Abstract
Editors’ Introduction
Deep and machine learning artificial intelligence (AI) platforms are increasingly being created to support health care, biomedical and environmental research. As we are watching this new wave of advanced tools being iterated and launched in research, we are interested in understanding the current impact of the adoption of advanced tools using AI across the continuum of biobanking, from communication and consent to data sharing. It is expected that AI will be integrated to enhance data management, operational efficiencies, and research capabilities in biobanking. However, this application may involve risks and ethical concerns. The intent of this Expert Speaks Forum article is to seek information from key opinion leaders across the globe on the extent of early implementation and/or impact of advanced tools in biobanking, including those utilizing AI.
Experts from the World Health Organization (WHO), the United States of America (USA), the United Kingdom (UK), China, and Japan participated in this informal perspectives article.
The Editors asked the authors to introduce themselves and answer the following questions:
In your organization or region, what are the implementations of AI technologies that impact your biobanking practices? How is AI being used to optimize various biobank processes (e.g., collection, storage, retrieval) in your organization or region? Please provide examples. What are the main ethical challenges, whether regional or global, associated with AI applications for data management in biobanking?
Expert Response: Zisis Kozlakidis (IARC, WHO)
Zisis Kozlakidis, PhD, is the Head of Laboratory Services and Biobanking at the International Agency for Research on Cancer, World Health Organization (IARC/WHO) and coordinator of the Biobanking and Cohort Building Network (BCNet). He is responsible for one of the largest international collections of clinical samples worldwide, focusing on gene–environment interactions and disease-based collections. He has significant expertise in the field of biobanking and has served as President of the International Society for Biological and Environmental Repositories (ISBER), and as a board member.
In our organization, the implementation of AI technologies that is gaining ground is the application to digital pathology images held as part of the existing biobank collections. These are images that have been taken at the same time as biological samples and associated clinical information, and as such beyond the annotation of the image itself, can be associated with a limited volume of additional data. However, this is not a routine implementation of AI, but rather project-based and ad hoc. 1 In recent years, the frequency at which such implementations are taking place has increased. Currently, we are trying to expand this capacity, in two distinct routes: one, to increase the images that are available locally within the organization (e.g., by scanning and digitizing existing formalin-fixed paraffin-embedded [FFPE] slides) and two, by applying for targeted grant funding that will allow a similar scanning and digitization approach to take place within low- and middle-income countries (LMICs) and with our colleagues at BCNet. We do believe that this is possible and should result in a more representative global availability of research-relevant images.
How is AI being used to optimize various biobank processes (e.g., collection, storage, retrieval) in your organization or region? Please provide examples
AI is not used directly to optimize biobanking processes, but indirectly. Specifically, there is consideration for the biobank processes relating to new prospective population cohorts to be AI-ready by design. In other words, for the data to be collected in such a way that they are structured, as completely as possible (including their validation for detection of errors), and eventually available for sharing with the research community. There are aspects of AI that are attractive for potential implementation and have been considered by the biobank, but only theoretically, that is, they have not been implemented in any of our biobank processes yet. These are: the use of AI for the creation of more patient-friendly informed consent forms; the use of AI for the description of existing cohorts targeted to different stakeholder groups (e.g., patients, research community, policymakers); the stratification and targeting of patient phenotypes for collection for specific studies; the creation of ‘digital twin’ models of some of our existing collections to allow for training of algorithms before implementing to the existing data sets, and others. 2 The main reasons for not being able to implement any of the above are threefold: the lack of appropriately trained staff, the lack of resources to support such staff, and the lack of computing infrastructure that would be amenable to large-scale implementation of AI processes. Thus, the need for biobanks’ capacities to be supported in becoming AI-able are both urgent and multi-faceted, and this is equally true for biobanks in both high-income and low- and middle-income countries (LMIC) settings.
What are the main ethical challenges, whether regional or global, associated with AI applications for data management in biobanking?
In our view, the main ethical challenges associated with AI applications for data management in biobanking are largely the same as with traditional data management in biobanking. Specifically, privacy is a paramount concern, as biobanks store data that may be sensitive. For example, AI algorithms that analyze and link data at scale are able to re-identify anonymized data that would not have been possible as easily with the current analytical programs. Related to privacy is data security, which has been well addressed so far on a practical level. Another ethical challenge is that of the transparency of use, as informed consents of older collections did not include an explicit mention to the use of data by AI applications. Moreover, there is mention in the scientific literature that AI trained on biased data sets is likely to reflect that bias. How to detect such a bias remains a consideration. However, perhaps the strongest ethical challenge is the possibility that AI becomes a barrier for LMIC-based biobanks to participate in cutting-edge research opportunities. Considering the routes through which our collaborating LMIC-based biobanks, for example, from the BCNet, 3 will be able to develop capacities that will allow them to be AI-ready is both a practical and ethical imperative.
