Generative Artificial Intelligence marks a critical inflection point in the evolution of machine learning systems, enabling the autonomous synthesis of content across text, image, audio, and biomedical domains. While these capabilities are advancing at pace, their deployment raises profound ethical, security, and privacy concerns that remain inadequately addressed by existing governance mechanisms. This study undertakes a systematic inquiry into these challenges, combining a PRISMA-guided literature review with thematic and quantitative analyses to interrogate the socio-technical implications of generative Artificial Intelligence. The article develops an integrated theoretical framework, grounded in established models of technology adoption, cybersecurity resilience, and normative governance. Structured across five lifecycle stages (design, implementation, monitoring, compliance, and feedback) the framework offers a practical schema for evaluating and guiding responsible AI deployment. The analysis reveals a disconnection between the fast adoption of generative systems and the maturity of institutional safeguards, resulting with new risks from the shadow Artificial Intelligence, and underscoring the need for adaptive, sector-specific governance. This study offers a coherent pathway towards ethically aligned and secure application of Artificial Intelligence in national critical infrastructure.

This review examines the various ways artificial intelligence (AI) has been utilized in medical education (MedEd)and presents ideas that will ethically and effectively leverage AI in enhancing the learning experience of medical trainees.

AI has improved accessibility to learning material in a manner that engages the wider population. It has utility as a reference tool and can assist academic writing by generating outlines, summaries and identifying relevant reference articles. As AI is increasingly integrated into MedEd and practice, its regulation should become a priority to prevent drawbacks to the education of trainees. By involving physicians in AI design and development, we can best preserve the integrity, quality, and clinical relevance of AI-generated content. In adopting the best practices for AI use, we can maximize its benefits while preserving the ethical standards of MedEd with the goal of improving learning outcomes.

Anonymized biomedical data sharing faces several challenges. This systematic review analyzes 1084 PubMed-indexed studies (2018-2022) using anonymized biomedical data to quantify usage trends across geographic, regulatory, and cultural regions to identify effective approaches and inform implementation agendas. We identified a significant yearly increase in such studies with a slope of 2.16 articles per 100,000 when normalized against the total number of PubMed-indexed articles (p = 0.021). Most studies used data from the US, UK, and Australia (78.2%). This trend remained when normalized by country-specific research output. Cross-border sharing was rare (10.5% of studies). We identified twelve common data sources, primarily in the US (seven) and UK (three), including commercial (seven) and public entities (five). The prevalence of anonymization in the US, UK, and Australia suggests their practices could guide broader adoption. Rare cross-border anonymized data sharing and differences between countries with comparable regulations underscore the need for global standards.

Artificial intelligence (AI) has the potential to revolutionize the management of heart failure. AI-based tools can guide the diagnosis and treatment of known risk factors, identify asymptomatic structural heart disease, improve cardiomyopathy diagnosis and symptomatic heart failure treatment, and uncover patients transitioning to advanced disease. By integrating multimodal data, including omics, imaging, signals, and electronic health records, state-of-the-art algorithms allow for a more tailored approach to patient care, addressing the unique needs of the individual. The past decade has led to the development of numerous AI solutions targeting each aspect of the heart failure syndrome. However, significant barriers to implementation remain and have limited clinical uptake. Data-privacy concerns, real-world model performance, integration challenges, trust in AI, model governance, and concerns about fairness and bias are some of the topics requiring additional research and the development of best practices. This review highlights progress in the use of AI to guide the diagnosis and management of heart failure while underscoring the importance of overcoming key implementation challenges that are currently slowing progress.

There is an urgent need to develop tools to enable older adults to live healthy, independent lives for as long as possible. To address this need, the National Institute on Aging (NIA) Artificial Intelligence and Technology Collaboratories (AITCs) for Aging Research were created to identify, develop, evaluate, commercialize, and disseminate innovative technologies and artificial intelligence (AI) methods to promote healthy aging and to support persons with Alzheimer's disease and Alzheimer's disease-related dementias (AD/ADRD). In 2023, AITC pilot grant applicants were required to answer questions about how, if at all, they would safeguard older adults' data privacy and confidentiality, advance health equity, address bias, and protect vulnerable participants. Our team analyzed applicants' answers to these ethics-focused questions using a constructivist grounded theory approach. In this article, we present what we learned and discuss modifications to our approach moving forward.

In summer 2024, the Children's Bureau revised the rules that govern data suppression in datasets from the National Child Abuse and Neglect Data System (NCANDS). To minimize the risk of re-identification, researchers had previously been unable to identify counties with fewer than 1000 annual maltreatment cases. Under the new data suppression rule, county identifiers will only be suppressed for counties with fewer than 700 cases. In this report, we document the consequences of this shift for research data access and re-identification risks, showing that reducing the data suppression threshold increased the number of identifiable counties from 835 to 1096 (a 31.3% increase) and doubled the number of identifiable rural counties. The percentage of reported children who face a particularly elevated re-identification risk due to having unique demographic characteristics increased from 0.7% to 1.0%.

With the introduction of the European Health Data Space (EHDS), the secondary use of health data for research purposes is attracting more attention. Secondary health data processing promises to address novel research questions, inform the design of future research and improve healthcare delivery generally. To comply with the existing data protection regulations, the secondary data use must be fair, among other things. However, there is no clear understanding of what fairness means in the context of secondary use of health data for scientific research purposes. In response, we conducted a scoping review of argument-based literature to explore how fairness in the secondary use of health data has been conceptualized. A total of 35 publications were included in the final synthesis after abstract and full-text screening. Using an inductive approach and a thematic analysis, our review has revealed that balancing individual and public interests, reducing power asymmetries, setting conditions for commercial involvement, and implementing benefit sharing are essential to guarantee fair secondary use research. The findings of this review can inform current and future research practices and policy development to adequately address concerns about fairness in the secondary use of health data.

