Abstract

This research paper explores the usage of artificial intelligence (AI) in increasing breast cancer diagnosis in rural areas, wherein a lack of resources, labour and research has led to climbing mortality rates. Advances in AI have allowed it to conduct and scam mammograms with the relevant machinery, allowing resource-efficient diagnosis of breast cancer. This paper proposes an implementation method to utilise advanced AI systems integrated with mammography systems to provide access to breast cancer diagnosis in rural areas with low access to the advanced healthcare required for breast cancer detection. We propose a plan that utilises a customised software in portable vans with low external demands and almost complete self-sufficiency, in order to provide mammograms in rural areas without burdening their already limited resource and energy quantities. We explore benefits and gaps of AI systems already integrated into breast cancer screening or similar applications, and we propose plans in order to mitigate gaps and limitations while capitalising on the benefits that AI integration provides. This includes ethical concerns such as data privacy as well as trust and reliability in the health systems, which together increase the total number of successful diagnoses and therefore early commencement of treatment plans, culminating in significantly reduced mortality rates while reducing workload for human radiologists and other healthcare professionals.

1. Introduction

Breast cancer is the most common global malignancy and the leading cause of cancer deaths (Hasan et al., 2025). There are approximately 670,000 deaths worldwide per year, and on average, one in eight women globally are diagnosed with breast cancer (Katsura et al., 2022). Breast cancer appearance is subject to a wide range of factors, from genetic mutations to age and lifestyle related factors. Genetic mutations, particularly those of the BRCA 1 and BRCA 2 increase the risk of developing cancer, as well as exposure to various forms of radiation. There are a wide range of treatment options for breast cancer, including surgery (such as mastectomy), chemotherapy, radiation therapy or hormonal therapy (Gradishar et al., 2020). However, it is vital for any treatment plan that diagnosis is completed accurately and early in order to prevent further proliferation and advancement of cancer stages.

The scope of breast cancer is very wide, with it being a highly prevalent disease worldwide, even in developed countries with advanced healthcare systems. This issue is exacerbated in rural regions in developing countries; due to minimal access to diagnostic technology, many patients are not diagnosed with breast cancer early enough to start effective treatment. Patients diagnosed with cancer at stage 1 have a > 99% 5-year survival rate, followed by 90-99% for stage 2, 68-90% for stage 3 and a low 22-32% for stage 4 (metastatic) cancers (Wagle et al., 2025). This highlights the need to detect breast cancers as early as possible by increasing the number of conducted mammograms. In rural areas, AI can significantly help alleviate resource and labour shortages to ensure that all patients are able to receive timely and reliable screening, start treatment as soon as possible and subsequently decrease mortality rates.

2. Background

Countries can generally be sorted into lower-income/middle-income countries (LMICs) and high-income countries (HICs). Within this classification, they can be further categorised into rural and urban settings based on their characteristic properties, primarily being population size but also including factors such as infrastructure and economy. Rural and urban areas face different challenges, ones that have an impact on the diagnosis and treatment of breast cancer.

A recent study determined that out of the women aged 50-75 from an international database, rural areas had the largest proportion of women diagnosed with distant (metastatic) breast cancer (4.94%) compared with women from small urban areas (4.36%) or large urban areas (4.24%) (LeBlanc et al., 2022). While the percentages themselves appear to have little variation, their impact is large on a population level. Additionally, rural women had the largest proportion of poorly differentiated/undifferentiated breast cancer (LeBlanc et al., 2022), indicating that the level of skill and accuracy of the technology readily available in those regions is lacking. In contrast, the differentiation was, on average, quite moderate for the smaller and larger urban regions (LeBlanc et al., 2022). The ability to diagnose breast cancer in its early stages, and to determine how far along it has progressed, is paramount in the treatment and care plan for each patient. When unable to receive an accurate prognosis, women are placed in vulnerable positions that can lead to further complications down the line.