Address correspondence to: Zisis Kozlakidis, PhD, MBA, Head, Laboratory Services and Biobanking, International Agency for Research on Cancer/World Health Organization 25 Avenue Tony Garnier, Lyon, France
Expert Response: Imon Banerjee and Gouri Mahajan (USA)
Dr. Imon Banerjee, Associate Professor at Mayo Clinic Arizona and Co-Director of the AI Innovation Hub, specializes in AI and data mining. Her research focuses on integrating multimodal medical data to develop fair predictive models for clinical diagnosis and treatment. She leads an NCI-funded project creating natural language processing (NLP) tools to extract clinical data from medical notes, collaborating with cancer registries in Georgia and California.
Dr. Gouri Mahajan is Director of the Biorepository and Pathology Research Core and Assistant Professor at Mayo Clinic Arizona. She is ISBER’s regional ambassador for the Americas region. With over 23 years of research experience, her work focuses on biobanking and genomic biomarkers discovery in cancer and mental health disorders.
In your organization or region, what are the implementations of AI technologies that impact your biobanking practices?
In Mayo Clinic, Phoenix, Arizona, AI is being utilized on the de-identified data from the clinic.
In the clinical setting, registration and patient recruitment occur via REDCap, a secure and robust data management system. The registration process consists of the consent and relevant questionnaires completion. This digitally captured data is stored in REDCap, which meets regulatory compliance and protects patient confidentiality.
Once data is secured in REDCap, it undergoes the de-identification process, maintaining the privacy and anonymity of the patients involved. The anonymized data is then transferred to an AI-powered tool for further processing. This AI tool leverages a large language model (LLM) to interpret the input data gathered in REDCap. The LLM generates a structured clinical note or summary, based on variables collected in REDCap.
By automation of this relevant clinical summary generation process, the AI tool helps to streamline clinical workflows, reduce administrative burdens, and enhance the overall efficiency of patient care documentation.
How is AI being used to optimize various biobank processes (e.g., collection, storage, retrieval) in your organization or region? Please provide examples
To optimize biobanking, we aim to create a diverse cancer database by curating long-term clinical and socioeconomic data from patient visits. Manual curation is impractical due to the complexity of integrating multimodal data. This Mayo database will add more depth of clinical information, since existing U.S. cancer registries, such as Surveillance, Epidemiology, and End Results (SEER), focus only on first-course treatment and lack continuous follow-up, essential for tracking long-term outcomes such as cancer recurrence.
We developed a flexible natural language processing (NLP) toolset that extracts clinical and patient-centered data from clinical notes, radiology, and pathology reports at the institutional level. In collaboration with Georgia SEER and California’s cancer registry, we curated data from Emory University Hospital and Stanford Medical Center. This NLP system automatically converts clinical outcomes (e.g., recurrence, treatment, and patient well-being) into structured formats, making them query able and integrable into cancer registry databases.
Research shows racial and ethnic minorities face worse breast cancer outcomes due to systemic biases and health care disparities, though most studies focus on small cohorts. Our NLP toolkit, deployed at Mayo Clinic, Stanford Healthcare, Emory, and Kaiser Permanente, will explore these disparities by analyzing factors such as age, race, and socioeconomic status. This collaborative study aims to provide new insights into long-term breast cancer disparities without requiring expert-level data curation.
This effort will help improve understanding of health disparities and enhance long-term cancer outcomes research.
What are the main ethical challenges, whether regional or global, associated with AI applications for data management in biobanking?
Biobanks, which may store sensitive personal data such as genetic information and health records, can pose significant privacy and security risks, especially with AI-driven systems that aggregate vast datasets. Even when anonymized, these systems can potentially re-identify individuals, highlighting the need for robust data protection measures such as advanced encryption and federated learning. Additionally, if biobank data is not representative of diverse populations, AI models may generate biased outcomes, reinforcing health disparities. To address this, it’s crucial to ensure that biobanks include diverse groups, conduct regular bias audits, and develop transparent AI systems.
