Bridging Data Gaps in Healthcare: A Scoping Review of Transfer Learning in Structured Data Analysis

In many regions globally, including low-resource settings, there is a growing trend towards using mHealth technology, such as wearable sensors, to enhance health behaviors and outcomes. However, adoption of such devices in research conducted in low-resource settings lags behind use in high-resource areas. Moreover, there is a scarcity of research that specifically examines the user experience, readiness for and challenges of integrating wearable sensors into health research and community interventions in low-resource settings specifically. This study summarizes the reactions and experiences of young women (N = 57), ages 18 to 24 years, living in poverty in Kampala, Uganda, who wore Garmin vívoactive 3 smartwatches for five days for a research project. Data collected from the Garmins included participant location, sleep, and heart rate. Through six focus group discussions, we gathered insights about the participants' experiences and perceptions of the wearable devices. Overall, the wearable devices were met with great interest and enthusiasm by participants. The findings were organized across 10 domains to highlight reactions and experiences pertaining to device settings, challenges encountered with the device, reports of discomfort/comfort, satisfaction, changes in daily activities, changes to sleep, speculative device usage, community reactions, community dynamics and curiosity, and general device comfort. The study sheds light on the introduction of new technology in a low-resource setting and also on the complex interplay between technology and culture in Kampala's slums. We also learned some insights into how wearable devices and perceptions may influence behaviors and social dynamics. These practical insights are shared to benefit future research and applications by health practitioners and clinicians to advance and enhance the implementation and effectiveness of wearable devices in similar contexts and populations. These insights and user experiences, if incorporated, may enhance device acceptance and data quality for those conducting research in similar settings or seeking to address population-specific needs and health issues.

Annotated medical images are typically more rare than labeled natural images, since they are limited by domain knowledge and privacy constraints. Recent advances in transfer and contrastive learning have provided effective solutions to tackle such issues from different perspectives. The state-of-the-art transfer learning (e.g., Big Transfer (BiT)) and contrastive learning (e.g., Simple Siamese Contrastive Learning (SimSiam)) approaches have been investigated independently, without considering the complementary nature of such techniques. It would be appealing to accelerate contrastive learning with transfer learning, given that slow convergence speed is a critical limitation of modern contrastive learning approaches. In this paper, we investigate the feasibility of aligning BiT with SimSiam. From empirical analyses, different normalization techniques (Group Norm in BiT vs. Batch Norm in SimSiam) is a key hurdle of adapting BiT to SimSiam. When combining BiT with SimSiam, we evaluated the performance of using BiT, SimSiam, and BiT+SimSiam on CIFAR-10 and HAM10000 datasets. The results suggest that the BiT models accelerate the convergence speed of SimSiam. When used together, the model gives superior performance over both of its counterparts. We hope this study will motivate researchers to revisit the task of aggregating big pretrained models with contrastive learning models for image analysis.

Clinical trials conducted in low-resource settings have unique challenges associated with their conduct. This is mainly attributed to the power and resource discrepancy between actors in the clinical trial. This thesis provides an in-depth look at clinical trials in low-resource settings and the effects of the resource discrepancy on the actors. It aims to answer what the ethical challenges are when conducting research in low-resource settings and the subsequent implications for research design. It focuses on capturing both the experience of caregivers of pediatric participants and the frontline researchers in a malaria vaccine clinical trial. Through exploring these two stories and bridging the relational with the formal, it provides a novel approach to address the challenges with research in low-resource settings. This approach employs the lens of complexity theory to evaluate the outcome of two systems, a human community and a clinical trial, merging. I will begin by outlining a general introduction of clinical trials in low-resource settings and the case study of a pediatric malaria vaccine clinical trial, here I detail the need to generate a vaccine against malaria and outline why such research should take place. This situates the reasons for the study and provides familiarity with the contextual reality. Then I will move into detailing the caregiver experiences, researcher experiences, and the application of complexity theory to bridge together the different experiences. This thesis is a result of qualitative data gathered from 78 interviews with caregivers of pediatric participants and 11 interviews with researchers involved on the frontline of a pediatric malaria vaccine clinical trial. The final part of the thesis is a theoretical reflection that explores the realities faced by researchers and argues for an approach that embraces the non-linearity of research taking place in human communities. Here I identify the challenges associated with choice and structural inequity, the conflict between beneficence and autonomy, and being a frontline researcher in low-resource settings.

