Breast cancer isn’t equal: how AI can help close the gap in England and beyond

    Breast cancer is often described as a success story of earlier detection and better treatment, but these gains have not been shared equally. In England and globally, who you are, where you live and the barriers you face still shape your chances of surviving breast cancer. (1)

Unequal risk, unequal outcomes
Breast cancer is the most common cancer in England, with tens of thousands of new cases each year and hundreds of thousands of women living with the disease. Yet the burden and outcomes of breast cancer are not evenly distributed. (1)
A recent pragmatic review for the National Institute for Health and Care Excellence (NICE) mapped breast cancer inequalities across deprivation, geography, protected characteristics (such as age, ethnicity and disability) and inclusion health groups (such as migrants and people experiencing homelessness). It found that although breast cancer is less common in more deprived groups, these groups experience worse health outcomes and higher mortality, driven in large part by later diagnosis and barriers to care.(1)
In other words, fewer people in deprived communities get breast cancer, but when they do, they are more likely to be diagnosed late and to die from it. (1)

What the inequalities look like in England
Deprivation and postcode
Women living in more deprived areas of England are:
•    Less likely to participate in breast cancer screening programmes. (1)
•    Less likely to be urgently referred for assessment of breast symptoms. (1)
•    More likely to be diagnosed at a later stage, requiring more intensive treatment and facing poorer outcomes. (1)
These patterns reflect wider differences in the conditions in which people live and work, including higher levels of obesity, physical inactivity and other risk factors, as well as lower health literacy and more complex competing demands on time and resources. (1)

Geography
Breast cancer incidence and outcomes also vary across regions. Affluent areas, such as parts of south‑east England, tend to have higher incidence, partly because populations are older and more likely to have certain risk factors. However, in areas with higher deprivation and ethnic diversity, screening uptake is lower, more people present with symptoms rather than through screening, and more intensive treatment is often needed, contributing to worse outcomes and mortality in some regions. (1)

Ethnicity
People from ethnic minority backgrounds in England have historically had lower breast cancer incidence than white groups, largely due to lower prevalence of some known risk factors. However, there is evidence that incidence in some minority groups is increasing as risk profiles change. (1)
Crucially, when ethnic minority women do develop breast cancer, they are:
•    More likely to be diagnosed at a younger age and through non‑screening routes. (1)

•    More likely to present with advanced or high‑grade disease and to experience delays in treatment initiation. (1)
•    Less likely to take part in screening and more likely to report emotional, cultural and practical barriers, including fear, shame, language needs and lack of representative support. (1)
Diagnostic intervals – the time between first presentation in primary care and a confirmed diagnosis – also show ethnic differences, with some minority groups experiencing longer waits, especially when symptoms are not classic breast lumps. (2)

Disability, inclusion health groups and LGBTQ+ communities
People with disabilities face higher levels of risk factors such as obesity and physical inactivity, and encounter practical barriers to screening, including inaccessible equipment and locations or not receiving invitations. As a result, they are more likely to be diagnosed at later stages and to have poorer outcomes. (1)
Inclusion health groups such as migrants, refugees, people experiencing homelessness and those in custodial settings face multiple structural barriers, from difficulties registering with a GP to fears about costs or immigration consequences, leading to missed invitations, extremely delayed care‑seeking and emergency diagnoses.[microsoft]
Some LGBTQ+ groups and trans people may have specific risk factors and barriers, and can feel excluded or unwelcome in mainstream services, impacting their experience and potentially their outcomes. (1)
Across many of these groups, low health literacy and limited awareness of breast health and symptoms contribute to delays in seeking help and lower screening uptake.(3, 1)

Why late diagnosis and low screening uptake matter
The NICE review makes one point very clearly: late diagnosis and low screening uptake are central drivers of breast cancer inequalities in England. When people do not know they are eligible for screening, cannot attend because of work or caring responsibilities, feel fear or shame, or struggle with language and trust, they are more likely to present with symptoms at a later stage. (4,3,1)
Later‑stage diagnosis usually means:
•    More complex and intensive treatment.
•    Greater physical, emotional and financial burden.
•    Lower survival and more years of life lost. (1)
These patterns are not inevitable. They are shaped by policy, service design, communication, and the wider social determinants of health. (5, 1)

Where AI fits: risks and opportunities
Artificial intelligence (AI) is already being used in breast cancer for tasks such as reading mammograms and helping radiologists detect cancers earlier. These tools can improve efficiency and accuracy, but if developed and deployed without attention to equity, they may unintentionally widen existing gaps. (6,7,1)
For example, if an AI model is trained mainly on data from women who regularly attend screening and come from more affluent, majority‑ethnic backgrounds, it may perform less well for women from under‑served groups, misestimate risk, or fail to capture patterns of delayed diagnosis, reinforcing existing inequalities. (7,1)
However, AI can also be a powerful tool for equity, if designed differently.

