How AI is transforming cancer diagnosis and treatment planning in 2025

Artificial intelligence is no longer a speculative tool in the world of oncology—it is now actively shaping real-world decisions across cancer diagnosis, prognosis, and treatment planning. From multinational drugmakers like AstraZeneca plc (LSE: AZN) investing in biomarker-AI combinations to health systems like the UK’s National Health Service (NHS) embedding AI into clinical workflows, the oncology ecosystem is undergoing a foundational shift.

This transformation is taking place at the intersection of several converging trends. Globally, the cancer burden is rising—more than 20 million new cancer cases are expected annually by 2025, putting immense strain on healthcare infrastructure. At the same time, AI models trained on rich datasets from electronic health records, imaging archives, and genomics are showing tangible improvements in speed, accuracy, and outcomes. As these tools begin to move from pilot phases into daily practice, they are prompting oncologists, hospital administrators, and investors to rethink the architecture of cancer care.

Why Is AI Becoming Critical to Cancer Diagnosis in 2025?

AI’s first and most visible contribution to oncology has come in diagnostic enhancement. In one study funded by AstraZeneca, AI-assisted algorithms helped pathologists improve their accuracy in identifying HER2 protein levels in breast cancer tissue—from 66.7% to 88.5%. This shift not only increases access to HER2-targeted therapies but also reduces the misallocation of high-cost drugs, improving both outcomes and health economics.

In the UK, the NHS has deployed AI tools to interpret “liquid biopsy” tests, which detect circulating tumor DNA (ctDNA) through a simple blood draw. Over 2,000 patients have already benefitted from faster diagnosis using this AI-powered workflow, with decisions on treatment reportedly accelerated by more than two weeks on average. In the U.S., companies like Tempus, Inc. are leading the commercialization of similar diagnostics by merging machine learning with genetic sequencing platforms—attracting attention from both institutional investors and life science-focused venture capital.

The implications are clear: AI is cutting down diagnostic lag, minimizing invasive procedures, and driving earlier treatment decisions. As a result, diagnostic labs and radiology departments—long constrained by manpower bottlenecks—are increasingly viewed as key AI deployment zones in healthcare.

How Are AI Tools Personalizing Cancer Treatment Decisions?

Beyond diagnosis, AI is also proving instrumental in guiding personalized treatment planning. In a recent international clinical study, researchers used AI to predict which subset of men with high-risk, localized prostate cancer would benefit from the drug abiraterone. The algorithm identified a 25% biomarker-positive group for whom the five-year mortality risk dropped from 17% to 9%—a statistically significant and clinically actionable result.

Such risk-stratification approaches represent a paradigm shift. Traditional treatment protocols often generalize across populations; AI is enabling oncologists to tailor regimens based on micro-level biological signatures. At Stanford Medicine, a multimodal AI system is now combining medical imaging with natural language processing to forecast both cancer progression and treatment response. Internal evaluations show that this model outperforms existing clinical staging tools, particularly in solid tumors.

Radiation oncology is another frontier. Cloud-based AI platforms are now automating radiation dose planning, contouring tumor boundaries, and identifying organs at risk—all within hours, compared to manual processes that often span multiple days. This directly boosts throughput in high-volume cancer centers and could help address oncologist shortages in emerging markets like India, Brazil, and South Africa.

Which Institutions Are Already Using AI for Real-World Cancer Care?

The integration of AI into daily oncology practice is no longer limited to elite academic hospitals. In India, AIIMS Patna has deployed AI-based diagnostic equipment for improved cancer detection across imaging and pathology. Doctors at the institute report faster and more consistent diagnostic turnaround times, especially in rural referrals.

In the UK, Chelsea and Westminster Hospital is now using an AI tool called “Derm” to instantly analyze skin lesions for signs of melanoma. Patients receive results within seconds—an improvement that reduces anxiety and eliminates weeks of waiting for dermatology appointments. These deployments showcase AI’s ability to democratize access to cancer screening, particularly in systems with long backlogs or workforce constraints.

Private-sector platforms are also reshaping oncology workflows. IBM‘s earlier Watson for Oncology paved the way for newer entrants like Guardant Health’s AI-assisted liquid biopsy platform and Tempus’ xT test, both of which use real-world evidence to suggest treatments based on a patient’s molecular profile. These companies are driving AI adoption in the U.S. not only by offering clinical utility but also by aligning their platforms with payer reimbursement structures—a key factor in sustainable AI adoption.

What Are the Institutional Challenges and Ethical Debates Surrounding AI in Oncology?

Despite growing uptake, the path to AI maturity in cancer care is not without resistance. One of the most cited concerns is algorithmic bias. AI tools trained primarily on North American or European datasets often underperform when applied in underrepresented populations—a challenge that global firms like AstraZeneca, Roche Holding AG (SWX: RO), and Novartis AG (SWX: NOVN) are now beginning to address by investing in more diverse training cohorts.

Explainability remains another sticking point. Many AI models, particularly those based on deep learning, function as “black boxes”—offering high-confidence outputs without transparent reasoning. For physicians, especially in oncology where life-altering decisions are at stake, this lack of interpretability limits trust and adoption.

From a regulatory perspective, agencies like the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) are still refining frameworks to evaluate and certify clinical-grade AI tools. In markets like India, the regulatory ecosystem is even more nascent, often lagging behind the pace of commercial and academic innovation.

What Is the Future Outlook for AI in the Oncology Industry?

Investor interest in AI-powered cancer platforms remains robust. Firms like Tempus and Freenome have raised hundreds of millions of dollars, betting on AI’s role in early detection, patient stratification, and trial matching. The launch of Scout—a clinical decision support platform integrating trial databases with AI—is a recent example of how companies are merging machine learning with patient data to optimize therapy selection in real-time.

Forward-looking hospital systems are beginning to embed AI into longitudinal cancer care pathways. Northwell Health’s iNav tool is using AI to identify undiagnosed cases by analyzing imaging archives and patient histories, enabling intervention before symptomatic onset. Similarly, Stanford Medicine’s work in using AI to monitor recurrence risks and predict emotional distress signals shows promise for survivorship and quality-of-life monitoring.

Market analysts expect that by 2028, over 60% of cancer treatment centers globally will have adopted at least one AI decision support system in routine care. Venture-backed platforms that demonstrate clinical efficacy, regulatory approval, and payer alignment are likely to see continued M&A interest from large diagnostics firms and biopharma players alike.