Deep neural networks (DNNs) have remarkable potential to reconstruct ultrasound images. However, this promise can suffer from overfitting to training data, which is typically detected via loss function monitoring during an otherwise time-consuming training process or via access to new sources of test data. We present a method to detect overfitting with associated evaluation approaches that only require knowledge of a network architecture and associated trained weights. Three types of artificial DNN inputs (i.e., zeros, ones, and Gaussian noise), unseen during DNN training, were input to three DNNs designed for ultrasound image formation, trained on multi-site data, and submitted to the Challenge on Ultrasound Beamforming with Deep Learning (CUBDL). Overfitting was detected using these artificial DNN inputs. Qualitative and quantitative comparisons of DNN-created images to ground truth images immediately revealed signs of overfitting (e.g., zeros input produced mean output values ≥0.08, ones input produced mean output values ≤0.07, with corresponding image-to-image normalized correlations ≤0.8). The proposed approach is promising to detect overfitting without requiring lengthy network retraining or the curation of additional test data. Potential applications include sanity checks during federated learning, as well as optimization, security, public policy, regulation creation, and benchmarking.

Cardiovascular disease (CVD) remains the foremost cause of morbidity and mortality worldwide. Recent advancements in machine learning (ML) have demonstrated substantial potential in augmenting risk stratification for primary prevention, surpassing conventional statistical models in predictive performance. Thus, integrating ML with Electronic Health Records (EHRs) enables refined risk estimation by leveraging the granularity and breadth of longitudinal individual patient data. However, fundamental barriers persist, including limited generalizability, challenges in interpretability, and the absence of rigorous external validation, all of which impede widespread clinical deployment.

This review adheres to the methodological rigor of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and Scale for the Assessment of Narrative Review Articles (SANRA) guidelines. A systematic literature search was performed in March 2024, encompassing the Medline and Embase databases, to identify studies published since 2010. Supplementary references were retrieved from the Institute for Scientific Information (ISI) Web of Science, and manual searches were curated. The selection process, conducted via Rayyan, focused on systematic and narrative reviews evaluating ML-driven models for long-term CVD risk prediction within primary prevention contexts utilizing EHR data. Studies investigating short-term prognostication, highly specific comorbid cohorts, or conventional models devoid of ML components were excluded.

Following an exhaustive screening of 1757 records, 22 studies met the inclusion criteria. Of these, 10 were systematic reviews (four incorporating meta-analyses), while 12 constituted narrative reviews, with the majority published post-2020. The synthesis underscores the superiority of ML in modeling intricate EHR-derived risk factors, facilitating precision-driven cardiovascular risk assessment. Nonetheless, salient challenges endure heterogeneity in CVD outcome definitions, undermine comparability, data incompleteness and inconsistency compromise model robustness, and a dearth of external validation constrains clinical translatability. Moreover, ethical and regulatory considerations, including algorithmic opacity, equity in predictive performance, and the absence of standardized evaluation frameworks, pose formidable obstacles to seamless integration into clinical workflows.

Despite the transformative potential of ML-based CVD risk prediction, it remains encumbered by methodological, technical, and regulatory impediments that hinder its full-scale adoption into real-world healthcare settings. This review underscores the imperative circumstances for standardized validation protocols, stringent regulatory oversight, and interdisciplinary collaboration to bridge the translational divide. Our findings established an integrative framework for developing, validating, and applying ML-based CVD risk prediction algorithms, addressing both clinical and technical dimensions. To further advance this field, we propose a standardized, transparent, and regulated EHR platform that facilitates fair model evaluation, reproducibility, and clinical translation by providing a high-quality, representative dataset with structured governance and benchmarking mechanisms. Meanwhile, future endeavors must prioritize enhancing model transparency, mitigating biases, and ensuring adaptability to heterogeneous clinical populations, fostering equitable and evidence-based implementation of ML-driven predictive analytics in cardiovascular medicine.

Major depressive disorder (MDD), a psychiatric disorder characterized by functional brain deficits, poses considerable diagnostic and treatment challenges, especially in adolescents owing to varying clinical presentations. Biomarkers hold substantial clinical potential in the field of mental health, enabling objective assessments of physiological and pathological states, facilitating early diagnosis, and enhancing clinical decision-making and patient outcomes. Recent breakthroughs combine neuroimaging with machine learning (ML) to distinguish brain activity patterns between MDD patients and healthy controls, paving the way for diagnostic support and personalized treatment. However, the accuracy of the results depends on the selection of neuroimaging features and algorithms. Ensuring privacy protection, ML model accuracy, and fostering trust are essential steps prior to clinical implementation. Future research should prioritize the establishment of comprehensive legal frameworks and regulatory mechanisms for using ML in MDD diagnosis while safeguarding patient privacy and rights. By doing so, we can advance accuracy and personalized care for MDD.

This report addresses the technical aspects of de-identification of medical images of human subjects and biospecimens, such that re-identification risk of ethical, moral, and legal concern is sufficiently reduced to allow unrestricted public sharing for any purpose, regardless of the jurisdiction of the source and distribution sites. All medical images, regardless of the mode of acquisition, are considered, though the primary emphasis is on those with accompanying data elements, especially those encoded in formats in which the data elements are embedded, particularly Digital Imaging and Communications in Medicine (DICOM). These images include image-like objects such as Segmentations, Parametric Maps, and Radiotherapy (RT) Dose objects. The scope also includes related non-image objects, such as RT Structure Sets, Plans and Dose Volume Histograms, Structured Reports, and Presentation States. Only de-identification of publicly released data is considered, and alternative approaches to privacy preservation, such as federated learning for artificial intelligence (AI) model development, are out of scope, as are issues of privacy leakage from AI model sharing. Only technical issues of public sharing are addressed.