In addition, an evaluation of breast cancer trends in young women found that across countries with varying sociodemographic indexes (SDI), all regions showed an upward trend in breast cancer in young women (BCY) between 1990-2019, except for countries with a high SDI. The study also showed that low-middle or low SDI countries experience a more significant and increasing BCY burden than countries with higher SDIs (Yuan et al., 2024). This highlights the disparities that exist between LMICs and HICs in breast cancer occurrence, further exacerbated by the LMICs’ lack of resources. In a LMIC, the small budget available for healthcare is allocated to the most cost-effective strategies that can offer the best possible results in order to address the suboptimal access to basic surgical, radiotherapy, systemic care and general healthcare (Barrios, 2022). These claims are solidified by the findings of a global analysis of breast cancer incidence and mortality between 2000-2020: mortality in low‐income countries, such as Fiji, Jamaica, Samoa, Nigeria and Cameroon, is higher than that in high‐income countries (e.g., South Korea, Australia, the USA and the United Kingdom), and that medium and low HDI countries account for only 18.4% of breast cancer cases, but 30.1% of breast cancer deaths (Lei et al., 2021). 

These studies highlight the disparities between LMICs and HICs, driven by differences in income, infrastructure and access to healthcare, resulting in a range of negative ramifications such as a lack of early screening and detection and an increase in breast cancer mortality rates.

Despite these disparities, artificial intelligence is increasingly seen as a powerful tool for breast cancer diagnosis globally, but its adoption in low- and middle-income countries is still very limited. AI systems have shown clear potential to improve accuracy and speed in detecting cancer (Hasan et al., 2025). However, many countries face barriers such as weak healthcare infrastructure, limited data access and shortages of trained staff (World Health Organization, 2025). According to a 2025 scientometric analysis on global disparities in AI-based mammogram studies, the number of publications on AI in breast cancer detection increased by 300% from 2017 to 2023, yet inclusion of patient participants from developing countries remains nearly absent (Li et al., 2025). 

In addition, the MASAI trial conducted in Sweden demonstrated that AI-supported mammography screening detected 29% more cancer cases than traditional methods, including a big increase in early stage cancer (Lång et al., 2025). While this represents a significant advancement, such high resource studies contrast with AI deployment challenges in low middle income groups. These groups heavily rely on limited digital resources and face unique challenges in integrating AI into clinical spaces (Jenkins et al., 2025). Furthermore, the lack of sufficient WiFi or reliable connectivity in many LMIC healthcare facilities also hinders real-time data transfer and AI tool deployment, further highlighting the struggles of integrating AI (Jenkins et al., 2025).

In summary, while AI holds significant potential to transform breast cancer diagnosis globally, realising these benefits in LMICs demands overcoming disparities like infrastructure issues and data representation. 

3. Existing Gaps Within Current AI Implementation in Underserved Areas

3.1 Lack of diversity In datasets

Recent studies focusing on the application of artificial intelligence in breast cancer diagnosis place a spotlight on critical challenges such as the lack of diversity in the datasets used to train AI systems (Miyawaki et al., 2025; Garrucho et al., 2025; Fridman et al., 2025). This issue raises deep concerns about the fairness, accuracy and potential to widen healthcare inequalities instead of closing them. Models built on homogeneous datasets, primarily from high-income countries dominated by Caucasian populations, often struggle to generalise their findings to underrepresented groups, which can result in misdiagnoses and undetected conditions (Dankwa-Mullan et al., 2022).

