Browse Articles

Enhancing smart city sustainability with explainable federated learning for vehicular energy control

ez-CAZy a reference annotation database for linking glycoside hydrolase sequence to enzymatic activity

A decision-support framework for evaluating AI-enabled ESG strategies in the context of sustainable manufacturing systems

Correction: Temporal adjustment approach for high-resolution continental scale modeling of soil organic carbon

CNN based method for classifying cervical cancer cells in pap smear images

Abstract The absence of reliable early treatment serves as one of the main causes of cervical cancer. Hence, it is crucial to detect cervical cancer early. The biggest challenge in diagnosing cervical cancer early is that it is asymptomatic until it develops into invasive carcinoma. In medical applications, the use of machine learning and deep learning is successful as a classifier in the preliminary identification of cancerous cells in the cervical region. In our study, we present a CNN-based method for the classification of cervical cancer cells. We present a method for accurately classifying Pap smear images into abnormal or healthy cells by extracting essential information using a variety of deep-learning approaches. Experiments are performed using the SIPaKMeD and Herlev datasets. Several pre-trained convolutional neural network (CNN) models are used via transfer learning methods, hence predicting and evaluating the accurate classifier with the best optimal solution. Classification of cervical cell clusters in whole slide images (WSI) has usually comprised two stages: segmentation to extract individual cell patches, and subsequently single-cell categorization. As a result, segmentation accuracy determines the classification pipeline’s performance. We propose a direct classification of WSI cervical cell groups without segmentation and demonstrate that segmentation is not strictly necessary for good classification results. Our solution outperformed prior methods and benchmarks, with an accuracy of 96.74% for WSI patches and 97.55% for full-cell images for the SIPaKMeD dataset, and an accuracy of 90.42% for the Herlev dataset. The results show that the suggested approach may accurately distinguish cervical cancerous and non-cancerous cells.

Clinical and metabolic consequences of a historic pathogenic lamin A/C founder variant

Abstract A novel LMNA p.(Glu105Leu) variant was identified in five families with dilated cardiomyopathy (DCM), revealed as a local founder variant originating approximately 650 years ago. Genetic testing and clinical analysis of 795 DCM patients demonstrated that probands with this variant typically present with severe DCM in their sixties, characterized by high prevalence of late gadolinium enhancement, arrhythmias, and conduction disorders. Time-to-event analysis suggested a later onset of clinical symptoms compared to other LMNA variants, with a trend towards longer event-free survival. Microscopic imaging of patient fibroblasts, induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs), and heart tissue confirmed structural nuclear LMNA-associated abnormalities. Patient iPSC-CMs exhibited distinct sarcomeric disorganization, increased glucose uptake and glycogen content, reduced mitochondrial function and biogenesis, and delayed contractile function. These findings support the pathogenicity of the variant and demonstrate its profound impact on structural and metabolic functions in cardiomyocytes.

Fast and accurate RFD-like descriptor approximation for SIMD architectures

Metabolome alterations in pediatric metabolically unhealthy obesity are primarily linked to abnormal glucose homeostasis

Conditional autoregressive model based on next scale prediction for missing data reconstruction

Graphene patterning without plasma etching via SU-8 pattern peel-off

Abstract Although graphene has made its way into many areas of science and technology, proper tools for patterning graphene are not available to all researchers. Therefore, any new patterning method is useful. This research investigates the patterning of graphene layers on $${{Si{O_2}} \mathord{\left/ {\vphantom {{Si{O_2}} {Si}}} \right. \kern-0pt} {Si}}$$ substrates via the use of an SU-8 photoresist to produce micrometer-sized components such as electrodes. The new method is based on the sufficient adhesion of graphene to SU-8 after SU-8 cross-linking. First, SU-8 photolithography in the inverse form of the final pattern is carried out on the graphene layer. Then, both the SU-8 pattern and the graphene part attached to it are simultaneously removed, resulting in the final graphene pattern. This method can also be extended in a way that is compatible with imprint lithography, as its framework is described in this paper. In this way, the non-crosslinked SU-8 in the inverse pattern is transferred to the graphene layer via a premade stamp. This latter approach could benefit from the complete elimination of SU-8 effects, especially SU-8 contamination, from the final graphene pattern, as well as the simplicity of replication for mass production.