As artificial intelligence (AI) gains prominence in pathology and medicine, the ethical implications and potential biases within such integrated AI models will require careful scrutiny. Ethics and bias are important considerations in our practice settings, especially as an increased number of machine learning (ML) systems are being integrated within our various medical domains. Such ML-based systems have demonstrated remarkable capabilities in specified tasks such as, but not limited to, image recognition, natural language processing, and predictive analytics. However, the potential bias that may exist within such AI-ML models can also inadvertently lead to unfair and potentially detrimental outcomes. The source of bias within such ML models can be due to numerous factors but is typically categorized into 3 main buckets (data bias, development bias, and interaction bias). These could be due to the training data, algorithmic bias, feature engineering and selection issues, clinic and institutional bias (ie, practice variability), reporting bias, and temporal bias (ie, changes in technology, clinical practice, or disease patterns). Therefore, despite the potential of these AI-ML applications, their deployment in our day-to-day practice also raises noteworthy ethical concerns. To address ethics and bias in medicine, a comprehensive evaluation process is required, which will encompass all aspects of such systems, from model development through clinical deployment. Addressing these biases is crucial to ensure that AI-ML systems remain fair, transparent, and beneficial to all. This review will discuss the relevant ethical and bias considerations in AI-ML specifically within the pathology and medical domain.

In the current paper, we will focus on requirements to ensure big data can advance the outcomes of our patients suffering from kidney disease. The associated ethical question is whether and how we as a nephrology community can and should encourage the collection of big data of our patients. We identify some ethical reflections on the use of big data, and their importance and relevance. Furthermore, we balance advantages and pitfalls and discuss requirements to make legitimate and ethical use of big data possible.  The collection, organization, and curation of data come upfront in the pipeline before any analyses. Great care must therefore be taken to ensure quality of the data at this stage, to avoid the 'garbage in garbage out' problem and suboptimal patient care as a consequence of such analyses.  Access to the data should be organized so that correct and efficient use of data is possible. This means that data must be stored safely, so that only those entitled to do so can access them. At the same time, those who are entitled to access the data should be able to do so in an efficient way, so as not to hinder relevant research.  Analysis of observational data is itself prone to many errors and biases. Each of these biases can finally result in provision of low-quality medical care. Secure platforms should therefore also ensure correct methodology is used to interpret the available data. This requires close collaboration of a skilled workforce of experts in medical research and data scientists. Only then will our patients be able to benefit fully from the potential of AI and big data.

Quality improvement, clinical research, and patient care can be supported by medical predictive analytics. Predictive models can be improved by integrating more patient records from different healthcare centers (horizontal) or integrating parts of information of a patient from different centers (vertical). We introduce Distributed Cross-Learning for Equitable Federated models (D-CLEF), which incorporates horizontally- or vertically-partitioned data without disseminating patient-level records, to protect patients' privacy. We compared D-CLEF with centralized/siloed/federated learning in horizontal or vertical scenarios. Using data of more than 15,000 patients with COVID-19 from five University of California (UC) Health medical centers, surgical data from UC San Diego, and heart disease data from Edinburgh, UK, D-CLEF performed close to the centralized solution, outperforming the siloed ones, and equivalent to the federated learning counterparts, but with increased synchronization time. Here, we show that D-CLEF presents a promising accelerator for healthcare systems to collaborate without submitting their patient data outside their own systems.

Precision medicine involves a paradigm shift toward personalized data-driven clinical decisions. The concept of a medical "digital twin" has recently become popular to designate digital representations of patients as a support for a wide range of data science applications. However, the concept is ambiguous when it comes to practical implementations. Here, we propose a medical digital twin framework with a data-centric approach. We argue that a single digital representation of patients cannot support all the data uses of digital twins for technical and regulatory reasons. Instead, we propose a data architecture leveraging three main families of digital representations: (1) multimodal dashboards integrating various raw health records at points of care to assist with perception and documentation, (2) virtual patients, which provide nonsensitive data for collective secondary uses, and (3) individual predictions that support clinical decisions. For a given patient, multiple digital representations may be generated according to the different clinical pathways the patient goes through, each tailored to balance the trade-offs associated with the respective intended uses. Therefore, our proposed framework conceives the medical digital twin as a data architecture leveraging several digital representations of patients along clinical pathways.

Data science is an emerging field that provides new analytical methods. It incorporates novel data sources (eg, internet data) and methods (eg, machine learning) that offer valuable and timely insights into public health issues, including injury and violence prevention. The objective of this research was to describe ethical considerations for public health data scientists conducting injury and violence prevention-related data science projects to prevent unintended ethical, legal, and social consequences, such as loss of privacy or loss of public trust. We first reviewed foundational bioethics and public health ethics literature to identify key ethical concepts relevant to public health data science. After identifying these ethics concepts, we held a series of discussions to organize them under broad ethical domains. Within each domain, we examined relevant ethics concepts from our review of the primary literature. Lastly, we developed questions for each ethical domain to facilitate the early conceptualization stage of the ethical analysis of injury and violence prevention projects. We identified 4 ethical domains: privacy, responsible stewardship, justice as fairness, and inclusivity and engagement. We determined that each domain carries equal weight, with no consideration bearing more importance than the others. Examples of ethical considerations are clearly identifying project goals, determining whether people included in projects are at risk of reidentification through external sources or linkages, and evaluating and minimizing the potential for bias in data sources used. As data science methodologies are incorporated into public health research to work toward reducing the effect of injury and violence on individuals, families, and communities in the United States, we recommend that relevant ethical issues be identified, considered, and addressed.