<h3>Background</h3> To improve health in low-resource settings, it is imperative to enhance (a) design of interventions, (b) piloting of interventions, and (c) analyse the processes through which interventions are implemented and how they are used and perceived by key stakeholders. Applied qualitative research is a potential approach. <h3>Objectives</h3> identify processes through which collaborations can be built in order to undertake focused studies in applied qualitative health research in low–resource settings; provide examples of how such collaborations have resulted in applied qualitative research that has impacted on policy and practice across a range of health issues and contexts. <h3>Methods</h3> Much research to improve health in low-resource settings includes embedded qualitative studies, which may be exploratory, to support the design of interventions, may form part of pilot studies, or may form part of process evaluations or other types of explanatory work. They are typically constrained by relatively short timeframes and tight budgets. We have developed a collaborative approach to high quality focused qualitative studies that involves a process of co-designing protocols and ethics applications; co-designing data collection tools; collecting, transcribing and analysing data in an iterative and collaborative manner, which enhances the quality of the data collected; and co-writing publications and policy briefs. We have identified processes that have provided both high quality research findings and findings of direct relevance for researchers, practitioners, and policy makers. <h3>Results</h3> Applied qualitative health research in low-resource settings has contributed to major impact on policy and practice at scale in settings such as China, India, Pakistan, Bangladesh, Nepal, Ghana and Sierra Leone, across a range of communicable and non-communicable diseases. <h3>Conclusions</h3> Despite difficult timeframes and tight budgets, applied qualitative research can contribute substantially to impact on health policy and practice in low-resource settings.

BackgroundThe disproportionate burden of injuries in low-resource settings is a global health challenge demanding context-specific and locally driven solutions. Bangladesh, more specifically, Centre for Injury Prevention and Research, Bangladesh (CIPRB)...

Active Learning is a machine learning and data mining technique that selects the most informative samples for labeling and uses them as training data; it is especially useful when there are large amount of unlabeled data and labeling them is expensive. Recently, batch-mode active learning, where a set of samples are selected concurrently for labeling, based on their collective merit, has attracted a lot of attention. The objective of batch-mode active learning is to select a set of informative samples so that a classifier learned on these samples has good generalization performance on the unlabeled data. Most of the existing batch-mode active learning methodologies try to achieve this by selecting samples based on certain criteria. In this article we propose a novel criterion which achieves good generalization performance of a classifier by specifically selecting a set of query samples that minimize the difference in distribution between the labeled and the unlabeled data, after annotation. We explicitly measure this difference based on all candidate subsets of the unlabeled data and select the best subset. The proposed objective is an NP-hard integer programming optimization problem. We provide two optimization techniques to solve this problem. In the first one, the problem is transformed into a convex quadratic programming problem and in the second method the problem is transformed into a linear programming problem. Our empirical studies using publicly available UCI datasets and two biomedical image databases demonstrate the effectiveness of the proposed approach in comparison with the state-of-the-art batch-mode active learning methods. We also present two extensions of the proposed approach, which incorporate uncertainty of the predicted labels of the unlabeled data and transfer learning in the proposed formulation. In addition, we present a joint optimization framework for performing both transfer and active learning simultaneously unlike the existing approaches of learning in two separate stages, that is, typically, transfer learning followed by active learning. We specifically minimize a common objective of reducing distribution difference between the domain adapted source, the queried and labeled samples and the rest of the unlabeled target domain data. Our empirical studies on two biomedical image databases and on a publicly available 20 Newsgroups dataset show that incorporation of uncertainty information and transfer learning further improves the performance of the proposed active learning based classifier. Our empirical studies also show that the proposed transfer-active method based on the joint optimization framework performs significantly better than a framework which implements transfer and active learning in two separate stages.

Social norms research in low resource settings: Opportunities ahead.

Little is known about researchers' practices regarding the provision of ancillary care (AC) in public health intervention studies they have conducted and the factors that influence their decisions about whether to provide ancillary care in low-resource settings. We conducted 52 in-person in-depth interviews with public health researchers. Data analysis was iterative and led to the identification of themes and patterns among themes. We found that researchers who conduct their research in the community setting are more likely to identify and plan for the AC needs of potential research subjects before a study begins, whereas those affiliated with a permanent facility are more likely to deliver AC to research subjects on an ad hoc basis. Our findings suggest that on the whole, at least for public health intervention research in low-resource settings, researchers conducting research in the community setting confront more complex ethical and operational challenges in their decision-making about AC than do researchers conducting facility-based studies.

The gap between need and access to mental health care is widest in low-resource settings. Health systems in these contexts devote few resources to expanding mental health care, and it is missing from the agenda of most global health donors. This is partially explained by the paucity of data regarding the nature and extent of the mental health burden in these settings, so accurate and comparable measurement is essential to advocating for, developing, and implementing appropriate policies and services. Inaccurate estimation of mental illness prevalence, and misunderstandings regarding its etiologies and expressions, are associated with unnecessary costs to health systems and people living with mental illness. This paper presents a selective literature review of the challenges associated with mental health measurement in these settings globally, presents several case studies, and suggests three strategies for researchers to improve their assessments: utilize qualitative data, conduct cognitive interviews and train research teams with a focus on inter-rater reliability. These three strategies presented, added to the complement of tools used by mental health researchers in low-resource settings, will enable more researchers to conduct culturally valid work, improve the quality of data available, and assist in narrowing the treatment gap.

