The Promise of AI for Cancer Care Showed No Sign of Slowing in 2024

8 MIN READ

Throughout 2024, Artificial Intelligence (AI) continued to gather momentum across the healthcare landscape. The technology, which has skyrocketed to the mainstream over the past few years, particularly showed promise in the field of oncology.

In this article:

• What were the macro developments in AI for healthcare?

• What were the key updates in AI-powered tools for cancer care?

• What's ahead for AI in cancer care in 2025?

AI Developments from a Macro Healthcare Perspective

In January 2024, the White House released a briefing stating that, following a landmark Executive Order by President Biden in October 2023 to “ensure that America leads the way in seizing the promise and managing the risks of artificial intelligence”, an AI Task Force had been established at the Department of Health and Human Services. The Task Force’s remit is to “develop policies to provide regulatory clarity and catalyze AI innovation in health care.” As an example, the Task Force will work on developing and evaluating AI-enabled tools and frameworks for use in advancing drug development, health care delivery, and other public health endeavors.” This update highlights how federal policy could be a significant enabler of future adoption of AI in healthcare, though it remains to be seen how the imminent change in presidential administration may impact this momentum.

Additionally, health care organization leaders have been showing strong signs of embracing digital transformation with new technologies, including Generative AI. In fact, according to the Deloitte AI Institute, a 2024 Life Sciences and Health Care Generative AI Outlook Survey showed that “75% of leading health care companies are experimenting with Generative AI or attempting to scale use cases”.

Key Updates in AI-Powered Tools for Cancer Care

2024 saw a number of notable developments as AI continues to find use cases across the care continuum. Below are five prominent examples.

The First FDA-Approved AI-Powered Skin Cancer Diagnostic Tool

To kick off the year, the Food and Drug Administration (FDA) this past January approved the first AI-powered noninvasive tool to diagnose skin cancer, including the three most common types - melanoma, basal cell carcinoma, and squamous cell carcinoma. The handheld device, called DermaSensor, “uses spectroscopy technology to examine lesions at cellular and subcellular levels, then analyze those characteristics using an FDA-cleared algorithm.” This tool enables providers to investigate suspicious lesions at the point of care.

According to Targeted Oncology, in a Mayo Clinic-led validation study of over 1000 patients (“DERM-SUCCESS”), the device demonstrated ~96% sensitivity across all 224 types of skin cancers, and had a negative predictive value of ~97%, meaning that there was only a 3% chance of malignancy when there is a negative result. Also noteworthy is the fact that the device sensitivity was superior to the investigator-reported sensitivity of 83%, highlighting how its AI algorithm might help PCPs refer more of the right patients to dermatologists.

The First Guidelines-Included AI-enabled Cancer Prognostic Tool

According to Business Wire, Artera, a precision medicine company developing AI tests to personalize cancer therapy, announced in March the inclusion of its flagship ArteraAI Prostate Test in the NCCN Clinical Practice Guidelines in Oncology February 2024 update, making it the first and only AI-enabled predictive and prognostic test in guidelines for localized prostate cancer.

According to Artera’s website, the test is “intended to identify patients who will benefit from therapy intensification and help guide treatment decisions for men with localized prostate cancer.” The AI-powered test specifically works by leveraging a unique algorithm that analyzes digital images from a patient’s biopsy, learns from the patient’s clinical data, and combines the different insights to predict whether a patient will benefit from hormone therapy and estimate long-term outcomes.

This update comes after another significant update for the ArteraAI Prostate Test earlier this year in January when The Centers for Medicare & Medicaid Services (CMS) established a payment rate for the test, according to Urology Times.

The NCCN guidelines inclusion marks a significant point of validation for AI in not only prostate cancer but in the broader oncology space and clinical practice in general.

The AI Model that Predicts Patients’ Response to Checkpoint Inhibitors

The National Institutes of Health (NIH) announced in June that a research team from the National Cancer Institute and Memorial Sloan Kettering Cancer Center developed an AI scoring model that could help predict how a patient’s likelihood to benefit from checkpoint inhibitor treatment. The model, named “Logistic Regression-based Immunotherapy-response Score”, or LORIS, was a product of a machine learning-powered data analyses of over 2880 cancer patients treated with checkpoint inhibitors, across 18 different types of solid tumors.

LORIS is based on an algorithm of only six clinical data points, including tumor mutational burden (TMB) - a well validated biomarker, patient’s age, cancer type, history of cancer therapy, blood albumin level, and blood neutrophil-to-lymphocyte ratio (NLR). Notably, it outperformed other more complex models with many more clinical feature variables in predicting patient likelihood of responding to treatment.

While tools like LORIS show great potential, adopting them widely into routine clinical workflow may still be further out in the future until they are successfully validated in larger clinical studies. Other factors, such as clinical guidelines, policy, and regulation will also play significant roles in broader adoption.

The AI Model that Performs Multiple Cancer Evaluation Tasks

The Harvard Gazette announced in September that a research team at Harvard Medical School, led by Kun-Hsing Yu, M.D, Ph.D. developed a new AI model called “Clinical Histopathology Imaging Evaluation Foundation”, or CHIEF, that completes multiple cancer evaluation tasks across multiple cancer types by reading digital slides of tumor tissues. Specifically, the model has the ability to detect cancer cells, predict a tumor’s molecular profile, forecast patient survival, and identify features in the tumor microenvironment that can inform a patient’s likely response to standard treatments, amongst other potential capabilities.

According to the The Gazette, the CHIEF AI model achieved nearly 94% cancer sensitivity and significantly outperformed current AI approaches across 15 datasets and 11 cancer types.

The model builds upon Dr. Yu’s previous research in AI systems that completed specific tasks in specific cancer types such as colon cancer, and highlights the potential for AI models to “enhance clinicians’ ability to evaluate cancers efficiently and accurately, including the identification of patients who might not respond well to standard cancer therapies.”

The AI Tool that Saves Oncologists Time on Analyzing Patient Records

This past October, GE Healthcare announced a new cloud-based AI application called CareIntellect for Oncology intended to help oncologists quickly get up to speed on a patient’s history and disease progression by presenting them with key actionable data points sourced from disparate sources. This helps clinicians to determine potential next steps for each unique patient.

As reported by CNBC, CareIntellect for Oncology may also help identify relevant clinical trials for which patients might be eligible based on their history and adherence to treatment protocols. This would save oncologists time, notably because the process of finding eligible patients for specific clinical trials is particularly tedious.

This development comes at a time where clinicians are increasingly burdened by high volumes of patient data from disparate sources, which makes the process of sorting through and understanding each unique patient’s history time-consuming and sometimes frustrating.

Looking Ahead into 2025

With all the AI momentum gathered in cancer care in 2024, there are no signs of slowing down in 2025 and beyond. From enhancing clinical decision making to easing the burden of documentation, AI will most likely continue to see adoption in very specific use cases across cancer care. That said, several critical success factors including favorable policy and regulation, patient and caregiver trust in AI, and continued funding of AI innovation and R&D, will be critical for future advancements.

Subscribe here to receive Oncology AI Trends updates directly in your inbox.