Beyond the Algorithm: Using AI Across the Breast Cancer Care Continuum

Each October, Breast Cancer Awareness Month brings renewed attention to the importance of early detection, advances in treatment, and the growing population of breast cancer survivors. This year, it also provides an opportunity to consider how artificial intelligence (AI) may help cancer programs improve care across each of these phases.

AI is already being evaluated and implemented throughout oncology, but its value may be particularly evident in breast cancer, where care generates an enormous amount of imaging, pathology, genomic, clinical, and patient-reported data. The opportunity is not to replace the expertise of radiologists, pathologists, oncologists, nurses, or other members of the multidisciplinary cancer care team. Instead, AI can help clinicians interpret complex information, identify patterns, streamline workflows, and direct attention to the patients who may need it most.

Improving Detection and Risk Assessment

Breast imaging has become one of the most visible applications of AI in cancer care. AI-enabled decision-support systems can analyze mammography and digital breast tomosynthesis to flag suspicious findings, support cancer detection, quantify breast density, and help prioritize examinations for radiologist review.

AI is also expanding the information that can be derived from a mammogram. Emerging image-based risk models can analyze features within a screening mammogram to estimate a patient's future breast cancer risk. This could eventually help identify patients who may benefit from more comprehensive risk assessment, prevention counseling, or individualized screening strategies. However, current evidence remainslargely retrospective, and prospective evaluation across diverse patient populations will be critical before these models can routinely guide risk-based screening.

Another promising application is workflow triage. In a recent study, researchers used an AI risk model to identify patients with abnormal mammograms who were most likely to benefit from expedited diagnostic evaluation. Rather than diagnosing cancer, the model helped determine which patients should move through the diagnostic pathway more quickly. This shows how AI may help reduce delays without removing the radiologist from the process.

Supporting Increasingly Complex Treatment Decisions

Once breast cancer is diagnosed, the amount of information needed to develop an individualized treatment plan can grow rapidly. Histology, stage, receptor status, genomic findings, prior therapies, comorbidities, imaging, patient preferences, and an expanding evidence base may all influence treatment decisions.

AI offers an opportunity to help organize and synthesize this complexity. In digital pathology, AI-enabled tools are being studied for tasks such as identifying tumor features, quantifying biomarkers, and flagging cases for additional pathologist review. Other tools may help clinicians integrate pathology, molecular findings, treatment history, and clinical characteristics with relevant guidelines, evidence, or clinical trials.

AI is also being evaluated in radiation oncology to assist with segmentation and contouring, treatment planning, toxicity prediction, and other repetitive or data-intensive parts of care. Similarly, algorithms that incorporate electronic health record data, laboratory values, patient-reported symptoms, and treatment information may help identify patients at greater risk of toxicity, acute care utilization, or treatment interruption.

Clinical trial matching is another important opportunity, particularly as breast cancer treatment becomes increasingly biomarker driven. AI tools can extract relevant clinical and genomic information and compare it with complex eligibility criteria, allowing the care team to identify candidate trials more efficiently.

Across all of these applications, however, AI should remain a decision-support tool. It should not be used to determine whether a biomarker is actionable, select a systemic therapy or dose, diagnose progression, or establish clinical trial eligibility without confirmation by the appropriate clinicians and/or the research team.

Extending Support Into Survivorship

AI's role in breast cancer care does not have to end when active treatment ends. Breast cancer survivors may experience persistent or late effects, including fatigue, pain, neuropathy, lymphedema, cognitive changes, sexual health concerns, anxiety, and other symptoms that evolve between scheduled visits. Electronic patient-reported outcomes (ePROs) already give patients a way to report these concerns directly. AI could make these systems more actionable by synthesizing repeated symptom assessments, free-text comments, treatment exposures, and changes over time.

Rather than relying on a stand-alone chatbot to manage survivorship concerns, cancer programs could develop AI-enhanced symptom and navigation pathways built on validated ePROs. AI could help identify concerning symptom trajectories or unmet needs and route that information to the appropriate member of the care team for assessment and intervention. Emerging research suggests that AI-supported analysis of patient-reported data may eventually help predict symptom deterioration, treatment-related toxicity, and unplanned health care utilization.

Keeping People at the Center of AI-Enabled Care

Across screening, treatment, and survivorship, the most meaningful question may not be whether AI can perform a particular task, but whether its use improves care for patients and the teams caring for them.

Answering that question requires thoughtful implementation. Cancer programs will need to evaluate tools within their own patient populations and workflows, monitor performance and potential bias, protect patient privacy, and establish clear processes for clinical oversight and escalation.

Breast cancer care has long depended on multidisciplinary collaboration. AI adds another powerful resource to that model, but not another member of the care team. Used responsibly, it can help clinicians find important information sooner, manage growing complexity, and spend more of their time where human expertise matters most: interpreting that information in the context of the individual patient and delivering personalized, compassionate cancer care.