Global collaborations in biobank research raise concerns about unequal benefits, particularly between high- and low-income countries. Wealthier nations may dominate access to AI tools, leading to potential exploitation of underrepresented populations. Solutions include implementing international governance frameworks, establishing fair benefit-sharing models, and empowering LMICs with AI infrastructure and resources.
Ethical challenges in cross-border data sharing require harmonized data governance principles, culturally sensitive informed consent, and clear guidelines on ownership and control of data. To promote fairness, international cooperation must prioritize equitable access to both data and the benefits derived from AI-driven health research, ensuring global health disparities are addressed and mitigated.
Address correspondence to: Imon Banerjee, PhD, Associate Professor, Department of Radiology, Department of Artificial Intelligence and Informatics (AI&I), Mayo Clinic Arizona, 6161 E. Mayo Blvd #319, Phoenix, AZ 85054, USA
Gouri Mahajan, MBBS, PhD, Assistant Professor, Department of Laboratory Medicine and Pathology Director—Arizona Biorepository Lab, Mayo Clinic, Arizona, 5777 E. Mayo Blvd, Phoenix, AZ 85054, USA
Expert Response: Weiye Charles Wang (China)
Dr. Weiye Charles Wang is a professor and Deputy Director of the National Engineering Center for Biochip in Shanghai, leading biobanking operations, IT applications, and data-driven innovation for translational medicine. He founded and directed the Xinhua Biobank and contributed to implementing ISO 20387 certification in China.
In your organization or region, what are the implementations of AI technologies that impact your biobanking practices?
While we have not yet fully implemented AI technologies, we are actively designing and developing a system that integrates AI capabilities into biobanking practices. This system focuses on two core aspects of AI: prediction and decision-making, applied across various biobanking scenarios. These scenarios are digitally defined and evaluated, covering the construction, management, and application of biobanks. A typical example is our adoption of the ISO 20387 quality management system through a digital construction approach, which focuses on three key aspects. First, by defining user behavior (the U Route) to assess staff capabilities and performance in real time. Second, by ensuring sample quality (the S Route) through continuous monitoring of key parameters such as biosafety, ethical compliance, and operational conditions. By combining these two routes, AI technologies will help optimize biobank operations and enhance overall quality management in our hospital setting:
U Route (User): This route focuses on the individual as a user, leveraging relevant data features to elucidate the user’s capabilities, suitability for specific roles, and mechanisms for training and assessment. It also includes tracking the user’s daily activities within the biobank and implementing a digital evaluation of their work performance. S Route (Sample): This route revolves around the biological sample as the subject of testing and analysis. It incorporates elements such as personnel, equipment, materials, methods, environment, ethical standards, and biosafety. The quality of the biological samples reflects the overall effectiveness of the quality management system. Integration of U and S Routes: This aspect focuses on how the U Route and S Route can be integrated and balanced to ensure the overall quality of the biobank.
Besides our response to the first question as a relevant example, I would like to provide a different perspective. A practical case we are currently developing focuses on compliance with the national Regulations on the Administration of Human Genetic Resources.1 Our aim is to leverage digital capabilities to ensure the effective construction and operation of biobanks while applying AI technologies to predict risk levels in key processes or scenarios. Additionally, through learning from and analyzing multiple business scenarios, the system can make decisions that balance operational efficiency with compliance with ethical and legal standards, for the “Go or Not Go” determinations.
For example, during the collection and storage of biological samples, the digital and intelligent system verifies each sample against the pre-approved permissions and scope for storage in the biobank. If a sample fails to meet the requirements, the system halts its acceptance and prevents it from entering the biobank for storage.
What are the main ethical challenges, whether regional or global, associated with AI applications for data management in biobanking?
The application of AI technologies in data management for biobanking indeed presents significant ethical challenges, both regionally and globally. These challenges include ensuring compliance with diverse ethical standards, maintaining data privacy and security, and addressing potential biases in AI algorithms. To address these challenges, we should adopt a three-element strategy that balances maximizing AI applications with minimizing ethical and regulatory risks.
