143174 research outputs found
Sort by
Adaptive economic model predictive control: performance guarantees for nonlinear systems
No full textWe consider the problem of optimizing the economic performance of nonlinear constrained systems subject to uncertain time-varying parameters and bounded disturbances. In particular, we propose an adaptive economic model predictive control (MPC) framework that: (i) directly minimizes transient economic costs, (ii) addresses parametric uncertainty through online model adaptation, (iii) determines optimal setpoints online, and (iv) ensures robustness by using a tube-based approach. The proposed design ensures recursive feasibility, robust constraint satisfaction, and a transient performance bound. In case the disturbances have a finite energy and the parameter variations have a finite path length, the asymptotic average performance is (approximately) not worse than the performance obtained when operating at the best reachable steady-state. We highlight performance benefits in a numerical example involving a chemical reactor with unknown time-invariant and time-varying parameters
Engineering whole-cell catalysts to use plastic waste as a feedstock
No full textThe extensive production, durability and waste mismanagement of plastic polymers have led to a highly concerning environmental problem. Recycling methods aim to reduce the amount of plastic pollution and, among them, biological processes have emerged as an interesting
alternative for the management of plastic waste that is difficult to collect or can not be recycled by other methods. While there has been significant progress in the field, in particular related to the enzymatic hydrolysis of polyesters, most biological methods rely on the use of enzymes in vitro, using collected plastics. In this review we explore the status of technologies using whole-cell catalysts that could be used for in vivo upcycling of plastic waste – with plastic becoming a microbial feedstock – and for the development of biodegradation strategies in relevant environments. We have identified a number of barriers related to polymer bioavailability, enzyme activity and secretion, and the use of strains and microbial communities that need to be overcome to materialise a much-needed solution to plastic pollution
Rotating night shifts and physical well-being in nurses: cross-sectional associations consistent with a sleep-quality pathway
No full textBackground: Rotating and night-including shifts disrupt circadian alignment, impair sleep, and may reduce nurses’ physiological recovery. Objectives: This study aimed (1) to compare sleep quality and physical well-being across four shift schedules among hospital nurses; and (2) to examine whether the association between rotating shifts and physical well-being was statistically consistent with an indirect association via sleep quality. Methods: In this cross-sectional study, 173 nurses from a ter-tiary hospital in Zagreb, Croatia, completed validated measures of sleep quality and physical well-being. Four shift patterns were analyzed—fixed morning, morning–afternoon, extended 12-h, and rotating three-shift—using Welch ANOVA and regression models. A bootstrapped mediation analysis (10,000 resamples; BCa method), interpreted as a statistical decomposition, estimated an indirect association consistent with sleep quality. Results: Rotating-shift nurses reported the poorest sleep (PSQI = 10.2 ± 2.6; p = 0.003). Physical well-being did not differ significantly across shift types (p = 0.08), although rotating-shift nurses had the lowest mean physical scores (24.3 ± 4.4). The rotat-ing-shift subgroup was small (n = 16), limiting precision. The mediation analysis was statistically consistent with an indirect association between rotating shifts and physical well-being via sleep quality (ACME = −1.85, 95% CI −3.05 to −0.88; p < 0.001), while the proportion of the total association was imprecisely estimated. Conclusions: In this single-site cross-sectional sample, rotating night shifts were associated with poorer sleep and, on average, lower physical well-being; patterns were statistically consistent with an indirect association via sleep quality. Because exposure, mediator, and outcome were measured concurrently, these findings are hypothesis-generating and do not estab-lish causality
Elective induction versus expectant management for suspected large-for-gestational-age fetuses: a systematic review and meta-analysis
No full textBackground
Suspected large-for-gestational-age (LGA) fetuses present a clinical dilemma: early induction may reduce birth trauma but raise intervention risks. Previous reviews lacked recent data.
Objectives
To assess whether elective induction at 37–39 weeks reduces adverse maternal and neonatal outcomes compared with expectant management in pregnancies with suspected LGA fetuses.
Methods
We systematically searched PubMed, Cochrane Library, Scopus, Web of Science, and ClinicalTrials.gov through May 2025, with no language restrictions. We included randomised controlled trials (RCTs) comparing elective induction (37–39 weeks) with expectant management in singleton pregnancies with suspected LGA fetuses. Two reviewers independently screened studies, extracted data, and assessed the risk of bias using RoB 2.0. Outcomes were pooled using fixed- or random-effects meta-analysis, and the certainty of evidence was evaluated using the GRADE framework.
Main Results
Three RCTs (n = 3,984) met the inclusion criteria. Induction significantly reduced shoulder dystocia (RR 0.65, 95% CI 0.46–0.91), caesarean birth (RR 0.87, 95% CI 0.79–0.95), and increased spontaneous vaginal birth (RR 1.12, 95% CI 1.06–1.19). No differences were seen in instrumental delivery, severe perineal trauma, or perinatal death. Induction lowered mean birthweight (–177 g, 95% CI –279 to –76) but was associated with increased neonatal phototherapy (RR 1.63, 95% CI 1.19–2.23). Certainty of evidence was moderate for most primary outcomes.