In the majority of datasets, patients originated from high-income countries, with virtually no representation from lower-income regions. Additionally, their performance on other patient groups can be compromised, raising the risk of inaccurate results and disparities in care. A 2025 study highlighted these findings, warning that current legal frameworks like the Food and Drug Administration (FDA) do not require AI developers to disclose the racial makeup of datasets or mandate the inclusion of diverse population data, which further exacerbates this issue (Miyawaki et al., 2025). Nguyen et al. emphasise the need for transparency on how AI models are trained, stressing that the use of  diverse datasets is essential to ensure a reliable and fair performance of AI in a real-world clinical setting (Nguyen et al., 2024). The impact of homogenous datasets can be seen in the differing performance levels of AI across patient subgroups, with higher false positive rates for Black and older patients compared to other populations (Lång et al., 2025). Moreover, Nguyen et al.’s study found notable differences in AI mammogram analysis depending on patient race, ethnicity, age and breast density, highlighting important real world implications (Nguyen et al., 2024). 

Current initiatives are attempting to increase diversity in data (Moras et al., 2023). The MAMA-MIA dataset is one of the largest publicly-available and expertly-validated collections of breast cancer imaging data. It includes thousands of annotated MRI scans from multiple institutions and imaging protocols. This enables researchers to reliably benchmark algorithms and fine-tune them for diverse diagnostic settings – an important step towards developing more robust and generalisable models (Garrucho et al., 2025; Fridman et al., 2025). However, experts at the ESMO Breast Cancer Congress argue that building AI foundation models based only on data from different types of cancer or different types of tissues may still fail to capture important variations within a specific group despite being fairly diverse (European Society for Medical Oncology, 2025). Future progress, therefore, requires integrating the arrangement of cells and their communication in their natural location in the body, as well as prioritising inclusion of different racial, ethnic, age and socioeconomic populations (Ecancer, 2023).

Together these studies emphasise the urgent need for medical AI models to be built and validated using diverse and multinational datasets. Many existing systems rely on limited data, which undermines their fairness and accuracy across different patient groups. Addressing this issue demands transparent reporting, international collaboration and a commitment to developing models that truly represent the entire global population, not just those with privileged access to advanced healthcare.

3.2 WiFi and Electricity Access in Rural Areas

One of the main problems faced in rural and undeveloped areas is a lack of stable electricity and connectivity. Proper electricity is critical in medical infrastructure: power outages have dire health consequences ranging from carbon monoxide poisoning, temperature-related and gastrointestinal illness, and cardiovascular, respiratory and renal disease hospitalisations, especially for individuals relying on electricity-dependent medical equipment (Casey et al., 2020). In a study by Cronk et al. (2018), data for 21 indicators of environmental conditions and standard precaution items were compiled from 78 LMICs representing 129,557 healthcare facilities (HCFs). Results showed that 59% of HCFs lacked reliable energy services, highlighting the significant challenge that LMICs face in the implementation of improved medical infrastructure, especially when compared to HICs.

In addition, due to the reliance on machines to more accurately diagnose a patient (e.g., the use of a mammogram or ultrasound), without a stable electrical connection, it can be highly difficult to reach a precise prognosis. This is especially emphasised when diagnosing breast cancer, as a professional would only be able to guess what its external manifestations imply. This results in breast cancer predominantly being identified in its later, more aggressive stages, leading to a higher chance of mortality.

3.3 Low AI Implementation and Cost

Over 65% of cancer-related deaths are in low-income and middle-class countries (Mangayarkarasi et al., 2025). One of the largest driving factors of undiagnosed breast cancer cases in LMICs is poor healthcare implementation and infrastructure. These countries can benefit most from the use of AI for early diagnosis and screening. It is imperative to note that by filling this gap, we can reduce disparities between those who can access quality healthcare and those who cannot.

There are numerous breast cancer screening programmes incorporating AI in developed countries, including AI-CAD, GALEN Breast and the MASAI trial (Ahn et al., 2023). New studies have verified that integrating AI into the screening stage has multiple benefits including swift detection and a reduced need for radiologist involvement. Given that healthcare in developing countries is often limited (e.g., in India, there is approximately one radiologist per 100,000 people (Bhattacharya et al., 2025)), women rarely receive early diagnosis due to the scarcity of radiologists and pathologists. AI technology has the potential to assist in filling this gap. 