Modeling earthquake-induced wavefields and stresses in alpine mountains with extreme topography

Abstract Earthquakes can trigger slope instabilities such as rockfalls, landslides, and avalanches, posing a significant hazard for residents and infrastructures, particularly in mountainous regions. This risk is further exacerbated by global warming and permafrost degradation, which destabilize surfaces. Hence, our study investigates earthquake-induced wave dynamics at mountain summits, particularly at the Matterhorn (Switzerland) and Tre Cime di Lavaredo (Italy). The selected sites represent exceptional cases of isolated peaks. While these landforms are rare even within alpine environments, they offer crucial boundary cases to explore the upper limits of topographic amplification under seismic excitation. Full wavefield modeling is utilized to simulate the induced resonant oscillations and amplification of seismic signals at the summits compared to adjacent valleys. The simulated amplification (up to 10 times) in the summit depends on the characteristics of motion direction, topography, and presence of permafrost. Major resonance modes are identified at Matterhorn at frequencies of 0.4 Hz and 1.4 Hz. Higher resonance frequencies above 2 Hz are obtained at the smaller rock formation Tre Cime di Lavaredo, indicating mountain-specific resonances. We demonstrate that the presence of a permafrost body inside the mountain tends to reduce seismic amplification by up to 30%. However, this effect is dependent on the amount of permafrost and the wavelength of the seismic waves. Locations of potential slope instabilities on the mountain’s surface are identified based on the dynamic stress changes during the simulated earthquake. We find that locations of stress amplification are mainly at the mountain flanks and are influenced by azimuthal characteristics of the incoming wave. The approach and findings presented in our study have the potential to improve hazard assessments for earthquake-induced slope instabilities, focusing on mountains with extreme geometries.

A bioelectric router for adaptive isochronous neurostimulation enables multipolar bioelectric stimulation from a single source

Circadian and temporal eating patterns in relation to metabolic syndrome in Iranian women

Relation between vitamin D deficiency and diabetic maculopathy

Abstract Vitamin D deficiency has been linked to DR progression, its specific role in diabetic maculopathy remains underexplored. This study aimed to evaluate the relationship between serum vitamin D levels and diabetic maculopathy. In this cross-sectional study, 68 patients with diabetic macular edema (DME) underwent comprehensive ophthalmic examinations, including OCTA to assess superficial and deep vascular density and foveal avascular zone (FAZ), alongside measurement of serum vitamin D levels. Patients with renal impairment, granulomatous diseases, or vitamin D supplementation were excluded. Of the 68 patients (54.4% female, 45.6% male; mean age 58 ± 7 years), vitamin D levels showed significant positive correlations with superficial vascular density in foveal (r = 0.711, P < 0.001), parafoveal (r = 0.852, P < 0.001), and perifoveal zones (r = 0.832, P < 0.001), and with deep vascular density in foveal (r = 0.868, P < 0.001), parafoveal (r = 0.790, P < 0.001), and perifoveal zones (r = 0.645, P < 0.001). Negative correlations were observed with FAZ in superficial (r = − 0.806, P < 0.001) and deep layers (r = − 0.801, P < 0.001). Multivariate linear regression, controlling for age, gender, and diabetes duration, confirmed vitamin D as a significant predictor of superficial and deep vascular density and FAZ parameters. Diabetic macular ischemia is closely linked to vitamin D deficiency, as shown by reduced vascular density and enlarged FAZ. These findings suggest that vitamin D may help prevent retinal microvascular damage in diabetic retinopathy.

Celebrating researchers who make the scientific workplace more inclusive

Daily briefing: Neanderthals boiled bones in ‘fat factories’

Audio long read: How to speak to a vaccine sceptic — research reveals what works

I help to build support systems for Latina researchers

Japan requires name change after marriage — with big effects on female scientists

Cognitive abstraction increases prosociality when loyalty is valued lowly, but decreases prosociality when loyalty is valued highly

Abstract Many studies show that people donate more to charitable causes that are presented in concrete (vs. abstract) terms; yet other research suggests that cognitive abstraction (vs. concreteness) encourages prosocial behavior. We propose that abstract cognition facilitates prosocial behavior among people who lowly value loyalty (i.e., those who value impartiality); concrete cognition should facilitate prosocial behaviors among people who highly value loyalty. Across three experiments and one cross-sectional survey in which we operationalize cognitive abstraction (vs. concreteness), valuing loyalty, and prosocial behavior in different ways, we consistently find that abstraction facilitates prosocial behaviors among people who lowly value loyalty. In two of the four studies, we also find that concreteness facilitates prosocial behavior among people who highly value loyalty. These findings help resolve theoretical ambiguity about the cognitive underpinnings of prosociality, and they have important practical implications for optimal framing of charity appeals to potential donors.