Defining Business Scenarios: We begin by defining the biobanking activities and scenarios. For instance, when samples are scanned and cataloged for storage, the system is designed to evaluate key ethical and regulatory elements relevant to each activity, ensuring compliance from the outset. Identifying Potential Risks: For each scenario, we analyze and identify aspects that may pose ethical or regulatory risks. This includes evaluating sensitive data and ensuring transparency and accountability in decision-making. Implementing Digital Monitoring and Risk Mitigation: This aims to detect potential issues promptly and assess their severity. Based on risk assessment, the system provides alerts and suggests mitigation strategies, such as pausing, adjusting, or terminating specific processes to avoid violations of ethical or regulatory standards.
Address correspondence to: Weiye Charles Wang, MD, PhD, Professor and Director of Xinhua Biobank, Xinhua Hospital, Shanghai Jiao Tong University, School of Medicine, 1665 Kong Jiang Rd. Shanghai 200092, PR, China
Expert Response: Gregory H. Grossman (USA)
Dr. Gregory H. Grossman is Chief Scientific Officer of Advancing Sight Network and Executive Director of the Precision Ocular Biobank, with over 20 years of clinical and research experience. An adjunct professor at UAB, he is active in ISBER leadership and specializes in retinal disorders, biobanking, clinical trials, and medical affairs.
As a co-lead of the International Society for Biological and Environmental Repositories (ISBER) AI and Advanced Technologies Taskforce, I have given extensive thought to this subject. Although there is limited use of this nascent technology in biobanking currently, several areas of impact within our field have been imagined.1 From a role in ethics to governance to operational efficiency, a role of AI was discussed by a panel of experts from each domain at the ISBER 2024 Annual Meeting in Melbourne, Symposium 2A.2 However, in my region and organization, it is in data management associated with specimens that is the most fertile ground for deployment. As biobanks generate and utilize a large volume of data, AI is well-suited to revolutionize how data is processed, managed, transformed, and analyzed. In this regard, AI’s implementation in biobanking parallels that in adjacent medical and research fields. Within the Americas, several companies are in active stages of deployment of AI embedded into laboratory information management systems (LIMS) or as standalone platforms that access LIMS data. As an associate editor of the ISBER Best Practices (5th Edition), focusing on the Information Management section, I recognize that data management has become a growing challenge, which I believe AI can aid, as biobanks increasingly deal with a variety of data types, data platforms, and interoperability issues that arise from these complex interactions.
How is AI being used to optimize various biobank processes (e.g., collection, storage, retrieval) in your organization or region? Please provide examples
My organization, Advancing Sight Network, first employed AI on a pilot basis in 2022. As a non-profit that recovers postmortem human eye tissue for both surgical transplantation and research, we are challenged with evaluating donor eligibility and suitability within a narrow window of time. Many biobanks are confronted with this scenario, as fit-for-purpose samples begin with appropriate sample collection. Collection of inappropriate samples can be costly and lead to low utilization rates—ultimately impacting sustainability. In our operations, manually matching donor data to prospective research project criteria has proven to be time-consuming, inefficient, and error prone. Our biggest hurdle in developing a fully automated matching system is that a portion of the donor medical history we access is often provided in free-text form. This unstructured data does not allow for comparison and matching to the standardized data of the criteria. We used natural language processing (a type of AI), to convert this unstructured data into useful and actionable medical data.3 We showed a high rate of success in medical terminology keyword extraction, which was achieved through the correction and standardization of unstructured data, that included medical entity recognition (decoding of medical abbreviations). This ultimately led to optimal donor matching in a fraction of the manual time we previously experienced.
What are the main ethical challenges, whether regional or global, associated with AI applications for data management in biobanking?
The ethical challenges of AI use in biobanking are, in general, the same as those in the medical and scientific fields. I do not see this largely as a regional issue, but a global one, and thus ISBER as an international organization is well-positioned to address. Whether through whitepapers, webinars, taskforces, or the ISBER Best Practices, guidelines need to be developed and employed to protect the security, quality, and accountability of AI-managed biobank data. There will be specific regional regulatory guardrails to ensure ethical use, but measures to promote the greater principles of accountability, fairness, and transparency will be universal. I believe there are many connections between these areas. Fairness in AI aims to address bias and discrimination in data. Transparency of AI seeks to understand how decisions are made and the human oversight of data outcomes or processing. Accountability is the principle that the organization utilizing the AI is ultimately responsible for ensuring ethical risk management and using transparency to build and maintain fairness.