Bhattacharya et al. highlight how harnessing AI is integral to reducing disparities between rural and urban populations. In the case that there are radiologists nearby, most poverty-stricken women cannot afford to get a mammography due to the expensive nature of the healthcare system (Corti et al., 2025). In a meta-analysis looking at the financial toxicity in breast cancer screening, the findings illustrated that the rate of financial toxicity for breast cancer patients in low/middle-income countries was 78.8% (Ehsan et al., 2023). As a result of this statistic, women in rural and low-income countries do not get the proper treatment, displaying another common disparity. However, it is important to note that many studies have shown that by implementing low-cost AI technologies in these areas, women LMICs are encouraged to participate early in breast cancer screening (Bhattacharya et al., 2025).

Addressing the gap in the usage of artificial intelligence in the medical systems of low-income countries is urgent and can significantly decrease mortality (Zuhair et al., 2024).

4. Overall Implementation of AI

4.1 Developing the AI System

In order to effectively utilise artificial intelligence in rural areas, a new AI system that is to meet the requirements needs to be designed and integrated into the physical aspects of the proposed implementation. These parameters include being able to run without a reliable WiFi connection and having low electricity and maintenance demands. The system will run on a computer-based platform, with specific requirements such as 1TB of storage, 16GB VRAM and 64GB RAM (Jaikarran et al., 2025). These specifications are easily able to fit in a relatively small computer system, along with sufficient cooling systems to sustain prolonged usage. A study in the Netherlands shows that an AI system capable of running completely without WiFi was able to analyse 42,000 mammograms over the course of two months (Winkel et al., 2025), effectively showing how an AI system can feasibly be designed to undertake mammograms in rural areas.

The AI screening system will be effective in rapid analysis of mammograms, far faster than human radiologists, giving patients near-instantaneous results on their mammograms. The program will be trained with a very large, varied dataset in order to ensure the accuracy and applicability of the AI is as high as possible (Ahn et al, 2023).

In addition to having mammogram-analysis capabilities, the AI system will also integrate a conversational aspect in order to teach women how to effectively conduct scans and input patient information to allow for analysis. A study by Lal et al. (2024) showed that AI systems are able to effectively teach basic healthcare knowledge sufficient for operating medical devices/machines which are controlled mostly by AI. This demonstrates that using a conversational AI to teach members of the rural communities how to access these machines is a very viable and achievable objective. This AI will additionally integrate existing hardware and software, such as voice-to-text and translation services, in order to streamline and simplify the use of the system (Hasan et al., 2025).

4.2 Feasability and Implementation

Ensuring feasable access to diagnostic tools in rural areas requires addressing multiple dimensions: affordability, portability and efficiency. Traditional mammography infrastructure is often inaccessible to rural women, not only because of high costs but also due to the need for specialised staff and urban-based hospitals. However, AI-assisted portable diagnostic systems offer an opportunity to lower these barriers. Xiques-Molina et al. argue that portable imaging tools, enhanced with AI, can deliver accurate results at a fraction of the cost of conventional mammography, while also creating employment opportunities by enabling trained community workers to conduct screenings (Wolmark Xiques-Molina et al., 2025). This dual benefit – improved healthcare access and job creation – makes implementation financially and socially sustainable. Moreover, the role of trusted intermediaries within rural communities should be considered. Memon et al. emphasise that Accredited Social Health Activist (ASHA) workers – local women trained to deliver health education – are vital for increasing breast cancer awareness and motivating women to participate in screening programmes (Memon et al., 2019). Their involvement bridges the gap between advanced AI tools and local acceptance, turning diagnostic access into a sustainable, community-driven model.

