• Multi-sector investigation into data quality dimensions and impacting factors. • A new theoretical framework to support organisational digitalisation or diversification. • A three-step process to apply the framework is presented to evaluate data quality and identify areas for improvement. • Thematic analysis of interviews with 31 practitioners from five sectors. • Four segments of data quality dimensions and five impacting factors are identified. The quality of data is central to decision making across all sectors. However, the many single-application studies, proposing hundreds of dimensions, make data quality assurance a daunting prospect. The diversification of industries is compounding this challenge, making a multi-sector classification essential. We propose a new classification for data quality – the DaTUM framework – which can act as a starting point for data quality assurance across diverse sectors. The framework was developed from reflective thematic analysis of 31 in-depth semi-structured interviews with academics and industry professionals from engineering, policy, economics, computer science, and psychology domains who generate, process, or analyse data. Eleven data quality dimensions and five factors that impact data quality were identified. The DaTUM framework, along with the operationalisation process, will simplify digitalisation and diversification efforts within organisations and support targeted data management and quality improvement strategies which are vital to achieving the true value of data.

Mitochondrial respirometry, the measurement of oxygen consumption rate (OCR) by the electron transport chain (ETC), is a cornerstone of mitochondrial biology and the gold standard for measurements of mitochondrial function. However, existing respirometry methodologies are poorly suited for large-scale studies and high-throughput applications, ultimately limiting the applicability of these methods. This limitation necessitates new methodologies, which are more easily scaled as mitochondrial studies become more complex and diverse. In this study, we detail a respirometry approach we have developed for high-throughput applications including optimized plate layouts, volume-based sample normalization, robust control selection, and automated data processing and quality control. Furthermore, we validate these methodologies across a respirometry study running 703 human brain samples, totaling more than 10,000 data points, which underwent our automated data processing and quality control techniques. Our workflow streamlines assay preparation, execution, and analysis to make respirometry scalable, while reducing operator burden and preserving data integrity. With this study, we provide a transferable blueprint for high-throughput respirometry as the mitochondrial biology field and the studies within it continue to expand in scale.

A scientometric overview of 75 years of epigenetics research in poultry

The online version contains supplementary material available at 10.1007/s41666-026-00229-9.

Artificial intelligence (AI) is reshaping pharmaceutical research by enabling data-intensive tasks to be performed with unprecedented speed and accuracy. While oral delivery remains the most common route of administration, it is dominated by a "one-size-fits-all" paradigm that fails to accommodate inter-individual variability, often leading to suboptimal efficacy or adverse reactions. AI offers a path toward personalised delivery by integrating patient-specific data to predict dose requirements and guide the development of bespoke dosage forms. When coupled with three-dimensional printing (3DP), AI-driven workflows enable decentralised, on-demand production of personalised medicines. This review examines advances of machine learning (ML) in enabling dose prediction, formulation optimisation, and digital manufacturing. It highlights emerging opportunities alongside challenges in data quality, regulatory acceptance, and clinical implementation. We discuss the need for collaboration between academia, industry, and regulators to establish interoperable datasets and robust quality-by-design frameworks. Together, these developments point toward a future in which oral drug delivery is increasingly precise, adaptive, and patient-centred.

Citizen science data reliability enhancing scientific research: insights from an 11-year study in the Mediterranean Sea

The AI Act marks a new chapter in AI governance, affecting companies around the world seeking to offer their services within the European Union. This study focuses on the comprehensive AI Act requirements set out for high-risk AI systems. We explored the perceived compliance challenge for the AI Act’s high-risk requirements and associated contributing factors; the AI Act’s impact on industry in terms of positive and negative side effects; and the sentiment of industry practitioners towards the AI Act’s Codes of Conduct for the voluntary application of the act’s high-risk AI requirements. A multiple case study encompassing six case companies supplemented by three independent experts with a total of 16 respondents was conducted. A ranking represents the different perceived levels of challenge for each AI Act high-risk requirement. The ranking is led by the following requirements, starting with the most challenging one: (1) data quality and governance (Art 10), (2) accuracy, robustness, and cybersecurity (Art 15), (3) risk and quality management system (Art 9, 17), and (4) transparency (Art 13). Moreover, four contributing factors emerged that impact the perceived compliance challenge: (1) industry and brand values, (2) existing regulatory environment, (3) AI maturity level and proficiency, and (4) company size. We identified several general key factors for the AI Act’s impact on industry and outlined strong arguments both for and against the AI Act voiced by practitioners. The sentiment towards the AI Act’s Codes of Conduct turned out very positive. This study offers a valuable primary research contribution to software engineering, where the state-of-the-art remains short of compliance-oriented studies with a focus on the operationalization of certain AI Act aspects. Future work is advised to develop artifacts facilitating AI Act operationalization and to validate them with industry partners. • Different perceived levels of challenge for each AI Act high-risk requirement. • Four contributing factors identified for the perceived AI Act compliance challenge. • Key factors of the AI Act’s impact on industry identified. • Strong arguments both for and against the AI Act voiced by practitioners. • Very positive sentiment towards the AI Act Codes of Conduct.

