Earlier in the week, there was a post from Andrew Ng on the topic of using Artificial Intelligence in the world of Radiology.
In response, Dr. Eric Topol said the following,
“That’s great Andrew. But Radiologists are going to lose their jobs.”
In repose, Sherry Reynolds asked the follow-up question and weighed in with a great article on the topic from Carestream entitled, “WILL RADIOLOGISTS BE REPLACED BY COMPUTERS? DEBUNKING THE HYPE OF AI” (I encourage you to read it for perspective)
The conversation continued amongst other radiologists, interventional radiologists, referring physicians, and myself.
And then this appeared:
In which Dr. Stephen Borstelmann (Stephen Borstelmann) responded
Which prompted my response….
Which brings me to the significance of my post and thoughts.
We live in 2017vand yet the majority of healthcare continues to try and practice as if it were the 80’s.
If we take the simplified meaning of Dr. Topol’s comment to mean that the work radiologists do today is being taken on by computers, algorithms, machine learning, and eventually artificial intelligence, then yes, the radiologist as we define her work today will soon be over.
The problem is that Dr. Topol and many other physicians in healthcare are defining the role of the radiologist as if it were 1980.
In today’s world, you have medical technology vendors creating bigger, better, faster, and more sensitive equipment. The medical technology world promises the ability to see more, see sooner, and transform our world from late to earlier diagnosis of patients.
The challenge is that there are now 2–4x as many images to manage.
Yet the problem is not in the sensitivity and number of images, it is in the correlation of all the other information which drives the definitive diagnosis, and gets to the specificity.
Information from EHR and EMR’s.
Information from PACS, HIS, RIS systems.
Information on CDs from other hospitals and clinics.
Information in paper form via fax and carbon copy.
Information from the referring physicians via phone calls.
In 1980, the role of the radiologist was to be the physician’s physician. The friendly doctor you could pick up the phone, call, and discuss what they are seeing inside the body to help guide treatment decisions.
In the last decade, the role of the radiologist has evolved from being the physician’s physician to being the gatekeeper and growth engine for healthcare.
Simply stated by one healthcare CEO,
“If you can’t see it, you can’t treat it. If you can’t treat it, you can’t bill for it. Radiology is our growth engine.”
In today’s world, the role of the radiologist is evolving yet again.
This time, the evolution of the Radiologist is not being driven by the referring physician or the hospital CEO, it is being driven by the market.
Patients today have the ability to access and know more than in any other era in history. Often times patients know more about their own conditions than there primary care physicians, internists, or even their oncologists.
Patients are learning that if they can get the right imaging performed, and when they share as much of their history and previous information with the radiologist, that it will lead to a definitive diagnosis (or monitoring of condition).
As artificial intelligence begins to be built on the backs of machine learning and deep learning, and it begins to be adopted inside of healthcare, an obvious use case is radiology.
The statement of Dr. Topol’s (as well as Dr. Ezekiel Emmanuel) is wrong.
Artificial Intelligence will not replace the radiologist.
Artificial Intelligence will augment the work of the radiologist.
The radiologist of today (and tomorrow) will be responsible for defining the use cases of artificial intelligence.
The radiologist will be building the library of patient images and clinical history that need to be loaded into machine learning and deep learning systems to build AI platforms.
The radiologist will collaborate and educate the data scientist and technology companies to help them refine, deploy, and iterate the AI platforms.
The radiologist will own the decision on which algorithm needs to be used for which patient based on a number of variables, inputs, and clinical decisions.
The radiologist understands and corrects the anomalies that occur when the algorithm begins to find things that a human eye misses or that we haven’t correlated before with the data.
The radiologist will spend more time interpreting findings that are the most suspicious and deferring the “easy or probable” work to the artificial intelligence.
Most importantly, radiologists will be able to spend more their time on their “why” moments,
“To make a difference with each individual patient. My obligation to contribute to the well being of others.” — Dr. Geraldine Mcginty (@DrGMcGinty)
What Dr. Topol and many other healthcare experts, advisors, futurists, and physicians do is define radiology in terms of the IQ.
Artificial Intelligence will own the IQ.
AI will begin slowly, but within the next 8 years, AI will evolve on an exponential trajectory until it owns the majority of the IQ in healthcare. Not just the IQ of radiology, but the IQ of all the general and specialty care physicians.
What radiologists “see” is that healthcare is undergoing a renaissance.
A renaissance in which AI will own the IQ and radiologists must own the EQ.
The vast majority of physicians, healthcare systems, medical technology companies, startups, and consultants believe the renaissance of healthcare is in the development, curation, and deployment of content.
Unfortunately, they are wrong.
The truth, if they understand the market, is that the healthcare renaissance is in the context.
“If the content is king, then context is God.”
Context will allow us to develop, deliver, and deploy care at the N of 1.
Context determines the right treatment for the right diagnosis for the right patient at the right time.
It is time to start practicing radiology like its 2018.
Develop, deliver, and deploy the context of radiology.
Practice delivering care to the N of 1.
As always feel free to email me at email@example.com or follow me on Twitter and Instagram as @cancergeek