Exploring The Impact of AI Technologies on Oncology Practice

Artificial intelligence may act as a force multiplier, with the automation of menial tasks enabling more time for clinicians to engage with patients.

Arturo Loaiza-Bonilla, MD, MSEd, FACP, systemwide chief of Hematology and Oncology at Saint Luke’s University Health Network, spoke with CancerNetwork® about the impact of emerging artificial intelligence (AI)-powered technologies on clinical practice, highlighting relevant literature performing real-world analyses of these novel technologies with established standards of care.

Loaiza-Bonilla initially outlined AI's impact on his practice, explaining how he has striven to incorporate as many newly developed AI tools into his practice as possible. Furthermore, he described how his approach to oncology has been modified through the use of AI in precision medicine for clinical trials, which can help expedite trial matching and cohort allocation for patients, as well as automate documentation for clinicians. AI can also create practical and sensible integration into clinical practice, which may help bolster interpersonal interaction between patients and clinicians and create a more empathetic approach to care.

Over the last few years, there has been an increased emergence of real-world data surrounding many FDA-approved AI-powered tools, explained Loaiza-Bonilla. He also expressed that he has seen data relevant to the diagnostics aspect of cancer treatment. The results of the Mammography Screening with Artificial Intelligence (MASAI) trial (NCT04838756) showcased that AI-assisted mammography was able to better optimize for recall than mammography without AI assistance.¹ Furthermore, Loaiza-Bonilla discussed the launch of another trial inspired by MASAI results, the UK-based Early Detection using Information Technology in Health (EDITH) trial, which purports to enroll approximately 700,000 women to participate in a prospective analysis of 5 AI platforms used in mammograms.²

Overall, Loaiza-Bonilla described the potential for AI to act as a force multiplier, which can automate many of the menial processes that enable clinicians to spend more time caring for patients. Emphasizing AI’s ability to augment many menial tasks associated with clinical care, he expressed that using these tools can result in time saved for both clinicians and their patients and a reduction in communication burden. Loaiza-Bonilla then outlined potential applications for AI in prognosis and early detection of cancers before they become greater concerns.

CancerNetwork: How has the emergence of AI-powered technologies impacted your practice?

Loaiza-Bonilla: AI has taken all of us by storm. We have seen almost daily releases of new tools and news about the different stakeholders since the advent of ChatGPT and others [like it]. We are still in the process of learning, but [regarding] my practice itself, I am an AI enthusiast, so we are trying to incorporate as many of those algorithms as possible daily. It has changed my approach to oncology, particularly around what we call precision medicine in clinical trials. I have been using it for several years in clinical trials matching from a patient-centric perspective because all that is data organization. We can use those technologies to summarize the patient notes, [assessing biomarkers that the patient has been tested for], and then find opportunities for optimization of treatments, or at least make my life easier documenting.

There has been a dramatic shift in how we approach these tools to make our lives easier. We were burned out—and we still are—by menial tasks such as documentation, just for billing, instead of focusing on patients. Saving hours of a day by using these tools has made a [significant] difference. Oncology is a multidisciplinary specialty where the patient is at the center of the whole experience. We have radiologists, pathologists, surgeons, genetic counselors with laboratory processes around that, infusion chairs, and clinical trials, and all of them have been touched by the AI Cambrian explosion that has been happening over the last couple of years.

The focus, now, is how we integrate things that make sense to the patient and to the use cases to make our lives easier, and [how] we can trust those AI tools in a meaningful way so we can devote more time to talk to patients, be more empathetic, and [spent more time on the] vocations that we went [to school] for and see what the ground truth is and what we can leverage. [The focus is also] to be better physicians and become not only better human beings, but also ethically trust that these technologies will help us in our daily practice.

Are there any ongoing clinical trials evaluating the application of AI in clinical practice that you would like to highlight?

Loaiza-Bonilla: One of the things that has been emerging from [AI’s growth] is that many of these tools are released as FDA-approved, and you may want to test [them], but many of them have not been compared with reality. Many of the solutions can be important, but we need to have cross-comparison with what happens when we have the AI tool vs the AI tool plus the clinician vs the clinician alone. The first few studies that have come up, at least with some results that I am interested in, are on the diagnostic side.