AI techniques are increasingly being used to identify individuals both offline and online. However, quantifying their effectiveness at scale and, by extension, the risks they pose remains a significant challenge. Here, we propose a two-parameter Bayesian model for exact matching techniques and derive an analytical expression for correctness (κ), the fraction of people accurately identified in a population. We then generalize the model to forecast how κ scales from small-scale experiments to the real world, for exact, sparse, and machine learning-based robust identification techniques. Despite having only two degrees of freedom, our method closely fits 476 correctness curves and strongly outperforms curve-fitting methods and entropy-based rules of thumb. Our work provides a principled framework for forecasting the privacy risks posed by identification techniques, while also supporting independent accountability efforts for AI-based biometric systems.

 This study aimed to (1) empirically investigate current practices and analyze ethical dimensions of clinical data sharing by health care organizations for uses other than treatment, payment, and operations; and (2) make recommendations to inform research and policy for health care organizations to protect patients' privacy and autonomy when sharing data with unrelated third parties.

 Semistructured interviews and surveys involving 24 informatics leaders from 22 U.S. health care organizations, accompanied by thematic and ethical analyses.

 We found considerable heterogeneity across organizations in policies and practices. Respondents understood "data sharing" and "research" in very different ways. Their interpretations of these terms ranged from making data available for academic and public health uses, and to health information exchanges; to selling data for corporate research; and to contracting with aggregators for future resale or use. The nine interview themes were that health care organizations: (1) share clinical data with many types of organizations, (2) have a variety of motivations for sharing data, (3) do not make data-sharing policies readily available, (4) have widely varying data-sharing approval processes, (5) most commonly rely on Health Insurance and Portability and Accountability Act (HIPAA) de-identification to protect privacy, (6) were concerned about clinical data use by electronic health record vendors, (7) lacked data-sharing transparency to the general public, (8) allowed individual patients little control over sharing of their data, and (9) had not yet changed data-sharing practices within the year following the U.S. Supreme Court 2022 decision denying rights to abortion.

 Our analysis identified gaps between ethical principles and health care organizations' data-sharing policies and practices. To better align clinical data-sharing practices with patient expectations and biomedical ethical principles, we recommend updating HIPAA, including re-identification and upstream sharing restrictions in data-sharing contracts, better coordination across data-sharing approval processes, fuller transparency and opt-out options for patients, and accountability for data-sharing and consequent harms.

The role of immunization in public health is crucial, offering widespread protection against infectious diseases and underpinning societal well-being. However, achieving optimal vaccination coverage is impeded by vaccine hesitancy, a significant challenge that necessitates comprehensive strategies to understand and mitigate its effects. We propose the integration of Population Health Management principles with Immunization Information Systems (IISs) to address vaccine hesitancy more effectively. Our approach leverages systematic health determinants analysis to identify at-risk populations and tailor interventions, thereby promoting vaccination coverage and public health responses. We call for the development of an enhanced version of the Italian National Vaccination Registry, which aims to facilitate real-time tracking of individuals' vaccination status while improving data accuracy and interoperability among healthcare systems. This registry is designed to overcome current barriers by ensuring robust data protection, addressing cultural and organizational challenges, and integrating behavioral insights to foster informed public health campaigns. Our proposal aligns with the Italian National Vaccination Prevention Plan 2023-2025 and emphasizes proactive, evidence-based strategies to increase vaccination uptake and contrast the spread of vaccine-preventable diseases. The ultimate goal is to establish a data-driven, ethically sound framework that enhances public health outcomes and addresses the complexities of vaccine hesitancy within the Italian context and beyond.

We take up the challenge of developing an international network with capacity to survey the world's scientists on an ongoing basis, providing rich datasets regarding the opinions of scientists and scientific sub-communities, both at a time and also over time. The novel methodology employed sees local coordinators, at each institution in the network, sending survey invitation emails internally to scientists at their home institution. The emails link to a '10 second survey', where the participant is presented with a single statement to consider, and a standard five-point Likert scale. In June 2023, a group of 30 philosophers and social scientists invited 20,085 scientists across 30 institutions in 12 countries to participate, gathering 6,807 responses to the statement Science has put it beyond reasonable doubt that COVID-19 is caused by a virus. The study demonstrates that it is possible to establish a global network to quickly ascertain scientific opinion on a large international scale, with high response rate, low opt-out rate, and in a way that allows for significant (perhaps indefinite) repeatability. Measuring scientific opinion in this new way would be a valuable complement to currently available approaches, potentially informing policy decisions and public understanding across diverse fields.

Epilepsy care generates multiple sources of high-dimensional data, including clinical, imaging, electroencephalographic, genomic, and neuropsychological information, that are collected routinely to establish the diagnosis and guide management. Thanks to high-performance computing, sophisticated graphics processing units, and advanced analytics, we are now on the cusp of being able to use these data to significantly improve individualized care for people with epilepsy. Despite this, many clinicians, health care providers, and people with epilepsy are apprehensive about implementing Big Data and accompanying technologies such as artificial intelligence (AI). Practical, ethical, privacy, and climate issues represent real and enduring concerns that have yet to be completely resolved. Similarly, Big Data and AI-related biases have the potential to exacerbate local and global disparities. These are highly germane concerns to the field of epilepsy, given its high burden in developing nations and areas of socioeconomic deprivation. This educational paper from the International League Against Epilepsy's (ILAE) Big Data Commission aims to help clinicians caring for people with epilepsy become familiar with how Big Data is collected and processed, how they are applied to studies using AI, and outline the immense potential positive impact Big Data can have on diagnosis and management.