Topological analysis is a valuable tool for understanding the interactions, age relationships and kinematics of faults, and, hence,their role in making predictions about seismic hazard and subsurface fluid flow. Its application, however, requires caution. This paper examines what fault maps can reveal about the underlying topology of fault networks, highlighting the limitations imposed by data quality, mapping techniques and resolution. Three fault maps from central and southern Germany, representing different scales, are analysed to illustrate how topological methods can be used to assess map quality and resolution. Variations in geometry and topology, including differing proportions of I-, Y-, and X-nodes, are evident across maps of varying scale and resolution. The apparent connectivity of fault networks increases with resolution, as demonstrated using published maps clipped according to fault throw. Topology provides a means to assess, validate and compare fault maps, helping to identify issues such as improbable fault intersections and inconsistent abutting relationships. A high proportion of X-nodes, for example, may indicate problems with mapping or digitisation. Differences in network topology across adjacent areas may reflect variations in mapping techniques or strategies. Caution is advised when extrapolating fault network topology across scales or regions, particularly without accounting for differences in geological settings, lithologies and mapping methods. • Fault map topology is highly sensitive to resolution and interpretation style. • Higher resolution maps show greater fault connectivity and range of orientations. • Node type proportions (I, Y, X) reflect map resolution and mapping techniques. • Topological analysis can identify mapping inconsistencies and errors. • Fault maps must be critically evaluated before use in hazard or flow models.

Population-based cancer registry (PBCR) of Bhutan was established in Jigme Dorji Wangchuk National Referral Hospital (JDWNRH) in 2014 with the support of the Ministry of Health (Bhutan) and IARC Regional Hub, Tata Memorial Centre (TMC), Mumbai, India. This PBCR provides nationwide coverage (0.7 million population). We aim to present the cancer patterns in Bhutan for the years 2014-2022 using PBCR data. Trained registry staff collect cancer patient information by visiting various sources such as hospitals and diagnostic facilities. Data is entered into CanReg5 software. Data quality and consistency are checked by the IARC Regional Hub-TMC, Mumbai. The age-specific rate and age-adjusted rate were calculated using CanReg5 software. In the 9-year period (2014-2022), the PBCR registered 5906 incidence cancer cases, of which 2659 (45%) were males and 3247 (55%) were females. The age-adjusted incidence rate for males and females was 88.3 and 113.2 per 100,000 population, respectively. Age-adjusted mortality rates for males and females were 42.7 and 44.6 per 100,000 population, respectively. The leading cancer sites among males are stomach, esophagus, liver, lung, and rectum, and for females, cervix uteri, stomach, breast, lung, and thyroid. Cancer registry will play a pivotal role in boosting and monitoring screening program initiatives in Bhutan. Through effective linkages, it will build a robust database providing a cancer profile of the Bhutanese population which can be employed to devise effective cancer control activities in Bhutan.

The role of artificial intelligence in sarcopenia: Advances, applications, and future directions.

Methodological review: Prioritizing a future research agenda for overcoming immunization implementation barriers in Pakistan.