Conclusions
For suspected LGA fetuses, induction around 38 weeks reduces birth trauma and caesarean risk without increasing major maternal or neonatal morbidity. Clinical discussions should weigh these benefits against patient preferences and contextual factors.
Funding
This study received no external funding
Polarisation-based approaches to quantitative phase and hyperspectral imaging
No full textThis thesis presents the development and application of polarisation-based hyperspectral and quantitative phase imaging (QPI) techniques featuring fast acquisition and overall simple and cost-effective implementation. These techniques, either already open-source or in preparation for public sharing, are suitable for both manual microscopy, including in lower resourced settings, and integration in high-content analysis (HCA) platforms.
PolSpec, a polarisation-based spectral imaging technique, was developed to provide fast, versatile, cost-effective and data-efficient hyperspectral imaging with improved signal levels compared to conventional “push-broom” approaches. PolSpec employs Lyot filter-based set-ups to directly acquire data for calculating spectral modulation vectors (SMVs), which provide an efficient graphical representation of spectral signatures to facilitate spectral classification and linear unmixing. PolSpec data acquisition requires only several images using normal cameras, or a single shot using a polarisation-resolving (Polarsens™) camera. Multiple potential PolSpec configurations are proposed, of which two have been implemented, with their measured instrument response functions demonstrating consistency with the theoretical curve shapes. Spectral classification and linear unmixing were performed using PolSpec data acquired from test objects and biological samples.
Open-source software and hardware tools were also developed for polarisation-resolved differential phase contrast (pDPC) microscopy, a single-shot, wavelength-agnostic (semi)quantitative phase imaging technique using the Polarsens™ camera. Utilising these tools, condenser-based pDPC was implemented on a customized openFrame microscope. The phase reconstruction accuracy and spatial resolution were benchmarked using a quantitative phase target, and its applications to diverse biological samples were demonstrated. pDPC was also integrated into an existing HCA platform configured on a commercial microscope frame to provide single-cell tracking during long time-lapse multiwell plate fluorescence assays probing cancer cell response to drug treatment and, combined with machine learning, to realise label-free identification of cell cycle stages. Noting the spatial resolution limits of condenser-based pDPC, low-cost condenser-less pDPC and Polarsens™-based phase-stepping holography were explored for high-NA single-shot QPI.Open Acces
Ligand-like lipid interactions with membrane proteins: simulations and machine learning
No full textMembrane lipids can bind to specific sites on membrane proteins in a ligand-like manner and modulate protein structure and function. Molecular dynamics simulations encompass a suite of approaches to identify, characterise, and explain the atomic-level mechanisms that underlie the functional effects of ligand-like lipids on membrane proteins. Simulations have shown good agreement with available structural data on lipid-protein interactions. Building on successes, simulations are now used to identify new interactions and mechanisms de novo for a given membrane protein. In this age of abundance, it is increasingly possible to analyse patterns across large groups of proteins and in ever more complex membrane environments. The dawn of machine learning approaches in lipid-protein cofolding holds considerable promise to synergistically capitalise on this availability of simulation data and uncover new facets of ligand-like lipid biology
Optimizing innovation failure rates and intelligence: why 95% failure isn’t failing enough
No full textRecent public discourse has interpreted high organizational failure rates in artificial intelligence (AI) adoption as evidence of limited economic value, most notably following claims that up to 95% of firms derive no return from AI investments. This article argues that such interpretations misunderstand the structural economics of innovation. Drawing on a rapid review of empirical literature across digital transformation, entrepreneurship, pharmaceuticals, and product development, we demonstrate that extreme attrition is a persistent and necessary feature of high-return innovation systems. Failure rates exceeding 90% are not anomalous but represent the natural outcome of funnel-based experimentation and portfolio selection processes.
We further situate generative AI within historical waves of enterprise technology adoption, including ERP, cloud computing, and data analytics, and show that contemporary AI pilots exhibit comparable or higher failure rates at substantially lower capital risk. Using a comparative framework of replacement, augmentation, and symbiotic human–AI deployment models, we analyze how organizational integration mediates economic outcomes. While replacement and augmentation approaches typically yield limited returns, symbiotic configurations—treating AI systems as peer collaborators embedded in core workflows—exhibit orders-of-magnitude performance improvements.
Drawing on published experimental evidence and longitudinal venture formation data, we present indicative cases in which symbiotic AI deployment produces exponential gains in productivity and venture success rates. These findings suggest that optimal innovation performance requires deliberately engineering high early-stage failure in conjunction with disciplined portfolio governance and organizational adaptation.