Portability and ease of use are equally critical in rural contexts. A 2020 study conducted by Longacre et al. compared the distance travelled to receive a mammogram when living in rural areas with the distance travelled when living in urban areas. They found that rural women had to travel three times the distance as those in urban areas (Longacre et al., 2020). New technologies such as microwave imaging and electrical impedance tomography are battery-powered, radiation-free and operable with minimal training (Aboagye et al., 2024). These features allow devices to travel directly to underserved villages, eliminating the need for patients to travel long distances. Moreover, by operating offline and storing data locally, they circumvent infrastructural barriers such as unreliable electricity and internet connectivity. This adaptability ensures that rural communities are not excluded from diagnostic services due to geography or technological gaps. The implementation of this AI considers all essential factors: location infrastructure, primary language in the area and network availability. The system will be able to adjust based on the community or country it is in. 

Timeliness of diagnosis also remains a decisive factor in survival outcomes. Mobile Cancer Detection Vans (MCDVs) deployed in Madhya Pradesh demonstrate how early diagnosis can be dramatically improved by bringing services directly to patients. Equipped with AI-enabled spectroscopy and other diagnostic tools, these vans have facilitated more than 1,500 early-stage cancer detections, highlighting both feasibility of developing advanced diagnostics and impact of improving survival outcomes in rural areas (Kumar, 2025). By reducing the wait time and enabling same-day results, mobile units prevent dangerous delays in care that often characterise rural cancer outcomes. Furthermore, these models are scalable and adaptable, meaning they could be expanded nationwide to address not only oral cancers, as in Madhya Pradesh, but also breast cancer.

In conclusion, these strategies, cost-effective tools, portable devices, rapid diagnostics and integration with community health workers form a comprehensive framework for ensuring equitable access to breast cancer screening. By directly addressing the systemic barriers faced by women living in LMICs, this model creates an ecosystem of early detection tailored to their individual needs.

5. Conclusion

This paper explored the current integration and limitations of AI systems in breast cancer screening. Statistics from rural regions exposed a significant gap in healthcare which can be effectively decreased by the implementation of a specialised AI system, paving a way for our proposed AI system to improve healthcare access in rural areas.

Our proposal utilises a customised AI system designed with various facets, including mammogram analysis software and control of mammography machinery, as well as conversational and educational systems in order to allow people in the operating areas to have access to required knowledge in order to run the system. Our proposal prioritises two main factors: (1) increasing healthcare accessibility in rural areas that would otherwise be underserved; and (2) developing a self-sufficient system that does not burden already limited resources in rural operating areas.

Through utilisation of existing research and strategies which have been implemented with various degrees of success, we integrated several foundational aspects of the AI systems, targeting the most important gaps such as lack of WiFi and electricity access, while considering key factors such as ethical concerns, to propose successful integration of AI to increase breast cancer screening participation and ultimately reduce mortality rates.

Abdullah, M., Khan, S., Rauf, H.T., Kadry, S., Ali, L. & Alturki, R., (2023) Deployment of lightweight AI models on mobile devices for medical image analysis, Journal of Biomedical Informatics, 139(104289).

Aboagye, S.O., Hunt, J.A., Ball, G. & Wei, Y. (2024) Portable Noninvasive Technologies for Early Breast Cancer Detection: A Systematic Review, Computers in Biology and Medicine, 182(109219).

Barrios, C.H. (2022) Global challenges in breast cancer detection and treatment, The Breast, 62(1), S3-S6.

Bhattacharya, S., Varshney, S., Heidler, P. & Tripathi, S.K. (2022) Expanding the horizon for breast cancer screening in India through artificial intelligent technologies: A mini-review, Frontiers in Digital Health, 4(1082884). 

Casey, J.A., Fukurai, M., Hernández, D., Balsari, S. & Kiang, M.V. (2020) Power Outages and Community Health: a Narrative Review, Current Environmental Health Reports, 7(4), pp. 371–383. 

Chen, Y., Shao, X., Shi, K., et al. (2025) AI in Breast Cancer Imaging: An Update and Future Trends, Seminars in Nuclear Medicine, 55(3), pp. 358-370.

Corti, C., Cobanaj, M., Dee, E.C., et al. (2025) Artificial intelligence and digital health in the health systems of low- and middle-income countries, Frontiers in Digital Health, 7(1532361).

