The increasing complexity of environmental systems and the imperative for transparent decision-making have necessitated the incorporation of Explainable Artificial Intelligence (XAI) into environmental monitoring and pollutant detection. This study undertakes a Systematic Literature Review (SLR), adhering to the PRISMA methodology, to examine scientific advancements in the application of XAI models for the detection and localization of pollution sources. A total of 26 studies published between 2020 and 2025 were included in the final review corpus. The studies were identified through systematic searches conducted in Web of Science, Scopus, IEEE Xplore, ScienceDirect, and ACM Digital Library, complemented by a backward and forward snowballing process to ensure comprehensive coverage of the literature. The most commonly employed techniques were SHAP, LIME, Grad-CAM, and Layer-wise Relevance Propagation (LRP), which were applied to the evaluation of water, air, and soil quality. The findings indicate that XAI facilitates the identification of the most influential environmental factors, such as pH, dissolved oxygen, SO2, PM2.5, and temperature, and enables the generation of visual and quantitative explanations that are consistent with physical-chemical processes. Despite these advancements, there remains a lack of methodological or regulatory consensus to guide its practical implementation, thereby limiting its adoption in environmental governance. This study underscores opportunities to develop integrated and auditable frameworks that enhance transparency, scientific traceability, and sustainable and ethically responsible environmental decision-making.

Access to the paper in the following link

Early identification of Mild Cognitive Impairment remains a critical challenge in ageing populations, particularly where access to neuroimaging and specialist assessment is limited. Behavioural digital biomarkers derived from unobtrusive smart home monitoring offer a scalable and non invasive alternative; however, model development in this context is constrained by small, imbalanced datasets and minority class scarcity. This study presents a methodological evaluation of imbalance aware machine learning strategies for Mild Cognitive Impairment detection using nocturia derived behavioural features under extreme minority representation. A unified experimental framework integrates classical models, ensemble methods, autoencoder based transfer learning, adaptive synthetic augmentation, and active learning within a leakage controlled cross validation protocol. Performance is assessed using diagnostic metrics and 99% confidence intervals to quantify uncertainty. Results indicate that augmentation and tree based ensemble strategies improve minority detection compared to baseline classifiers, although performance variability remains substantial. The highest achieved minority-class Sensitivity reached 0.53 under ADASYN-based augmentation, with substantial variability across folds (standard deviation up to 0.39), reflecting instability inherent to extreme minority scarcity. Rather than claiming clinical validation, this work characterises reproducibility and uncertainty under constrained data conditions, providing a reproducible framework for scalable Internet of Things based cognitive screening research.

Access to the paper in this link

Training AI models requires high-volume, high-quality data annotation. Contractors in the Global South often meet this demand under questionable conditions. Meanwhile, people with Autism Spectrum Disorder (ASD) have above-average abilities in this field. As part of a pilot project, a global technology company created an inclusive in-house team to annotate traffic images. This paper introduces the Training Station, an accessible learning and practice platform, as well as a digital aptitude test for image annotation. According to feedback from the inclusive annotation team, the Training Station effectively conveys an understanding of the image annotation task to applicants. Recent efforts aim to integrate machine learning tasks into the platform to provide a learning ecosystem for new AI-related jobs.

Access to the paper in this link

Early detection plays a vital role in improving oral cancer outcomes, yet it remains a clinical challenge due to the inherent limitations of conventional screening methods, which often rely on subjective assessment and demonstrate poor consistency. Although histopathological analysis of a biopsy remains the gold standard for definitive diagnosis, Artificial intelligence (AI)-assisted tools can facilitate preliminary risk stratification and prompt earlier clinical suspicion, thereby supporting timely intervention and enhancing diagnostic efficiency. Recent advances in AI, particularly Deep Learning, offer promising solutions by enabling automated image-based diagnostics. This study proposes a comparative framework that evaluates the performance of Convolutional Neural Networks (CNN) architectures and hybrid models combining CNN-based feature extraction with traditional classifiers. Using two public datasets (Kaggle and Roboflow), the models were assessed across clinically relevant metrics: accuracy, sensitivity, specificity, loss, and diagnostic odds ratio. Inception-v3 achieved the most consistent diagnostic performance, with high accuracy, sensitivity, and diagnostic odds ratio, demonstrating strong suitability for clinical deployment. This makes it ideal for early screening scenarios, despite moderate specificity. Hybrid models improved specificity but underperformed in overall diagnostic balance, suggesting their complementary role in confirmatory diagnostics. Statistical analyses confirmed significant performance differences among models, reinforcing the reliability of deep learning approaches for oral cancer detection. These findings validate the potential of deep learning architectures for integration into preliminary diagnostic workflows and population-oriented telehealth platforms. They also highlight the need for further model optimisation, dataset expansion, and clinical validation to ensure generalisability and safe deployment in real-world healthcare environments.