The use of real-world data has become increasingly popular, also in the field of infectious disease (ID), particularly since the COVID-19 pandemic emerged. While much useful data for research is being collected, these data are generally stored across different sources. Privacy concerns limit the possibility to store the data centrally, thereby also limiting the possibility of fully leveraging the potential power of combined data. Federated learning (FL) has been suggested to overcome privacy issues by making it possible to perform research on data from various sources without those data leaving local servers. In this review, we discuss existing applications of FL in ID research, as well as the most relevant opportunities and challenges of this method.

References for this review were identified through searches of MEDLINE/PubMed, Google Scholar, Embase and Scopus until July 2023. We searched for studies using FL in different applications related to ID.

Thirty references were included and divided into four sub-topics: disease screening, prediction of clinical outcomes, infection epidemiology, and vaccine research. Most research was related to COVID-19. In all studies, FL achieved good accuracy when predicting diseases and outcomes, also in comparison to non-federated methods. However, most studies did not make use of real-world federated data, but rather showed the potential of FL by using data that was manually partitioned.

FL is a promising methodology which allows using data from several sources, potentially generating stronger and more generalisable results. However, further exploration of FL application possibilities in ID research is needed.

The rise of large, publicly shared functional magnetic resonance imaging (fMRI) data sets in human neuroscience has focused on acquiring either a few hours of data on many individuals ('wide' fMRI) or many hours of data on a few individuals ('deep' fMRI). In this opinion article, we highlight an emerging approach within deep fMRI, which we refer to as 'intensive' fMRI: one that strives for extensive sampling of cognitive phenomena to support computational modeling and detailed investigation of brain function at the single voxel level. We discuss the fundamental principles, trade-offs, and practical considerations of intensive fMRI. We also emphasize that intensive fMRI does not simply mean collecting more data: it requires careful design of experiments to enable a rich hypothesis space, optimizing data quality, and strategically curating public resources to maximize community impact.

Pancreatic cancer remains one of the most lethal malignancies, with conventional treatment options providing limited efficacy. Recent advancements in immunotherapy have offered new hope, yet the unique tumor microenvironment (TME) of pancreatic cancer poses significant challenges to its successful application. This review explores the transformative impact of single-cell technology on the understanding and treatment of pancreatic cancer. By enabling high-resolution analysis of cellular heterogeneity within the TME, single-cell approaches have elucidated the complex interplay between various immune and tumor cell populations. These insights have led to the identification of predictive biomarkers and the development of innovative, personalized immunotherapeutic strategies. The review discusses the role of single-cell technology in dissecting the intricate immune landscape of pancreatic cancer, highlighting the discovery of T cell exhaustion profiles and macrophage polarization states that influence treatment response. Moreover, it outlines the potential of single-cell data in guiding the selection of immunotherapy drugs and optimizing treatment plans. The review also addresses the challenges and prospects of translating these single-cell-based innovations into clinical practice, emphasizing the need for interdisciplinary research and the integration of artificial intelligence to overcome current limitations. Ultimately, the review underscores the promise of single-cell technology in driving therapeutic strategy innovation and improving patient outcomes in the battle against pancreatic cancer.

Variables related to socioeconomic status (SES), including income, ethnicity, and education, shape contact structures and affect the spread of infectious diseases. However, these factors are often overlooked in epidemic models, which typically stratify social contacts by age and interaction contexts. Here, we introduce and study generalized contact matrices that stratify contacts across multiple dimensions. We demonstrate a lower-bound theorem proving that disregarding additional dimensions, besides age and context, might lead to an underestimation of the basic reproductive number. By using SES variables in both synthetic and empirical data, we illustrate how generalized contact matrices enhance epidemic models, capturing variations in behaviors such as heterogeneous levels of adherence to nonpharmaceutical interventions among demographic groups. Moreover, we highlight the importance of integrating SES traits into epidemic models, as neglecting them might lead to substantial misrepresentation of epidemic outcomes and dynamics. Our research contributes to the efforts aiming at incorporating socioeconomic and other dimensions into epidemic modeling.

Routine surgical video recording has multiple benefits. Video acts as an objective record of the operative record, allows video-based coaching and is integral to the development of digital technologies. Despite these benefits, adoption is not widespread. To date, only questionnaire studies have explored this failure in adoption. This study aims to determine the barriers and provide recommendations for the implementation of routine surgical video recording.

A pre- and post-pilot questionnaire surrounding a real-world implementation of a C-SATS©, an educational recording and surgical analytics platform, was conducted in a university teaching hospital trust. Usage metrics from the pilot study and descriptive analyses of questionnaire responses were used with the non-adoption, abandonment, scale-up, spread, sustainability (NASSS) framework to create topic guides for semi-structured interviews. Transcripts of interviews were evaluated in an inductive thematic analysis.

Engagement with the C-SATS© platform failed to reach consistent levels with only 57 videos uploaded. Three attending surgeons, four surgical residents, one scrub nurse, three patients, one lawyer, and one industry representative were interviewed, all of which perceived value in recording. Barriers of 'change,' 'resource,' and 'governance,' were identified as the main themes. Resistance was centred on patient misinterpretation of videos. Participants believed availability of infrastructure would facilitate adoption but integration into surgical workflow is required. Regulatory uncertainty was centred around anonymity and data ownership.

Barriers to the adoption of routine surgical video recording exist beyond technological barriers alone. Priorities for implementation include integration recording into the patient record, engaging all stakeholders to ensure buy-in, and formalising consent processes to establish patient trust.