Forensic toxicology focuses on the detection, quantification, and interpretation of medicinal and recreational drugs, other chemicals or poisons, and their metabolites in biological matrices. Chromatography, combined with mass spectrometry (MS), is the most widely used analytical technique. However, forensic toxicology faces increasing analytical challenges due to a continuously changing drug landscape. In particular, the emergence of new psychoactive substances (NPS) has driven the development of more complex analytical methods (e.g., high-resolution mass spectrometry), novel markers (e.g., metabolomics), or innovative screening approaches (e.g., activity-based), which collectively generate vast amounts of data. These challenges include rapid market dynamics with the constant emergence of new chemical scaffolds and modifications, complex fragmentation and metabolic behavior, and limited or delayed access to reference materials- These developments are not limited to NPS alone. Consequently, machine learning (ML) algorithms have increasingly found their way into forensic toxicology. This review discusses various applications of ML methods related to bioanalysis, metabolomics, and toxicodynamics in the context of forensic toxicology. Currently, a major limitation is the compilation of sufficiently large and suitable datasets, which is often constrained by limited availability of real case data, inhomogeneous analytical data, in vivo study designs with small group size (< 10 animals per group), or a low number of included substances. Ultimately, the quality of an ML model relies not only on data quality but also on a thorough understanding of analytical chemistry, biochemistry, pharmacology, medical case history, and ML design, highlighting the importance of interdisciplinary collaboration in these studies.

Mobile EEG as a valid alternative to high-resolution laboratory EEG measures.

Electrical resistivity tomography (ERT) is a widely used technique for imaging subsurface resistivity distributions, but conventional data acquisition strategies face trade-offs between resolution, efficiency, and data quality. Comprehensive datasets that maximize subsurface information are impractical to measure directly, motivating the development of ERT data reconstruction approaches that can generate any four-electrode datasets from a limited subset of measurements. In this study, two reconstruction methods: the pseudo–Pole–Pole (pdPP) approach and a modified pseudo–Pole–Dipole (pdPD) method, were systematically evaluated. First, a linear error model is applied to quantify noise propagation in the reconstructed datasets under both absolute and relative noise conditions. We then use synthetic modelling to test the imaging performance of the reconstructed datasets. Finally, field experiments at two test sites in Sweden are analysed to validate the numerical findings and to assess practical performance in real environments. Results show that, under relative errors, large readings of base measurements lead to high errors in reconstructed data, with pdPP generally yielding lower reconstruction errors than pdPD. Under absolute errors, the number of base measurements governs error accumulation, and both methods perform similarly. Imaging results show that plausible inversion results can be achieved using reconstructed datasets with estimated errors as data weighting for inversion. Field experiments validated the numerical findings and further demonstrated that pdPP method is preferred for ERT data reconstruction given that it provides more data with small reconstruction error and is better suited for efficient data acquisition. • Systematic evaluation of ERT data reconstruction using pdPP and pdPD methods. • Linear error model analysis of noise propagation under absolute and relative errors. • Synthetic and field validation confirming reliability of reconstructed datasets.

This study examined the relationship between mobbing and quiet quitting attitudes among nursing and midwifery academics. A descriptive cross-sectional design was used, and data were collected online between June and December 2024 from 209 academics via social media platforms. The instruments included a Data Collection Form, the Quiet Quitting Attitude Scale (QQAS) and the Academicians Mobbing Scale (AMS). Statistical analyses were applied to assess group differences and relationships. Academics at private universities reported higher mobbing exposure. Nurse academics had higher overall QQAS and 'Personal Thought' scores, whereas midwife academics scored higher in the 'Positive Attitude' subdimension. Doctoral students experienced more professional attacks. Mobbing exposure varied significantly by academic status and was notably linked to deteriorations in social relationships and psychological well-being. A strong positive correlation was found between mobbing and quiet quitting attitudes (t = 24.239, p < 0.001). Midwifery academics reported greater mobbing, while nurse academics showed stronger quiet quitting tendencies. Findings suggest that early-career academics are especially at risk. Institutions should prioritise anti-mobbing strategies and foster academic engagement to promote a healthier work environment. This study highlights that mobbing is strongly associated with quiet quitting attitudes among nursing and midwifery academics, particularly affecting early-career professionals. Implementing anti-mobbing strategies and fostering academic engagement are essential to support well-being and productivity in academic settings. The study's online survey was conducted and reported following the CHERRIES guidelines to ensure transparency, completeness and quality of web-based research data. This study did not involve any direct patient or public contribution in its design, data collection or analysis.

