Paper short abstract

Drawing on ethnography in Czechia, this paper shows how AI-driven diabetes technologies selectively translate metabolic knowledge into algorithmic temporalities, sidelining menstrual hormonal rhythms and producing gendered care labor that reveals epistemic inequalities in automated healthcare.

Paper long abstract

This paper explores how AI-driven diabetes technologies produce gendered inequalities through the selective epistemic incorporation of bodily temporalities into algorithmic care. Drawing on ethnographic research with patients, diabetologists, and open-source developers conducted in Czechia—a post-socialist healthcare context shaped by public provision, marketized medical technologies, and strong patient communities—I analyze how automated insulin delivery systems translate metabolic knowledge into algorithms and how cyclical hormonal rhythms related to menstruation remain only partially legible within them.

These systems rely on continuous glucose monitoring data to model metabolic responsiveness over time, privileging linear temporal patterns that implicitly reflect a non-menstruating, male-coded metabolic norm. Although clinical knowledge about hormonal effects on insulin sensitivity exists, it is unevenly operationalized: menstrual hormonal cycles introduce recurring yet variable changes that exceed the temporal “learning horizons” of many commercial algorithms. As a result, users who menstruate must perform additional temporal labor—resetting profiles, or adjusting sensitivities—to maintain metabolic stability.

Rather than framing this as an absence of knowledge, the paper argues that menstrual hormonal rhythms are epistemically marginalized through algorithmic design choices that shape what counts as actionable data, redistributing responsibility and labor along gendered lines. Open-source systems developed within the Czech diabetes community partially counter these dynamics through transparency and collective tinkering.

Methodologically, the paper thinks with temporality as an ethnographic and conceptual lens for examining epistemic inequalities in AI healthcare, contributing to the anthropology of AI by tracing how care, knowledge, and responsibility are unevenly reconfigured through the temporal infrastructures of algorithmic systems.

Paper short abstract

Using the case of a health diplomacy site, we discuss social considerations of autonomy and clinical experiences to inform the design of ethically grounded data-sharing frameworks across contexts, including systems transitioning to digital platforms and those merging datasets for AI applications.

Paper long abstract

U.S. Venture-capital investments in AI in healthcare are projected to reach $11 billion in 2025. Central to the return on investment are the biospecimens and health information within data systems such as biodata-tracking technologies connected to personal devices, biobanks, and Electronic Health Records. The more access, system integration, and the greater number of people and data feeding the system, the better the learning and innovation to fulfill the promise of AI in healthcare. Investors know the value of this data and premise returns on the assumption of increasing access to these sources. Globally, personal health information, once protected by pre-AI regulatory systems, is now accessed through new means and frontiers of regulatory landscapes, most of which the average public is unaware of. Many national systems haven’t yet connected social and sensitive health data in cross-communicating networks. However, stakeholders from the Global North have long experimented with interoperable systems in health diplomacy recipient sites where they’ve piloted technologies gathering, tracing, and digitizing health information. Because of the nature of funding and the scientific satellite-metropole collaborations, datasets are fundamentally designed for sharing across systems and borders. Drawing on qualitative data from 30 interviews with health providers in Lesotho experienced in sharing health-related information with international partners for care, research, and product development we present data informing the social considerations for designing transparent, and ethically grounded data-sharing frameworks across diverse contexts, including large-scale research collaborations, smaller systems transitioning to digital platforms, and the merging of independent databases for interoperability and artificial intelligence applications.

Paper short abstract

This paper examines how rare disease families in China build open genetic databases and confront “structural rareness,” as institutional authority and geopolitical borders restrict data sharing and fragment research futures.

Paper long abstract

Genes are small, but they carry large worlds. They encode inheritance, disease mechanisms, and therapeutic futures while simultaneously tracing kinship and belonging. Since the Human Genome Project, genetic research has relied on international collaboration and data sharing. With advancements in AI technology, high-quality and large-scale data are especially significant. However, nowadays, genetic data are increasingly treated as strategic resources with states, companies, and institutions restricting circulation in the name of biosecurity and technological competition.