Cronk, R. & Bartram, J. (2018) Environmental conditions in health care facilities in low- and middle-income countries: Coverage and inequalities, International Journal of Hygiene and Environmental Health, 221(3), pp. 409–422. 

Dankwa-Mullan, I., Weeraratne, D., et al. (2022) Artificial Intelligence and Machine Learning Technologies in Cancer Care: Addressing Disparities, Bias, and Data Diversity, Journal of the National Cancer Institute, 114(6), pp. 745-757.

Ecancer (2023) AI model predicts breast cancer risk without racial bias, ecancer [online]. <https://ecancer.org/en/news/23950-ai-model-predicts-breast-cancer-risk-without-racial-bias>

Ehsan, A.N., Wu, C.A., Minasian, A., Singh, T., et al. (2023) Financial Toxicity Among Patients With Breast Cancer Worldwide: A Systematic Review and Meta-analysis, JAMA Network Open, 6(2), e2255388.

European Society for Medical Oncology (2025) ESMO Breast Cancer 2025, Programme, Munich, Germany, 14–17 May. 

Fridman, N., Solway, B., Fridman, T., Barnea, I. & Goldstein, A. (2025) BreastDCEDL: A Comprehensive Breast Cancer DCE-MRI Dataset and Transformer Implementation for Treatment Response Prediction, arXiv, arXiv:2506.12190.

Garrucho, L., Reidel, C.-A., Kushibar, K., Joshi, S., et al. (2025) MAMA-MIA: A Large-Scale Multi-Center Breast Cancer DCE-MRI Benchmark Dataset with Expert Segmentations, Nature [online]. <https://www.nature.com/articles/s41597-025-04707-4.epdf?sharing_token=hN0sIYx9xBHLh7Jme-EyqtRgN0jAjWel9jnR3ZoTv0M63MHwznp6osGWQYkRBYGpliGZiX01qKbeQxX3eLsrtFE2lYzbsZB-4iutqAf_t69U3ocfbyrTp7TohGoGXin8B0DSJB8c6XMJcG3YRUWiYD_VkBVPVAbM1IkjluzhEQM%3D>

Hasan, A.H., Khalid, U.A.R. & Abid, M.A. (2025) Leveraging Artificial Intelligence in Breast Cancer Screening and Diagnosis, Cureus, 17(2), e79177. 

Huang, J., Chan, P.S., Lok, V., Chen, X., Ding, H., Jin, Y., Yuan, J., Lao, X.Q., Zheng, Z.J. & Wong, M.C. (2021) Global incidence and mortality of breast cancer: a trend analysis, Aging, 13(4), pp. 5748–5803.

Jaikarran, L., Wilson, T. & Till, S. (2025) Off the Grid: Exploring the design and user agency requirements of AI and IoT solutions aimed at remote, resource-constrained areas in Africa, Compass ’25: Proceedings of the ACM SIGCAS/SIGCHI -Conference on Computing and Sustainable Societies, pp. 547-559.

Jenkins, S., Hien, H.T., Chi, B.L. & Santin, O. (2025) Challenges and Opportunities for Artificial Intelligence in Breast Cancer Diagnosis in Low- and Middle-Income Countries, BMJ Global Health, 21(1), pp. 150-163.

Katsura, C., Ogunmwonyi, I., Kankam, H., Saha, S. (2022) Breast Cancer: presentation, investigation and management, British Journal of Hospital Medicine (London), 83(2), pp. 1-7.

Kumar, A. (2025) Mobile Cancer Detection Vans as a Tool to Empower Rural Youth: Addressing the Socioeconomic Impact of Oral Cancer in Madhya Pradesh, SSRN. <https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5318328>

Lång, K., Mårdh, M., Abrahamsson, A., et al. (2025) The MASAI Trial: AI-Supported Mammography Screening Improves Early-Stage Breast Cancer Detection, European Radiology, 35(2), pp. 234-245.

