Skin cancer is among the most prevalent cancers globally, emphasizing the need for early detection and accurate diagnosis to improve outcomes. Traditional diagnostic methods, based on visual examination, are subjective, time-intensive, and require specialized expertise. Current artificial intelligence (AI) approaches for skin cancer detection face challenges such as computational inefficiency, lack of interpretability, and reliance on standalone CNN architectures. To address these limitations, this study proposes a comprehensive pipeline combining transfer learning, feature selection, and machine-learning algorithms to improve detection accuracy. Multiple pretrained CNN models were evaluated, with Xception emerging as the optimal choice for its balance of computational efficiency and performance. An ablation study further validated the effectiveness of freezing task-specific layers within the Xception architecture. Feature dimensionality was optimized using Particle Swarm Optimization, reducing dimensions from 1024 to 508, significantly enhancing computational efficiency. Machine-learning classifiers, including Subspace KNN and Medium Gaussian SVM, further improved classification accuracy. Evaluated on the ISIC 2018 and HAM10000 datasets, the proposed pipeline achieved impressive accuracies of 98.5% and 86.1%, respectively. Moreover, Explainable-AI (XAI) techniques, such as Grad-CAM, LIME, and Occlusion Sensitivity, enhanced interpretability. This approach provides a robust, efficient, and interpretable solution for automated skin cancer diagnosis in clinical applications.

Minimal model analysis of glucose and insulin data from an IVGTT (intravenous glucose tolerance test) is widely used to estimate insulin sensitivity; however, the use of the model often requires intervention by a trained operator and some problems can occur in the estimation of model parameters. In the present study, a new method for minimal model analysis, termed GAMMOD, was developed based on genetic algorithms for the estimation of model parameters. Such an algorithm does not require the fixing of initial values for the parameters (that may lead to unreliable estimates). Our method also implements an automated weighting scheme not requiring manual intervention of the operator, thus improving the usability of the model. We studied a group of 170 women with a history of previous gestational diabetes. Results obtained by GAMMOD were compared with those obtained by MINMOD (a traditional gradient-based algorithm for minimal model analysis). Insulin sensitivity by GAMMOD was (3.86±0.19) compared with (4.33±0.20)×10−4 μ-units·ml−1·min−1 by MINMOD; glucose effectiveness was 0.0236±0.0005 compared with 0.0229±0.0005 min−1 respectively. The difference in the estimation by the two methods was within the precision expected for such metabolic parameters and is probably of no clinical relevance. Moreover, both the coefficient of variation of the estimated parameters and the error of fit were generally lower in GAMMOD, despite the fact that it does not require manual intervention. In conclusion, the GAMMOD approach for parameter estimation in the minimal model provides a reliable estimation of the model parameters and improves the usability of the model, thus facilitating its further use and application in a clinical context.

The development of fetal cardiac surgery, considered the ultimate goal in the treatment of congenital cardiac malformations, needs to be supported by detailed knowledge of the blood circulation in the fetal cardiovascular system. The hemodynamic behavior in distal territories is usually inferred from vessel resistance indices, which give limited physiological information. This study presents a mathematical model of the human fetal global cardiovascular system, developed to clarify the relationships and differences existing between upper and lower body circulation. We modelled the heart with two time-varying capacitances, each representing the respective ventricle's pressure-volume relationship. The fetal vascular system was represented using two six-element Windkessel models, for the upper and lower body respectively. We obtained the identification of the set of circuital and elastance function parameters of the model using Genetic Algorithms (GAs), which follow the laws of evolutionary theory. We compared the results of our numerical study on the model identified with data collected from measurements and literature, to validate the proposed global cardiovascular system model of the human fetus. This model is intended as an instrument to investigate the differences in blood distribution between the different vascular districts in the upper and lower fetal body and the role of the aortic isthmus, the small tract of vessel connecting upper and lower fetal vascular beds; it may also represent a useful tool in the assessment of dynamic balance during mechanical assistance of circulation.

Understanding the intricate interplay between structural features and signal-processing events is crucial for unravelling the mechanisms of biomolecular systems. G protein-coupled receptors (GPCRs), a pervasive protein family in humans, serve a wide spectrum of vital functions. TAS2Rs, a subfamily of GPCRs, play a primary role in recognizing bitter molecules and triggering events leading to the perception of bitterness, a crucial defence mechanism against spoiled or poisonous food. Beyond taste, TAS2Rs function is associated with many diseases as they are expressed in several extra-oral tissues. Since the precise mechanism of TAS2R activation is poorly understood, this work aims to characterize the mechanisms underlying the signal transduction on the recently experimentally solved human TAS2R46 bitter taste receptor using molecular dynamics simulations coupled with network-based analysis. The results show that the allosteric activation of the receptor is associated with more correlated dynamics of the receptor and the formation of an interaction between two helices which mainly convey the signal transferring from the extracellular to the intracellular region. By elucidating the hallmarks of the allosteric network of TAS2R46 under varying conditions (ligand-bound, ligand-free, and transition states), this study has enabled the identification of the unique functional mechanisms of this receptor, thereby establishing a foundation for a more profound characterisation of this intriguing class of receptors. Highlights: The dynamics of TAS2R46 in different states were studied through molecular dynamics and network analysis. The presence of the bitter agonist increases intra-protein correlations. TM3 and TM6 helices mediate the allosteric network in Holo TAS2R46. The rotation of Y241 ^6.48 residue is pivotal in the allosteric network for the Holo TAS2R46.

Plant diseases increasingly threaten global agriculture due to climate change, yet manual diagnosis remains challenging. We introduce B2-GraftingNet, a lightweight deep-learning framework for automated grape-leaf disease detection that combines a VGG16 backbone with Inception-style blocks to learn robust multi-scale cues. Binary Particle Swarm Optimization selects the most informative features before classification. On the public Kaggle grape-leaf dataset, a cubic SVM classifier achieves 99.56% peak accuracy, surpassing standard pretrained CNNs (VGG16/VGG19: 34.04%, Xception: 97.95%, Darknet: 94.91%, ResNet-50: 98.44%) while being faster and lighter. For transparency, we incorporate Grad-CAM, LIME, and occlusion-sensitivity with a local 20B-parameter LLM via Ollama, which processes ROI-aware XAI summaries and provides plain-language guidance in structured sections (verdict, where to look, what to do next, caution) without suggesting specific treatments.