Archivio della ricerca della Scuola Superiore Sant'Anna
Not a member yet
26957 research outputs found
Sort by
Integration of artificial intelligence and remote sensing for crop yield prediction and crop growth parameter estimation in Mediterranean agroecosystems: Methodologies, emerging technologies, research gaps, and future directions
No full textContext: Crop yield prediction (CYP) along with crop growth parameter estimation (CGPE) recently gained
prominence as essential means for optimizing agricultural resource use and addressing global food security
challenges, particularly in regions with vulnerable climates and diverse agricultural systems, such as the Mediterranean one. Artificial intelligence (AI) and remote sensing (RS) play an important role in achieving such
objectives.
Objective: To identify present methodologies and frameworks, emerging trends, research gaps and future directions in the integrated use of AI and RS in the Mediterranean area for CYP and CGPE.
Methods: We systematically reviewed the published scientific literature on the topic (106 studies) by means of the
PRISMA methodology.
Result and conclusions: We found that integration of AI, particularly machine learning methods such as Random
Forest, Support Vector Machine, and Artificial Neural Networks, along with satellite-based RS platforms such as
Sentinel-2, Sentinel-1, MODIS, and Landsat-8, demonstrated strong potential to enhance monitoring and support
adaptive agricultural decision-making. Deep learning models, such as Convolutional Neural Networks and Long
Short Term Memories, are emerging tools for spatio-temporal modelling, although their use is limited, likely due
to data and computational constraints. Wheat is the most frequently analyzed crop, alongside high-value
perennial crops like olives and vineyards. Data acquisition relies predominantly on satellite imagery, though
hybrid approaches incorporating unmanned aerial vehicle and ground-based data are promising in improving
prediction accuracy. Despite these advancements, significant challenges persist, including uneven geographical
research coverage, limited model transferability, and insufficient consideration of crop phenology. A critical lack
of standardized validation datasets and the underrepresentation of North African and Middle Eastern countries
further constrain progress.
Significance: To fully harness AI-RS integration for sustainable agriculture and food security in the Mediterranean
area, and similar agroecosystems, future efforts should aim at i) prioritizing cross-regional collaboration, ii)
focusing on hybrid AI-RS methods, iii) developing phenology-aware models, and iv) widening access to data
Bio-Adaptive Robot Control: Integrating Biometric Feedback and Gesture-Based Interfaces for Intuitive Human–Robot Interaction (HRI)
No full textAI-driven assistance can help the user perform complex teleoperated tasks, introduce autonomous patterns, or adapt the workbench to objects of interest. On the other hand, the level of assistance should be responsive to the user’s response and adapt accordingly to promote a positive and effective experience. Envisaging this final goal, this article investigates whether physiological signals can be used to estimate the user’s performance and response in a teleoperation setup, with and without AI-driven assistance. In more detail, a teleoperated pick-and-place task was performed with or without AI-driven assistance during the grasping phase. A deep-learning algorithm for affordance detection provided assistance, helping participants align the robotic hand with the target object. Physiological and kinematic data were measured and processed by machine learning models to predict the effects of AI assistance on task performance during teleoperation. Results showed that AI-driven assistance, as expected, affected pick-and-place performance. Beyond this, the assistance affected the participant’s fatigue level, which the machine learning models could predict with an average accuracy of 84% based on the physiological response. In addition, the success or failure of the pick-and-place task could be predicted with an average accuracy of 88%. These findings highlight the potential of integrating deep learning with biometric feedback and gesture-based control to create more intuitive and adaptive HRI systems
Mental health in conflict and post-conflict settings: An analysis of UN human rights treaty bodies’ Concluding Observations
No full textPopulation-based screening for conditions associated with juvenile sudden cardiac death: a systematic review and meta-analysis
No full textThe World of the (Far) Right: Cultural Foundations and Political Strategies at the Global Level
No full textThis review article discusses World of the Right: Radical Conservatism and Global Order, a collective volume that offers a compelling and original analysis of the global far right. Moving beyond electoral performance and national case studies, the book investigates the ideological foundations and cultural strategies that underpin the success of the far right. Drawing on intellectual history and political sociology, the authors argue that the radical conservative right has developed a coherent worldview, rooted in a critique of liberalism and sustained by a Gramscian-inspired counter-hegemonic project. The review highlights three key contributions of the volume: the strategic appropriation of Gramsci's thought, the far right's cultural activism through publishing and education, and the transnational cooperation among far right parties, particularly within the European Parliament. While acknowledging internal divisions and national specificities, the book convincingly shows that the far right is increasingly global in scope and ambition. By taking the far right seriously as an object of scholarly inquiry, World of the Right provides a valuable framework for understanding its ideological coherence and political strategies. The book is an important reading for both academics and policymakers seeking to grasp—and respond to—the far right's growing influence in contemporary global politics