Address correspondence to: Gregory H. Grossman, PhD, CCRP, BCMAS, CEBT, Chief Scientific Officer, Advancing Sight Network, 500 Robert Jemison Rd. Birmingham, AL, USA
Expert Response: Soichi Ogishima (Japan)
Dr. Soichi Ogishima is a Professor at Tohoku University, leading the development of the dbTMM integrated database for genomic, health, clinical, and phenotypic data. He conducts research on deep phenotyping using Electronic Health Records (EHR) data and leads Japan’s biobank network. He holds a PhD in Medicine from Tokyo Medical and Dental University.
In your organization or region, what are the implementations of AI technologies that impact your biobanking practices?
The AI technology influencing biobanking operations is generative AI, which leverages large-scale language models, and it exerts a profound impact across a wide range of biobanking practices. OpenAI o1, for instance, has surpassed an IQ of 120 and is beginning to demonstrate “emergent” capabilities. In the context of biobanking operations, a critical component is the collection of clinical data. While progress has been made in utilizing structured clinical data—such as disease classifications, prescription records, and laboratory test results, the application of unstructured natural language medical records authored by physicians remains costly and has not been effectively scaled. To address this challenge, generative AI is increasingly employed to structure natural language medical records. Generative AI demonstrates significant efficacy in transforming unstructured natural language medical records into structured formats. Medical records often include free form text written by health care professionals, documenting the patient’s name, daily symptom progression, subjective complaints, and other details, which are typically unstructured. By leveraging generative AI, it is now feasible to analyze the content of SOAP (Subjective, Objective, Assessment, and Plan) notes within medical records, extract critical information, and convert it into structured data. Furthermore, generative AI can validate the accuracy of collected data, flag discrepancies, and issue alerts for missing or incorrect data, thereby enhancing the quality of data collection. This capability minimizes the burden on health care professionals by enabling the structuring, collection, and accumulation of clinical data through advanced AI technologies.
How is AI being used to optimize various biobank processes (e.g., collection, storage, retrieval) in your organization or region? Please provide examples.
As mentioned above, generative AI is playing a very important role in the utilization of unstructured natural language medical records in the collection of clinical data. Large-scale language models are also essential for responding to a wide variety of queries. Modern biobanks house not only biological samples donated by participants but also an array of complex datasets, including epidemiological data, genomic data, proteomic data, and other omics data, alongside clinical data. Maximizing the value of biobank samples and data requires researchers to gain in-depth knowledge of their background, including collection methods, associated metadata, and any relevant clinical or demographic context. However, there are limitations to the extent to which personnel at biobanks can respond to such complex inquiries and searches. By utilizing large-scale language models, it becomes possible to perform sophisticated responses to complex inquiries and searches across diverse topics—ranging from biological samples to genomic/omics, epidemiological, and clinical data—within complex contextual frameworks. For users, navigating the complex processes involved in biobank utilization—ranging from biosamples/data understanding to access procedures—can be daunting. Large-scale language models, however, offer an intuitive and efficient interface to bridge this complexity. By enabling users to interact with biobank inquiry contacts through natural language, large language models lower the barriers to accessibility to biosamples/data in biobank. Researchers, for instance, can pose highly specific or interdisciplinary queries in natural language. This capability not only accelerates research but also democratizes access to biobank resources, making them more accessible to a wider range of users, including those with limited technical expertise. Generative AI, underpinned by large-scale language models, is ushering in transformative advancements in scientific domains, including biobanking. In Japan, research and development efforts are actively progressing to integrate such generative AI technologies into biobanking operations, aiming to further enhance their efficiency and potential applications. At present, the Biobank Network Japan is working on prototyping a system that uses a large-scale language model to respond to the various and complex inquiries from users.
What are the main ethical challenges, whether regional or global, associated with AI applications for data management in biobanking?
The integration of AI into biobanking for data management poses significant ethical challenges due to the sensitive nature of the data, the complexity of AI technologies, and diverse regulatory and cultural contexts. Key issues include safeguarding data privacy and security against re-identification risks caused by AI systems and improving transparency in AI decision-making. Additional challenges involve clarifying data control, navigating fragmented global regulatory frameworks, and building public trust through transparent communication and ethical governance. Broader concerns include promoting justice and equity, particularly in low-resource settings, and managing global disparities in ethical standards and governance. Tackling these challenges requires robust governance, interdisciplinary collaboration, public engagement, and ongoing oversight to balance scientific innovation with societal welfare.