Leveraging artificial intelligence (AI) for the analysis of electrocardiograms (ECGs) has the potential to transform diagnosis and estimate the prognosis of not only cardiac but, increasingly, noncardiac conditions. In this review, we summarize clinical studies and AI-enhanced ECG-based clinical applications in the early detection, diagnosis, and estimating prognosis of cardiovascular diseases in the past 5 years (2019-2023). With advancements in deep learning and the rapid increased use of ECG technologies, a large number of clinical studies have been published. However, most of these studies are single-centre, retrospective, proof-of-concept studies that lack external validation. Prospective studies that progress from development toward deployment in clinical settings account for < 15% of the studies. Successful implementations of ECG-based AI applications that have received approval from the Food and Drug Administration have been developed through commercial collaborations, with approximately half of them being for mobile or wearable devices. The field is in its early stages, and overcoming several obstacles is essential, such as prospective validation in multicentre large data sets, addressing technical issues, bias, privacy, data security, model generalizability, and global scalability. This review concludes with a discussion of these challenges and potential solutions. By providing a holistic view of the state of AI in ECG analysis, this review aims to set a foundation for future research directions, emphasizing the need for comprehensive, clinically integrated, and globally deployable AI solutions in cardiovascular disease management.

Sharing neuroimaging datasets enables reproducibility, education, tool development, and new discoveries. Neuroimaging from many studies are publicly available, providing a glimpse into progressive disorders and human development. In contrast, few stroke studies are shared, and these datasets lack longitudinal sampling of functional imaging, diffusion imaging, as well as the behavioral and demographic data that encourage novel applications. This is surprising, as stroke is a leading cause of disability, and acquiring brain imaging is considered standard of care. The first release of the Aphasia Recovery Cohort includes imaging data, demographics and behavioral measures from 230 chronic stroke survivors who experienced aphasia. We also share scripts to illustrate how the imaging data can predict impairment. In conclusion, recent advances in machine learning thrive on large, diverse datasets. Clinical data sharing can contribute to improvements in automated detection of brain injury, identification of white matter hyperintensities, measures of brain health, and prognostic abilities to guide care.

In settings requiring synthetic data generation based on a clinical cohort, e.g., due to data protection regulations, heterogeneity across individuals might be a nuisance that we need to control or faithfully preserve. The sources of such heterogeneity might be known, e.g., as indicated by sub-groups labels, or might be unknown and thus reflected only in properties of distributions, such as bimodality or skewness. We investigate how such heterogeneity can be preserved and controlled when obtaining synthetic data from variational autoencoders (VAEs), i.e., a generative deep learning technique that utilizes a low-dimensional latent representation. To faithfully reproduce unknown heterogeneity reflected in marginal distributions, we propose to combine VAEs with pre-transformations. For dealing with known heterogeneity due to sub-groups, we complement VAEs with models for group membership, specifically from propensity score regression. The evaluation is performed with a realistic simulation design that features sub-groups and challenging marginal distributions. The proposed approach faithfully recovers the latter, compared to synthetic data approaches that focus purely on marginal distributions. Propensity scores add complementary information, e.g., when visualized in the latent space, and enable sampling of synthetic data with or without sub-group specific characteristics. We also illustrate the proposed approach with real data from an international stroke trial that exhibits considerable distribution differences between study sites, in addition to bimodality. These results indicate that describing heterogeneity by statistical approaches, such as propensity score regression, might be more generally useful for complementing generative deep learning for obtaining synthetic data that faithfully reflects structure from clinical cohorts.

Artificial Intelligence (AI) has shown remarkable potential to revolutionise healthcare by enhancing diagnostics, improving treatment outcomes, and streamlining administrative processes. In the global regulatory landscape, several countries are working on regulating AI in healthcare. There are five key regulatory issues that need to be addressed: (i) data security and protection-measures to cover the "digital health footprints" left unknowingly by patients when they access AI in health services; (ii) data quality-availability of safe and secure data and more open database sources for AI, algorithms, and datasets to ensure equity and prevent demographic bias; (iii) validation of algorithms-mapping of the explainability and causability of the AI system; (iv) accountability-whether this lies with the healthcare professional, healthcare organisation, or the personified AI algorithm; (v) ethics and equitable access-whether fundamental rights of people are met in an ethical manner. Policymakers may need to consider the entire life cycle of AI in healthcare services and the databases that were used for the training of the AI system, along with requirements for their risk assessments to be publicly accessible for effective regulatory oversight. AI services that enhance their functionality over time need to undergo repeated algorithmic impact assessment and must also demonstrate real-time performance. Harmonising regulatory frameworks at the international level would help to resolve cross-border issues of AI in healthcare services.

Advances in biomedical data science and artificial intelligence (AI) are profoundly changing the landscape of healthcare. This article reviews the ethical issues that arise with the development of AI technologies, including threats to privacy, data security, consent, and justice, as they relate to donors of tissue and data. It also considers broader societal obligations, including the importance of assessing the unintended consequences of AI research in biomedicine. In addition, this article highlights the challenge of rapid AI development against the backdrop of disparate regulatory frameworks, calling for a global approach to address concerns around data misuse, unintended surveillance, and the equitable distribution of AI's benefits and burdens. Finally, a number of potential solutions to these ethical quandaries are offered. Namely, the merits of advocating for a collaborative, informed, and flexible regulatory approach that balances innovation with individual rights and public welfare, fostering a trustworthy AI-driven healthcare ecosystem, are discussed.

