Artificial intelligence enables earlier detection, faster treatment decisions and precision treatment for individuals. (Photo by Travis Dove for The Washington Post via Getty Images)
Cancer is a major global issue, affecting nearly everyone in some way and ranking as the second leading cause of death worldwide after heart disease.
What fascinates me most is the complexity of the disease, characterised by the uncontrolled division of abnormal cells that can spread rapidly throughout the body. Fortunately, several types of cancer can be cured if they are diagnosed and treated early.
Unfortunately, the global cancer burden is expected to increase in the coming years, driven by aging populations and lifestyle changes, emphasising the need for more innovative approaches.
This is where artificial intelligence (AI) — the ability of machines to perform tasks typically associated with human intelligence such as learning, reasoning and problem-solving — is making exciting breakthroughs, transforming how we understand, diagnose and treat cancer. With AI transforming healthcare, we can begin to envision a future where cancer diagnoses are instantaneous and personalised treatments save lives with minimal side effects.
In cancer research, AI is reshaping personalised clinical care by improving diagnostics, treatment planning and, ultimately, patient outcomes. To mark World Cancer Day on 4 February, this article explores how AI is not just aiding in cancer diagnosis but also shaping personalised treatments and addressing longstanding challenges in the study of cancer (oncology).
When it comes to diagnosing cancer, the study of the disease and its causes and effects (pathology) has always been the gold standard. Yet, we can’t help but notice how much this process relies on human judgment, which can lead to inconsistencies and varying diagnostic accuracy. For clinicians in low-income countries, limited access to pathology services adds another layer of complexity.
Second, the timing and accuracy of cancer detection are key in determining a tumour’s aggressiveness and in guiding treatment strategies. AI has made great strides in this area, sometimes matching the performance of human experts. It also offers the advantage of being scalable and automated, which can improve cancer care.
One powerful AI tool, deep neural networks (DNNs), can analyse large images like biopsy slides to detect cancer cells with high accuracy. These networks can also distinguish between similar cancer types and identify whether tissue is healthy or cancerous.
Deep neural networks are also not limited to analysing biopsy slides — they also work with other medical images such as CT scans, MRIs, mammograms and even photos of skin conditions.
One notable study trained a DNN to classify skin conditions, demonstrating superior performance in diagnosing melanoma (a more aggressive skin cancer that starts in pigment-producing cells) and carcinoma (less aggressive skin cancers that develop in the outer layers of the skin) compared to 21 dermatologists.
Additionally, Google’s AI software outperformed trained radiologists in detecting breast cancer from mammograms.
AI’s role in diagnostics extends beyond DNNs.
Liquid biopsies, for example, use AI to detect cancer-specific biomarkers (biological indicators of disease) such as DNA released by the tumour cells into the blood (circulating tumour DNA), providing a less invasive and more accessible diagnostic tool.
AI also accelerates genomic analyses (studies that look at a person’s DNA to understand their genes), identifying mutations that traditional methods may overlook.
Additionally, AI’s scalability has the potential to address healthcare disparities by providing high-quality diagnostics in regions with limited services to test samples from the body to diagnose cancer, understand how it progresses and guide treatment.
Beyond detection, AI is proving equally transformative in developing personalised treatment strategies for cancer patients.
Personalised medicine has always seemed like the future of cancer care. Integrating genetics, environment, and lifestyle to tailor treatments for individual patients isn’t just innovative, it’s necessary. By analysing large datasets, personalised medicine helps track disease progression, monitor treatment responses and even uncover the molecular basis of drug resistance.
Combining AI and precision medicine is revolutionising personalised care by addressing complex issues. AI algorithms process large, complex datasets to uncover hidden patterns, which are often missed by traditional methods.
These insights can reveal new biomarkers, pathways and therapeutic targets, significantly advancing our understanding of the disease. For clinicians, adopting these technologies will soon be as essential as learning traditional diagnostic skills.
AI also offers practical benefits such as predicting treatment toxicity. For instance, a study showed that skin cancer patients following AI-recommended treatment plans had a 20% improvement in survival rates compared with those receiving standard care.
In rare cancers, AI tools are repurposing existing drugs by identifying new targets. This approach not only broadens treatment options but also accelerates the development of tailored therapies.
While AI offers incredible promise, it is not without problems. Issues such as algorithmic biases, data quality and high implementation costs must be addressed before widespread adoption can occur.
Algorithmic biases also remain a critical concern, particularly when AI systems are trained on datasets that lack diversity. This can lead to discrepancies in diagnostic accuracy across populations. Additionally, ensuring patient data privacy and meeting regulatory standards are significant hurdles.
Addressing these issues will require collaborations between AI developers, clinicians and policymakers.
Looking ahead, AI-driven drug discovery and predictive models accounting for tumour evolution hold transformative potential.
By integrating real-time data from patients, AI could adapt treatment plans dynamically, ensuring therapies remain effective as tumours evolve. These advancements promise a future where oncology is not just reactive but anticipates and prevents cancer progression.
By embracing AI with principles of security, actionable insights, and collaboration, we can reimagine the future of oncology — one where no patient is left behind.
Dr Carla Eksteen is a postdoctoral research fellow and a member of the Cancer Research Group in the Department of Physiological Sciences at Stellenbosch University.