Essential insights from the latest AIMed Champions Connect Quarterly meeting on harnessing the potential of AI to transform cancer care

 

Oncology was the focus of the latest AIMed Champions Connect Quarterly meeting, which took place on May 18th with support from Jvion, Siemens Healthineers, and Zebra Medical Vision.

The two-hour webinar consisted of a keynote session on harnessing the potential of AI to transform cancer care, two roundtable discussions on the use of AI to act on potential risk factors and early detection of cancers, and a concluding open forum panel.

 The AIMed Champions Connect Q1 Report

AIMed’s Director of Content, Alexis May, took the opportunity to share some of the key findings from the recently published AIMed Champions Connect Q1 Report. The report sought to examine applications of AI in population health, understand the areas that AI is successfully impacting, how AI is complementing or replacing existing analytics to better inform population health strategies and the ability of AI in addressing inequity driven by socio-economic gaps and other barriers.

In terms of adoption, a majority (56%) of respondents are actively working on AI at a population health level. 31% have already deployed AI solutions in this space, with 19% evaluating performance. Cardiovascular disease, behavioral and mental health, and cancer are the top three areas in which respondents feel AI can impact population health most successfully. An overwhelming percentage of respondents see AI as complementary to other analytics being leveraged to inform population health strategies. While 59% believe AI can help address inequity driven by socioeconomic gaps and barriers.

Delving a little into inequity, some respondents believe if AI is implemented in the right way, it could have the ability to address the challenge. This is especially so if AI is provided with broad and in-depth population level data that has been accounted for biases. AI may even be able to identify root causes of inequity not previously considered and provide novel solutions for them.

However, there will be significant detriments on a large scale where the least affluent demographics are digitally excluded, and healthcare provision is limited to those with a personal and curated “data wallet”.

Deployment and implementation of AI, standardization of global population health surveillance models, reconcile population health and personalized medicine, preventive care, and using AI to mitigate the mental health crisis are the top five areas respondents have highlighted which population health research should head towards next. The full report is available here

 

AI is impacting each step of the clinical pathway

Liana Romero, Head of Global Marketing, Clinical Decision Support within Digital Health at Siemens Healthineers, kick-started the keynote session by highlighting that AI-powered solutions are continuously bringing significant transformations into healthcare by playing a crucial role to improve each step of the clinical pathway. “AI’s implications are considerable, and its potential is limitless,” she commented.

“I am fascinated by what I am witnessing now. From enabling or increasing screening, early identification of high-risk members, to enhancing accuracy, speeding-up diagnoses and preventing the advancement of the disease. AI solutions are making a huge difference in survival rates. Furthermore, clinical details and genomic profiles of individual patients can also be incorporated to provide a more personal treatment option or one that better fits the preference of the patients.”

Dr. John Frownfelter, Jvion’s Chief Medical Officer also noted AI will eventually lead us to a “patient pathway” or the medical journey of individual patients driven by precision medicine. Dr. Frownfelter observed that it’s fairly narrow to define precision medicine based solely on genomics because there’s also a need to consider socio-economic factors, environmental factors, and any others that fall outside clinical context.

“There’s no other tool, other than AI, which combines the vast amount of data from various disparate sources and ties them together to transform medicine,” said Dr. Frownfelter. “I ventured into risk stratification or predictive analytics seven or eight years ago. But I believe AI will bring us from forecasting what’s going to happen next to prescriptive analytics or providing wisdom as to what to do next. This is gradually shifting now.”

Delivering value is where AI should come in

Despite the possibilities, Dr. Frownfelter urged the audience to be realistic. “Things aren’t going to change this year. They will come in a series of steps and it may take place over the next several years,” he said. “In healthcare, there are incentives for care providers and physicians to do as much as possible because that’s how the reimbursement system was set up. While this has been overtaken by the concept of ‘fee for value’ or trying to arrive at better outcomes with lower costs. Oncology is not a straightforward disease. I believe trying to solve a problem is just a starting point; delivering value is where AI should come in.”

Dr. Ajeet Gajra, Vice-President and Chief Medical Officer of Specialty Solutions at Cardinal added that medicine needs high-value interventions. Yet, what we are doing right now is getting cancer patients to undergo expensive treatments. Instead, each patient should be individually assessed before deciding whether they should undergo further interventions or enter palliative care. “We need to spare patients from dangerous interventions,” he warned. “We also need to spare the system by better utilizing the resources we have. That’s why identifying at-risk patients and modifying their course of treatment is what makes AI the most exciting innovation right now.”

Find patients before they become patients

Dr. Gajra felt there was still a lot for AI to catch up with. “I wish to see more work being done in matching the right treatment to patients,” he said. “For the longest time, oncologists relied on histology and cancer staging to determine patients’ treatment pathways. I think interventions need to flexibly adapt and cater to every single mutation and provide insights to oncologists on whether they are most appropriate, based on patients’ progressions. That means oncologists no longer work with tumor characteristics only but also characteristics of individual patients, including their underlying risks, lifestyle, and work. Decision making in oncology can be relatively complex and AI can step in by giving appropriate feedback and support.”

Michael Callahan, US Director of Value Based Care at Zebra Medical Vision agreed. He firmly believes in the need to find the patients before they become patients. “I met a group of healthcare executives a few weeks ago and they mentioned the importance of population health,” he said. “If we can deliver earlier interventions for cancer, it will not cost us loads in the long run. At the same time, we can financially prepare individuals who are most at-risk or the need to have additional treatment.”

Callahan added that presently we rely on risk factors to single out individuals for opportunistic screenings and we need to do the same for AI too. “For example, lung cancer, the first question we are likely to ask is ‘are you a smoker?’” he explained. “Because smoking is a risk factor and the individual who happens to smoke will undergo opportunistic screenings. Likewise, we need to know when AI should be applied to identify individuals for opportunistic screening. We can no longer rely on risk factors alone since AI is an entirely different tool but what are the early indicators? What are we looking for here? These are what we need to know as AI takes up a more prominent role in medicine.”

Find the right treatment option and monitor disease progression

Liana Romero regards AI as inseparable from making care more targeted and affordable. She described simulating therapy options on digital twins created using patients’ physiology before administering them to patients. “The essence of a digital twin lies in helping the patients to recover without putting them under too much risk and minimizing tissue extraction and damage,” she explained. “We can closely monitor the patients to prevent relapse, which is relatively common in some cancers. The entire simulation process is an opportunity for education. I believe education drives behavioral change and from a public health perspective, that’s the most challenging thing we could imagine.”

“Indeed, but cancer, unfortunately, is a very dynamic process,” Dr. Gajra added. “A patient may be well this week but start deteriorating the week after. It would be great if AI can provide a warning driven by certain biomarkers or performance status, to the oncology team or some kind of comprehensive review leveraging all social determinants of health, to determine the outcomes or day-by-day progression of patients. All these ideas where AI can be utilized excite me.”

Dr. Frownfelter agreed. “We are likely going to have a new definition of preventive care and detail what’s going to happen to a patient,” he said. “’What can we do differently for this patient that we didn’t realize that we could do early on?” will be the question for many clinicians. Last year was what we called ‘accelerated disruption’. We need courageous leaders to continue what can be achieved and have the vision to drive through some of the bumps along the way and not feel disheartened. We need dynamic disruptions; rapid change is never going to be perfect.”