Radiologists ill-prepared for how AI will impact the profession

Many radiologists are not ready for the changes that big data and artificial intelligence will bring to their lives, says Paul Chang, MD.


Many radiologists are not ready for the changes that big data and artificial intelligence will bring to their lives, says Paul Chang, MD

For now, Chang—a professor of radiology and vice chair of radiology informatics at University of Chicago Pritzker School of Medicine—asks radiologists to consider a simple yet crucial question: Who is in control of the custody of images of patient care? It needs to be the radiologist protecting the integrity of images and data, Chang contended during a recent web seminar.

Currently, some radiologists are more open to change than others, but for all of those in the profession, incorporating big data and AI in radiology will be disruptive, and some will become disillusioned. What’s needed is some impetus—a driver to foster interest in disruptive technology—but it will take considerable time to leverage big data and AI in current health IT infrastructures.


But this is where the hype comes in, Chang notes. “We as radiologists need help, we’re barely hanging on,” he says. "We have too much ‘busy work,’ and we’re consuming too much time on image manipulation.

“But we have to improve quality in our outcomes, with actionable intelligence moving from just a report to actionable information and decision support. With help from machine learning technology, precision radiology is possible,” he contends.

With reimbursements declining, radiologists further need to demonstrate that they can cut costs while improving quality and efficiency.

“Because no one knows what the next imaging technology will be, we must become irreplaceable,” Chang cautions colleagues. "We need more IT support. Health IT systems are still relatively immature, but because we are 20 years behind other industries, we can

Chang contends that there are four levels of interoperability that providers will need to better master data sharing. They include:
  • Basic interconnectivity: Having the tools for systems to actually share data.
  • Structural connectivity: Defining the structures and formats of data exchange.
  • Semantic: Having two or more systems to exchange, interpret and use the information.
  • What Chang terms the The Holy Grail, which is actually understanding the data and using it.

More for you

Loading data for hdm_tax_topic #better-outcomes...