Access to the paper in this link

Cancer presents multiple challenges for its study, which is why mathematical models have become essential tools to understand its dynamics and reduce reliance on costly biological experiments. This investigation explores the use of Physics-Informed Neural Networks (PINNs) to approximate and predict cancer progression based on a simplified ordinary differential system mathematical model, which describes the interactions among tumor, normal, and immune cells. Synthetic data are generated using the implicit Euler method, incorporating noise to simulate real clinical measurements. The study evaluates how the amount of data, temporal spacing, and noise level affect the network’s performance. Results show that having at least 40 days of data enables accurate predictions in most evaluated scenarios. A comparative analysis with a Multi-Layer Perceptron (MLP) and a Least Squares (LS) approach using RK45 demonstrated that the PINN is significantly more robust for learning and predicting future dynamics, especially under limited or noisy data conditions. The inclusion of the physical loss allowed the model to extrapolate beyond the observed domain, although it did not fully compensate for data scarcity. Accurately modeling the immune cell population proved particularly challenging. These findings help identify the limitations and obstacles that such techniques must overcome to be effectively applied in real-world clinical settings, ultimately supporting data-driven medical decision-making through robust, model-based predictions.

Access to the paper in this link

Early identification of Mild Cognitive Impairment remains a critical challenge in ageing populations, particularly where access to neuroimaging and specialist assessment is limited. Behavioural digital biomarkers derived from unobtrusive smart home monitoring offer a scalable and non invasive alternative; however, model development in this context is constrained by small, imbalanced datasets and minority class scarcity. This study presents a methodological evaluation of imbalance aware machine learning strategies for Mild Cognitive Impairment detection using nocturia derived behavioural features under extreme minority representation. A unified experimental framework integrates classical models, ensemble methods, autoencoder based transfer learning, adaptive synthetic augmentation, and active learning within a leakage controlled cross validation protocol. Performance is assessed using diagnostic metrics and 99% confidence intervals to quantify uncertainty. Results indicate that augmentation and tree based ensemble strategies improve minority detection compared to baseline classifiers, although performance variability remains substantial. The highest achieved minority-class Sensitivity reached 0.53 under ADASYN-based augmentation, with substantial variability across folds (standard deviation up to 0.39), reflecting instability inherent to extreme minority scarcity. Rather than claiming clinical validation, this work characterises reproducibility and uncertainty under constrained data conditions, providing a reproducible framework for scalable Internet of Things based cognitive screening research.

Access to the paper in this link

Information security within telemedicine systems is essential to advancing the digital transformation of healthcare. Telemedicine encompasses diverse modalities, including teleconsultation, telehealth, and remote patient monitoring, all of which depend on digital platforms, secure communication networks, and internet-connected devices. Although these systems have progressed in aligning with information security standards and regulations, there remains a shortage of comprehensive, practice-oriented studies evaluating which aspects of security are effectively addressed and which remain insufficiently managed, particularly within the Chilean context. This study aims to examine how effectively telemedicine systems in Chile address the core security attributes of confidentiality, availability, and integrity. Data were analysed from an evaluation tool designed to assess the quality of telemedicine systems in Chile. Over a six-year period, 25 telemedicine systems from different providers were assessed, and an in-depth examination of how companies manage key information security sub-characteristics within their systems was undertaken. The findings indicate that 52% of telemedicine systems optimally implement cryptographic techniques to protect confidentiality. In contrast, 44% lack robust strategies for adapting to, recovering from, and mitigating security-related incidents. Fault tolerance mechanisms are frequently integrated to minimise service disruption caused by system failures. However, the prioritisation of data integrity varies: while some companies treat it as a critical requirement, others assign it limited importance. This study offers an understanding of the security priorities and practices adopted by telemedicine providers. It highlights a prevailing tendency to prioritise security measures over usability, underscoring the need for a balanced approach that safeguards patient information while supporting efficient clinical workflows.

This study will published in https://medinform.jmir.org

Health information systems (HISs) are integral in enhancing clinical operations and improving patient care. To fulfill this role, these systems require a comprehensive design capable of addressing essential health quality attributes such as security. This design, typically embodied in software architecture, must incorporate secure design decisions that adhere to established software security policies and guidelines. Such design decisions are frequently represented by security control (also known as security tactics). Despite the significance of implementing and developing security control to protect information within HISs, there is a paucity of empirical studies that examine which security control are actually used in these systems. This gap significantly hinders the reuse and acceleration of secure design decisions within the software architecture of a system. In this paper, we report a study aimed at identifying security controls in health software projects by utilizing a CodeBERT model. We applied the trained model to 10 open-source projects related to HISs, and classified the identified security tactics.
The findings suggest that the security controls identified in HISs predominantly focus on security-by-design prevention strategies, whereas detection and recovery strategies remain largely unaddressed in the context of attacks. Our study represents an initial effort to elucidate which secure design decisions are prioritized in the development of HISs.

This research will be presented in International Conference of the Chilean Computer Science Society (SCCC)

Hidden risk factors in cancer are elements that contribute to the development or progression of cancer but are not immediately apparent or easily detectable. These factors may encompass genetic alterations, environmental exposure to carcinogens, and socio-demographic variables. Although early detection strategies exist to identify hidden risk factors at more treatable stages of cancer, there is limited discussion on the application of artificial intelligence models to support the identification of these hidden risk factors. This paper presents a study focused on the identification of potential hidden risk factors in cancer through the use of machine learning and explainable artificial intelligence techniques.
We analyzed a breast cancer database and employed support vector machine, random forest, and extreme gradient boosting to classify the data. Subsequently, we utilized four explainable artificial intelligence techniques to examine the positive, neutral, and negative features of the dataset. The findings of our study suggest that explainable artificial intelligence facilitates the identification of positive features within the dataset that are considered potential hidden risk factors for breast cancer.
These results can significantly contribute to the enhancement and support of cancer-screening strategies.

This research will be presented in International Conference of the Chilean Computer Science Society (SCCC)