We conclude that innovation systems generating only modest failure rates are structurally underperforming. In the age of synthetic intelligence, maximizing economic value depends not on minimizing failure, but on accelerating intelligent attrition while scaling symbiotic intelligence architectures. This reframes AI investment from episodic experimentation toward the systematic accumulation of intelligence capital
Glucocorticoid-induced adrenal insufficiency: physiological dose tapering promotes recovery
No full textObjective
Glucocorticoid discontinuation is complicated by glucocorticoid-induced adrenal insufficiency. Guidelines discourage tapering below physiological doses (prednisolone 3-6 mg) when morning cortisol is ≤300 nmol/L, with values <150 nmol/L thought to
indicate persistent adrenal insufficiency, though this may underestimate hypothalamic-pituitary-adrenal axis suppression from such doses. We aim to evaluate how
hypothalamic-pituitary-adrenal axis function evolves during physiological dose tapering and assess whether current cortisol thresholds restrict successful
discontinuation.
Design
Retrospective cohort study.
Methods
Adults (n=65) with long-term glucocorticoid use for inflammatory disease undergoing prednisolone tapering between 2019 and 2024 were included. Serial short Synacthen tests (n=52) on reducing prednisolone doses (≤5 mg) were analysed using linear mixed-effects modelling. Nadir morning cortisol values at doses ≤5 mg from successful weans were compared with guideline thresholds.
Results
At referral, mean age was 55.4±16.4 years, with median prednisolone dose and duration of therapy being 5 [3.5-5] mg and 23 [6.5-66.5] months, respectively. For each
1 mg dose reduction, morning and post-Synacthen cortisol rose by 48.8 nmol/L and 57.5 nmol/L (both p2 mg producing larger cortisol increases than 1 mg reductions (both p<0.05). Among completed wean attempts (n=47), 81% (n=38) were successful. Of these, 42% (n=16) had a nadir
morning cortisol <150 nmol/L, including six with values <28 nmol/L. No adrenal crises occurred.
Conclusions
Physiological dose tapering in glucocorticoid-induced adrenal insufficiency enables, rather than follows, hypothalamic-pituitary-adrenal axis recovery, with structured, symptom-led tapering being safe and effective. Future guidelines should recognise the
axis suppression from physiological doses
Shaping the UK radiology workforce: two decades of diversity, progression, and training evolution – a national observational study
No full textAims
The UK continues to face radiology workforce pressures. This study examined how major policy events have shaped diversity and career timelines within the senior UK radiology workforce over nearly two decades.
Methods
All radiologists added to the General Medical Council (GMC) Specialist Register between 2006 and 2023 were identified via a Freedom of Information request. Gender and country of primary medical qualification (PMQ) were analysed, classifying radiologists as UK medical graduates (UKMGs) or international medical graduates (IMGs). The interval between PMQ and specialty registration was a surrogate proxy for overall progression. Temporal trends were assessed using Poisson, binomial logistic, and quantile regression, with interrupted time-series models assessing inflection points at 2012 (MMC), 2016 (Brexit), and 2020 (COVID-19).
Results
Between 2006 and 2023, 4409 radiologists entered the GMC Specialist Register; 39.2% were women and 39.8% were IMGs from 80 countries. Annual registrations rose by +3.85%/year (p < 0.001) without significant change in female (β:+0.0045, p = 0.44) or IMG (β: 0.0045, p = 0.44) representation. Median time from PMQ to registration was 10 years (IQR: 9–13), 1 year longer for females (p < 0.001) and 4 years longer for IMGs (p < 0.001). The gender gap remained stable (interaction β:+0.029/year, p = 0.51), while the UKMG–IMG gap widened (interaction β:+0.224/year, p < 0.001).
Conclusions
The senior radiology workforce has expanded and remains internationally diverse, though gender representation has plateaued. Female radiologists and IMG radiologists had longer intervals to registration. Further work is needed to understand these differences, which may be due to differences in career breaks, migration and global changes in training pathways
Contribution of Mg-templated porosity to activity and durability in Fe-N-C O2 reduction catalysts
No full textAtomically dispersed Fe in N-doped carbon (Fe-N-C) catalysts are leading platinum-group-metal-free candidates for the O2 reduction reaction in proton exchange membrane fuel cells (PEMFCs). Zeolitic
imidazolate framework (ZIF-8) derived Fe-N-C present the most promising performance; however, they possess a narrow distribution of small micropores, which limits active site accessibility. Here, to induce hierarchical porosity in Fe-N-C, we report a systematic study on MgCl₂·6H₂O-templated ZIF-8-derived Fe-N-C catalysts for the O2 reduction reaction. MgCl₂·6H₂O addition
induced complete Zn removal, collapse of the ZIF-8 framework, and formation of large micro- and mesopores, with graphene-like structures. N content was markedly reduced, with conversion from pyridinic to pyrrolic N species. Rotating disc electrode tests showed a progressive increase in O2 reduction activity with MgCl₂·6H₂O, which is strongly correlated (R2 = 0.98) to the formation of large micropores and small mesopores (1-4 nm). This introduces an indirect structure-activity design principle for Fe-N-Cs. The enhanced Fe-N-C porosity also leads to increased degradation rates under accelerated stress test conditions, which we attributed to the oxidation of disordered carbon domains and active Fe loss. This study highlights a key trade-off between porosity-driven O2 reduction activity and durability in Fe-N-C catalysts