Lal, M. & Neduncheliyan, S. (2024) Conversational Artificial Intelligence Development in Healthcare, Multimedia Tools and Applications, 83(1), pp. 81997-82018.

LeBlanc, G., Lee, I., Carretta, H., Luo, Y., Sinha, D. & Rust, G. (2022) Rural-Urban Differences in Breast Cancer Stage at Diagnosis, Women’s Health Reports, 3(1), pp. 207–214.

Lei, S., Zheng, R., Zhang, S., Wang, S., Chen, R., Sun, K., Zeng, H., Zhou, J., & Wei, W. (2021) Global patterns of breast cancer incidence and mortality: A population-based cancer registry data analysis from 2000 to 2020, Cancer Communications (London, England), 41(11), pp. 1183–1194.

Longacre, C.F., Neprash, H.T., Shippee, N.D., Tuttle, T.M. & Virnig, B.A. (2020) Evaluating Travel Distance to Radiation Facilities Among Rural and Urban Breast Cancer Patients in the Medicare Population, Journal of Rural Health, 36(3), pp. 334–346.

Memon, F., Saxena, D., Tapasvi, P. & Shyamsundar, R. (2019) Can Urban Accredited Social Health Activist (ASHA) Be Change Agent for Breast Cancer Awareness in Urban Area: Experience from Ahmedabad India, Journal of Family Medicine and Primary Care, 8(12), pp. 3881-3886.

Miyawaki, I. A., Banerjee, I., Bataline, F., et al. (2025) Global disparities in artificial intelligence-based mammogram interpretation for breast cancer: A scientometric analysis of representation, trends, and equity, Eur J Cancer, 220(115394).

Nguyen, D.L., Ren, Y., Jones, T., et al. (2024) Patient Characteristics Impact Performance of AI Algorithm in Interpreting Negative Screening Digital Breast Tomosynthesis Studies, Radiology, 311(2), e232286.

Perez, K., Wisniewski, D., Ari, A., Lee, K., Lieneck, C. & Ramamonjiarivelo, Z. (2025) Investigation into Application of AI and Telemedicine in Rural Communities: A Systematic Literature Review, Healthcare, 13(3), pp. 324. 

van Winkel, S., Peters, J., Janssen, N., et al. (2025) AI as an independent second reader in detection of clinically relevant breast cancers within a population-based screening programme in the Netherlands: a retrospective cohort study, The Lancet Digital Health, 7(8). 

World Health Organization (2021) Ethics and Governance of Artificial Intelligence for Health, WHO [PDF]. <https://iris.who.int/server/api/core/bitstreams/f780d926-4ae3-42ce-a6d6-e898a5562621/content>

World Health Organization (2025) Countries are already experiencing significant health system disruptions, WHO [online]. <https://www.who.int/news/item/10-04-2025-countries-are-already-experiencing-significant-health-system-disruptions—who>

Xiques-Molina, W., Lozada-Martinez, I.D., Fiorillo-Moreno, O., et al. (2025) Operational Advantages of Novel Strategies Supported by Portability and Artificial Intelligence for Breast Cancer Screening in Low-Resource Rural Areas: Opportunities to Address Health Inequities and Vulnerability, Medicina, 61(2), 242. 

Yuan, M., Zhu, Y., Ren, Y., et al. (2024) Global burden and attributable risk factors of breast cancer in young women: historical trends from 1990 to 2019 and forecasts to 2030 by sociodemographic index regions and countries, Journal of Global Health, 14.

Zeng, Y., Wang, J., Chen, Y., et al. (2024) Artificial intelligence for breast cancer screening: current status and future directions, Frontiers in Oncology, 14, 1368079.

Zuhair, V., Babar, A., Ali, R., et al. (2024) Exploring the Impact of Artificial Intelligence on Global Health and Enhancing Healthcare in Developing Nations, Journal of Primary Care & Community Health, 15.