This viewpoint article first explores the ethical challenges associated with the future application of large language models (LLMs) in the context of medical education. These challenges include not only ethical concerns related to the development of LLMs, such as artificial intelligence (AI) hallucinations, information bias, privacy and data risks, and deficiencies in terms of transparency and interpretability but also issues concerning the application of LLMs, including deficiencies in emotional intelligence, educational inequities, problems with academic integrity, and questions of responsibility and copyright ownership. This paper then analyzes existing AI-related legal and ethical frameworks and highlights their limitations with regard to the application of LLMs in the context of medical education. To ensure that LLMs are integrated in a responsible and safe manner, the authors recommend the development of a unified ethical framework that is specifically tailored for LLMs in this field. This framework should be based on 8 fundamental principles: quality control and supervision mechanisms; privacy and data protection; transparency and interpretability; fairness and equal treatment; academic integrity and moral norms; accountability and traceability; protection and respect for intellectual property; and the promotion of educational research and innovation. The authors further discuss specific measures that can be taken to implement these principles, thereby laying a solid foundation for the development of a comprehensive and actionable ethical framework. Such a unified ethical framework based on these 8 fundamental principles can provide clear guidance and support for the application of LLMs in the context of medical education. This approach can help establish a balance between technological advancement and ethical safeguards, thereby ensuring that medical education can progress without compromising the principles of fairness, justice, or patient safety and establishing a more equitable, safer, and more efficient environment for medical education.

Though the rise of big data in the field of occupational health offers new opportunities especially for cross-cutting research, they raise the issue of privacy and security of data, especially when linking sensitive data from the field of insurance, occupational health or compensation claims. We aimed to validate a large, blinded synthesized database developed from the CONSTANCES cohort by comparing associations between three independently selected outcomes, and various exposures.

From the CONSTANCES cohort, a large synthetic dataset was constructed using the avatar method (Octopize) that is agnostic to the data primary or secondary data uses. Three main analyses of interest were chosen to compare associations between the raw and avatar dataset: risk of stroke (any stroke, and subtypes of stroke), risk of knee pain and limitations associated with knee pain. Logistic models were computed, and a qualitative comparison of paired odds ratio (OR) was made.

Both raw and avatar datasets included 162,434 observations and 19 relevant variables. On the 172 paired raw/avatar OR that were computed, including stratified analyses on sex, more than 77% of the comparisons had a OR difference ≤0.5 and less than 7% had a discrepancy in the statistical significance of the associations, with a Cohen's Kappa coefficient of 0.80.

This study shows the flexibility and the multiple usage of a synthetic database created with the avatar method in the particular field of occupational health, which can be shared in open access without risking re-identification and privacy issues and help bring new insights for complex phenomenon like return to work.

One of the most widely used data sources for research on foster care and adoption is the Adoption and Foster Care Analysis and Reporting System (AFCARS). County identifiers in AFCARS are suppressed for all counties with fewer than 1000 cases to prevent the re-identification of vulnerable children, but this also impacts researchers' ability to study smaller communities and analyze how local environments may affect out-of-home placements. This study uses non-public AFCARS datasets to assess, for the first time, how data suppression rules impact data access and re-identification risk. It compares the long-standing 1000-case threshold against a wide range of potential alternatives and finds substantial data access gains coupled with moderate risk increases for thresholds between 400 and 700. Adopting a 700-case threshold leads to a 50% increase in the number of identifiable counties while also keeping the percentage of fostered children who face an elevated risk of re-identification below 1%. Making data from a substantial number of rural counties available to researchers requires much larger threshold changes, which in turn increases re-identification risks.

Widespread adoption of digital health tools has the potential to improve health and health care for individuals and their communities, but realizing this potential requires anticipating and addressing numerous ethical and regulatory challenges. Here, we help digital health tool developers identify ethical and regulatory considerations - and opportunities to advance desirable outcomes - by organizing them within a general product-development lifecycle that spans generation of ideas to commercialization of a product.

This viewpoint article explores the transformative role of large language models (LLMs) in the field of medical education, highlighting their potential to enhance teaching quality, promote personalized learning paths, strengthen clinical skills training, optimize teaching assessment processes, boost the efficiency of medical research, and support continuing medical education. However, the use of LLMs entails certain challenges, such as questions regarding the accuracy of information, the risk of overreliance on technology, a lack of emotional recognition capabilities, and concerns related to ethics, privacy, and data security. This article emphasizes that to maximize the potential of LLMs and overcome these challenges, educators must exhibit leadership in medical education, adjust their teaching strategies flexibly, cultivate students' critical thinking, and emphasize the importance of practical experience, thus ensuring that students can use LLMs correctly and effectively. By adopting such a comprehensive and balanced approach, educators can train health care professionals who are proficient in the use of advanced technologies and who exhibit solid professional ethics and practical skills, thus laying a strong foundation for these professionals to overcome future challenges in the health care sector.

Information about us, our actions, and our preferences is created at scale through surveys or scientific studies or as a result of our interaction with digital devices such as smartphones and fitness trackers. The ability to safely share and analyze such data is key for scientific and societal progress. Anonymization is considered by scientists and policy-makers as one of the main ways to share data while minimizing privacy risks. In this review, we offer a pragmatic perspective on the modern literature on privacy attacks and anonymization techniques. We discuss traditional de-identification techniques and their strong limitations in the age of big data. We then turn our attention to modern approaches to share anonymous aggregate data, such as data query systems, synthetic data, and differential privacy. We find that, although no perfect solution exists, applying modern techniques while auditing their guarantees against attacks is the best approach to safely use and share data today.

