AI in Oncology: Transforming Haematology SCE Practice

Introduction: Why AI Literacy Matters for SCE Candidates

The intersection of artificial intelligence and clinical medicine is no longer a futuristic concept—it is a present-day reality reshaping how haematology and oncology are practised across the NHS and globally. For candidates preparing for the Specialty Certificate Examination (SCE) in Haematology and Oncology, understanding these AI-driven transformations is no longer optional. Recent exam cycles have begun incorporating questions that touch upon molecular profiling, predictive analytics, and AI-assisted diagnostics, reflecting the evolving landscape of precision medicine.

This blog explores the most clinically relevant AI advances in haematology and oncology, providing SCE candidates with a structured overview of what to know, why it matters, and how it may appear in exam scenarios.

AI in Haematology: From Morphology to Molecular Diagnostics

1. Automated Morphological Analysis

Traditional peripheral blood film review has long depended on the trained eye of a haematologist. However, AI-driven digital morphological analysers are now capable of:

• Detecting and classifying abnormal cells with sensitivity comparable to expert haematopathologists

• Flagging dysplastic features in myelodysplastic syndromes (MDS)

• Quantifying blast percentages with high reproducibility

Exam relevance: SCE candidates may encounter vignettes describing a newly diagnosed leukaemia case where AI-assisted morphology has flagged an abnormal cell population. Understanding the strengths and limitations of these tools—particularly that AI serves as an adjunct, not a replacement for expert review—is essential.

2. AI in Flow Cytometry Interpretation

Flow cytometry data is complex, and AI algorithms are increasingly used to:

• Identify aberrant immunophenotypes suggestive of acute leukaemia or lymphoproliferative disorders

• Classify subtypes of non-Hodgkin lymphoma based on surface marker patterns

• Detect minimal residual disease (MRD) with enhanced precision

💡 Clinical Pearl: While AI can enhance MRD detection sensitivity, SCE candidates should remember that clinical correlation with molecular results (e.g., PCR for fusion transcripts) remains the gold standard for treatment response assessment.

AI in Oncology: Precision Treatment and Predictive Modelling

3. Radiomics and Imaging-Based AI

Radiomics extracts quantitative features from imaging studies (CT, MRI, PET) using AI, enabling:

• Tumour characterisation without invasive biopsy in selected cases

• Prediction of treatment response in cancers such as glioblastoma and hepatocellular carcinoma

• Early detection of disease recurrence through serial imaging analysis

Exam relevance: Candidates may be presented with scenarios involving a patient with hepatocellular carcinoma where radiomics has been used to predict recurrence risk following transarterial chemoembolisation (TACE). Understanding that AI-derived data complements—not replaces—established staging systems (e.g., BCLC criteria) is key.

4. Genomic Profiling and AI-Driven Treatment Matching

Next-generation sequencing (NGS) generates vast datasets. AI platforms such as Watson for Oncology and FoundationOne CDx assist by:

• Matching tumour molecular profiles to targeted therapies

• Flagging relevant clinical trials for patients with rare mutations

• Predicting drug resistance patterns based on genomic signatures

In haematological malignancies, AI has been particularly impactful in:

• Acute myeloid leukaemia (AML): Predicting response to venetoclax-based regimens

• Multiple myeloma: Identifying high-risk cytogenetic profiles through machine learning integration of FISH and gene expression data

AI in Cancer Screening and Early Detection

5. Breast Cancer Screening

AI algorithms have demonstrated the ability to:

• Reduce false-positive rates in mammography screening

• Improve cancer detection rates when used as a second reader

The NHS is actively piloting AI-assisted mammography programmes, and SCE candidates should be aware of the ongoing debate regarding whether AI could eventually replace one of the two human readers in the UK double-reading protocol.

6. Colorectal Cancer Screening

AI-enhanced colonoscopy systems can:

• Detect adenomas in real-time (computer-aided detection, CADe)

• Characterise polyp morphology (computer-aided diagnosis, CADx) to support optical biopsy decisions

These advances align with the NHS Long Term Plan's ambition to diagnose 75% of cancers at stage I or II by 2028.