This paper examines how patients with rare genetic diseases and their families in China understand, live with, and imagine genetic data and data-driven futures. Facing conditions that affect one in thousands or even millions, they work to construct public databases from genetic reports, natural histories, and clinical records, hoping to render otherwise marginalized diseases legible to scientists.

Yet these efforts encounter significant constraints, producing what I call “structural rareness”: a condition where political and systemic boundaries make communities appear too small for research investment. First, medical professionals often question the validity of patient-generated databases, complicating negotiations over who may gather, analyze, and store data. Second, institutional policies could restrict patients’ access to their own genetic information, and many fear that sharing data externally might damage relationships with their physicians. Third, geopolitical tensions, particularly between China and the West, impose strict limits on international data sharing. Consequently, as biomedicine requires data collaboration under ethical review, essential genetic information remains bounded by borders, narrowing patients’ access to and control over their own data.

Paper short abstract

Wearable UV sensing is used to examine everyday sunlight exposure in dense urban housing in Karachi, Pakistan. The study focuses on ambient UVB availability across indoor and outdoor spaces and explores how housing form and routine shape health-relevant environmental exposure.

Paper long abstract

This paper presents an ongoing study using a wearable ultraviolet (UV) sensing device to examine everyday sunlight exposure and its implications for health knowledge in dense urban environments. Although the biomedical importance of UVB exposure for Vitamin D synthesis is well established, access to clinical testing and preventive care remains uneven.

The research draws on doctoral fieldwork in Karachi, Pakistan, focusing on women living in middle-income neighbourhoods shaped by rapid densification, vertical growth, and privacy-oriented housing forms. These spatial and social conditions significantly influence daily exposure to sunlight yet are rarely captured in standard health or environmental assessments. A wearable UV sensing device was developed using Arduino and Python and housed in a 3D-printed, watch-style casing worn on the wrist. Currently, the device functions as a research instrument rather than a diagnostic or consumer tool, enabling the measurement of ambient UV exposure as experienced by the wearer over the course of the day.

Although the device records the full UV spectrum, the analytical focus is on UVB availability and its variation across indoor and outdoor domestic spaces. Because the wrist is often partially shielded from direct sunlight, recorded values are adjusted using a calibrated coefficient to approximate effective exposure. The device is currently in a testing phase, but early trials indicate stable and consistent readings for ambient UVA, UVB, and UVC. By combining wearable sensing with spatial observation, the study offers a grounded approach to understanding how health-relevant environmental data are produced and constrained by everyday housing and routine.

Paper short abstract

AI-powered prediction models for severe mental illness used presymptomatically are believed to operate in future healthcare systems. Individuals with experience of mental illness and their support persons imagine the use of prediction tools in future healthcare systems accentuating polarisation.

Paper long abstract

Background. In psychiatry, there are attempts to develop predictive models identifying risk for severe mental illness before any clinical symptoms appear, thus enabling early intervention and reduction of long-term burden of mental illness. Such risk prediction tools are currently in the early stages of development, but potentially they will be operated in social contexts and care systems of different countries.

Research question. How do individuals with lived experience of mental illness and their support persons imagine the use of prediction tools in future healthcare systems?

Methods. Thirty semi-structured interviews with individuals with lived experience of mental illness and their support persons from 9 European countries conducted during 2024.

Results. The complicated experience of mental illness is lived through in different social and healthcare systems. Some systems are perceived as supportive and reduce othering, whereas others are difficult to access, associated with increased stigmatisation building a polarising divide of healthy and unhealthy. Research participants worried about the use of predictive models becoming mandatory to access healthcare and reproducing further inequalities.

Research participants imagine future trajectories of the potential predictive model as rooted in their past experiences (difficulties accessing support, prejudiced attitudes from professionals, etc). If the existing experience with the healthcare system is seen as responsive for needs, people are more likely to see the benefit of the predictive model since it allows timely prevention of the illness or its consequences. Meanwhile those who experienced healthcare systems as unsupportive, see the tool bringing more of a risk of inequality.