The Transformative Impact of Artificial Intelligence on Cancer Care Delivery

The integration of artificial intelligence/machine learning (AI/ML) into oncology is revolutionizing the delivery of cancer care. The development of scalable algorithms is slowly extending into real-world clinical applications. By leveraging modern health care technologies and data-driven insights, AI/ML has the potential to enhance screening, diagnosis, and personalized treatment strategies across multiple cancer types. In March 2024, the National Cancer Institute published a webinar as part of the Events: AI in Cancer Research series. During the session, William Lotter, PhD, assistant professor at Dana-Farber Cancer Institute, described the impact of AI/ML tools on daily clinical workflow and the delivery of care to patients with cancer. Below, I explore how AI/ML is reshaping oncology, with a focus on pivotal shifts and clinical applications.

The Digital Revolution in Oncology

The expanding role of AI/ML in oncology is driven by 2 key shifts: technological advancement and the digitization of health care data. The first involves the advent of sophisticated computational tools, especially deep learning. Over the past decade, deep learning has leveraged algorithmic models to identify intricate patterns in real-world data, enabling breakthrough applications in medical imaging, genomics, and molecular oncology. Coupled with advancements in graphical processing units and cloud computing, AI systems are becoming faster, more accurate, and scalable.

The second shift involves the digital transformation of oncology. Patient data is now stored in electronic health record (EHR) systems, radiology and pathology images have been digitized, and genomic profiling is increasingly being standardized with evidence-based guidelines to streamline clinical practice. This abundance of high-quality, longitudinal data is critical for building AI/ML models tailored to real-life patient care. Eventually, real-time access to such data will enable personalized AI predictions, further advancing the goals of precision oncology.

Clinical Applications: AI/ML Tools in Action

As of July 2023, 692 AI algorithms have been FDA-cleared for medical use, and over 500 of these are for radiology applications. Several categories of AI/ML algorithmic tools are particularly relevant to oncology:

• Computer-Aided Detection (CADe): Augmenting Early Cancer Detection

  CADe systems are designed to assist health care professionals with identifying potentially cancerous lesions in radiographic imaging (eg, mammograms, CT scans, and MRIs). The primary function of CADe systems is to act as a second pair of eyes, highlighting areas of concern for further detailed review by radiologists. For example, in breast cancer screening, CADe tools analyze mammograms to detect subtle abnormalities such as microcalcifications or tiny, aberrant masses that may be easily overlooked. These systems are particularly valuable during early-stage cancer detection, where the success of treatment heavily depends on timely diagnosis.

• Computer-Aided Diagnosis (CADx): Diagnostic Precision

  While CADe focuses on detection, CADx tools enhance the next step in the process: diagnosis. CADx systems can analyze detected abnormalities and comment on the likelihood of malignancy. Radiographic data is synthesized in the context of patient-specific parameters such as age, medical history, and biomarker profiles to describe a more complete diagnostic picture. For example, CADx systems can evaluate lung nodules identified on a CT scan to predict whether they are benign or malignant.

• Computer-Aided Triage (CADt): Prioritizing Critical Cases

  CADt models are designed to optimize workflow efficiency by automatically prioritizing cases that require urgent attention. Using real-time data analysis, these systems flag high-risk cases, enabling clinicians to address critical patients in a timely manner. In real-world cancer care, CADt is often employed by centers where the volume of imaging can be overwhelming. This prioritization reduces delays for patients with an aggressive disease while optimizing overall workflow of a radiology department.

• Computer-Aided Assessment (CADa): Guiding Treatment Decisions
  CADa systems focus on postdiagnostic analysis, providing insights that can inform treatment planning, disease monitoring, and other management considerations. CADa leverages data from multiple sources, including imaging, genomics, and EHR, to assess tumor characteristics and predict patient outcomes. One application of CADa in active development is for prostate cancer: MRI images are analyzed to grade tumor aggressiveness using AI-based scoring. This assessment can assist physicians in deciding between active surveillance and more invasive treatment modalities such as surgery and radiotherapy.

Looking to the Future

The integration of AI/ML technologies into daily clinical workflow is transforming the landscape of cancer care. As society continues to witness advancements in computational power, data accessibility, and algorithmic development, the role of AI/ML in oncology will only expand. However, realizing the full potential of these powerful tools requires collaboration among clinicians, researchers, administrators, computer scientists, and policymakers to address challenges such as data standardization, ethical concerns, and regulatory approvals. Ultimately, cancer care delivery will continue to be reshaped by AI/ML, ensuring that patients benefit from earlier detection, more accurate diagnoses, and better-informed treatment strategies. By embracing this digital revolution, the oncology community is taking a meaningful step toward the vision of precision medicine and improved outcomes for patients worldwide.

Nicole A Colwell, MD, is a senior editor/medical writer for the Association of Cancer Centers.