AI in Haematology and Oncology: Key Clinical Trials and Evidence

How AI Topics May Appear in SCE Questions

Based on recent exam trends and the evolving curriculum, AI-related concepts may appear in the following formats:

1. Best-of-Five questions: Describing a molecular profiling result and asking about the most appropriate targeted therapy

2. Data interpretation: Providing radiomic or genomic data and asking about prognostic implications

3. Ethical scenarios: Presenting a situation where AI has flagged an abnormality missed by a clinician, testing knowledge of governance and patient communication

4. Trial interpretation: Presenting summary statistics from an AI-enhanced screening trial and asking about clinical applicability

Key AI Concepts Every SCE Candidate Should Know

Precision Medicine

AI enables patient-specific treatment recommendations by integrating multi-omic data (genomic, transcriptomic, proteomic) with clinical parameters.

Explainable AI (XAI)

The push for transparency in AI decision-making is particularly relevant in oncology, where treatment decisions carry significant consequences. SCE candidates should understand that regulatory bodies (including the MHRA and FDA) increasingly require AI tools to provide interpretable outputs.

Bias and Generalisability

AI models trained predominantly on Caucasian populations may underperform in ethnically diverse cohorts—a critical consideration for the UK's diverse population. Candidates should be able to critically appraise this limitation.

Data Governance

The use of patient data for AI model training raises important GDPR and confidentiality considerations. The NHS's National Data Guardian principles and the Department of Health's Code of Conduct for AI in Healthcare provide the framework.

Practical Study Recommendations

Resources for SCE Preparation

1. NHS AI Lab publications: Provides authoritative summaries of AI initiatives in cancer care

2. The Lancet Digital Health: High-yield for recent clinical validation studies

3. NICE Evidence Standards Framework for AI: Essential for understanding how AI tools are evaluated for NHS adoption

4. BSH (British Society for Haematology) guidelines: Increasingly reference AI-assisted diagnostics

Study Strategy

• Do not memorise algorithms: SCE tests clinical reasoning, not computer science

• Focus on clinical applications: Understand what AI tools exist, their validated uses, and their limitations

• Stay current with guidelines: NICE, ESMO, and ASH guidelines increasingly incorporate AI-assisted recommendations

• Practise with scenario-based questions: Familiarise yourself with vignette formats that integrate AI findings

The Future: What's on the Horizon?

Liquid Biopsies and AI

Circulating tumour DNA (ctDNA) analysis generates enormous datasets. AI is being used to:

• Detect cancer earlier in asymptomatic individuals

• Monitor treatment response in real-time

• Identify emerging resistance mutations before clinical progression

Federated Learning

This approach allows AI models to learn across multiple NHS trusts without sharing patient data, addressing privacy concerns while improving model accuracy—a concept SCE candidates may encounter in questions about data governance.

Conversational AI in Oncology

Large language models are being explored for:

• Generating patient information leaflets in multiple languages

• Summarising complex multidisciplinary team (MDT) discussions

• Assisting with chemotherapy regimen verification

Conclusion: Embracing the AI Era in SCE Preparation

Artificial intelligence is fundamentally altering the practice of haematology and oncology. For SCE candidates, this represents both a challenge and an opportunity. By understanding the clinical applications, limitations, and ethical considerations of AI tools, candidates can demonstrate not only factual knowledge but also the critical appraisal skills that define a competent specialist.

The key message for exam preparation is clear: AI in medicine is not about replacing the clinician—it is about augmenting clinical decision-making. Understanding this distinction will serve candidates well, both in the examination hall and in their future practice as consultants.

📖 Final Tip: When encountering AI-related content in SCE questions, always anchor your answer in established clinical guidelines. AI outputs are adjunctive; clinical judgement remains the cornerstone of specialist practice.

Good luck with your SCE preparation!