices/software. Later macro utilities and pre-build reports were used to augment workflow. The next thing is AI & Machine learning. Ultimately, many years from now, machine learning will replace the radiologist who only reads film, but it will be a while before that happens, and in the time between now and absolute automation, the demand for radiographic interpretation will increase as the cost per read declines.

-Mycoplasma pneumonia is not diagnosed by wheezing.
-Pediatricians don't order unnecessary x-rays not because of cost but because of radiation.
-Decreasing cost of x-rays will not increase demand. They are dirt cheap already.

 

[Tiger100]

Another aspect that should be considered is the growing need for radiographic interpretations. At stroke centers there are specific turn-around-times for radiographic reads. This consequently has changed the practice of radiology from being an historical "ROAD" speciality (lifestyle friendly), to one where providers now maintain call schedules and work longer hours. Furthermore, medical clinicians historically have used more clinical judgement to diagnose medical conditions, but the future of medicine will expand the need and use of radiographic technology to diagnose disease. I see a short-to-medium term future where machine learning will advance in parallel to the need for radiographic interpretations. Or as the advances of machine learning progress and augment radiology, the clinical protocols to diagnose a disease will become more lenient on when to order medical imaging. An example of this is pediatric providers will routinely diagnose mycoplasma "walking" pneumonia by end-inspiratory wheezing on auscultation. They wont order an x-ray because of added cost to the patient. But as efficiency goes up, the cost-per-interpretation will go down, all at the same time increasing the demand for timely, less costly radiographic reads. History teaches us this: first with the decreasing cost of compensation for radiographic interpretations, radiologies started to employ transcription devices/software. Later macro utilities and pre-build reports were used to augment workflow. The next thing is AI & Machine learning. Ultimately, many years from now, machine learning will replace the radiologist who only reads film, but it will be a while before that happens, and in the time between now and absolute automation, the demand for radiographic interpretation will increase as the cost per read declines.

What is radiographic technology? What is radiographic read and interpretation?

In what part of the world mycoplasma pneumonia is diagnosed by end-inspiratory wheezing?

Your argument is incorrect because you see things only from business/supply-demand curve which is far from the practice of medicine. If I decrease the cost of MRI liver to 10 dollars per scan, the number of liver MRIs won't go up significantly.