Electronic Health Record (EHR) data are increasingly used in cancer research, yet the fidelity of this data when exchanged between systems remains poorly quantified. This study investigated the agreement in essential biomarker data after they are passed from the EHR into the cancer registry and Fast Healthcare Interoperability Resources (FHIR) extracts. This single-institution retrospective study compared demographics and 6 biomarkers from 30 lung cancer patients seen between July 2020 and July 2022. Manual review from the EHR served as the gold standard, with concordance tested between the source EHR, Institutional Cancer Registry, and FHIR exports. Demographics showed high concordance across databases. In contrast, biomarker data present in the source EHR were missing in 80%-100% of FHIR extracts. The demographic registry variables were highly concordant. This study reports a significant loss in biomarker data availability across real-world data (RWD) sources. Results underscore critical gaps in RWD extraction or exchange methods and highlight risks of relying on RWD without validation.

XGNN: A chemometric dual-tower model for predicting aqueous solubility.

Microstructure modelling quantifies subvoxel tissue features by combining an MRI acquisition with a mathematical model, which is typically fitted voxel-by-voxel with least-squares (LSQ) minimisation to give voxelwise maps of microstructural quantities such as diffusivity and compartmental fractions. Such approaches are susceptible to voxelwise noise, which can lead to erroneous values in parameter maps. Hierarchical Bayesian modelling (HBM) can address this limitation but has only been demonstrated for simple models. We previously derived an HBM approach for an arbitrary microstructure model with flexible parameter constraints, utilising a Markov chain Monte Carlo algorithm for parameter estimation; here, the method is demonstrated and evaluated using simulated and human data for two previously unexplored diffusion MRI techniques, namely, diffusion kurtosis imaging and blood-brain barrier filter exchange imaging. When compared with LSQ minimisation, HBM increased the accuracy, precision, contrast-to-noise ratio and parameter map quality in both simulated and human data. HBM was also able to resolve local parameter variations associated with white matter lesions in a small sample of cerebral small vessel disease subjects, which were obscured by high noise levels in the LSQ-derived parameter maps. Finally, a noise sensitivity assessment in simulations showed that HBM improved the contrast-to-noise ratio and parameter map quality even at low signal-to-noise ratios. This generalised HBM framework can improve parameter estimation for more complex diffusion MRI microstructural models that extend beyond linear combinations of exponentials.

Well logging petrophysical data uncertainty analysis and quality assurance driven by robust machine learning models

Emergency medicine (EM) interns have been required to complete a minimum of 25 eFAST (Extended Focused Assessment with Sonography in Trauma) scans during residency training, despite limited evidence supporting this numeric benchmark. We hypothesized that learners will achieve initial eFAST competency in less than 25 scans. We performed a multicenter retrospective review of prospectively gathered quality assurance (QA) data from eFAST scans performed by EM interns during their one-month ultrasound rotation at three sites using the same eFAST curriculum and QA rubric. Pre-data collection inter-rater reliability (IRR) of the QA rubric was assessed among four faculty using 30 sample eFAST scans. Post-data collection IRR was performed on a 10% randomized sample with a faculty from another site. Interns received an anonymous survey on prior point-of-care ultrasound (POCUS) experience. Cumulative summation (CUSUM) analysis was employed to calculate the mean number of eFAST scans required to achieve initial competency. There were 29 EM interns, 31% (9/29) with no prior POCUS experience. The pre-data collection IRR among the four faculty was 1 for the eFAST QA rubric. The post-data collection IRR on a 10% randomized sample between two faculty was moderate to excellent. The median number of scans to achieve competency per LC-CUSUM analysis was 8 (IQR [5, 10]) with a range of 5-15. Using the Wilcoxon rank sum analysis, there was no significant difference in scans required for competency between interns with and without prior eFAST experience (p = 0.85). LC-CUSUM analysis demonstrated that our EM interns achieved initial eFAST competency after a median of 8 scans (range 5-15). This finding may help differentiate faster learners from those who may require additional time, support, and targeted instruction. CUSUM analysis provides a valuable framework for assessing initial eFAST competency and may be one component of a multifaceted, longitudinal competency evaluation strategy.