Address correspondence to: Soichi Ogishima, PhD, Professor at INGEM/ToMMo, Tohoku University Seiryo 2-1, Aoba, Sendai, Miyagi 980-8575, Japan
Expert Response: Philip Quinlan (UK)
Professor Philip Quinlan is a Digital Engineering and Health Informatics expert at the University of Nottingham, co-leading the federated analytics programme at Health Data Research UK and coordinating the TREvolution programme. His work, which builds on tools deployed during the COVID-19 pandemic, focuses on health data science and digital solutions.
In your organization or region, what are the implementations of AI technologies that impact your biobanking practices?
We are not a biobank, but we have been providing services to support biobanks and data assets for some time. Common data models and common vocabularies have dictated much of the standards work to seek to make biobanks interoperable. The challenge with these standards is always in the ability for a biobank to convert its data to the latest version. Large language models (LLMs) are changing this dynamic, with the potential for AI to assist in the data curation processes, especially when the source data is in natural language. We have some pilot tooling that can help transform informal language into formal data standards.
How is AI being used to optimize various biobank processes (e.g., collection, storage, retrieval) in your organization or region? Please provide examples.
I am not aware of any applications of AI in this area of biobanking yet, within my organization.
What are the main ethical challenges, whether regional or global, associated with AI applications for data management in biobanking?
In the example I gave with LLMs, they often require significant computation and are often provided as central services. Therefore, to use these services, as a biobank, you must send the data to the provider of LLM. There is much debate about the various policies that are applied to those centralized models and how the data supplied may be reused. One of the larger concerns is whether LLMs simply re-enforce previous bias, and if we rely too heavily on LLMs to convert data to standards, we might inadvertently introduce that bias into the datasets. In the feedback loops to help support the development of LLMs, there is also an opportunity for the introduction of bias. None of these challenges are caused by AI but are an artifact of how subjective data curation/conversion is, so we must be careful about always believing the output. Therefore, we must be cautious in the use of models, and consider how to use them most effectively, as they can (1) change the data sharing relationship in unknown ways, and (2) re-enforce previous bias.
Address correspondence to: Philip Quinlan, BSc (Hons), PhD, Professor, University of Nottingham, Queen’s Medical Centre, East Block, Lenton, Nottingham NG7 2UH, UK
Editors’ Conclusions
From the input of experts around the world, we are seeing that AI is beginning to impact biobanking in various ways, including data management, digital image analysis, data curation, decision-making, and supporting biobank operations. Additionally, AI can optimize biobank processes by improving efficiency and enabling interactive data retrieval. Technologies such as NLP and LLMs can play a substantial role in processing, structuring, and analyzing complex datasets, thus streamlining biobank operations. Investment in AI infrastructure is likely to occur centrally in organizations, based on technical resources needed and cost, although training at the biobank level will enhance readiness to adopt and maximize efficiencies.
However, the implementation of AI is associated with several ethical challenges that should be considered with caution. These include concerns about data privacy and security, the potential for reinforcing existing biases, affecting fairness in sample selection, analysis, and decision-making, as well as risks related to accountability, transparency, and regulatory compliance.
It is essential to develop clear governance frameworks and ethical guidelines to ensure AI is used responsibly in biobanking. Organizations such as ISBER, with the input of key opinion leaders in AI, are well-positioned to establish global guidelines for the ethical use of these new advanced technologies in biobanking. These guidelines should address issues related to data privacy, bias, and decision-making transparency. Ongoing monitoring and education for biobank professionals are necessary to address ethical concerns and ensure a responsible approach to AI tool implementation in biobanking.
By investing in AI infrastructure and training, developing ethical frameworks, and promoting global equity, biobanks can harness the full potential of AI while safeguarding privacy, security, and fairness. These are exciting times to increase the impact of quality and use of biospecimens and data through biobanking.
Address correspondence to: Ahmed Samir Abdelhafiz, MD, PhD, MBA, Department of Clinical Pathology, National Cancer Institute, Cairo University, Cairo, Egypt
Marianne K. Henderson, MS, PMP, Division of Cancer Epidemiology and Genetics, National Cancer Institute, NIH, DHHS, Bethesda, MD, USA
Footnotes
Disclaimer
Where authors are identified as personnel of the International Agency for Research on Cancer/WHO, the authors alone are responsible for the views expressed in this article, and they do not necessarily represent the decisions, policy, or views of the International Agency for Research on Cancer/WHO.