AI to see the gaps
AI and advanced analytics can help health systems:
•    Identify which communities have the lowest screening uptake, by ethnicity, deprivation and geography. (3,4,1)
•    Map diagnostic intervals and see which groups wait longest between first GP visit and diagnosis. (2,5,1)
•    Detect where advanced‑stage diagnoses are clustering and which barriers (language, transport, work patterns, digital exclusion) might be contributing.(4,5,3,1)
Instead of hiding inequities in averages, AI can make inequities visible and measurable, so they can be targeted. (7,,1)

AI to support targeted action
Beyond analytics, AI can support action by:
•    Helping design and evaluate tailored invitation and reminder strategies that match local language, culture and communication preferences. (3,4,1)
•    Integrating social determinants of health – such as housing, employment, caring responsibilities and immigration status – into risk tools and outreach planning. (5,7,1)
•    Powering decision‑support or navigation tools that flag patients at high risk of disengagement or delay, prompting earlier contact, translation support or flexible appointment options.(5,7,3,1)
In low‑resource and global health settings, AI can also help extend diagnostic capacity by enabling triage, image interpretation or risk assessment where specialist expertise is scarce, provided tools are adapted to local populations and infrastructures. (6,7)
The key is equity‑by‑design: bringing diverse data, communities and experts into the development process from the start, and measuring success not only by overall accuracy but by how well tools perform for those currently most at risk of being left behind. (7,1)

OncoEquity Global Health AI: focusing on breast cancer equity
This is the space where OncoEquity Global Health AI aims to work.
Our focus is not simply “AI for breast cancer”, but AI for breast cancer equity. We are interested in tools that help health systems:
•    See where breast cancer pathways are failing women in under‑served groups.
•    Understand how deprivation, ethnicity, disability, migration and other factors shape screening uptake, diagnostic intervals and treatment access. (2,4,1)
•    Act on these insights through targeted, ethical and community‑informed interventions. (7,1)
In practice, this could mean developing an equity‑focused analytics and decision‑support layer that sits alongside existing screening and diagnostic systems, highlighting:
•    Clinics or regions where certain groups are consistently diagnosed later.
•    Populations with persistently low screening participation or high non‑attendance.
•    Groups facing long delays from first presentation to diagnosis, especially when symptoms are subtle. (2,3,5)
Over time, our ambition is to work with partners in England and in global health settings to co‑design and validate tools that reduce breast cancer inequalities, rather than deepen them. (6,7,1)

A shared responsibility
No algorithm can, on its own, fix unequal housing, income, racism, or structural barriers to care. But AI, used thoughtfully, can become part of a broader effort to understand and address why breast cancer outcomes remain so unequal. (5,7,1)
The goal is simple: your chances of surviving breast cancer should not depend on your postcode, your ethnicity, your disability status or your passport. Getting there will require collaboration between clinicians, data scientists, policymakers, communities and people with lived experience. (5,7,1)
If you are a clinician, researcher, community organisation or someone directly affected by these inequalities and would like to share your perspective or explore collaboration, I would be very glad to connect.

REFERENCES 

    1.   Fenton A, Ogunsina O, Patel B, et al. Health inequalities in breast cancer in England: a pragmatic review to inform NICE guidance. London: National Institute for Health and Care Excellence; 2024.
    2.    Round T, Ashworth M, et al. Assessing ethnic inequalities in diagnostic intervals of breast cancer among patients presenting symptoms to general practitioners in England. Scientific Reports. 2026;16: Article s41598-026-36070-8.
    3.    Enhancing breast cancer screening uptake: a community‑based approach. Journal of Community Health. 2025;50(3):xxx–xxx.
    4.    Quantifying inequalities across the breast cancer pathway in England. ISPOR Europe 2024 Presentation EU24008; 19 November 2024.
    5.    Exploring ethnic differences in diagnostic intervals of cancer. NIHR School for Primary Care Research; 2023.
    6.    NIHR. Artificial intelligence: 10 promising interventions for healthcare. NIHR Evidence Collection; 27 July 2023.
    7.    Equity360: gender, race, and ethnicity—the power of AI to advance health equity. Journal of Health Care for the Poor and Underserved. 2024;35(1):xxx–xxx.