The gap between need and access to mental health care is widest in low-resource settings. Health systems in these contexts devote few resources to expanding mental health care, and it is missing from the agenda of most global health donors. This is partially explained by the paucity of data regarding the nature and extent of the mental health burden in these settings, so accurate and comparable measurement is essential to advocating for, developing, and implementing appropriate policies and services. Inaccurate estimation of mental illness prevalence, and misunderstandings regarding its etiologies and expressions, are associated with unnecessary costs to health systems and people living with mental illness. This paper presents a selective literature review of the challenges associated with mental health measurement in these settings globally, presents several case studies, and suggests three strategies for researchers to improve their assessments: utilize qualitative data, conduct cognitive interviews and train research teams with a focus on inter-rater reliability. These three strategies presented, added to the complement of tools used by mental health researchers in low-resource settings, will enable more researchers to conduct culturally valid work, improve the quality of data available, and assist in narrowing the treatment gap.

80% of people with disabilities worldwide live in low resourced settings, rural areas, informal settlements and in multidimensional poverty. ICT4D leverages technological innovations to deliver programs for international development. But very few do so with a focus on and involving people with disabilities in low resource settings. Also, most studies largely focus on publishing the results of the research with a focus on the positive stories and not the learnings and recommendations regarding research processes. In short, researchers rarely examine what was challenging in the process of collaboration. We present reflections from the field across four studies. Our contributions are: (1) an overview of past work in computing with a focus on disability in low resource settings and (2) learnings and recommendations from four collaborative projects in Uganda, Jordan and Kenya over the last two years, that are relevant for future HCI studies in low resource settings with communities with disabilities. We do this through a lens of Disability Interaction and ICT4D.

Transfer (machine) learning approaches coupled with target data augmentation to predict the mechanical properties of concrete

Progress towards a world where all can enjoy optimal health and health care is crucially dependent on all kinds of research including research involving humans. Involving humans in medical research is necessary to improve the knowledge base on which medicine should be based. At the same time, individuals participating in health-related research have individual human rights and have a right to be protected against the risks that research may bring to them. The tension between these two considerations has led the medical community to endorse ethical guidelines for health-related research. Research Ethics Committees can use these guidelines to evaluate whether a given research protocol is ethically acceptable or not. -- In the late 1970s, CIOMS set out, in cooperation with WHO, to prepare guidelines to indicate how the ethical principles set forth in the Declaration of Helsinki of the World Medical Association, could be effectively applied, particularly in low-resource settings, given their socio-economic circumstances, laws and regulations, and executive and administrative arrangements. Since then, revised editions of the CIOMS ethical guidelines were published in 1993 and 2002. New developments in research prompted CIOMS to again revise their ethical guidelines. The result is available in this publication. -- In the 2016 version of the ethical Guidelines, CIOMS provides answers to a number of pressing issues in research ethics. The Council does so by stressing the need for research having scientific and social value, by providing special guidelines for health-related research in low-resource settings, by detailing the provisions for involving vulnerable groups in research and for describing under what conditions biological samples and health-related data can be used for research. In providing this revised version, CIOMS hopes to ensure that the ethical Guidelines remain a living document that provides reasoned conditions for research in order to meet the challenges of modern research.

Obesity in pregnancy is a growing global health challenge, with increasing prevalence across high-income, low- and middle-income countries (LMICs). Maternal obesity is associated with significant risks, including gestational diabetes, hypertensive disorders, and delivery complications, as well as adverse fetal outcomes such as macrosomia, stillbirth, and neonatal metabolic dysfunction. These complications not only pose immediate threats to maternal and fetal health but also have long-term implications, influencing fetal genetic programming and perpetuating a cycle of obesity and metabolic diseases. Epigenetic mechanisms, such as DNA methylation and histone modification, play a crucial role in altering fetal gene expression, predisposing offspring to obesity, insulin resistance, and cardiometabolic disorders. The burden of maternal obesity is particularly concerning in LMICs like Nigeria, where limited healthcare resources and shifting dietary patterns exacerbate the problem. Despite advancements in understanding the mechanisms underlying fetal programming, research gaps remain, particularly regarding region-specific data and long term trans-generational effects. Effective prevention and management strategies must integrate pre-pregnancy, antenatal, and postnatal interventions, alongside public health initiatives that promote healthy nutrition, physical activity, and community engagement. Multidisciplinary approaches, incorporating evidence-based clinical practices with culturally tailored interventions, are essential in mitigating the impact of maternal obesity. Addressing this public health concern requires concerted efforts from healthcare providers, policymakers, and researchers to enhance maternal and neonatal outcomes. Strengthening healthcare systems, improving awareness, and fostering research in low resource settings will be critical to breaking the cycle of obesity and securing a healthier future for subsequent generations.

Reproductive health research in low-resource settings poses unique and complex challenges that must be addressed to ensure that global research is conducted with strict adherence to ethical principles, offers direct benefit to the research subjects, and has the potential for adoption of positive findings to the target population. This article addresses challenges to conducting reproductive health research in low-resource settings in the following areas: (1) establishment and maintenance of global collaboration, (2) community partnerships, (3) ethical issues, including informed consent and the role of incentives, (4) staff training and development, (5) data collection and management, and (6) infrastructure and logistics. Particular attention to these challenges is important to ensure that research is culturally appropriate and methodologically sound and enhances the adoption of health-promoting behaviors. Rigorous evaluation of interventions in low-resource settings may be a cost-effective and time-efficient way to identify interventions for large-scale program replication to improve women's health.