There is no denying that artificial intelligence (AI) has been a seemingly ubiquitous buzzword in healthcare for some time now, but it has taken longer for machine learning to infuse the radiation therapy market, which has been focused on making treatment more precise in order to spare healthy tissue.

For the past two years, you couldn’t turn a corner at Chicago’s McCormick Place during RSNA without seeing or hearing those two letters but there has been a bit less focus on AI at previous American Society for Radiation Oncology (ASTRO) meetings. That changed at this year’s 61st meeting, which also took place at McCormick.

Aside from AI, a lot of discussion centered around the future of flash therapy — which targets tumors using high doses at ultra-fast speeds, potentially reducing toxicity in healthy tissue — and a continued look at how to streamline workflow for radiation oncologists.

Here are our key takeaways from the booth tours, presentations and research presented at the show.

AI to adapt treatment
Perhaps the buzziest product release was Ethos, Varian's AI-powered adaptive radiation therapy system

The company promoted an AI-driven technology that's able to alter a patient's treatment plan based on the tumor and anatomical changes.

Chris Toth, president of Varian Oncology Systems, said the AI technology allows for a more streamlined workflow and shorter treatment time.

"I think we're leapfrogging anything that's out there," Toth told HCB News. "Other devices have been designed to do adaptive radiation therapy, but I think of AI in the context of aiding the physician. If I'm sitting at a console marking outlines ... it's not the best use of my time."

AI to predict outcomes
During the show, Dr. Jay Reddy, an assistant professor of radiation oncology at the University of Texas MD Anderson Cancer Center, presented a look at a computer model that was used to predict side effects associated with radiation therapy for head and neck cancer.

The computer models analyzed large data sets — more than 700 clinical and treatment variables for patients who received radiation therapy from 2016 to 2018 — from electronic health records, an internal web-based charting tool called Brocade and Elekta's MOSAIQ oncology information system.

The researchers used the models to accurately predict the patients who would benefit from interventions to prevent side effects like significant weight loss, feeding tube placement and unplanned hospitalizations.

“Machine learning can make doctors more efficient and treatment safer by reducing the risk of error,” Reddy said in a statement announcing the results. “It has the potential for influencing all aspects of radiation oncology today — anything where a computer can look at data and recognize a pattern.”

The limits of AI
Despite the promise of machine learning to impact care, some experts cast doubt on the ethics of using patient information for machine learning.

During his keynote, David Magnus, the Thomas A. Raffin professor of Medicine and Biomedical Ethics and professor of pediatrics and medicine at Stanford University, talked about the ethics of sharing de-identified data with third parties without consent.

“We need to think about a new model of data stewardship, recognizing the duties to protect patients and the data entrusted to their providers,” Magnus said.

Magnus also warned that use of AI can be limited when data that is missing from underrepresented populations, which can reinforce the existing biases in current clinical practice.

Flash forward
While there were a few more education sessions about AI, flash therapy was also a popular buzzword at ASTRO this year.

The technology is still in the research and exploration phase, with Thomas Schmid of Klinikum Rechts der Isar in Munich, Germany, speaking about results from laser-accelerated proton irradiations during an education session.

Varian also promoted its ProBeam 360° proton therapy system — the company announced at the show that the small-footprint system was now available in a multi-room configuration — which it says provides customers with a pathway to flash therapy.

Working on workflow
Improving workflow continues to be a key part of new radiation therapy planning tools.

Philips continued to promote its IntelliSpace Radiation Oncology system that it unveiled in April at the European Society of Radiation & Oncology 2019 Annual Meeting in Milan, Italy.

The product is designed to integrate patient data from various sources into the planning software, which Ardie Ermers, general manager of radiation oncology at Philips, told HCB News would allow treatment to start earlier.