Quantifying the fatal and non-fatal burden of disease associated with child growth failure, 2000–2023: a systematic analysis from the Global Burden of Disease Study 2023
No full textBackground: Child growth failure (CGF), which includes underweight, wasting, and stunting, is among the factors most strongly associated with mortality and morbidity in children younger than 5 years worldwide. Poor height and bodyweight gain arise from a variety of biological and sociodemographic factors and are associated with increased vulnerability to infectious diseases. We used data from the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2023 to estimate CGF prevalence, the risk of infectious diseases associated with CGF, and the disease mortality, morbidity, and overall burden associated with CGF. Methods: In this analysis we estimated the all-cause and cause-specific (diarrhoea, lower respiratory tract infections, malaria, and measles) disability-adjusted life-years (DALYs) lost and mortality associated with stunting, wasting, underweight, and CGF in aggregate. We combined the burden associated with mild, moderate, and severe forms of CGF: stunting was defined as height-for-age Z scores (HAZ) less than –1, underweight was defined as weight-for-age Z scores (WAZ) less than –1, and wasting was defined as weight-for-height Z scores (WHZ) less than –1, according to WHO Child Growth Standards. Population-level continuous distributions of HAZ, WAZ, and WHZ were estimated for 2000 to 2023 using data from surveys, literature, and individual-level study data. The risk of incidence of, and mortality due to, diarrhoea, lower respiratory infections, malaria, and measles was separately estimated in a meta-regression framework from longitudinal cohort data for Z scores less than –1. Finally, fatal outcomes associated with these diseases were estimated with vital registration, verbal autopsy, and case-fatality data, while non-fatal outcomes were estimated with surveys as well as health-care utilisation and case reporting data. The exposure prevalence and relative risk estimates were from continuous distributions, allowing for direct assessment of the attributable fractions for mild, moderate, and severe stunting, underweight, wasting, and the combined impact of child growth failure within populations. All estimates were age-specific, sex-specific, geography-specific, and year-specific. Findings: We estimated that, in children younger than 5 years in 2023, CGF was associated with 79·4 million (95% uncertainty interval [UI] 47·0–106) DALYs lost and 880 000 (517 000–1 170 000) deaths. This represented 17·9% (10·6–23·8) of 444 million (434–457) total under-5 DALYs and 18·8% (11·1–25·0) of all 4·67 million (4·59–4·75) under-5 deaths. Compared to stunting (33·0 million [24·1–42·2] DALYs, 373 000 [272 000–477 000] deaths) and wasting (39·2 million [23·8–53·0] DALYs, 428 000 [256 000–583 000] deaths), childhood underweight was associated with the largest share of CGF-related disease burden: 52·2 million (21·9–75·1) DALYs and 573 000 (236 000–824 000) deaths in children younger than 5 years in 2023. Interpretation: CGF remains a leading factor associated with death and disability in children younger than 5 years, despite global attention and focused interventions to reduce the prevalence of associated CGF indicators. Our findings underscore the need for policies, strategies, and interventions that focus on all indicators of CGF to reduce its associated health burden. Funding: Gates Foundation
Fast, robust, and accurate anomaly detection for multivariate time series
No full textAnomaly detection is a topic widely studied both in Statistics and Computer Science, with an ever growing literature in both disciplines. We present a novel, fast, robust, accurate, and widely applicable semi-supervised procedure for anomaly detection in multivariate time series, (Fast, Robust, and Accurate ANomaly detection). It comprises 5 steps: smoothing, multicollinearity mitigation, dissimilarity measurement, threshold selection, identification of the causes of the anomalies. can tackle issues from different challenging contexts, where signals can be highly multicollinear, have unknown distributions, and intertwine short-lived noise with longer-lived anomalies. Using several experiments, we demonstrate the generality, low computational cost, precision, and interpretability of. In particular: (i) Using public benchmark datasets from anomaly detection, we evaluate the computational cost and performance of against the semi-supervised methods from a recent literature review, finding that is effective, broadly applicable, and that it outperforms existing approaches in anomaly detection and runtime; (ii) Using such datasets we also show that can explain the causes of the discovered anomalies; (iii) Using simulation studies, we show that is robust to several possible issues in the data; (iv) Using a case study from an industrial partner, we show that is effective