The MASAI trial, was recently published in The Lancet. It showed that mammograms, digitally assisted by AI, were able to detect with optimized recalls compared with the readings on their own, making them more efficient. That is now leading, for example, to another study that is about to launch in the UK called the Early Detection using Information Technology in Health [EDITH] trial. We still do not have the full details, but it was supposedly announced by the UK Government. [It will be] one of the largest AI prospective evaluations in mammography. They will plan to enroll [700,000] women in 5 AI platforms––not just 1 AI platform, but 5––across 30 National Health Service sites. The government is giving [£11 million] to fund this.

This is all based on the MASAI trial from Sweden, which is the precursor study to this, that which showed [patients who received a mammogram] that AI increased cancer detection by 29% compared with standard screening and reduced the interval cancers by 19% at 2-year follow up. This is quite promising, showing a non-inferiority to double reading in AI, maintaining a detection rate of [6.4] cancers per 1000 screenings vs [5.0] manually. You may say, “Oh, [it is only 1.4%],” but if you compound that by the hundreds of thousands of mammograms we do, that helps [many] patients. I am interested [in following] those studies.

There is other analytical work [done as well] in the pathology space, and other things that are happening that I feel are relevant would be utilizing the tools and comparing [them in the real world]. I know the vendors of Ambient AI are doing some studies as well, comparing internally with real-world data on how patients feel with the ambient AI experience compared with those who are just seen by the physician without the help of those tools, many of them integrated into the electronic medical records [EMRs].

Is there anything else that you would like to highlight regarding the application of AI into clinical practice?

Loaiza-Bonilla: I’m a human but also an AI enthusiast. I believe that AI is not a replacement for us; it is a force multiplier, the same way that we have agents that [perform] certain things. For me, I can do grocery shopping myself, but I may use an app instead because it is easier and more convenient, so I can use my time to take care of patients, so I do not have to drive to the supermarket. Someone else can do it for me. In this case, an agent can do that ordering for me.

The true value of that is that there are going to be some tasks that we should relinquish to AI, and there are some others that are going to be part of the human experience, and some others that we need to combine AI with humans to [enhance]. I doubt that there is going to be a 100% AI robot that is going to be a physician, because the skill set to be [one requires] having an empathetic relationship with a patient, giving them comfort when they are in the time of need, or understanding the social determinants of health of a patient [that] is not going to be solved by AI or a robot with an algorithm inside no matter how advanced it is. That takes a lot of time. We may get there at some point, but not over the next 50 years.

The true value is in augmentation. Whether through faster clinical trial matching and helping us [create] a model that helps the patients to get access [to them] in real time; sharper diagnostics, such as prognostication and multimodal approaches, so we do not waste money and resources on a patient whose time is the only major asset they have. We do not want to waste their time, right? That is the thing we cannot buy. We can reinvest efforts into a physician’s time or a patient’s time. It is a well-invested effort, reducing the communication burden, making streamlined processes so we [primarily can talk with patients]. It is changing information meaningfully that helps us to get to a better place as a whole; as a cancer community or society as a whole, that trickles down across [clinical practice].

The key is ethical deployment and continuous validation in making sure that it is representative of all of us. I want an algorithm that knows me and can [perform] multimodal [approaches]––now we have potential quantum computing, or we have these advanced agentic models that are able to understand all these different nuances in real-time and run simultaneously in a parallelizable approach. [There are] potential journeys that I should run so at least I can [make] an informed decision [to ensure] these patient outcomes are always the forefront of what we're doing so we can advance science.

Perhaps in the future, we get diagnosed with a multimodal scan that detects cancer before it shows up, with a blood draw and a scan. We can use the large language models [LLMs] and [alternative] models, transformers and beyond––AlphaFold and things like AI-based companies for drug discovery are doing recursion. [Others are] finding molecules that are matched to a specific pocket of [a microtumor] area and generate a vaccine by mRNA or other technologies emerging at scale. You get it off the shelf, take it, and you can target multiple agents and then be done with your cancer before it even becomes a problem. That will be amazing. But for that [to happen], still a lot to work [to do].

References

1. Hernström V, Josefsson V, Sartor H, et al. Screening performance and characteristics of breast cancer detected in the Mammography Screening with Artificial Intelligence trial (MASAI): a randomised, controlled, parallel-group, non-inferiority, single-blinded, screening accuracy study. The Lancet. Published online February 3, 2025. doi:10.1016/S2589-7500(24)00267-X
2. World-leading AI trial to tackle breast cancer launched. News release. Gov.UK. February 4, 2025. Accessed February 7, 2025. https://tinyurl.com/3hna99we