Chitosan is a promising natural biopolymer for food preservation because of its antimicrobial, antioxidant, and film-forming properties. Its cationic structure enables direct interaction with microbial cell membranes, providing broad-spectrum activity against bacteria, yeasts, and molds. This review examines the mechanisms underlying chitosan’s preservative effects, its interactions with food matrices, modification strategies, synergistic combinations with natural preservatives, and applications in food systems. Evidence indicates that chitosan inhibits microorganisms through membrane disruption, metal-ion chelation, and interference with intracellular processes. Its antioxidant activity helps reduce oxidative deterioration and maintain food quality. In addition, chitosan-based films and coatings act as barriers to moisture transfer and gas exchange, contributing to shelf-life extension. Preservation efficacy depends on factors such as molecular weight, degree of deacetylation, pH, water activity, and food composition. To overcome performance limitations, recent studies have developed composite materials and functionalized derivatives with improved antimicrobial, antioxidant, and barrier properties. Overall, chitosan’s effectiveness is governed by both its intrinsic characteristics and its interactions within specific food systems. Although challenges remain regarding solubility, mechanical strength, safety assessment of modified forms, and large-scale commercialization, advances in chitosan modification and active packaging continue to strengthen its potential as a sustainable alternative to synthetic food preservatives.
Context. Type Ia supernovae (SNe Ia) are fundamental probes of cosmic expansion, whose luminosities are empirically standardized using correlations with light-curve stretch and color. A common assumption in cosmological analyses is that the standardization coefficients are independent of redshift. Aims. We test the internal consistency of this assumption by searching for a possible redshift dependence in the effective empirical calibration of SNe Ia. Methods. We analyzed a compilation of type Ia supernovae, Pantheon+. We allowed the stretch and color coefficients of the SALT2 framework to vary linearly with redshift and performed a full covariance likelihood analysis using Markov chain Monte Carlo sampling. The model comparison was carried out using χ2 statistics and information criteria. Results. The redshift-dependent model yields a statistically significant improvement in the goodness of fit relative to the constant-coefficient scenario. Allowing for redshift dependence introduces a non-negligible degeneracy between empirical calibration parameters and the inferred matter density parameter Ωm. Conclusions. Our results highlight the sensitivity of precision supernova cosmology to assumptions about redshift-independent standardization. The detected trends should be interpreted phenomenologically and may reflect astrophysical evolution, selection effects, or residual systematics. Future analyses combining forward simulations and independent cosmological probes will be essential to clarify their origin and implications.
The prevalence of virtual reality (VR) is growing in many educational spaces. Yet while explored extensively in medical science research, there is a need for more studies on VR in educational and social science research. The present study explores the use of VR to teach higher education students about anti-racism content, specifically about privilege. Recognizing privilege is central to learning about anti-racism since it helps people understand each other’s social location, intersectionalities, and access to services. We implemented a two-staged mixed-methods research design, with the first stage consisting of a survey design and the second relying on interview data. Participants volunteered either for stage one or for both stages. Twenty-three students participated in stage one, while 11 participated in both stages. The findings reveal that the VR module promoted the participants’: 1) layered understanding of privilege; 2) critical reflection on their privilege; and 3) impetus to engage in more advocacy around the misuse of privilege. The findings point to a need for more strategies that specifically detail how to address the abuse of privilege in order to increase people’s comfort in conducting this type of advocacy. The research contributes to a growing body of studies on VR’s ability to improve participants’ understanding of complex humanistic principles, such as privilege.
Microbiome analysis has become an important area of study in biomedical and environmental sciences, offering knowledge regarding microbial communities and their interactions. Machine Learning (ML) and Deep Learning (DL) have also played a major role in improving the study of microbiomes by supporting effective processing and classification of high-dimensional metagenomic data. Here, the application of DL methods, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Autoencoders, is explored in the context of disease prognosis, microbial community typing, environmental microbiome research, and personalised medicine. The challenges, such as limited labelled data, model explainability, and the absence of generalisability, persist despite all the progress. The integration of multiomics data, Explainable AI (XAI), and Federated Learning becomes unavoidable in overcoming these bottlenecks. Future research needs to be focused on developing standardised datasets and scalable AI models for real-time monitoring of the microbiome. Overwhelming these encounters, ML and DL will continue to revolutionise microbiome analysis, foremost to innovation in accurate medicine and biotechnology, besides ecological sustainability.
The new advances in genome sequencing technologies have produced exponentially greater genomic data, which need scalable and accurate computational methods for genome annotation and assembly. Traditional annotation and assembly procedures are not scalable, accurate, and computationally efficient, which allows the utilisation of Machine Learning (ML) approaches. This chapter thoroughly reviews ML-based procedures for genomic data processing, annotation, and assembly. It describes some of the various ML approaches, from conventional algorithms to deep learning and combination models, and how they improve sequence alignment, feature extraction, and functional analysis of genetic components. The chapter also describes some of the challenges of integrating ML into genomics, including data heterogeneity, model interpretability, and scalability issues. In addition, a comparative assessment of ML-based and traditional genome annotation tools is illustrated, emphasising the superiority of ML models in terms of accuracy and efficiency.