Federated Learning (FL) is emerging in the medical field to address the need for diverse datasets while complying with data protection regulations. This decentralised learning paradigm allows hospitals (clients) to train machine learning models locally, ensuring that patient data remains within the confines of its originating institution. Nonetheless, FL by itself is not enough to guarantee privacy, as the central aggregation process may still be susceptible to identity-exposing attacks, potentially compromising data protection compliance. To strengthen privacy, differential privacy (DP) is often introduced. In this work, we conduct a comprehensive comparative analysis to evaluate the impact of DP in both traditional Centralised Learning (CL) frameworks and FL for polyp segmentation, a common medical image analysis task. Experiments are performed in PolypGen, a multi-centre publicly available dataset designed for polyp segmentation. The results show a clear drop in performance with the introduction of DP, exposing the trade-off between privacy and performance and highlighting the need to develop novel privacy-preserving techniques.

Open science principles are revolutionizing the transparency, reproducibility, and accessibility of research. Meta-analysis has become a key technique for synthesizing data across studies in a principled way; however, its impact is contingent on adherence to open science practices. Here, we outline 9 quick tips for open meta-analyses, aimed at guiding researchers to maximize the reach and utility of their findings. We advocate for outlining preregistering clear protocols, opting for open tools and software, and the use of version control systems to ensure transparency and facilitate collaboration. We further emphasize the importance of reproducibility, for example, by sharing search syntax and analysis scripts, and discuss the benefits of planning for dynamic updating to enable living meta-analyses. We also recommend publication in open-access formats, as well as open data, open code, and open access publication. We close by encouraging active promotion of research findings to bridge the gap between complex syntheses and public discourse, and provide a detailed submission checklist to equip researchers, reviewers and journal editors with a structured approach to conducting and reporting open meta-analyses.

More than 5 billion people in the world own a smartphone. More than half of these have been used to collect and process health-related data. As such, the existing volume of potentially exploitable health data is unprecedentedly large and growing rapidly. Mobile health applications (apps) on smartphones are some of the worst offenders and are increasingly being used for gathering and exchanging significant amounts of personal health data from the public. This data is often utilized for health research purposes and for algorithm training. While there are advantages to utilizing this data for expanding health knowledge, there are associated risks for the users of these apps, such as privacy concerns and the protection of their data. Consequently, gaining a deeper comprehension of how apps collect and crowdsource data is crucial. To explore how apps are crowdsourcing data and to identify potential ethical, legal, and social issues (ELSI), we conducted an examination of the Apple App Store and the Google Play Store in North America and Europe to identify apps that could potentially gather health data through crowdsourcing. Subsequently, we analyzed their privacy policies, terms of use, and other related documentation to gain insights into the utilization of users' data and the possibility of repurposing it for research or algorithm training purposes. More specifically, we reviewed privacy policies to identify clauses pertaining to the following key categories: research, data sharing, privacy/confidentiality, commercialization, and return of findings. Based on the results of these app search, we developed an App Atlas that presents apps which crowdsource data for research or algorithm training. We identified 46 apps available in the European and Canadian markets that either openly crowdsource health data for research or algorithm training or retain the legal or technical capability to do so. This app search showed an overall lack of consistency and transparency in privacy policies that poses challenges to user comprehensibility, trust, and informed consent. A significant proportion of applications presented contradictions or exhibited considerable ambiguity. For instance, the vast majority of privacy policies in the App Atlas contain ambiguous or contradictory language regarding the sharing of users' data with third parties. This raises a number of ethico-legal concerns which will require further academic and policy attention to ensure a balance between protecting individual interests and maximizing the scientific utility of crowdsourced data. This article represents a key first step in better understanding these concerns and bringing attention to this important issue.

The online version contains supplementary material available at 10.1007/s41649-024-00296-3.

Access to individual-level health data is essential for gaining new insights and advancing science. In particular, modern methods based on artificial intelligence rely on the availability of and access to large datasets. In the health sector, access to individual-level data is often challenging due to privacy concerns. A promising alternative is the generation of fully synthetic data, i.e., data generated through a randomised process that have similar statistical properties as the original data, but do not have a one-to-one correspondence with the original individual-level records. In this study, we use a state-of-the-art synthetic data generation method and perform in-depth quality analyses of the generated data for a specific use case in the field of nutrition. We demonstrate the need for careful analyses of synthetic data that go beyond descriptive statistics and provide valuable insights into how to realise the full potential of synthetic datasets. By extending the methods, but also by thoroughly analysing the effects of sampling from a trained model, we are able to largely reproduce significant real-world analysis results in the chosen use case.

To develop and validate a natural language processing (NLP) pipeline that detects 18 conditions in French clinical notes, including 16 comorbidities of the Charlson index, while exploring a collaborative and privacy-enhancing workflow.

The detection pipeline relied both on rule-based and machine learning algorithms, respectively, for named entity recognition and entity qualification, respectively. We used a large language model pre-trained on millions of clinical notes along with annotated clinical notes in the context of 3 cohort studies related to oncology, cardiology, and rheumatology. The overall workflow was conceived to foster collaboration between studies while respecting the privacy constraints of the data warehouse. We estimated the added values of the advanced technologies and of the collaborative setting.

The pipeline reached macro-averaged F1-score positive predictive value, sensitivity, and specificity of 95.7 (95%CI 94.5-96.3), 95.4 (95%CI 94.0-96.3), 96.0 (95%CI 94.0-96.7), and 99.2 (95%CI 99.0-99.4), respectively. F1-scores were superior to those observed using alternative technologies or non-collaborative settings. The models were shared through a secured registry.

We demonstrated that a community of investigators working on a common clinical data warehouse could efficiently and securely collaborate to develop, validate and use sensitive artificial intelligence models. In particular, we provided an efficient and robust NLP pipeline that detects conditions mentioned in clinical notes.